diff options
Diffstat (limited to 'drivers/s390/block')
24 files changed, 4523 insertions, 1032 deletions
diff --git a/drivers/s390/block/Kconfig b/drivers/s390/block/Kconfig index 07883197f47..4a3b6232618 100644 --- a/drivers/s390/block/Kconfig +++ b/drivers/s390/block/Kconfig @@ -2,7 +2,8 @@ comment "S/390 block device drivers" depends on S390 && BLOCK config BLK_DEV_XPRAM - tristate "XPRAM disk support" + def_tristate m + prompt "XPRAM disk support" depends on S390 && BLOCK help Select this option if you want to use your expanded storage on S/390 @@ -12,13 +13,15 @@ config BLK_DEV_XPRAM xpram. If unsure, say "N". config DCSSBLK - tristate "DCSSBLK support" + def_tristate m + prompt "DCSSBLK support" depends on S390 && BLOCK help Support for dcss block device config DASD - tristate "Support for DASD devices" + def_tristate y + prompt "Support for DASD devices" depends on CCW && BLOCK select IOSCHED_DEADLINE help @@ -27,28 +30,32 @@ config DASD natively on a single image or an LPAR. config DASD_PROFILE - bool "Profiling support for dasd devices" + def_bool y + prompt "Profiling support for dasd devices" depends on DASD help Enable this option if you want to see profiling information in /proc/dasd/statistics. config DASD_ECKD - tristate "Support for ECKD Disks" + def_tristate y + prompt "Support for ECKD Disks" depends on DASD help ECKD devices are the most commonly used devices. You should enable this option unless you are very sure to have no ECKD device. config DASD_FBA - tristate "Support for FBA Disks" + def_tristate y + prompt "Support for FBA Disks" depends on DASD help Select this option to be able to access FBA devices. It is safe to say "Y". config DASD_DIAG - tristate "Support for DIAG access to Disks" + def_tristate y + prompt "Support for DIAG access to Disks" depends on DASD help Select this option if you want to use Diagnose250 command to access @@ -56,9 +63,28 @@ config DASD_DIAG say "N". config DASD_EER - bool "Extended error reporting (EER)" + def_bool y + prompt "Extended error reporting (EER)" depends on DASD help This driver provides a character device interface to the DASD extended error reporting. This is only needed if you want to use applications written for the EER facility. + +config SCM_BLOCK + def_tristate m + prompt "Support for Storage Class Memory" + depends on S390 && BLOCK && EADM_SCH && SCM_BUS + help + Block device driver for Storage Class Memory (SCM). This driver + provides a block device interface for each available SCM increment. + + To compile this driver as a module, choose M here: the + module will be called scm_block. + +config SCM_BLOCK_CLUSTER_WRITE + def_bool y + prompt "SCM force cluster writes" + depends on SCM_BLOCK + help + Force writes to Storage Class Memory (SCM) to be in done in clusters. diff --git a/drivers/s390/block/Makefile b/drivers/s390/block/Makefile index 0a89e080b38..c2f4e673e03 100644 --- a/drivers/s390/block/Makefile +++ b/drivers/s390/block/Makefile @@ -17,3 +17,9 @@ obj-$(CONFIG_DASD_ECKD) += dasd_eckd_mod.o obj-$(CONFIG_DASD_FBA) += dasd_fba_mod.o obj-$(CONFIG_BLK_DEV_XPRAM) += xpram.o obj-$(CONFIG_DCSSBLK) += dcssblk.o + +scm_block-objs := scm_drv.o scm_blk.o +ifdef CONFIG_SCM_BLOCK_CLUSTER_WRITE +scm_block-objs += scm_blk_cluster.o +endif +obj-$(CONFIG_SCM_BLOCK) += scm_block.o diff --git a/drivers/s390/block/dasd.c b/drivers/s390/block/dasd.c index fb613d70c2c..1eef0f58695 100644 --- a/drivers/s390/block/dasd.c +++ b/drivers/s390/block/dasd.c @@ -1,5 +1,4 @@ /* - * File...........: linux/drivers/s390/block/dasd.c * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Horst Hummel <Horst.Hummel@de.ibm.com> * Carsten Otte <Cotte@de.ibm.com> @@ -17,10 +16,12 @@ #include <linux/ctype.h> #include <linux/major.h> #include <linux/slab.h> -#include <linux/buffer_head.h> #include <linux/hdreg.h> #include <linux/async.h> #include <linux/mutex.h> +#include <linux/debugfs.h> +#include <linux/seq_file.h> +#include <linux/vmalloc.h> #include <asm/ccwdev.h> #include <asm/ebcdic.h> @@ -37,19 +38,17 @@ */ #define DASD_CHANQ_MAX_SIZE 4 -#define DASD_SLEEPON_START_TAG (void *) 1 -#define DASD_SLEEPON_END_TAG (void *) 2 - /* * SECTION: exported variables of dasd.c */ debug_info_t *dasd_debug_area; +static struct dentry *dasd_debugfs_root_entry; struct dasd_discipline *dasd_diag_discipline_pointer; void dasd_int_handler(struct ccw_device *, unsigned long, struct irb *); MODULE_AUTHOR("Holger Smolinski <Holger.Smolinski@de.ibm.com>"); MODULE_DESCRIPTION("Linux on S/390 DASD device driver," - " Copyright 2000 IBM Corporation"); + " Copyright IBM Corp. 2000"); MODULE_SUPPORTED_DEVICE("dasd"); MODULE_LICENSE("GPL"); @@ -70,6 +69,8 @@ static void dasd_return_cqr_cb(struct dasd_ccw_req *, void *); static void dasd_device_timeout(unsigned long); static void dasd_block_timeout(unsigned long); static void __dasd_process_erp(struct dasd_device *, struct dasd_ccw_req *); +static void dasd_profile_init(struct dasd_profile *, struct dentry *); +static void dasd_profile_exit(struct dasd_profile *); /* * SECTION: Operations on the device structure. @@ -77,6 +78,7 @@ static void __dasd_process_erp(struct dasd_device *, struct dasd_ccw_req *); static wait_queue_head_t dasd_init_waitq; static wait_queue_head_t dasd_flush_wq; static wait_queue_head_t generic_waitq; +static wait_queue_head_t shutdown_waitq; /* * Allocate memory for a new device structure. @@ -120,7 +122,7 @@ struct dasd_device *dasd_alloc_device(void) device->state = DASD_STATE_NEW; device->target = DASD_STATE_NEW; mutex_init(&device->state_mutex); - + spin_lock_init(&device->profile.lock); return device; } @@ -158,6 +160,7 @@ struct dasd_block *dasd_alloc_block(void) init_timer(&block->timer); block->timer.function = dasd_block_timeout; block->timer.data = (unsigned long) block; + spin_lock_init(&block->profile.lock); return block; } @@ -221,19 +224,44 @@ static int dasd_state_known_to_new(struct dasd_device *device) return 0; } +static struct dentry *dasd_debugfs_setup(const char *name, + struct dentry *base_dentry) +{ + struct dentry *pde; + + if (!base_dentry) + return NULL; + pde = debugfs_create_dir(name, base_dentry); + if (!pde || IS_ERR(pde)) + return NULL; + return pde; +} + /* * Request the irq line for the device. */ static int dasd_state_known_to_basic(struct dasd_device *device) { - int rc; + struct dasd_block *block = device->block; + int rc = 0; /* Allocate and register gendisk structure. */ - if (device->block) { - rc = dasd_gendisk_alloc(device->block); + if (block) { + rc = dasd_gendisk_alloc(block); if (rc) return rc; - } + block->debugfs_dentry = + dasd_debugfs_setup(block->gdp->disk_name, + dasd_debugfs_root_entry); + dasd_profile_init(&block->profile, block->debugfs_dentry); + if (dasd_global_profile_level == DASD_PROFILE_ON) + dasd_profile_on(&device->block->profile); + } + device->debugfs_dentry = + dasd_debugfs_setup(dev_name(&device->cdev->dev), + dasd_debugfs_root_entry); + dasd_profile_init(&device->profile, device->debugfs_dentry); + /* register 'device' debug area, used for all DBF_DEV_XXX calls */ device->debug_area = debug_register(dev_name(&device->cdev->dev), 4, 1, 8 * sizeof(long)); @@ -242,7 +270,8 @@ static int dasd_state_known_to_basic(struct dasd_device *device) DBF_DEV_EVENT(DBF_EMERG, device, "%s", "debug area created"); device->state = DASD_STATE_BASIC; - return 0; + + return rc; } /* @@ -251,7 +280,11 @@ static int dasd_state_known_to_basic(struct dasd_device *device) static int dasd_state_basic_to_known(struct dasd_device *device) { int rc; + if (device->block) { + dasd_profile_exit(&device->block->profile); + if (device->block->debugfs_dentry) + debugfs_remove(device->block->debugfs_dentry); dasd_gendisk_free(device->block); dasd_block_clear_timer(device->block); } @@ -259,6 +292,9 @@ static int dasd_state_basic_to_known(struct dasd_device *device) if (rc) return rc; dasd_device_clear_timer(device); + dasd_profile_exit(&device->profile); + if (device->debugfs_dentry) + debugfs_remove(device->debugfs_dentry); DBF_DEV_EVENT(DBF_EMERG, device, "%p debug area deleted", device); if (device->debug_area != NULL) { @@ -295,8 +331,10 @@ static int dasd_state_basic_to_ready(struct dasd_device *device) if (block->base->discipline->do_analysis != NULL) rc = block->base->discipline->do_analysis(block); if (rc) { - if (rc != -EAGAIN) + if (rc != -EAGAIN) { device->state = DASD_STATE_UNFMT; + goto out; + } return rc; } dasd_setup_queue(block); @@ -304,14 +342,29 @@ static int dasd_state_basic_to_ready(struct dasd_device *device) block->blocks << block->s2b_shift); device->state = DASD_STATE_READY; rc = dasd_scan_partitions(block); - if (rc) + if (rc) { device->state = DASD_STATE_BASIC; + return rc; + } } else { device->state = DASD_STATE_READY; } +out: + if (device->discipline->basic_to_ready) + rc = device->discipline->basic_to_ready(device); return rc; } +static inline +int _wait_for_empty_queues(struct dasd_device *device) +{ + if (device->block) + return list_empty(&device->ccw_queue) && + list_empty(&device->block->ccw_queue); + else + return list_empty(&device->ccw_queue); +} + /* * Remove device from block device layer. Destroy dirty buffers. * Forget format information. Check if the target level is basic @@ -321,6 +374,11 @@ static int dasd_state_ready_to_basic(struct dasd_device *device) { int rc; + if (device->discipline->ready_to_basic) { + rc = device->discipline->ready_to_basic(device); + if (rc) + return rc; + } device->state = DASD_STATE_BASIC; if (device->block) { struct dasd_block *block = device->block; @@ -355,19 +413,18 @@ static int dasd_state_unfmt_to_basic(struct dasd_device *device) static int dasd_state_ready_to_online(struct dasd_device * device) { - int rc; struct gendisk *disk; struct disk_part_iter piter; struct hd_struct *part; - if (device->discipline->ready_to_online) { - rc = device->discipline->ready_to_online(device); - if (rc) - return rc; - } device->state = DASD_STATE_ONLINE; if (device->block) { dasd_schedule_block_bh(device->block); + if ((device->features & DASD_FEATURE_USERAW)) { + disk = device->block->gdp; + kobject_uevent(&disk_to_dev(disk)->kobj, KOBJ_CHANGE); + return 0; + } disk = device->block->bdev->bd_disk; disk_part_iter_init(&piter, disk, DISK_PITER_INCL_PART0); while ((part = disk_part_iter_next(&piter))) @@ -392,8 +449,9 @@ static int dasd_state_online_to_ready(struct dasd_device *device) if (rc) return rc; } + device->state = DASD_STATE_READY; - if (device->block) { + if (device->block && !(device->features & DASD_FEATURE_USERAW)) { disk = device->block->bdev->bd_disk; disk_part_iter_init(&piter, disk, DISK_PITER_INCL_PART0); while ((part = disk_part_iter_next(&piter))) @@ -492,11 +550,11 @@ static void dasd_change_state(struct dasd_device *device) if (rc) device->target = device->state; - if (device->state == device->target) - wake_up(&dasd_init_waitq); - /* let user-space know that the device status changed */ kobject_uevent(&device->cdev->dev.kobj, KOBJ_CHANGE); + + if (device->state == device->target) + wake_up(&dasd_init_waitq); } /* @@ -598,26 +656,22 @@ void dasd_enable_device(struct dasd_device *device) dasd_set_target_state(device, DASD_STATE_NEW); /* Now wait for the devices to come up. */ wait_event(dasd_init_waitq, _wait_for_device(device)); + + dasd_reload_device(device); + if (device->discipline->kick_validate) + device->discipline->kick_validate(device); } /* * SECTION: device operation (interrupt handler, start i/o, term i/o ...) */ -#ifdef CONFIG_DASD_PROFILE -struct dasd_profile_info_t dasd_global_profile; -unsigned int dasd_profile_level = DASD_PROFILE_OFF; +unsigned int dasd_global_profile_level = DASD_PROFILE_OFF; -/* - * Increments counter in global and local profiling structures. - */ -#define dasd_profile_counter(value, counter, block) \ -{ \ - int index; \ - for (index = 0; index < 31 && value >> (2+index); index++); \ - dasd_global_profile.counter[index]++; \ - block->profile.counter[index]++; \ -} +#ifdef CONFIG_DASD_PROFILE +struct dasd_profile_info dasd_global_profile_data; +static struct dentry *dasd_global_profile_dentry; +static struct dentry *dasd_debugfs_global_entry; /* * Add profiling information for cqr before execution. @@ -628,30 +682,122 @@ static void dasd_profile_start(struct dasd_block *block, { struct list_head *l; unsigned int counter; - - if (dasd_profile_level != DASD_PROFILE_ON) - return; + struct dasd_device *device; /* count the length of the chanq for statistics */ counter = 0; - list_for_each(l, &block->ccw_queue) - if (++counter >= 31) - break; - dasd_global_profile.dasd_io_nr_req[counter]++; - block->profile.dasd_io_nr_req[counter]++; + if (dasd_global_profile_level || block->profile.data) + list_for_each(l, &block->ccw_queue) + if (++counter >= 31) + break; + + if (dasd_global_profile_level) { + dasd_global_profile_data.dasd_io_nr_req[counter]++; + if (rq_data_dir(req) == READ) + dasd_global_profile_data.dasd_read_nr_req[counter]++; + } + + spin_lock(&block->profile.lock); + if (block->profile.data) { + block->profile.data->dasd_io_nr_req[counter]++; + if (rq_data_dir(req) == READ) + block->profile.data->dasd_read_nr_req[counter]++; + } + spin_unlock(&block->profile.lock); + + /* + * We count the request for the start device, even though it may run on + * some other device due to error recovery. This way we make sure that + * we count each request only once. + */ + device = cqr->startdev; + if (device->profile.data) { + counter = 1; /* request is not yet queued on the start device */ + list_for_each(l, &device->ccw_queue) + if (++counter >= 31) + break; + } + spin_lock(&device->profile.lock); + if (device->profile.data) { + device->profile.data->dasd_io_nr_req[counter]++; + if (rq_data_dir(req) == READ) + device->profile.data->dasd_read_nr_req[counter]++; + } + spin_unlock(&device->profile.lock); } /* * Add profiling information for cqr after execution. */ + +#define dasd_profile_counter(value, index) \ +{ \ + for (index = 0; index < 31 && value >> (2+index); index++) \ + ; \ +} + +static void dasd_profile_end_add_data(struct dasd_profile_info *data, + int is_alias, + int is_tpm, + int is_read, + long sectors, + int sectors_ind, + int tottime_ind, + int tottimeps_ind, + int strtime_ind, + int irqtime_ind, + int irqtimeps_ind, + int endtime_ind) +{ + /* in case of an overflow, reset the whole profile */ + if (data->dasd_io_reqs == UINT_MAX) { + memset(data, 0, sizeof(*data)); + getnstimeofday(&data->starttod); + } + data->dasd_io_reqs++; + data->dasd_io_sects += sectors; + if (is_alias) + data->dasd_io_alias++; + if (is_tpm) + data->dasd_io_tpm++; + + data->dasd_io_secs[sectors_ind]++; + data->dasd_io_times[tottime_ind]++; + data->dasd_io_timps[tottimeps_ind]++; + data->dasd_io_time1[strtime_ind]++; + data->dasd_io_time2[irqtime_ind]++; + data->dasd_io_time2ps[irqtimeps_ind]++; + data->dasd_io_time3[endtime_ind]++; + + if (is_read) { + data->dasd_read_reqs++; + data->dasd_read_sects += sectors; + if (is_alias) + data->dasd_read_alias++; + if (is_tpm) + data->dasd_read_tpm++; + data->dasd_read_secs[sectors_ind]++; + data->dasd_read_times[tottime_ind]++; + data->dasd_read_time1[strtime_ind]++; + data->dasd_read_time2[irqtime_ind]++; + data->dasd_read_time3[endtime_ind]++; + } +} + static void dasd_profile_end(struct dasd_block *block, struct dasd_ccw_req *cqr, struct request *req) { long strtime, irqtime, endtime, tottime; /* in microseconds */ long tottimeps, sectors; + struct dasd_device *device; + int sectors_ind, tottime_ind, tottimeps_ind, strtime_ind; + int irqtime_ind, irqtimeps_ind, endtime_ind; - if (dasd_profile_level != DASD_PROFILE_ON) + device = cqr->startdev; + if (!(dasd_global_profile_level || + block->profile.data || + device->profile.data)) return; sectors = blk_rq_sectors(req); @@ -666,29 +812,392 @@ static void dasd_profile_end(struct dasd_block *block, tottime = ((cqr->endclk - cqr->buildclk) >> 12); tottimeps = tottime / sectors; - if (!dasd_global_profile.dasd_io_reqs) - memset(&dasd_global_profile, 0, - sizeof(struct dasd_profile_info_t)); - dasd_global_profile.dasd_io_reqs++; - dasd_global_profile.dasd_io_sects += sectors; - - if (!block->profile.dasd_io_reqs) - memset(&block->profile, 0, - sizeof(struct dasd_profile_info_t)); - block->profile.dasd_io_reqs++; - block->profile.dasd_io_sects += sectors; - - dasd_profile_counter(sectors, dasd_io_secs, block); - dasd_profile_counter(tottime, dasd_io_times, block); - dasd_profile_counter(tottimeps, dasd_io_timps, block); - dasd_profile_counter(strtime, dasd_io_time1, block); - dasd_profile_counter(irqtime, dasd_io_time2, block); - dasd_profile_counter(irqtime / sectors, dasd_io_time2ps, block); - dasd_profile_counter(endtime, dasd_io_time3, block); + dasd_profile_counter(sectors, sectors_ind); + dasd_profile_counter(tottime, tottime_ind); + dasd_profile_counter(tottimeps, tottimeps_ind); + dasd_profile_counter(strtime, strtime_ind); + dasd_profile_counter(irqtime, irqtime_ind); + dasd_profile_counter(irqtime / sectors, irqtimeps_ind); + dasd_profile_counter(endtime, endtime_ind); + + if (dasd_global_profile_level) { + dasd_profile_end_add_data(&dasd_global_profile_data, + cqr->startdev != block->base, + cqr->cpmode == 1, + rq_data_dir(req) == READ, + sectors, sectors_ind, tottime_ind, + tottimeps_ind, strtime_ind, + irqtime_ind, irqtimeps_ind, + endtime_ind); + } + + spin_lock(&block->profile.lock); + if (block->profile.data) + dasd_profile_end_add_data(block->profile.data, + cqr->startdev != block->base, + cqr->cpmode == 1, + rq_data_dir(req) == READ, + sectors, sectors_ind, tottime_ind, + tottimeps_ind, strtime_ind, + irqtime_ind, irqtimeps_ind, + endtime_ind); + spin_unlock(&block->profile.lock); + + spin_lock(&device->profile.lock); + if (device->profile.data) + dasd_profile_end_add_data(device->profile.data, + cqr->startdev != block->base, + cqr->cpmode == 1, + rq_data_dir(req) == READ, + sectors, sectors_ind, tottime_ind, + tottimeps_ind, strtime_ind, + irqtime_ind, irqtimeps_ind, + endtime_ind); + spin_unlock(&device->profile.lock); +} + +void dasd_profile_reset(struct dasd_profile *profile) +{ + struct dasd_profile_info *data; + + spin_lock_bh(&profile->lock); + data = profile->data; + if (!data) { + spin_unlock_bh(&profile->lock); + return; + } + memset(data, 0, sizeof(*data)); + getnstimeofday(&data->starttod); + spin_unlock_bh(&profile->lock); +} + +void dasd_global_profile_reset(void) +{ + memset(&dasd_global_profile_data, 0, sizeof(dasd_global_profile_data)); + getnstimeofday(&dasd_global_profile_data.starttod); +} + +int dasd_profile_on(struct dasd_profile *profile) +{ + struct dasd_profile_info *data; + + data = kzalloc(sizeof(*data), GFP_KERNEL); + if (!data) + return -ENOMEM; + spin_lock_bh(&profile->lock); + if (profile->data) { + spin_unlock_bh(&profile->lock); + kfree(data); + return 0; + } + getnstimeofday(&data->starttod); + profile->data = data; + spin_unlock_bh(&profile->lock); + return 0; +} + +void dasd_profile_off(struct dasd_profile *profile) +{ + spin_lock_bh(&profile->lock); + kfree(profile->data); + profile->data = NULL; + spin_unlock_bh(&profile->lock); +} + +char *dasd_get_user_string(const char __user *user_buf, size_t user_len) +{ + char *buffer; + + buffer = vmalloc(user_len + 1); + if (buffer == NULL) + return ERR_PTR(-ENOMEM); + if (copy_from_user(buffer, user_buf, user_len) != 0) { + vfree(buffer); + return ERR_PTR(-EFAULT); + } + /* got the string, now strip linefeed. */ + if (buffer[user_len - 1] == '\n') + buffer[user_len - 1] = 0; + else + buffer[user_len] = 0; + return buffer; +} + +static ssize_t dasd_stats_write(struct file *file, + const char __user *user_buf, + size_t user_len, loff_t *pos) +{ + char *buffer, *str; + int rc; + struct seq_file *m = (struct seq_file *)file->private_data; + struct dasd_profile *prof = m->private; + + if (user_len > 65536) + user_len = 65536; + buffer = dasd_get_user_string(user_buf, user_len); + if (IS_ERR(buffer)) + return PTR_ERR(buffer); + + str = skip_spaces(buffer); + rc = user_len; + if (strncmp(str, "reset", 5) == 0) { + dasd_profile_reset(prof); + } else if (strncmp(str, "on", 2) == 0) { + rc = dasd_profile_on(prof); + if (!rc) + rc = user_len; + } else if (strncmp(str, "off", 3) == 0) { + dasd_profile_off(prof); + } else + rc = -EINVAL; + vfree(buffer); + return rc; +} + +static void dasd_stats_array(struct seq_file *m, unsigned int *array) +{ + int i; + + for (i = 0; i < 32; i++) + seq_printf(m, "%u ", array[i]); + seq_putc(m, '\n'); +} + +static void dasd_stats_seq_print(struct seq_file *m, + struct dasd_profile_info *data) +{ + seq_printf(m, "start_time %ld.%09ld\n", + data->starttod.tv_sec, data->starttod.tv_nsec); + seq_printf(m, "total_requests %u\n", data->dasd_io_reqs); + seq_printf(m, "total_sectors %u\n", data->dasd_io_sects); + seq_printf(m, "total_pav %u\n", data->dasd_io_alias); + seq_printf(m, "total_hpf %u\n", data->dasd_io_tpm); + seq_printf(m, "histogram_sectors "); + dasd_stats_array(m, data->dasd_io_secs); + seq_printf(m, "histogram_io_times "); + dasd_stats_array(m, data->dasd_io_times); + seq_printf(m, "histogram_io_times_weighted "); + dasd_stats_array(m, data->dasd_io_timps); + seq_printf(m, "histogram_time_build_to_ssch "); + dasd_stats_array(m, data->dasd_io_time1); + seq_printf(m, "histogram_time_ssch_to_irq "); + dasd_stats_array(m, data->dasd_io_time2); + seq_printf(m, "histogram_time_ssch_to_irq_weighted "); + dasd_stats_array(m, data->dasd_io_time2ps); + seq_printf(m, "histogram_time_irq_to_end "); + dasd_stats_array(m, data->dasd_io_time3); + seq_printf(m, "histogram_ccw_queue_length "); + dasd_stats_array(m, data->dasd_io_nr_req); + seq_printf(m, "total_read_requests %u\n", data->dasd_read_reqs); + seq_printf(m, "total_read_sectors %u\n", data->dasd_read_sects); + seq_printf(m, "total_read_pav %u\n", data->dasd_read_alias); + seq_printf(m, "total_read_hpf %u\n", data->dasd_read_tpm); + seq_printf(m, "histogram_read_sectors "); + dasd_stats_array(m, data->dasd_read_secs); + seq_printf(m, "histogram_read_times "); + dasd_stats_array(m, data->dasd_read_times); + seq_printf(m, "histogram_read_time_build_to_ssch "); + dasd_stats_array(m, data->dasd_read_time1); + seq_printf(m, "histogram_read_time_ssch_to_irq "); + dasd_stats_array(m, data->dasd_read_time2); + seq_printf(m, "histogram_read_time_irq_to_end "); + dasd_stats_array(m, data->dasd_read_time3); + seq_printf(m, "histogram_read_ccw_queue_length "); + dasd_stats_array(m, data->dasd_read_nr_req); +} + +static int dasd_stats_show(struct seq_file *m, void *v) +{ + struct dasd_profile *profile; + struct dasd_profile_info *data; + + profile = m->private; + spin_lock_bh(&profile->lock); + data = profile->data; + if (!data) { + spin_unlock_bh(&profile->lock); + seq_printf(m, "disabled\n"); + return 0; + } + dasd_stats_seq_print(m, data); + spin_unlock_bh(&profile->lock); + return 0; +} + +static int dasd_stats_open(struct inode *inode, struct file *file) +{ + struct dasd_profile *profile = inode->i_private; + return single_open(file, dasd_stats_show, profile); +} + +static const struct file_operations dasd_stats_raw_fops = { + .owner = THIS_MODULE, + .open = dasd_stats_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, + .write = dasd_stats_write, +}; + +static ssize_t dasd_stats_global_write(struct file *file, + const char __user *user_buf, + size_t user_len, loff_t *pos) +{ + char *buffer, *str; + ssize_t rc; + + if (user_len > 65536) + user_len = 65536; + buffer = dasd_get_user_string(user_buf, user_len); + if (IS_ERR(buffer)) + return PTR_ERR(buffer); + str = skip_spaces(buffer); + rc = user_len; + if (strncmp(str, "reset", 5) == 0) { + dasd_global_profile_reset(); + } else if (strncmp(str, "on", 2) == 0) { + dasd_global_profile_reset(); + dasd_global_profile_level = DASD_PROFILE_GLOBAL_ONLY; + } else if (strncmp(str, "off", 3) == 0) { + dasd_global_profile_level = DASD_PROFILE_OFF; + } else + rc = -EINVAL; + vfree(buffer); + return rc; +} + +static int dasd_stats_global_show(struct seq_file *m, void *v) +{ + if (!dasd_global_profile_level) { + seq_printf(m, "disabled\n"); + return 0; + } + dasd_stats_seq_print(m, &dasd_global_profile_data); + return 0; +} + +static int dasd_stats_global_open(struct inode *inode, struct file *file) +{ + return single_open(file, dasd_stats_global_show, NULL); +} + +static const struct file_operations dasd_stats_global_fops = { + .owner = THIS_MODULE, + .open = dasd_stats_global_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, + .write = dasd_stats_global_write, +}; + +static void dasd_profile_init(struct dasd_profile *profile, + struct dentry *base_dentry) +{ + umode_t mode; + struct dentry *pde; + + if (!base_dentry) + return; + profile->dentry = NULL; + profile->data = NULL; + mode = (S_IRUSR | S_IWUSR | S_IFREG); + pde = debugfs_create_file("statistics", mode, base_dentry, + profile, &dasd_stats_raw_fops); + if (pde && !IS_ERR(pde)) + profile->dentry = pde; + return; +} + +static void dasd_profile_exit(struct dasd_profile *profile) +{ + dasd_profile_off(profile); + if (profile->dentry) { + debugfs_remove(profile->dentry); + profile->dentry = NULL; + } +} + +static void dasd_statistics_removeroot(void) +{ + dasd_global_profile_level = DASD_PROFILE_OFF; + if (dasd_global_profile_dentry) { + debugfs_remove(dasd_global_profile_dentry); + dasd_global_profile_dentry = NULL; + } + if (dasd_debugfs_global_entry) + debugfs_remove(dasd_debugfs_global_entry); + if (dasd_debugfs_root_entry) + debugfs_remove(dasd_debugfs_root_entry); +} + +static void dasd_statistics_createroot(void) +{ + umode_t mode; + struct dentry *pde; + + dasd_debugfs_root_entry = NULL; + dasd_debugfs_global_entry = NULL; + dasd_global_profile_dentry = NULL; + pde = debugfs_create_dir("dasd", NULL); + if (!pde || IS_ERR(pde)) + goto error; + dasd_debugfs_root_entry = pde; + pde = debugfs_create_dir("global", dasd_debugfs_root_entry); + if (!pde || IS_ERR(pde)) + goto error; + dasd_debugfs_global_entry = pde; + + mode = (S_IRUSR | S_IWUSR | S_IFREG); + pde = debugfs_create_file("statistics", mode, dasd_debugfs_global_entry, + NULL, &dasd_stats_global_fops); + if (!pde || IS_ERR(pde)) + goto error; + dasd_global_profile_dentry = pde; + return; + +error: + DBF_EVENT(DBF_ERR, "%s", + "Creation of the dasd debugfs interface failed"); + dasd_statistics_removeroot(); + return; } + #else #define dasd_profile_start(block, cqr, req) do {} while (0) #define dasd_profile_end(block, cqr, req) do {} while (0) + +static void dasd_statistics_createroot(void) +{ + return; +} + +static void dasd_statistics_removeroot(void) +{ + return; +} + +int dasd_stats_generic_show(struct seq_file *m, void *v) +{ + seq_printf(m, "Statistics are not activated in this kernel\n"); + return 0; +} + +static void dasd_profile_init(struct dasd_profile *profile, + struct dentry *base_dentry) +{ + return; +} + +static void dasd_profile_exit(struct dasd_profile *profile) +{ + return; +} + +int dasd_profile_on(struct dasd_profile *profile) +{ + return 0; +} + #endif /* CONFIG_DASD_PROFILE */ /* @@ -744,10 +1253,6 @@ struct dasd_ccw_req *dasd_smalloc_request(int magic, int cplength, char *data; int size; - /* Sanity checks */ - BUG_ON(datasize > PAGE_SIZE || - (cplength*sizeof(struct ccw1)) > PAGE_SIZE); - size = (sizeof(struct dasd_ccw_req) + 7L) & -8L; if (cplength > 0) size += cplength * sizeof(struct ccw1); @@ -853,9 +1358,8 @@ int dasd_term_IO(struct dasd_ccw_req *cqr) rc = ccw_device_clear(device->cdev, (long) cqr); switch (rc) { case 0: /* termination successful */ - cqr->retries--; cqr->status = DASD_CQR_CLEAR_PENDING; - cqr->stopclk = get_clock(); + cqr->stopclk = get_tod_clock(); cqr->starttime = 0; DBF_DEV_EVENT(DBF_DEBUG, device, "terminate cqr %p successful", @@ -905,6 +1409,16 @@ int dasd_start_IO(struct dasd_ccw_req *cqr) return rc; } device = (struct dasd_device *) cqr->startdev; + if (((cqr->block && + test_bit(DASD_FLAG_LOCK_STOLEN, &cqr->block->base->flags)) || + test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags)) && + !test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) { + DBF_DEV_EVENT(DBF_DEBUG, device, "start_IO: return request %p " + "because of stolen lock", cqr); + cqr->status = DASD_CQR_ERROR; + cqr->intrc = -EPERM; + return -EPERM; + } if (cqr->retries < 0) { /* internal error 14 - start_IO run out of retries */ sprintf(errorstring, "14 %p", cqr); @@ -913,9 +1427,14 @@ int dasd_start_IO(struct dasd_ccw_req *cqr) cqr->status = DASD_CQR_ERROR; return -EIO; } - cqr->startclk = get_clock(); + cqr->startclk = get_tod_clock(); cqr->starttime = jiffies; cqr->retries--; + if (!test_bit(DASD_CQR_VERIFY_PATH, &cqr->flags)) { + cqr->lpm &= device->path_data.opm; + if (!cqr->lpm) + cqr->lpm = device->path_data.opm; + } if (cqr->cpmode == 1) { rc = ccw_device_tm_start(device->cdev, cqr->cpaddr, (long) cqr, cqr->lpm); @@ -928,35 +1447,53 @@ int dasd_start_IO(struct dasd_ccw_req *cqr) cqr->status = DASD_CQR_IN_IO; break; case -EBUSY: - DBF_DEV_EVENT(DBF_DEBUG, device, "%s", + DBF_DEV_EVENT(DBF_WARNING, device, "%s", "start_IO: device busy, retry later"); break; case -ETIMEDOUT: - DBF_DEV_EVENT(DBF_DEBUG, device, "%s", + DBF_DEV_EVENT(DBF_WARNING, device, "%s", "start_IO: request timeout, retry later"); break; case -EACCES: - /* -EACCES indicates that the request used only a - * subset of the available pathes and all these - * pathes are gone. - * Do a retry with all available pathes. + /* -EACCES indicates that the request used only a subset of the + * available paths and all these paths are gone. If the lpm of + * this request was only a subset of the opm (e.g. the ppm) then + * we just do a retry with all available paths. + * If we already use the full opm, something is amiss, and we + * need a full path verification. */ - cqr->lpm = LPM_ANYPATH; - DBF_DEV_EVENT(DBF_DEBUG, device, "%s", - "start_IO: selected pathes gone," - " retry on all pathes"); + if (test_bit(DASD_CQR_VERIFY_PATH, &cqr->flags)) { + DBF_DEV_EVENT(DBF_WARNING, device, + "start_IO: selected paths gone (%x)", + cqr->lpm); + } else if (cqr->lpm != device->path_data.opm) { + cqr->lpm = device->path_data.opm; + DBF_DEV_EVENT(DBF_DEBUG, device, "%s", + "start_IO: selected paths gone," + " retry on all paths"); + } else { + DBF_DEV_EVENT(DBF_WARNING, device, "%s", + "start_IO: all paths in opm gone," + " do path verification"); + dasd_generic_last_path_gone(device); + device->path_data.opm = 0; + device->path_data.ppm = 0; + device->path_data.npm = 0; + device->path_data.tbvpm = + ccw_device_get_path_mask(device->cdev); + } break; case -ENODEV: - DBF_DEV_EVENT(DBF_DEBUG, device, "%s", + DBF_DEV_EVENT(DBF_WARNING, device, "%s", "start_IO: -ENODEV device gone, retry"); break; case -EIO: - DBF_DEV_EVENT(DBF_DEBUG, device, "%s", + DBF_DEV_EVENT(DBF_WARNING, device, "%s", "start_IO: -EIO device gone, retry"); break; case -EINVAL: /* most likely caused in power management context */ - DBF_DEV_EVENT(DBF_DEBUG, device, "%s", + DBF_DEV_EVENT(DBF_WARNING, device, "%s", "start_IO: -EINVAL device currently " "not accessible"); break; @@ -1093,17 +1630,12 @@ void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm, return; } - now = get_clock(); - - /* check for unsolicited interrupts */ + now = get_tod_clock(); cqr = (struct dasd_ccw_req *) intparm; - if (!cqr || ((scsw_cc(&irb->scsw) == 1) && - (scsw_fctl(&irb->scsw) & SCSW_FCTL_START_FUNC) && - ((scsw_stctl(&irb->scsw) == SCSW_STCTL_STATUS_PEND) || - (scsw_stctl(&irb->scsw) == (SCSW_STCTL_STATUS_PEND | - SCSW_STCTL_ALERT_STATUS))))) { - if (cqr && cqr->status == DASD_CQR_IN_IO) - cqr->status = DASD_CQR_QUEUED; + /* check for conditions that should be handled immediately */ + if (!cqr || + !(scsw_dstat(&irb->scsw) == (DEV_STAT_CHN_END | DEV_STAT_DEV_END) && + scsw_cstat(&irb->scsw) == 0)) { if (cqr) memcpy(&cqr->irb, irb, sizeof(*irb)); device = dasd_device_from_cdev_locked(cdev); @@ -1114,17 +1646,14 @@ void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm, dasd_put_device(device); return; } - device->discipline->dump_sense_dbf(device, irb, - "unsolicited"); - if ((device->features & DASD_FEATURE_ERPLOG)) - device->discipline->dump_sense(device, cqr, - irb); - dasd_device_clear_timer(device); - device->discipline->handle_unsolicited_interrupt(device, - irb); + device->discipline->dump_sense_dbf(device, irb, "int"); + if (device->features & DASD_FEATURE_ERPLOG) + device->discipline->dump_sense(device, cqr, irb); + device->discipline->check_for_device_change(device, cqr, irb); dasd_put_device(device); - return; } + if (!cqr) + return; device = (struct dasd_device *) cqr->startdev; if (!device || @@ -1164,25 +1693,19 @@ void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm, struct dasd_ccw_req, devlist); } } else { /* error */ - memcpy(&cqr->irb, irb, sizeof(struct irb)); - /* log sense for every failed I/O to s390 debugfeature */ - dasd_log_sense_dbf(cqr, irb); - if (device->features & DASD_FEATURE_ERPLOG) { - dasd_log_sense(cqr, irb); - } - /* * If we don't want complex ERP for this request, then just * reset this and retry it in the fastpath */ if (!test_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags) && cqr->retries > 0) { - if (cqr->lpm == LPM_ANYPATH) + if (cqr->lpm == device->path_data.opm) DBF_DEV_EVENT(DBF_DEBUG, device, "default ERP in fastpath " "(%i retries left)", cqr->retries); - cqr->lpm = LPM_ANYPATH; + if (!test_bit(DASD_CQR_VERIFY_PATH, &cqr->flags)) + cqr->lpm = device->path_data.opm; cqr->status = DASD_CQR_QUEUED; next = cqr; } else @@ -1210,13 +1733,13 @@ enum uc_todo dasd_generic_uc_handler(struct ccw_device *cdev, struct irb *irb) goto out; if (test_bit(DASD_FLAG_OFFLINE, &device->flags) || device->state != device->target || - !device->discipline->handle_unsolicited_interrupt){ + !device->discipline->check_for_device_change){ dasd_put_device(device); goto out; } - - dasd_device_clear_timer(device); - device->discipline->handle_unsolicited_interrupt(device, irb); + if (device->discipline->dump_sense_dbf) + device->discipline->dump_sense_dbf(device, irb, "uc"); + device->discipline->check_for_device_change(device, NULL, irb); dasd_put_device(device); out: return UC_TODO_RETRY; @@ -1262,11 +1785,11 @@ static void __dasd_device_process_ccw_queue(struct dasd_device *device, list_for_each_safe(l, n, &device->ccw_queue) { cqr = list_entry(l, struct dasd_ccw_req, devlist); - /* Stop list processing at the first non-final request. */ + /* Skip any non-final request. */ if (cqr->status == DASD_CQR_QUEUED || cqr->status == DASD_CQR_IN_IO || cqr->status == DASD_CQR_CLEAR_PENDING) - break; + continue; if (cqr->status == DASD_CQR_ERROR) { __dasd_device_recovery(device, cqr); } @@ -1335,6 +1858,13 @@ static void __dasd_device_check_expire(struct dasd_device *device) cqr = list_entry(device->ccw_queue.next, struct dasd_ccw_req, devlist); if ((cqr->status == DASD_CQR_IN_IO && cqr->expires != 0) && (time_after_eq(jiffies, cqr->expires + cqr->starttime))) { + if (test_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags)) { + /* + * IO in safe offline processing should not + * run out of retries + */ + cqr->retries++; + } if (device->discipline->term_IO(cqr) != 0) { /* Hmpf, try again in 5 sec */ dev_err(&device->cdev->dev, @@ -1366,8 +1896,14 @@ static void __dasd_device_start_head(struct dasd_device *device) cqr = list_entry(device->ccw_queue.next, struct dasd_ccw_req, devlist); if (cqr->status != DASD_CQR_QUEUED) return; - /* when device is stopped, return request to previous layer */ - if (device->stopped) { + /* when device is stopped, return request to previous layer + * exception: only the disconnect or unresumed bits are set and the + * cqr is a path verification request + */ + if (device->stopped && + !(!(device->stopped & ~(DASD_STOPPED_DC_WAIT | DASD_UNRESUMED_PM)) + && test_bit(DASD_CQR_VERIFY_PATH, &cqr->flags))) { + cqr->intrc = -EAGAIN; cqr->status = DASD_CQR_CLEARED; dasd_schedule_device_bh(device); return; @@ -1383,6 +1919,23 @@ static void __dasd_device_start_head(struct dasd_device *device) dasd_device_set_timer(device, 50); } +static void __dasd_device_check_path_events(struct dasd_device *device) +{ + int rc; + + if (device->path_data.tbvpm) { + if (device->stopped & ~(DASD_STOPPED_DC_WAIT | + DASD_UNRESUMED_PM)) + return; + rc = device->discipline->verify_path( + device, device->path_data.tbvpm); + if (rc) + dasd_device_set_timer(device, 50); + else + device->path_data.tbvpm = 0; + } +}; + /* * Go through all request on the dasd_device request queue, * terminate them on the cdev if necessary, and return them to the @@ -1417,7 +1970,7 @@ int dasd_flush_device_queue(struct dasd_device *device) } break; case DASD_CQR_QUEUED: - cqr->stopclk = get_clock(); + cqr->stopclk = get_tod_clock(); cqr->status = DASD_CQR_CLEARED; break; default: /* no need to modify the others */ @@ -1457,6 +2010,7 @@ static void dasd_device_tasklet(struct dasd_device *device) __dasd_device_check_expire(device); /* find final requests on ccw queue */ __dasd_device_process_ccw_queue(device, &final_queue); + __dasd_device_check_path_events(device); spin_unlock_irq(get_ccwdev_lock(device->cdev)); /* Now call the callback function of requests with final status */ __dasd_device_process_final_queue(device, &final_queue); @@ -1464,6 +2018,8 @@ static void dasd_device_tasklet(struct dasd_device *device) /* Now check if the head of the ccw queue needs to be started. */ __dasd_device_start_head(device); spin_unlock_irq(get_ccwdev_lock(device->cdev)); + if (waitqueue_active(&shutdown_waitq)) + wake_up(&shutdown_waitq); dasd_put_device(device); } @@ -1532,13 +2088,14 @@ void dasd_add_request_tail(struct dasd_ccw_req *cqr) /* * Wakeup helper for the 'sleep_on' functions. */ -static void dasd_wakeup_cb(struct dasd_ccw_req *cqr, void *data) +void dasd_wakeup_cb(struct dasd_ccw_req *cqr, void *data) { spin_lock_irq(get_ccwdev_lock(cqr->startdev->cdev)); cqr->callback_data = DASD_SLEEPON_END_TAG; spin_unlock_irq(get_ccwdev_lock(cqr->startdev->cdev)); wake_up(&generic_waitq); } +EXPORT_SYMBOL_GPL(dasd_wakeup_cb); static inline int _wait_for_wakeup(struct dasd_ccw_req *cqr) { @@ -1613,15 +2170,20 @@ static int _dasd_sleep_on(struct dasd_ccw_req *maincqr, int interruptible) continue; if (cqr->status != DASD_CQR_FILLED) /* could be failed */ continue; - + if (test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags) && + !test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) { + cqr->status = DASD_CQR_FAILED; + cqr->intrc = -EPERM; + continue; + } /* Non-temporary stop condition will trigger fail fast */ if (device->stopped & ~DASD_STOPPED_PENDING && test_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags) && (!dasd_eer_enabled(device))) { cqr->status = DASD_CQR_FAILED; + cqr->intrc = -ENOLINK; continue; } - /* Don't try to start requests if device is stopped */ if (interruptible) { rc = wait_event_interruptible( @@ -1634,7 +2196,9 @@ static int _dasd_sleep_on(struct dasd_ccw_req *maincqr, int interruptible) } else wait_event(generic_waitq, !(device->stopped)); - cqr->callback = dasd_wakeup_cb; + if (!cqr->callback) + cqr->callback = dasd_wakeup_cb; + cqr->callback_data = DASD_SLEEPON_START_TAG; dasd_add_request_tail(cqr); if (interruptible) { @@ -1653,7 +2217,7 @@ static int _dasd_sleep_on(struct dasd_ccw_req *maincqr, int interruptible) wait_event(generic_waitq, _wait_for_wakeup(cqr)); } - maincqr->endclk = get_clock(); + maincqr->endclk = get_tod_clock(); if ((maincqr->status != DASD_CQR_DONE) && (maincqr->intrc != -ERESTARTSYS)) dasd_log_sense(maincqr, &maincqr->irb); @@ -1666,6 +2230,77 @@ static int _dasd_sleep_on(struct dasd_ccw_req *maincqr, int interruptible) return rc; } +static inline int _wait_for_wakeup_queue(struct list_head *ccw_queue) +{ + struct dasd_ccw_req *cqr; + + list_for_each_entry(cqr, ccw_queue, blocklist) { + if (cqr->callback_data != DASD_SLEEPON_END_TAG) + return 0; + } + + return 1; +} + +static int _dasd_sleep_on_queue(struct list_head *ccw_queue, int interruptible) +{ + struct dasd_device *device; + int rc; + struct dasd_ccw_req *cqr, *n; + +retry: + list_for_each_entry_safe(cqr, n, ccw_queue, blocklist) { + device = cqr->startdev; + if (cqr->status != DASD_CQR_FILLED) /*could be failed*/ + continue; + + if (test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags) && + !test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) { + cqr->status = DASD_CQR_FAILED; + cqr->intrc = -EPERM; + continue; + } + /*Non-temporary stop condition will trigger fail fast*/ + if (device->stopped & ~DASD_STOPPED_PENDING && + test_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags) && + !dasd_eer_enabled(device)) { + cqr->status = DASD_CQR_FAILED; + cqr->intrc = -EAGAIN; + continue; + } + + /*Don't try to start requests if device is stopped*/ + if (interruptible) { + rc = wait_event_interruptible( + generic_waitq, !device->stopped); + if (rc == -ERESTARTSYS) { + cqr->status = DASD_CQR_FAILED; + cqr->intrc = rc; + continue; + } + } else + wait_event(generic_waitq, !(device->stopped)); + + if (!cqr->callback) + cqr->callback = dasd_wakeup_cb; + cqr->callback_data = DASD_SLEEPON_START_TAG; + dasd_add_request_tail(cqr); + } + + wait_event(generic_waitq, _wait_for_wakeup_queue(ccw_queue)); + + rc = 0; + list_for_each_entry_safe(cqr, n, ccw_queue, blocklist) { + if (__dasd_sleep_on_erp(cqr)) + rc = 1; + } + if (rc) + goto retry; + + + return 0; +} + /* * Queue a request to the tail of the device ccw_queue and wait for * it's completion. @@ -1676,6 +2311,15 @@ int dasd_sleep_on(struct dasd_ccw_req *cqr) } /* + * Start requests from a ccw_queue and wait for their completion. + */ +int dasd_sleep_on_queue(struct list_head *ccw_queue) +{ + return _dasd_sleep_on_queue(ccw_queue, 0); +} +EXPORT_SYMBOL(dasd_sleep_on_queue); + +/* * Queue a request to the tail of the device ccw_queue and wait * interruptible for it's completion. */ @@ -1693,11 +2337,20 @@ int dasd_sleep_on_interruptible(struct dasd_ccw_req *cqr) static inline int _dasd_term_running_cqr(struct dasd_device *device) { struct dasd_ccw_req *cqr; + int rc; if (list_empty(&device->ccw_queue)) return 0; cqr = list_entry(device->ccw_queue.next, struct dasd_ccw_req, devlist); - return device->discipline->term_IO(cqr); + rc = device->discipline->term_IO(cqr); + if (!rc) + /* + * CQR terminated because a more important request is pending. + * Undo decreasing of retry counter because this is + * not an error case. + */ + cqr->retries++; + return rc; } int dasd_sleep_on_immediatly(struct dasd_ccw_req *cqr) @@ -1706,17 +2359,26 @@ int dasd_sleep_on_immediatly(struct dasd_ccw_req *cqr) int rc; device = cqr->startdev; + if (test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags) && + !test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) { + cqr->status = DASD_CQR_FAILED; + cqr->intrc = -EPERM; + return -EIO; + } spin_lock_irq(get_ccwdev_lock(device->cdev)); rc = _dasd_term_running_cqr(device); if (rc) { spin_unlock_irq(get_ccwdev_lock(device->cdev)); return rc; } - cqr->callback = dasd_wakeup_cb; cqr->callback_data = DASD_SLEEPON_START_TAG; cqr->status = DASD_CQR_QUEUED; - list_add(&cqr->devlist, &device->ccw_queue); + /* + * add new request as second + * first the terminated cqr needs to be finished + */ + list_add(&cqr->devlist, device->ccw_queue.next); /* let the bh start the request to keep them in order */ dasd_schedule_device_bh(device); @@ -1731,6 +2393,12 @@ int dasd_sleep_on_immediatly(struct dasd_ccw_req *cqr) rc = cqr->intrc; else rc = -EIO; + + /* kick tasklets */ + dasd_schedule_device_bh(device); + if (device->block) + dasd_schedule_block_bh(device->block); + return rc; } @@ -1738,8 +2406,7 @@ int dasd_sleep_on_immediatly(struct dasd_ccw_req *cqr) * Cancels a request that was started with dasd_sleep_on_req. * This is useful to timeout requests. The request will be * terminated if it is currently in i/o. - * Returns 1 if the request has been terminated. - * 0 if there was no need to terminate the request (not started yet) + * Returns 0 if request termination was successful * negative error code if termination failed * Cancellation of a request is an asynchronous operation! The calling * function has to wait until the request is properly returned via callback. @@ -1765,7 +2432,7 @@ int dasd_cancel_req(struct dasd_ccw_req *cqr) "Cancelling request %p failed with rc=%d\n", cqr, rc); } else { - cqr->stopclk = get_clock(); + cqr->stopclk = get_tod_clock(); } break; default: /* already finished or clear pending - do nothing */ @@ -1776,7 +2443,6 @@ int dasd_cancel_req(struct dasd_ccw_req *cqr) return rc; } - /* * SECTION: Operations of the dasd_block layer. */ @@ -1863,7 +2529,7 @@ static void __dasd_process_request_queue(struct dasd_block *block) return; } /* Now we try to fetch requests from the request queue */ - while (!blk_queue_plugged(queue) && (req = blk_peek_request(queue))) { + while ((req = blk_peek_request(queue))) { if (basedev->features & DASD_FEATURE_READONLY && rq_data_dir(req) == WRITE) { DBF_DEV_EVENT(DBF_ERR, basedev, @@ -1873,6 +2539,16 @@ static void __dasd_process_request_queue(struct dasd_block *block) __blk_end_request_all(req, -EIO); continue; } + if (test_bit(DASD_FLAG_ABORTALL, &basedev->flags) && + (basedev->features & DASD_FEATURE_FAILFAST || + blk_noretry_request(req))) { + DBF_DEV_EVENT(DBF_ERR, basedev, + "Rejecting failfast request %p", + req); + blk_start_request(req); + __blk_end_request_all(req, -ETIMEDOUT); + continue; + } cqr = basedev->discipline->build_cp(basedev, block, req); if (IS_ERR(cqr)) { if (PTR_ERR(cqr) == -EBUSY) @@ -1911,8 +2587,10 @@ static void __dasd_process_request_queue(struct dasd_block *block) */ cqr->callback_data = (void *) req; cqr->status = DASD_CQR_FILLED; + req->completion_data = cqr; blk_start_request(req); list_add_tail(&cqr->blocklist, &block->ccw_queue); + INIT_LIST_HEAD(&cqr->devlist); dasd_profile_start(block, cqr, req); } } @@ -1926,8 +2604,17 @@ static void __dasd_cleanup_cqr(struct dasd_ccw_req *cqr) req = (struct request *) cqr->callback_data; dasd_profile_end(cqr->block, cqr, req); status = cqr->block->base->discipline->free_cp(cqr, req); - if (status <= 0) - error = status ? status : -EIO; + if (status < 0) + error = status; + else if (status == 0) { + if (cqr->intrc == -EPERM) + error = -EBADE; + else if (cqr->intrc == -ENOLINK || + cqr->intrc == -ETIMEDOUT) + error = cqr->intrc; + else + error = -EIO; + } __blk_end_request_all(req, error); } @@ -1993,7 +2680,7 @@ restart: } /* Rechain finished requests to final queue */ - cqr->endclk = get_clock(); + cqr->endclk = get_tod_clock(); list_move_tail(&cqr->blocklist, final_queue); } } @@ -2016,11 +2703,19 @@ static void __dasd_block_start_head(struct dasd_block *block) list_for_each_entry(cqr, &block->ccw_queue, blocklist) { if (cqr->status != DASD_CQR_FILLED) continue; + if (test_bit(DASD_FLAG_LOCK_STOLEN, &block->base->flags) && + !test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) { + cqr->status = DASD_CQR_FAILED; + cqr->intrc = -EPERM; + dasd_schedule_block_bh(block); + continue; + } /* Non-temporary stop condition will trigger fail fast */ if (block->base->stopped & ~DASD_STOPPED_PENDING && test_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags) && (!dasd_eer_enabled(block->base))) { cqr->status = DASD_CQR_FAILED; + cqr->intrc = -ENOLINK; dasd_schedule_block_bh(block); continue; } @@ -2070,6 +2765,8 @@ static void dasd_block_tasklet(struct dasd_block *block) __dasd_block_start_head(block); spin_unlock(&block->queue_lock); spin_unlock_irq(&block->request_queue_lock); + if (waitqueue_active(&shutdown_waitq)) + wake_up(&shutdown_waitq); dasd_put_device(block->base); } @@ -2079,6 +2776,26 @@ static void _dasd_wake_block_flush_cb(struct dasd_ccw_req *cqr, void *data) } /* + * Requeue a request back to the block request queue + * only works for block requests + */ +static int _dasd_requeue_request(struct dasd_ccw_req *cqr) +{ + struct dasd_block *block = cqr->block; + struct request *req; + unsigned long flags; + + if (!block) + return -EINVAL; + spin_lock_irqsave(&block->queue_lock, flags); + req = (struct request *) cqr->callback_data; + blk_requeue_request(block->request_queue, req); + spin_unlock_irqrestore(&block->queue_lock, flags); + + return 0; +} + +/* * Go through all request on the dasd_block request queue, cancel them * on the respective dasd_device, and return them to the generic * block layer. @@ -2127,7 +2844,7 @@ restart_cb: } /* call the callback function */ spin_lock_irq(&block->request_queue_lock); - cqr->endclk = get_clock(); + cqr->endclk = get_tod_clock(); list_del_init(&cqr->blocklist); __dasd_cleanup_cqr(cqr); spin_unlock_irq(&block->request_queue_lock); @@ -2171,6 +2888,82 @@ static void do_dasd_request(struct request_queue *queue) } /* + * Block timeout callback, called from the block layer + * + * request_queue lock is held on entry. + * + * Return values: + * BLK_EH_RESET_TIMER if the request should be left running + * BLK_EH_NOT_HANDLED if the request is handled or terminated + * by the driver. + */ +enum blk_eh_timer_return dasd_times_out(struct request *req) +{ + struct dasd_ccw_req *cqr = req->completion_data; + struct dasd_block *block = req->q->queuedata; + struct dasd_device *device; + int rc = 0; + + if (!cqr) + return BLK_EH_NOT_HANDLED; + + device = cqr->startdev ? cqr->startdev : block->base; + if (!device->blk_timeout) + return BLK_EH_RESET_TIMER; + DBF_DEV_EVENT(DBF_WARNING, device, + " dasd_times_out cqr %p status %x", + cqr, cqr->status); + + spin_lock(&block->queue_lock); + spin_lock(get_ccwdev_lock(device->cdev)); + cqr->retries = -1; + cqr->intrc = -ETIMEDOUT; + if (cqr->status >= DASD_CQR_QUEUED) { + spin_unlock(get_ccwdev_lock(device->cdev)); + rc = dasd_cancel_req(cqr); + } else if (cqr->status == DASD_CQR_FILLED || + cqr->status == DASD_CQR_NEED_ERP) { + cqr->status = DASD_CQR_TERMINATED; + spin_unlock(get_ccwdev_lock(device->cdev)); + } else if (cqr->status == DASD_CQR_IN_ERP) { + struct dasd_ccw_req *searchcqr, *nextcqr, *tmpcqr; + + list_for_each_entry_safe(searchcqr, nextcqr, + &block->ccw_queue, blocklist) { + tmpcqr = searchcqr; + while (tmpcqr->refers) + tmpcqr = tmpcqr->refers; + if (tmpcqr != cqr) + continue; + /* searchcqr is an ERP request for cqr */ + searchcqr->retries = -1; + searchcqr->intrc = -ETIMEDOUT; + if (searchcqr->status >= DASD_CQR_QUEUED) { + spin_unlock(get_ccwdev_lock(device->cdev)); + rc = dasd_cancel_req(searchcqr); + spin_lock(get_ccwdev_lock(device->cdev)); + } else if ((searchcqr->status == DASD_CQR_FILLED) || + (searchcqr->status == DASD_CQR_NEED_ERP)) { + searchcqr->status = DASD_CQR_TERMINATED; + rc = 0; + } else if (searchcqr->status == DASD_CQR_IN_ERP) { + /* + * Shouldn't happen; most recent ERP + * request is at the front of queue + */ + continue; + } + break; + } + spin_unlock(get_ccwdev_lock(device->cdev)); + } + dasd_schedule_block_bh(block); + spin_unlock(&block->queue_lock); + + return rc ? BLK_EH_RESET_TIMER : BLK_EH_NOT_HANDLED; +} + +/* * Allocate and initialize request queue and default I/O scheduler. */ static int dasd_alloc_queue(struct dasd_block *block) @@ -2186,12 +2979,12 @@ static int dasd_alloc_queue(struct dasd_block *block) elevator_exit(block->request_queue->elevator); block->request_queue->elevator = NULL; + mutex_lock(&block->request_queue->sysfs_lock); rc = elevator_init(block->request_queue, "deadline"); - if (rc) { + if (rc) blk_cleanup_queue(block->request_queue); - return rc; - } - return 0; + mutex_unlock(&block->request_queue->sysfs_lock); + return rc; } /* @@ -2201,8 +2994,20 @@ static void dasd_setup_queue(struct dasd_block *block) { int max; - blk_queue_logical_block_size(block->request_queue, block->bp_block); - max = block->base->discipline->max_blocks << block->s2b_shift; + if (block->base->features & DASD_FEATURE_USERAW) { + /* + * the max_blocks value for raw_track access is 256 + * it is higher than the native ECKD value because we + * only need one ccw per track + * so the max_hw_sectors are + * 2048 x 512B = 1024kB = 16 tracks + */ + max = 2048; + } else { + max = block->base->discipline->max_blocks << block->s2b_shift; + } + blk_queue_logical_block_size(block->request_queue, + block->bp_block); blk_queue_max_hw_sectors(block->request_queue, max); blk_queue_max_segments(block->request_queue, -1L); /* with page sized segments we can translate each segement into @@ -2241,15 +3046,14 @@ static void dasd_flush_request_queue(struct dasd_block *block) static int dasd_open(struct block_device *bdev, fmode_t mode) { - struct dasd_block *block = bdev->bd_disk->private_data; struct dasd_device *base; int rc; - if (!block) + base = dasd_device_from_gendisk(bdev->bd_disk); + if (!base) return -ENODEV; - base = block->base; - atomic_inc(&block->open_count); + atomic_inc(&base->block->open_count); if (test_bit(DASD_FLAG_OFFLINE, &base->flags)) { rc = -ENODEV; goto unlock; @@ -2282,22 +3086,25 @@ static int dasd_open(struct block_device *bdev, fmode_t mode) goto out; } + dasd_put_device(base); return 0; out: module_put(base->discipline->owner); unlock: - atomic_dec(&block->open_count); + atomic_dec(&base->block->open_count); + dasd_put_device(base); return rc; } -static int dasd_release(struct gendisk *disk, fmode_t mode) +static void dasd_release(struct gendisk *disk, fmode_t mode) { - struct dasd_block *block = disk->private_data; - - atomic_dec(&block->open_count); - module_put(block->base->discipline->owner); - return 0; + struct dasd_device *base = dasd_device_from_gendisk(disk); + if (base) { + atomic_dec(&base->block->open_count); + module_put(base->discipline->owner); + dasd_put_device(base); + } } /* @@ -2305,20 +3112,20 @@ static int dasd_release(struct gendisk *disk, fmode_t mode) */ static int dasd_getgeo(struct block_device *bdev, struct hd_geometry *geo) { - struct dasd_block *block; struct dasd_device *base; - block = bdev->bd_disk->private_data; - if (!block) + base = dasd_device_from_gendisk(bdev->bd_disk); + if (!base) return -ENODEV; - base = block->base; if (!base->discipline || - !base->discipline->fill_geometry) + !base->discipline->fill_geometry) { + dasd_put_device(base); return -EINVAL; - - base->discipline->fill_geometry(block, geo); - geo->start = get_start_sect(bdev) >> block->s2b_shift; + } + base->discipline->fill_geometry(base->block, geo); + geo->start = get_start_sect(bdev) >> base->block->s2b_shift; + dasd_put_device(base); return 0; } @@ -2353,6 +3160,7 @@ dasd_exit(void) debug_unregister(dasd_debug_area); dasd_debug_area = NULL; } + dasd_statistics_removeroot(); } /* @@ -2438,13 +3246,16 @@ void dasd_generic_remove(struct ccw_device *cdev) cdev->handler = NULL; - dasd_remove_sysfs_files(cdev); device = dasd_device_from_cdev(cdev); - if (IS_ERR(device)) + if (IS_ERR(device)) { + dasd_remove_sysfs_files(cdev); return; - if (test_and_set_bit(DASD_FLAG_OFFLINE, &device->flags)) { + } + if (test_and_set_bit(DASD_FLAG_OFFLINE, &device->flags) && + !test_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags)) { /* Already doing offline processing */ dasd_put_device(device); + dasd_remove_sysfs_files(cdev); return; } /* @@ -2455,7 +3266,6 @@ void dasd_generic_remove(struct ccw_device *cdev) dasd_set_target_state(device, DASD_STATE_NEW); /* dasd_delete_device destroys the device reference. */ block = device->block; - device->block = NULL; dasd_delete_device(device); /* * life cycle of block is bound to device, so delete it after @@ -2463,6 +3273,8 @@ void dasd_generic_remove(struct ccw_device *cdev) */ if (block) dasd_free_block(block); + + dasd_remove_sysfs_files(cdev); } /* @@ -2541,16 +3353,13 @@ int dasd_generic_set_offline(struct ccw_device *cdev) { struct dasd_device *device; struct dasd_block *block; - int max_count, open_count; + int max_count, open_count, rc; + rc = 0; device = dasd_device_from_cdev(cdev); if (IS_ERR(device)) return PTR_ERR(device); - if (test_and_set_bit(DASD_FLAG_OFFLINE, &device->flags)) { - /* Already doing offline processing */ - dasd_put_device(device); - return 0; - } + /* * We must make sure that this device is currently not in use. * The open_count is increased for every opener, that includes @@ -2574,10 +3383,57 @@ int dasd_generic_set_offline(struct ccw_device *cdev) return -EBUSY; } } + + if (test_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags)) { + /* + * safe offline already running + * could only be called by normal offline so safe_offline flag + * needs to be removed to run normal offline and kill all I/O + */ + if (test_and_set_bit(DASD_FLAG_OFFLINE, &device->flags)) { + /* Already doing normal offline processing */ + dasd_put_device(device); + return -EBUSY; + } else + clear_bit(DASD_FLAG_SAFE_OFFLINE, &device->flags); + + } else + if (test_bit(DASD_FLAG_OFFLINE, &device->flags)) { + /* Already doing offline processing */ + dasd_put_device(device); + return -EBUSY; + } + + /* + * if safe_offline called set safe_offline_running flag and + * clear safe_offline so that a call to normal offline + * can overrun safe_offline processing + */ + if (test_and_clear_bit(DASD_FLAG_SAFE_OFFLINE, &device->flags) && + !test_and_set_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags)) { + /* + * If we want to set the device safe offline all IO operations + * should be finished before continuing the offline process + * so sync bdev first and then wait for our queues to become + * empty + */ + /* sync blockdev and partitions */ + rc = fsync_bdev(device->block->bdev); + if (rc != 0) + goto interrupted; + + /* schedule device tasklet and wait for completion */ + dasd_schedule_device_bh(device); + rc = wait_event_interruptible(shutdown_waitq, + _wait_for_empty_queues(device)); + if (rc != 0) + goto interrupted; + } + + set_bit(DASD_FLAG_OFFLINE, &device->flags); dasd_set_target_state(device, DASD_STATE_NEW); /* dasd_delete_device destroys the device reference. */ block = device->block; - device->block = NULL; dasd_delete_device(device); /* * life cycle of block is bound to device, so delete it after @@ -2586,12 +3442,63 @@ int dasd_generic_set_offline(struct ccw_device *cdev) if (block) dasd_free_block(block); return 0; + +interrupted: + /* interrupted by signal */ + clear_bit(DASD_FLAG_SAFE_OFFLINE, &device->flags); + clear_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags); + clear_bit(DASD_FLAG_OFFLINE, &device->flags); + dasd_put_device(device); + return rc; +} + +int dasd_generic_last_path_gone(struct dasd_device *device) +{ + struct dasd_ccw_req *cqr; + + dev_warn(&device->cdev->dev, "No operational channel path is left " + "for the device\n"); + DBF_DEV_EVENT(DBF_WARNING, device, "%s", "last path gone"); + /* First of all call extended error reporting. */ + dasd_eer_write(device, NULL, DASD_EER_NOPATH); + + if (device->state < DASD_STATE_BASIC) + return 0; + /* Device is active. We want to keep it. */ + list_for_each_entry(cqr, &device->ccw_queue, devlist) + if ((cqr->status == DASD_CQR_IN_IO) || + (cqr->status == DASD_CQR_CLEAR_PENDING)) { + cqr->status = DASD_CQR_QUEUED; + cqr->retries++; + } + dasd_device_set_stop_bits(device, DASD_STOPPED_DC_WAIT); + dasd_device_clear_timer(device); + dasd_schedule_device_bh(device); + return 1; } +EXPORT_SYMBOL_GPL(dasd_generic_last_path_gone); + +int dasd_generic_path_operational(struct dasd_device *device) +{ + dev_info(&device->cdev->dev, "A channel path to the device has become " + "operational\n"); + DBF_DEV_EVENT(DBF_WARNING, device, "%s", "path operational"); + dasd_device_remove_stop_bits(device, DASD_STOPPED_DC_WAIT); + if (device->stopped & DASD_UNRESUMED_PM) { + dasd_device_remove_stop_bits(device, DASD_UNRESUMED_PM); + dasd_restore_device(device); + return 1; + } + dasd_schedule_device_bh(device); + if (device->block) + dasd_schedule_block_bh(device->block); + return 1; +} +EXPORT_SYMBOL_GPL(dasd_generic_path_operational); int dasd_generic_notify(struct ccw_device *cdev, int event) { struct dasd_device *device; - struct dasd_ccw_req *cqr; int ret; device = dasd_device_from_cdev_locked(cdev); @@ -2602,53 +3509,109 @@ int dasd_generic_notify(struct ccw_device *cdev, int event) case CIO_GONE: case CIO_BOXED: case CIO_NO_PATH: - /* First of all call extended error reporting. */ - dasd_eer_write(device, NULL, DASD_EER_NOPATH); - - if (device->state < DASD_STATE_BASIC) - break; - /* Device is active. We want to keep it. */ - list_for_each_entry(cqr, &device->ccw_queue, devlist) - if (cqr->status == DASD_CQR_IN_IO) { - cqr->status = DASD_CQR_QUEUED; - cqr->retries++; - } - dasd_device_set_stop_bits(device, DASD_STOPPED_DC_WAIT); - dasd_device_clear_timer(device); - dasd_schedule_device_bh(device); - ret = 1; + device->path_data.opm = 0; + device->path_data.ppm = 0; + device->path_data.npm = 0; + ret = dasd_generic_last_path_gone(device); break; case CIO_OPER: - /* FIXME: add a sanity check. */ - dasd_device_remove_stop_bits(device, DASD_STOPPED_DC_WAIT); - if (device->stopped & DASD_UNRESUMED_PM) { - dasd_device_remove_stop_bits(device, DASD_UNRESUMED_PM); - dasd_restore_device(device); - ret = 1; - break; - } - dasd_schedule_device_bh(device); - if (device->block) - dasd_schedule_block_bh(device->block); ret = 1; + if (device->path_data.opm) + ret = dasd_generic_path_operational(device); break; } dasd_put_device(device); return ret; } +void dasd_generic_path_event(struct ccw_device *cdev, int *path_event) +{ + int chp; + __u8 oldopm, eventlpm; + struct dasd_device *device; + + device = dasd_device_from_cdev_locked(cdev); + if (IS_ERR(device)) + return; + for (chp = 0; chp < 8; chp++) { + eventlpm = 0x80 >> chp; + if (path_event[chp] & PE_PATH_GONE) { + oldopm = device->path_data.opm; + device->path_data.opm &= ~eventlpm; + device->path_data.ppm &= ~eventlpm; + device->path_data.npm &= ~eventlpm; + if (oldopm && !device->path_data.opm) { + dev_warn(&device->cdev->dev, + "No verified channel paths remain " + "for the device\n"); + DBF_DEV_EVENT(DBF_WARNING, device, + "%s", "last verified path gone"); + dasd_eer_write(device, NULL, DASD_EER_NOPATH); + dasd_device_set_stop_bits(device, + DASD_STOPPED_DC_WAIT); + } + } + if (path_event[chp] & PE_PATH_AVAILABLE) { + device->path_data.opm &= ~eventlpm; + device->path_data.ppm &= ~eventlpm; + device->path_data.npm &= ~eventlpm; + device->path_data.tbvpm |= eventlpm; + dasd_schedule_device_bh(device); + } + if (path_event[chp] & PE_PATHGROUP_ESTABLISHED) { + if (!(device->path_data.opm & eventlpm) && + !(device->path_data.tbvpm & eventlpm)) { + /* + * we can not establish a pathgroup on an + * unavailable path, so trigger a path + * verification first + */ + device->path_data.tbvpm |= eventlpm; + dasd_schedule_device_bh(device); + } + DBF_DEV_EVENT(DBF_WARNING, device, "%s", + "Pathgroup re-established\n"); + if (device->discipline->kick_validate) + device->discipline->kick_validate(device); + } + } + dasd_put_device(device); +} +EXPORT_SYMBOL_GPL(dasd_generic_path_event); + +int dasd_generic_verify_path(struct dasd_device *device, __u8 lpm) +{ + if (!device->path_data.opm && lpm) { + device->path_data.opm = lpm; + dasd_generic_path_operational(device); + } else + device->path_data.opm |= lpm; + return 0; +} +EXPORT_SYMBOL_GPL(dasd_generic_verify_path); + + int dasd_generic_pm_freeze(struct ccw_device *cdev) { + struct dasd_device *device = dasd_device_from_cdev(cdev); + struct list_head freeze_queue; struct dasd_ccw_req *cqr, *n; + struct dasd_ccw_req *refers; int rc; - struct list_head freeze_queue; - struct dasd_device *device = dasd_device_from_cdev(cdev); if (IS_ERR(device)) return PTR_ERR(device); + + /* mark device as suspended */ + set_bit(DASD_FLAG_SUSPENDED, &device->flags); + + if (device->discipline->freeze) + rc = device->discipline->freeze(device); + /* disallow new I/O */ dasd_device_set_stop_bits(device, DASD_STOPPED_PM); - /* clear active requests */ + + /* clear active requests and requeue them to block layer if possible */ INIT_LIST_HEAD(&freeze_queue); spin_lock_irq(get_ccwdev_lock(cdev)); rc = 0; @@ -2668,7 +3631,6 @@ int dasd_generic_pm_freeze(struct ccw_device *cdev) } list_move_tail(&cqr->devlist, &freeze_queue); } - spin_unlock_irq(get_ccwdev_lock(cdev)); list_for_each_entry_safe(cqr, n, &freeze_queue, devlist) { @@ -2676,15 +3638,38 @@ int dasd_generic_pm_freeze(struct ccw_device *cdev) (cqr->status != DASD_CQR_CLEAR_PENDING)); if (cqr->status == DASD_CQR_CLEARED) cqr->status = DASD_CQR_QUEUED; - } - /* move freeze_queue to start of the ccw_queue */ - spin_lock_irq(get_ccwdev_lock(cdev)); - list_splice_tail(&freeze_queue, &device->ccw_queue); - spin_unlock_irq(get_ccwdev_lock(cdev)); - if (device->discipline->freeze) - rc = device->discipline->freeze(device); + /* requeue requests to blocklayer will only work for + block device requests */ + if (_dasd_requeue_request(cqr)) + continue; + + /* remove requests from device and block queue */ + list_del_init(&cqr->devlist); + while (cqr->refers != NULL) { + refers = cqr->refers; + /* remove the request from the block queue */ + list_del(&cqr->blocklist); + /* free the finished erp request */ + dasd_free_erp_request(cqr, cqr->memdev); + cqr = refers; + } + if (cqr->block) + list_del_init(&cqr->blocklist); + cqr->block->base->discipline->free_cp( + cqr, (struct request *) cqr->callback_data); + } + /* + * if requests remain then they are internal request + * and go back to the device queue + */ + if (!list_empty(&freeze_queue)) { + /* move freeze_queue to start of the ccw_queue */ + spin_lock_irq(get_ccwdev_lock(cdev)); + list_splice_tail(&freeze_queue, &device->ccw_queue); + spin_unlock_irq(get_ccwdev_lock(cdev)); + } dasd_put_device(device); return rc; } @@ -2720,6 +3705,7 @@ int dasd_generic_restore_device(struct ccw_device *cdev) if (device->block) dasd_schedule_block_bh(device->block); + clear_bit(DASD_FLAG_SUSPENDED, &device->flags); dasd_put_device(device); return 0; } @@ -2761,7 +3747,7 @@ static struct dasd_ccw_req *dasd_generic_build_rdc(struct dasd_device *device, cqr->memdev = device; cqr->expires = 10*HZ; cqr->retries = 256; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; return cqr; } @@ -2818,6 +3804,23 @@ char *dasd_get_sense(struct irb *irb) } EXPORT_SYMBOL_GPL(dasd_get_sense); +void dasd_generic_shutdown(struct ccw_device *cdev) +{ + struct dasd_device *device; + + device = dasd_device_from_cdev(cdev); + if (IS_ERR(device)) + return; + + if (device->block) + dasd_schedule_block_bh(device->block); + + dasd_schedule_device_bh(device); + + wait_event(shutdown_waitq, _wait_for_empty_queues(device)); +} +EXPORT_SYMBOL_GPL(dasd_generic_shutdown); + static int __init dasd_init(void) { int rc; @@ -2825,6 +3828,7 @@ static int __init dasd_init(void) init_waitqueue_head(&dasd_init_waitq); init_waitqueue_head(&dasd_flush_wq); init_waitqueue_head(&generic_waitq); + init_waitqueue_head(&shutdown_waitq); /* register 'common' DASD debug area, used for all DBF_XXX calls */ dasd_debug_area = debug_register("dasd", 1, 1, 8 * sizeof(long)); @@ -2839,6 +3843,8 @@ static int __init dasd_init(void) dasd_diag_discipline_pointer = NULL; + dasd_statistics_createroot(); + rc = dasd_devmap_init(); if (rc) goto failed; diff --git a/drivers/s390/block/dasd_3990_erp.c b/drivers/s390/block/dasd_3990_erp.c index 968c76cf712..d2613471368 100644 --- a/drivers/s390/block/dasd_3990_erp.c +++ b/drivers/s390/block/dasd_3990_erp.c @@ -1,9 +1,8 @@ /* - * File...........: linux/drivers/s390/block/dasd_3990_erp.c * Author(s)......: Horst Hummel <Horst.Hummel@de.ibm.com> * Holger Smolinski <Holger.Smolinski@de.ibm.com> * Bugreports.to..: <Linux390@de.ibm.com> - * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 2000, 2001 + * Copyright IBM Corp. 2000, 2001 * */ @@ -152,9 +151,9 @@ dasd_3990_erp_alternate_path(struct dasd_ccw_req * erp) spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags); opm = ccw_device_get_path_mask(device->cdev); spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags); - //FIXME: start with get_opm ? if (erp->lpm == 0) - erp->lpm = LPM_ANYPATH & ~(erp->irb.esw.esw0.sublog.lpum); + erp->lpm = device->path_data.opm & + ~(erp->irb.esw.esw0.sublog.lpum); else erp->lpm &= ~(erp->irb.esw.esw0.sublog.lpum); @@ -230,7 +229,7 @@ dasd_3990_erp_DCTL(struct dasd_ccw_req * erp, char modifier) dctl_cqr->expires = 5 * 60 * HZ; dctl_cqr->retries = 2; - dctl_cqr->buildclk = get_clock(); + dctl_cqr->buildclk = get_tod_clock(); dctl_cqr->status = DASD_CQR_FILLED; @@ -270,10 +269,11 @@ static struct dasd_ccw_req *dasd_3990_erp_action_1(struct dasd_ccw_req *erp) { erp->function = dasd_3990_erp_action_1; dasd_3990_erp_alternate_path(erp); - if (erp->status == DASD_CQR_FAILED) { + if (erp->status == DASD_CQR_FAILED && + !test_bit(DASD_CQR_VERIFY_PATH, &erp->flags)) { erp->status = DASD_CQR_FILLED; erp->retries = 10; - erp->lpm = LPM_ANYPATH; + erp->lpm = erp->startdev->path_data.opm; erp->function = dasd_3990_erp_action_1_sec; } return erp; @@ -1717,9 +1717,9 @@ dasd_3990_erp_action_1B_32(struct dasd_ccw_req * default_erp, char *sense) erp->startdev = device; erp->memdev = device; erp->magic = default_erp->magic; - erp->expires = 0; + erp->expires = default_erp->expires; erp->retries = 256; - erp->buildclk = get_clock(); + erp->buildclk = get_tod_clock(); erp->status = DASD_CQR_FILLED; /* remove the default erp */ @@ -1907,15 +1907,14 @@ dasd_3990_erp_compound_retry(struct dasd_ccw_req * erp, char *sense) static void dasd_3990_erp_compound_path(struct dasd_ccw_req * erp, char *sense) { - if (sense[25] & DASD_SENSE_BIT_3) { dasd_3990_erp_alternate_path(erp); - if (erp->status == DASD_CQR_FAILED) { + if (erp->status == DASD_CQR_FAILED && + !test_bit(DASD_CQR_VERIFY_PATH, &erp->flags)) { /* reset the lpm and the status to be able to * try further actions. */ - - erp->lpm = 0; + erp->lpm = erp->startdev->path_data.opm; erp->status = DASD_CQR_NEED_ERP; } } @@ -2207,7 +2206,7 @@ dasd_3990_erp_inspect_32(struct dasd_ccw_req * erp, char *sense) * DASD_3990_ERP_CONTROL_CHECK * * DESCRIPTION - * Does a generic inspection if a control check occured and sets up + * Does a generic inspection if a control check occurred and sets up * the related error recovery procedure * * PARAMETER @@ -2250,7 +2249,7 @@ dasd_3990_erp_inspect(struct dasd_ccw_req *erp) struct dasd_ccw_req *erp_new = NULL; char *sense; - /* if this problem occured on an alias retry on base */ + /* if this problem occurred on an alias retry on base */ erp_new = dasd_3990_erp_inspect_alias(erp); if (erp_new) return erp_new; @@ -2282,7 +2281,7 @@ dasd_3990_erp_inspect(struct dasd_ccw_req *erp) * DASD_3990_ERP_ADD_ERP * * DESCRIPTION - * This funtion adds an additional request block (ERP) to the head of + * This function adds an additional request block (ERP) to the head of * the given cqr (or erp). * For a command mode cqr the erp is initialized as an default erp * (retry TIC). @@ -2323,7 +2322,7 @@ static struct dasd_ccw_req *dasd_3990_erp_add_erp(struct dasd_ccw_req *cqr) DBF_DEV_EVENT(DBF_ERR, device, "%s", "Unable to allocate ERP request"); cqr->status = DASD_CQR_FAILED; - cqr->stopclk = get_clock (); + cqr->stopclk = get_tod_clock(); } else { DBF_DEV_EVENT(DBF_ERR, device, "Unable to allocate ERP request " @@ -2363,9 +2362,9 @@ static struct dasd_ccw_req *dasd_3990_erp_add_erp(struct dasd_ccw_req *cqr) erp->memdev = device; erp->block = cqr->block; erp->magic = cqr->magic; - erp->expires = 0; + erp->expires = cqr->expires; erp->retries = 256; - erp->buildclk = get_clock(); + erp->buildclk = get_tod_clock(); erp->status = DASD_CQR_FILLED; return erp; diff --git a/drivers/s390/block/dasd_alias.c b/drivers/s390/block/dasd_alias.c index 4155805dcdf..a2597e683e7 100644 --- a/drivers/s390/block/dasd_alias.c +++ b/drivers/s390/block/dasd_alias.c @@ -1,7 +1,7 @@ /* * PAV alias management for the DASD ECKD discipline * - * Copyright IBM Corporation, 2007 + * Copyright IBM Corp. 2007 * Author(s): Stefan Weinhuber <wein@de.ibm.com> */ @@ -189,14 +189,12 @@ int dasd_alias_make_device_known_to_lcu(struct dasd_device *device) unsigned long flags; struct alias_server *server, *newserver; struct alias_lcu *lcu, *newlcu; - int is_lcu_known; struct dasd_uid uid; private = (struct dasd_eckd_private *) device->private; device->discipline->get_uid(device, &uid); spin_lock_irqsave(&aliastree.lock, flags); - is_lcu_known = 1; server = _find_server(&uid); if (!server) { spin_unlock_irqrestore(&aliastree.lock, flags); @@ -208,7 +206,6 @@ int dasd_alias_make_device_known_to_lcu(struct dasd_device *device) if (!server) { list_add(&newserver->server, &aliastree.serverlist); server = newserver; - is_lcu_known = 0; } else { /* someone was faster */ _free_server(newserver); @@ -226,12 +223,10 @@ int dasd_alias_make_device_known_to_lcu(struct dasd_device *device) if (!lcu) { list_add(&newlcu->lcu, &server->lculist); lcu = newlcu; - is_lcu_known = 0; } else { /* someone was faster */ _free_lcu(newlcu); } - is_lcu_known = 0; } spin_lock(&lcu->lock); list_add(&device->alias_list, &lcu->inactive_devices); @@ -239,68 +234,7 @@ int dasd_alias_make_device_known_to_lcu(struct dasd_device *device) spin_unlock(&lcu->lock); spin_unlock_irqrestore(&aliastree.lock, flags); - return is_lcu_known; -} - -/* - * The first device to be registered on an LCU will have to do - * some additional setup steps to configure that LCU on the - * storage server. All further devices should wait with their - * initialization until the first device is done. - * To synchronize this work, the first device will call - * dasd_alias_lcu_setup_complete when it is done, and all - * other devices will wait for it with dasd_alias_wait_for_lcu_setup. - */ -void dasd_alias_lcu_setup_complete(struct dasd_device *device) -{ - struct dasd_eckd_private *private; - unsigned long flags; - struct alias_server *server; - struct alias_lcu *lcu; - struct dasd_uid uid; - - private = (struct dasd_eckd_private *) device->private; - device->discipline->get_uid(device, &uid); - lcu = NULL; - spin_lock_irqsave(&aliastree.lock, flags); - server = _find_server(&uid); - if (server) - lcu = _find_lcu(server, &uid); - spin_unlock_irqrestore(&aliastree.lock, flags); - if (!lcu) { - DBF_EVENT_DEVID(DBF_ERR, device->cdev, - "could not find lcu for %04x %02x", - uid.ssid, uid.real_unit_addr); - WARN_ON(1); - return; - } - complete_all(&lcu->lcu_setup); -} - -void dasd_alias_wait_for_lcu_setup(struct dasd_device *device) -{ - struct dasd_eckd_private *private; - unsigned long flags; - struct alias_server *server; - struct alias_lcu *lcu; - struct dasd_uid uid; - - private = (struct dasd_eckd_private *) device->private; - device->discipline->get_uid(device, &uid); - lcu = NULL; - spin_lock_irqsave(&aliastree.lock, flags); - server = _find_server(&uid); - if (server) - lcu = _find_lcu(server, &uid); - spin_unlock_irqrestore(&aliastree.lock, flags); - if (!lcu) { - DBF_EVENT_DEVID(DBF_ERR, device->cdev, - "could not find lcu for %04x %02x", - uid.ssid, uid.real_unit_addr); - WARN_ON(1); - return; - } - wait_for_completion(&lcu->lcu_setup); + return 0; } /* @@ -319,6 +253,9 @@ void dasd_alias_disconnect_device_from_lcu(struct dasd_device *device) private = (struct dasd_eckd_private *) device->private; lcu = private->lcu; + /* nothing to do if already disconnected */ + if (!lcu) + return; device->discipline->get_uid(device, &uid); spin_lock_irqsave(&lcu->lock, flags); list_del_init(&device->alias_list); @@ -447,6 +384,29 @@ static void _remove_device_from_lcu(struct alias_lcu *lcu, group->next = NULL; }; +static int +suborder_not_supported(struct dasd_ccw_req *cqr) +{ + char *sense; + char reason; + char msg_format; + char msg_no; + + sense = dasd_get_sense(&cqr->irb); + if (!sense) + return 0; + + reason = sense[0]; + msg_format = (sense[7] & 0xF0); + msg_no = (sense[7] & 0x0F); + + /* command reject, Format 0 MSG 4 - invalid parameter */ + if ((reason == 0x80) && (msg_format == 0x00) && (msg_no == 0x04)) + return 1; + + return 0; +} + static int read_unit_address_configuration(struct dasd_device *device, struct alias_lcu *lcu) { @@ -488,7 +448,7 @@ static int read_unit_address_configuration(struct dasd_device *device, ccw->count = sizeof(*(lcu->uac)); ccw->cda = (__u32)(addr_t) lcu->uac; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; /* need to unset flag here to detect race with summary unit check */ @@ -498,6 +458,8 @@ static int read_unit_address_configuration(struct dasd_device *device, do { rc = dasd_sleep_on(cqr); + if (rc && suborder_not_supported(cqr)) + return -EOPNOTSUPP; } while (rc && (cqr->retries > 0)); if (rc) { spin_lock_irqsave(&lcu->lock, flags); @@ -584,7 +546,7 @@ static void lcu_update_work(struct work_struct *work) * processing the data */ spin_lock_irqsave(&lcu->lock, flags); - if (rc || (lcu->flags & NEED_UAC_UPDATE)) { + if ((rc && (rc != -EOPNOTSUPP)) || (lcu->flags & NEED_UAC_UPDATE)) { DBF_DEV_EVENT(DBF_WARNING, device, "could not update" " alias data in lcu (rc = %d), retry later", rc); schedule_delayed_work(&lcu->ruac_data.dwork, 30*HZ); @@ -680,6 +642,9 @@ int dasd_alias_remove_device(struct dasd_device *device) private = (struct dasd_eckd_private *) device->private; lcu = private->lcu; + /* nothing to do if already removed */ + if (!lcu) + return 0; spin_lock_irqsave(&lcu->lock, flags); _remove_device_from_lcu(lcu, device); spin_unlock_irqrestore(&lcu->lock, flags); @@ -703,6 +668,16 @@ struct dasd_device *dasd_alias_get_start_dev(struct dasd_device *base_device) if (lcu->pav == NO_PAV || lcu->flags & (NEED_UAC_UPDATE | UPDATE_PENDING)) return NULL; + if (unlikely(!(private->features.feature[8] & 0x01))) { + /* + * PAV enabled but prefix not, very unlikely + * seems to be a lost pathgroup + * use base device to do IO + */ + DBF_DEV_EVENT(DBF_ERR, base_device, "%s", + "Prefix not enabled with PAV enabled\n"); + return NULL; + } spin_lock_irqsave(&lcu->lock, flags); alias_device = group->next; @@ -758,7 +733,7 @@ static int reset_summary_unit_check(struct alias_lcu *lcu, cqr->memdev = device; cqr->block = NULL; cqr->expires = 5 * HZ; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; rc = dasd_sleep_on_immediatly(cqr); diff --git a/drivers/s390/block/dasd_devmap.c b/drivers/s390/block/dasd_devmap.c index 8d41f3ed38d..2ead7e78c45 100644 --- a/drivers/s390/block/dasd_devmap.c +++ b/drivers/s390/block/dasd_devmap.c @@ -1,11 +1,10 @@ /* - * File...........: linux/drivers/s390/block/dasd_devmap.c * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Horst Hummel <Horst.Hummel@de.ibm.com> * Carsten Otte <Cotte@de.ibm.com> * Martin Schwidefsky <schwidefsky@de.ibm.com> * Bugreports.to..: <Linux390@de.ibm.com> - * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 1999-2001 + * Copyright IBM Corp. 1999,2001 * * Device mapping and dasd= parameter parsing functions. All devmap * functions may not be called from interrupt context. In particular @@ -208,6 +207,8 @@ dasd_feature_list(char *str, char **endp) features |= DASD_FEATURE_READONLY; else if (len == 4 && !strncmp(str, "diag", 4)) features |= DASD_FEATURE_USEDIAG; + else if (len == 3 && !strncmp(str, "raw", 3)) + features |= DASD_FEATURE_USERAW; else if (len == 6 && !strncmp(str, "erplog", 6)) features |= DASD_FEATURE_ERPLOG; else if (len == 8 && !strncmp(str, "failfast", 8)) @@ -300,7 +301,7 @@ dasd_parse_keyword( char *parsestring ) { /* * Try to interprete the first element on the comma separated parse string * as a device number or a range of devices. If the interpretation is - * successfull, create the matching dasd_devmap entries and return a pointer + * successful, create the matching dasd_devmap entries and return a pointer * to the residual string. * If interpretation fails or in case of an error, return an error code. */ @@ -409,8 +410,7 @@ dasd_add_busid(const char *bus_id, int features) struct dasd_devmap *devmap, *new, *tmp; int hash; - new = (struct dasd_devmap *) - kzalloc(sizeof(struct dasd_devmap), GFP_KERNEL); + new = kzalloc(sizeof(struct dasd_devmap), GFP_KERNEL); if (!new) return ERR_PTR(-ENOMEM); spin_lock(&dasd_devmap_lock); @@ -639,6 +639,7 @@ dasd_put_device_wake(struct dasd_device *device) { wake_up(&dasd_delete_wq); } +EXPORT_SYMBOL_GPL(dasd_put_device_wake); /* * Return dasd_device structure associated with cdev. @@ -671,6 +672,36 @@ dasd_device_from_cdev(struct ccw_device *cdev) return device; } +void dasd_add_link_to_gendisk(struct gendisk *gdp, struct dasd_device *device) +{ + struct dasd_devmap *devmap; + + devmap = dasd_find_busid(dev_name(&device->cdev->dev)); + if (IS_ERR(devmap)) + return; + spin_lock(&dasd_devmap_lock); + gdp->private_data = devmap; + spin_unlock(&dasd_devmap_lock); +} + +struct dasd_device *dasd_device_from_gendisk(struct gendisk *gdp) +{ + struct dasd_device *device; + struct dasd_devmap *devmap; + + if (!gdp->private_data) + return NULL; + device = NULL; + spin_lock(&dasd_devmap_lock); + devmap = gdp->private_data; + if (devmap && devmap->device) { + device = devmap->device; + dasd_get_device(device); + } + spin_unlock(&dasd_devmap_lock); + return device; +} + /* * SECTION: files in sysfs */ @@ -856,7 +887,7 @@ dasd_use_diag_store(struct device *dev, struct device_attribute *attr, spin_lock(&dasd_devmap_lock); /* Changing diag discipline flag is only allowed in offline state. */ rc = count; - if (!devmap->device) { + if (!devmap->device && !(devmap->features & DASD_FEATURE_USERAW)) { if (val) devmap->features |= DASD_FEATURE_USEDIAG; else @@ -869,6 +900,89 @@ dasd_use_diag_store(struct device *dev, struct device_attribute *attr, static DEVICE_ATTR(use_diag, 0644, dasd_use_diag_show, dasd_use_diag_store); +/* + * use_raw controls whether the driver should give access to raw eckd data or + * operate in standard mode + */ +static ssize_t +dasd_use_raw_show(struct device *dev, struct device_attribute *attr, char *buf) +{ + struct dasd_devmap *devmap; + int use_raw; + + devmap = dasd_find_busid(dev_name(dev)); + if (!IS_ERR(devmap)) + use_raw = (devmap->features & DASD_FEATURE_USERAW) != 0; + else + use_raw = (DASD_FEATURE_DEFAULT & DASD_FEATURE_USERAW) != 0; + return sprintf(buf, use_raw ? "1\n" : "0\n"); +} + +static ssize_t +dasd_use_raw_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + struct dasd_devmap *devmap; + ssize_t rc; + unsigned long val; + + devmap = dasd_devmap_from_cdev(to_ccwdev(dev)); + if (IS_ERR(devmap)) + return PTR_ERR(devmap); + + if ((kstrtoul(buf, 10, &val) != 0) || val > 1) + return -EINVAL; + + spin_lock(&dasd_devmap_lock); + /* Changing diag discipline flag is only allowed in offline state. */ + rc = count; + if (!devmap->device && !(devmap->features & DASD_FEATURE_USEDIAG)) { + if (val) + devmap->features |= DASD_FEATURE_USERAW; + else + devmap->features &= ~DASD_FEATURE_USERAW; + } else + rc = -EPERM; + spin_unlock(&dasd_devmap_lock); + return rc; +} + +static DEVICE_ATTR(raw_track_access, 0644, dasd_use_raw_show, + dasd_use_raw_store); + +static ssize_t +dasd_safe_offline_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + struct ccw_device *cdev = to_ccwdev(dev); + struct dasd_device *device; + int rc; + + device = dasd_device_from_cdev(cdev); + if (IS_ERR(device)) { + rc = PTR_ERR(device); + goto out; + } + + if (test_bit(DASD_FLAG_OFFLINE, &device->flags) || + test_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags)) { + /* Already doing offline processing */ + dasd_put_device(device); + rc = -EBUSY; + goto out; + } + + set_bit(DASD_FLAG_SAFE_OFFLINE, &device->flags); + dasd_put_device(device); + + rc = ccw_device_set_offline(cdev); + +out: + return rc ? rc : count; +} + +static DEVICE_ATTR(safe_offline, 0200, NULL, dasd_safe_offline_store); + static ssize_t dasd_discipline_show(struct device *dev, struct device_attribute *attr, char *buf) @@ -1111,7 +1225,7 @@ dasd_expires_store(struct device *dev, struct device_attribute *attr, if (IS_ERR(device)) return -ENODEV; - if ((strict_strtoul(buf, 10, &val) != 0) || + if ((kstrtoul(buf, 10, &val) != 0) || (val > DASD_EXPIRES_MAX) || val == 0) { dasd_put_device(device); return -EINVAL; @@ -1126,6 +1240,198 @@ dasd_expires_store(struct device *dev, struct device_attribute *attr, static DEVICE_ATTR(expires, 0644, dasd_expires_show, dasd_expires_store); +static ssize_t +dasd_retries_show(struct device *dev, struct device_attribute *attr, char *buf) +{ + struct dasd_device *device; + int len; + + device = dasd_device_from_cdev(to_ccwdev(dev)); + if (IS_ERR(device)) + return -ENODEV; + len = snprintf(buf, PAGE_SIZE, "%lu\n", device->default_retries); + dasd_put_device(device); + return len; +} + +static ssize_t +dasd_retries_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + struct dasd_device *device; + unsigned long val; + + device = dasd_device_from_cdev(to_ccwdev(dev)); + if (IS_ERR(device)) + return -ENODEV; + + if ((kstrtoul(buf, 10, &val) != 0) || + (val > DASD_RETRIES_MAX)) { + dasd_put_device(device); + return -EINVAL; + } + + if (val) + device->default_retries = val; + + dasd_put_device(device); + return count; +} + +static DEVICE_ATTR(retries, 0644, dasd_retries_show, dasd_retries_store); + +static ssize_t +dasd_timeout_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct dasd_device *device; + int len; + + device = dasd_device_from_cdev(to_ccwdev(dev)); + if (IS_ERR(device)) + return -ENODEV; + len = snprintf(buf, PAGE_SIZE, "%lu\n", device->blk_timeout); + dasd_put_device(device); + return len; +} + +static ssize_t +dasd_timeout_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + struct dasd_device *device; + struct request_queue *q; + unsigned long val, flags; + + device = dasd_device_from_cdev(to_ccwdev(dev)); + if (IS_ERR(device) || !device->block) + return -ENODEV; + + if ((kstrtoul(buf, 10, &val) != 0) || + val > UINT_MAX / HZ) { + dasd_put_device(device); + return -EINVAL; + } + q = device->block->request_queue; + if (!q) { + dasd_put_device(device); + return -ENODEV; + } + spin_lock_irqsave(&device->block->request_queue_lock, flags); + if (!val) + blk_queue_rq_timed_out(q, NULL); + else + blk_queue_rq_timed_out(q, dasd_times_out); + + device->blk_timeout = val; + + blk_queue_rq_timeout(q, device->blk_timeout * HZ); + spin_unlock_irqrestore(&device->block->request_queue_lock, flags); + + dasd_put_device(device); + return count; +} + +static DEVICE_ATTR(timeout, 0644, + dasd_timeout_show, dasd_timeout_store); + +static ssize_t dasd_reservation_policy_show(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct dasd_devmap *devmap; + int rc = 0; + + devmap = dasd_find_busid(dev_name(dev)); + if (IS_ERR(devmap)) { + rc = snprintf(buf, PAGE_SIZE, "ignore\n"); + } else { + spin_lock(&dasd_devmap_lock); + if (devmap->features & DASD_FEATURE_FAILONSLCK) + rc = snprintf(buf, PAGE_SIZE, "fail\n"); + else + rc = snprintf(buf, PAGE_SIZE, "ignore\n"); + spin_unlock(&dasd_devmap_lock); + } + return rc; +} + +static ssize_t dasd_reservation_policy_store(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) +{ + struct dasd_devmap *devmap; + int rc; + + devmap = dasd_devmap_from_cdev(to_ccwdev(dev)); + if (IS_ERR(devmap)) + return PTR_ERR(devmap); + rc = 0; + spin_lock(&dasd_devmap_lock); + if (sysfs_streq("ignore", buf)) + devmap->features &= ~DASD_FEATURE_FAILONSLCK; + else if (sysfs_streq("fail", buf)) + devmap->features |= DASD_FEATURE_FAILONSLCK; + else + rc = -EINVAL; + if (devmap->device) + devmap->device->features = devmap->features; + spin_unlock(&dasd_devmap_lock); + if (rc) + return rc; + else + return count; +} + +static DEVICE_ATTR(reservation_policy, 0644, + dasd_reservation_policy_show, dasd_reservation_policy_store); + +static ssize_t dasd_reservation_state_show(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct dasd_device *device; + int rc = 0; + + device = dasd_device_from_cdev(to_ccwdev(dev)); + if (IS_ERR(device)) + return snprintf(buf, PAGE_SIZE, "none\n"); + + if (test_bit(DASD_FLAG_IS_RESERVED, &device->flags)) + rc = snprintf(buf, PAGE_SIZE, "reserved\n"); + else if (test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags)) + rc = snprintf(buf, PAGE_SIZE, "lost\n"); + else + rc = snprintf(buf, PAGE_SIZE, "none\n"); + dasd_put_device(device); + return rc; +} + +static ssize_t dasd_reservation_state_store(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) +{ + struct dasd_device *device; + int rc = 0; + + device = dasd_device_from_cdev(to_ccwdev(dev)); + if (IS_ERR(device)) + return -ENODEV; + if (sysfs_streq("reset", buf)) + clear_bit(DASD_FLAG_LOCK_STOLEN, &device->flags); + else + rc = -EINVAL; + dasd_put_device(device); + + if (rc) + return rc; + else + return count; +} + +static DEVICE_ATTR(last_known_reservation_state, 0644, + dasd_reservation_state_show, dasd_reservation_state_store); + static struct attribute * dasd_attrs[] = { &dev_attr_readonly.attr, &dev_attr_discipline.attr, @@ -1134,10 +1440,16 @@ static struct attribute * dasd_attrs[] = { &dev_attr_vendor.attr, &dev_attr_uid.attr, &dev_attr_use_diag.attr, + &dev_attr_raw_track_access.attr, &dev_attr_eer_enabled.attr, &dev_attr_erplog.attr, &dev_attr_failfast.attr, &dev_attr_expires.attr, + &dev_attr_retries.attr, + &dev_attr_timeout.attr, + &dev_attr_reservation_policy.attr, + &dev_attr_last_known_reservation_state.attr, + &dev_attr_safe_offline.attr, NULL, }; @@ -1162,7 +1474,7 @@ dasd_get_feature(struct ccw_device *cdev, int feature) /* * Set / reset given feature. - * Flag indicates wether to set (!=0) or the reset (=0) the feature. + * Flag indicates whether to set (!=0) or the reset (=0) the feature. */ int dasd_set_feature(struct ccw_device *cdev, int feature, int flag) diff --git a/drivers/s390/block/dasd_diag.c b/drivers/s390/block/dasd_diag.c index 266b34b5540..c062f1620c5 100644 --- a/drivers/s390/block/dasd_diag.c +++ b/drivers/s390/block/dasd_diag.c @@ -1,15 +1,15 @@ /* - * File...........: linux/drivers/s390/block/dasd_diag.c * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Based on.......: linux/drivers/s390/block/mdisk.c * ...............: by Hartmunt Penner <hpenner@de.ibm.com> * Bugreports.to..: <Linux390@de.ibm.com> - * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 1999,2000 + * Copyright IBM Corp. 1999, 2000 * */ #define KMSG_COMPONENT "dasd" +#include <linux/kernel_stat.h> #include <linux/stddef.h> #include <linux/kernel.h> #include <linux/slab.h> @@ -23,7 +23,7 @@ #include <asm/debug.h> #include <asm/ebcdic.h> #include <asm/io.h> -#include <asm/s390_ext.h> +#include <asm/irq.h> #include <asm/vtoc.h> #include <asm/diag.h> @@ -184,14 +184,14 @@ dasd_start_diag(struct dasd_ccw_req * cqr) private->iob.bio_list = dreq->bio; private->iob.flaga = DASD_DIAG_FLAGA_DEFAULT; - cqr->startclk = get_clock(); + cqr->startclk = get_tod_clock(); cqr->starttime = jiffies; cqr->retries--; rc = dia250(&private->iob, RW_BIO); switch (rc) { case 0: /* Synchronous I/O finished successfully */ - cqr->stopclk = get_clock(); + cqr->stopclk = get_tod_clock(); cqr->status = DASD_CQR_SUCCESS; /* Indicate to calling function that only a dasd_schedule_bh() and no timer is needed */ @@ -222,13 +222,13 @@ dasd_diag_term_IO(struct dasd_ccw_req * cqr) mdsk_term_io(device); mdsk_init_io(device, device->block->bp_block, 0, NULL); cqr->status = DASD_CQR_CLEAR_PENDING; - cqr->stopclk = get_clock(); + cqr->stopclk = get_tod_clock(); dasd_schedule_device_bh(device); return 0; } /* Handle external interruption. */ -static void dasd_ext_handler(unsigned int ext_int_code, +static void dasd_ext_handler(struct ext_code ext_code, unsigned int param32, unsigned long param64) { struct dasd_ccw_req *cqr, *next; @@ -238,7 +238,7 @@ static void dasd_ext_handler(unsigned int ext_int_code, addr_t ip; int rc; - switch (ext_int_code >> 24) { + switch (ext_code.subcode >> 8) { case DASD_DIAG_CODE_31BIT: ip = (addr_t) param32; break; @@ -248,6 +248,7 @@ static void dasd_ext_handler(unsigned int ext_int_code, default: return; } + inc_irq_stat(IRQEXT_DSD); if (!ip) { /* no intparm: unsolicited interrupt */ DBF_EVENT(DBF_NOTICE, "%s", "caught unsolicited " "interrupt"); @@ -275,10 +276,10 @@ static void dasd_ext_handler(unsigned int ext_int_code, return; } - cqr->stopclk = get_clock(); + cqr->stopclk = get_tod_clock(); expires = 0; - if ((ext_int_code & 0xff0000) == 0) { + if ((ext_code.subcode & 0xff) == 0) { cqr->status = DASD_CQR_SUCCESS; /* Start first request on queue if possible -> fast_io. */ if (!list_empty(&device->ccw_queue)) { @@ -294,7 +295,7 @@ static void dasd_ext_handler(unsigned int ext_int_code, cqr->status = DASD_CQR_QUEUED; DBF_DEV_EVENT(DBF_DEBUG, device, "interrupt status for " "request %p was %d (%d retries left)", cqr, - (ext_int_code >> 16) & 0xff, cqr->retries); + ext_code.subcode & 0xff, cqr->retries); dasd_diag_erp(device); } @@ -358,6 +359,7 @@ dasd_diag_check_device(struct dasd_device *device) } device->default_expires = DIAG_TIMEOUT; + device->default_retries = DIAG_MAX_RETRIES; /* Figure out position of label block */ switch (private->rdc_data.vdev_class) { @@ -502,7 +504,7 @@ static struct dasd_ccw_req *dasd_diag_build_cp(struct dasd_device *memdev, struct dasd_diag_req *dreq; struct dasd_diag_bio *dbio; struct req_iterator iter; - struct bio_vec *bv; + struct bio_vec bv; char *dst; unsigned int count, datasize; sector_t recid, first_rec, last_rec; @@ -523,10 +525,10 @@ static struct dasd_ccw_req *dasd_diag_build_cp(struct dasd_device *memdev, /* Check struct bio and count the number of blocks for the request. */ count = 0; rq_for_each_segment(bv, req, iter) { - if (bv->bv_len & (blksize - 1)) + if (bv.bv_len & (blksize - 1)) /* Fba can only do full blocks. */ return ERR_PTR(-EINVAL); - count += bv->bv_len >> (block->s2b_shift + 9); + count += bv.bv_len >> (block->s2b_shift + 9); } /* Paranoia. */ if (count != last_rec - first_rec + 1) @@ -543,8 +545,8 @@ static struct dasd_ccw_req *dasd_diag_build_cp(struct dasd_device *memdev, dbio = dreq->bio; recid = first_rec; rq_for_each_segment(bv, req, iter) { - dst = page_address(bv->bv_page) + bv->bv_offset; - for (off = 0; off < bv->bv_len; off += blksize) { + dst = page_address(bv.bv_page) + bv.bv_offset; + for (off = 0; off < bv.bv_len; off += blksize) { memset(dbio, 0, sizeof (struct dasd_diag_bio)); dbio->type = rw_cmd; dbio->block_number = recid + 1; @@ -554,8 +556,8 @@ static struct dasd_ccw_req *dasd_diag_build_cp(struct dasd_device *memdev, recid++; } } - cqr->retries = DIAG_MAX_RETRIES; - cqr->buildclk = get_clock(); + cqr->retries = memdev->default_retries; + cqr->buildclk = get_tod_clock(); if (blk_noretry_request(req) || block->base->features & DASD_FEATURE_FAILFAST) set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags); @@ -581,7 +583,10 @@ dasd_diag_free_cp(struct dasd_ccw_req *cqr, struct request *req) static void dasd_diag_handle_terminated_request(struct dasd_ccw_req *cqr) { - cqr->status = DASD_CQR_FILLED; + if (cqr->retries < 0) + cqr->status = DASD_CQR_FAILED; + else + cqr->status = DASD_CQR_FILLED; }; /* Fill in IOCTL data for device. */ @@ -617,6 +622,7 @@ static struct dasd_discipline dasd_diag_discipline = { .ebcname = "DIAG", .max_blocks = DIAG_MAX_BLOCKS, .check_device = dasd_diag_check_device, + .verify_path = dasd_generic_verify_path, .fill_geometry = dasd_diag_fill_geometry, .start_IO = dasd_start_diag, .term_IO = dasd_diag_term_IO, @@ -639,8 +645,8 @@ dasd_diag_init(void) } ASCEBC(dasd_diag_discipline.ebcname, 4); - ctl_set_bit(0, 9); - register_external_interrupt(0x2603, dasd_ext_handler); + irq_subclass_register(IRQ_SUBCLASS_SERVICE_SIGNAL); + register_external_irq(EXT_IRQ_CP_SERVICE, dasd_ext_handler); dasd_diag_discipline_pointer = &dasd_diag_discipline; return 0; } @@ -648,8 +654,8 @@ dasd_diag_init(void) static void __exit dasd_diag_cleanup(void) { - unregister_external_interrupt(0x2603, dasd_ext_handler); - ctl_clear_bit(0, 9); + unregister_external_irq(EXT_IRQ_CP_SERVICE, dasd_ext_handler); + irq_subclass_unregister(IRQ_SUBCLASS_SERVICE_SIGNAL); dasd_diag_discipline_pointer = NULL; } diff --git a/drivers/s390/block/dasd_diag.h b/drivers/s390/block/dasd_diag.h index 4f71fbe60c8..a803cc73158 100644 --- a/drivers/s390/block/dasd_diag.h +++ b/drivers/s390/block/dasd_diag.h @@ -1,10 +1,9 @@ /* - * File...........: linux/drivers/s390/block/dasd_diag.h * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Based on.......: linux/drivers/s390/block/mdisk.h * ...............: by Hartmunt Penner <hpenner@de.ibm.com> * Bugreports.to..: <Linux390@de.ibm.com> - * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 1999,2000 + * Copyright IBM Corp. 1999, 2000 * */ diff --git a/drivers/s390/block/dasd_eckd.c b/drivers/s390/block/dasd_eckd.c index bf61274af3b..2e8e0755070 100644 --- a/drivers/s390/block/dasd_eckd.c +++ b/drivers/s390/block/dasd_eckd.c @@ -1,5 +1,4 @@ /* - * File...........: linux/drivers/s390/block/dasd_eckd.c * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Horst Hummel <Horst.Hummel@de.ibm.com> * Carsten Otte <Cotte@de.ibm.com> @@ -18,12 +17,13 @@ #include <linux/hdreg.h> /* HDIO_GETGEO */ #include <linux/bio.h> #include <linux/module.h> +#include <linux/compat.h> #include <linux/init.h> +#include <asm/css_chars.h> #include <asm/debug.h> #include <asm/idals.h> #include <asm/ebcdic.h> -#include <asm/compat.h> #include <asm/io.h> #include <asm/uaccess.h> #include <asm/cio.h> @@ -32,8 +32,6 @@ #include "dasd_int.h" #include "dasd_eckd.h" -#include "../cio/chsc.h" - #ifdef PRINTK_HEADER #undef PRINTK_HEADER @@ -54,6 +52,15 @@ #define ECKD_F7(i) (i->factor7) #define ECKD_F8(i) (i->factor8) +/* + * raw track access always map to 64k in memory + * so it maps to 16 blocks of 4k per track + */ +#define DASD_RAW_BLOCK_PER_TRACK 16 +#define DASD_RAW_BLOCKSIZE 4096 +/* 64k are 128 x 512 byte sectors */ +#define DASD_RAW_SECTORS_PER_TRACK 128 + MODULE_LICENSE("GPL"); static struct dasd_discipline dasd_eckd_discipline; @@ -63,7 +70,7 @@ static struct dasd_discipline dasd_eckd_discipline; static struct ccw_device_id dasd_eckd_ids[] = { { CCW_DEVICE_DEVTYPE (0x3990, 0, 0x3390, 0), .driver_info = 0x1}, { CCW_DEVICE_DEVTYPE (0x2105, 0, 0x3390, 0), .driver_info = 0x2}, - { CCW_DEVICE_DEVTYPE (0x3880, 0, 0x3390, 0), .driver_info = 0x3}, + { CCW_DEVICE_DEVTYPE (0x3880, 0, 0x3380, 0), .driver_info = 0x3}, { CCW_DEVICE_DEVTYPE (0x3990, 0, 0x3380, 0), .driver_info = 0x4}, { CCW_DEVICE_DEVTYPE (0x2105, 0, 0x3380, 0), .driver_info = 0x5}, { CCW_DEVICE_DEVTYPE (0x9343, 0, 0x9345, 0), .driver_info = 0x6}, @@ -78,6 +85,8 @@ MODULE_DEVICE_TABLE(ccw, dasd_eckd_ids); static struct ccw_driver dasd_eckd_driver; /* see below */ +static void *rawpadpage; + #define INIT_CQR_OK 0 #define INIT_CQR_UNFORMATTED 1 #define INIT_CQR_ERROR 2 @@ -90,6 +99,18 @@ static struct { } *dasd_reserve_req; static DEFINE_MUTEX(dasd_reserve_mutex); +/* definitions for the path verification worker */ +struct path_verification_work_data { + struct work_struct worker; + struct dasd_device *device; + struct dasd_ccw_req cqr; + struct ccw1 ccw; + __u8 rcd_buffer[DASD_ECKD_RCD_DATA_SIZE]; + int isglobal; + __u8 tbvpm; +}; +static struct path_verification_work_data *path_verification_worker; +static DEFINE_MUTEX(dasd_path_verification_mutex); /* initial attempt at a probe function. this can be simplified once * the other detection code is gone */ @@ -120,6 +141,10 @@ dasd_eckd_set_online(struct ccw_device *cdev) static const int sizes_trk0[] = { 28, 148, 84 }; #define LABEL_SIZE 140 +/* head and record addresses of count_area read in analysis ccw */ +static const int count_area_head[] = { 0, 0, 0, 0, 2 }; +static const int count_area_rec[] = { 1, 2, 3, 4, 1 }; + static inline unsigned int round_up_multiple(unsigned int no, unsigned int mult) { @@ -192,7 +217,7 @@ check_XRC (struct ccw1 *de_ccw, rc = get_sync_clock(&data->ep_sys_time); /* Ignore return code if sync clock is switched off. */ - if (rc == -ENOSYS || rc == -EACCES) + if (rc == -EOPNOTSUPP || rc == -EACCES) rc = 0; de_ccw->count = sizeof(struct DE_eckd_data); @@ -303,7 +328,7 @@ static int check_XRC_on_prefix(struct PFX_eckd_data *pfxdata, rc = get_sync_clock(&pfxdata->define_extent.ep_sys_time); /* Ignore return code if sync clock is switched off. */ - if (rc == -ENOSYS || rc == -EACCES) + if (rc == -EOPNOTSUPP || rc == -EACCES) rc = 0; return rc; } @@ -373,6 +398,23 @@ static void fill_LRE_data(struct LRE_eckd_data *data, unsigned int trk, data->length = reclen; data->operation.operation = 0x03; break; + case DASD_ECKD_CCW_WRITE_FULL_TRACK: + data->operation.orientation = 0x0; + data->operation.operation = 0x3F; + data->extended_operation = 0x11; + data->length = 0; + data->extended_parameter_length = 0x02; + if (data->count > 8) { + data->extended_parameter[0] = 0xFF; + data->extended_parameter[1] = 0xFF; + data->extended_parameter[1] <<= (16 - count); + } else { + data->extended_parameter[0] = 0xFF; + data->extended_parameter[0] <<= (8 - count); + data->extended_parameter[1] = 0x00; + } + data->sector = 0xFF; + break; case DASD_ECKD_CCW_WRITE_TRACK_DATA: data->auxiliary.length_valid = 0x1; data->length = reclen; /* not tlf, as one might think */ @@ -396,6 +438,12 @@ static void fill_LRE_data(struct LRE_eckd_data *data, unsigned int trk, case DASD_ECKD_CCW_READ_COUNT: data->operation.operation = 0x06; break; + case DASD_ECKD_CCW_READ_TRACK: + data->operation.orientation = 0x1; + data->operation.operation = 0x0C; + data->extended_parameter_length = 0; + data->sector = 0xFF; + break; case DASD_ECKD_CCW_READ_TRACK_DATA: data->auxiliary.length_valid = 0x1; data->length = tlf; @@ -439,10 +487,16 @@ static int prefix_LRE(struct ccw1 *ccw, struct PFX_eckd_data *pfxdata, ccw->cmd_code = DASD_ECKD_CCW_PFX; ccw->flags = 0; - ccw->count = sizeof(*pfxdata); - ccw->cda = (__u32) __pa(pfxdata); + if (cmd == DASD_ECKD_CCW_WRITE_FULL_TRACK) { + ccw->count = sizeof(*pfxdata) + 2; + ccw->cda = (__u32) __pa(pfxdata); + memset(pfxdata, 0, sizeof(*pfxdata) + 2); + } else { + ccw->count = sizeof(*pfxdata); + ccw->cda = (__u32) __pa(pfxdata); + memset(pfxdata, 0, sizeof(*pfxdata)); + } - memset(pfxdata, 0, sizeof(*pfxdata)); /* prefix data */ if (format > 1) { DBF_DEV_EVENT(DBF_ERR, basedev, @@ -476,6 +530,7 @@ static int prefix_LRE(struct ccw1 *ccw, struct PFX_eckd_data *pfxdata, dedata->mask.perm = 0x1; dedata->attributes.operation = basepriv->attrib.operation; break; + case DASD_ECKD_CCW_READ_TRACK: case DASD_ECKD_CCW_READ_TRACK_DATA: dedata->mask.perm = 0x1; dedata->attributes.operation = basepriv->attrib.operation; @@ -502,6 +557,11 @@ static int prefix_LRE(struct ccw1 *ccw, struct PFX_eckd_data *pfxdata, dedata->attributes.operation = DASD_BYPASS_CACHE; rc = check_XRC_on_prefix(pfxdata, basedev); break; + case DASD_ECKD_CCW_WRITE_FULL_TRACK: + dedata->mask.perm = 0x03; + dedata->attributes.operation = basepriv->attrib.operation; + dedata->blk_size = 0; + break; case DASD_ECKD_CCW_WRITE_TRACK_DATA: dedata->mask.perm = 0x02; dedata->attributes.operation = basepriv->attrib.operation; @@ -696,24 +756,13 @@ dasd_eckd_cdl_reclen(int recid) return sizes_trk0[recid]; return LABEL_SIZE; } - -/* - * Generate device unique id that specifies the physical device. - */ -static int dasd_eckd_generate_uid(struct dasd_device *device) +/* create unique id from private structure. */ +static void create_uid(struct dasd_eckd_private *private) { - struct dasd_eckd_private *private; - struct dasd_uid *uid; int count; - unsigned long flags; + struct dasd_uid *uid; - private = (struct dasd_eckd_private *) device->private; - if (!private) - return -ENODEV; - if (!private->ned || !private->gneq) - return -ENODEV; uid = &private->uid; - spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags); memset(uid, 0, sizeof(struct dasd_uid)); memcpy(uid->vendor, private->ned->HDA_manufacturer, sizeof(uid->vendor) - 1); @@ -736,6 +785,23 @@ static int dasd_eckd_generate_uid(struct dasd_device *device) private->vdsneq->uit[count]); } } +} + +/* + * Generate device unique id that specifies the physical device. + */ +static int dasd_eckd_generate_uid(struct dasd_device *device) +{ + struct dasd_eckd_private *private; + unsigned long flags; + + private = (struct dasd_eckd_private *) device->private; + if (!private) + return -ENODEV; + if (!private->ned || !private->gneq) + return -ENODEV; + spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags); + create_uid(private); spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags); return 0; } @@ -755,26 +821,42 @@ static int dasd_eckd_get_uid(struct dasd_device *device, struct dasd_uid *uid) return -EINVAL; } -static struct dasd_ccw_req *dasd_eckd_build_rcd_lpm(struct dasd_device *device, - void *rcd_buffer, - struct ciw *ciw, __u8 lpm) +/* + * compare device UID with data of a given dasd_eckd_private structure + * return 0 for match + */ +static int dasd_eckd_compare_path_uid(struct dasd_device *device, + struct dasd_eckd_private *private) { - struct dasd_ccw_req *cqr; - struct ccw1 *ccw; + struct dasd_uid device_uid; - cqr = dasd_smalloc_request(DASD_ECKD_MAGIC, 1 /* RCD */, ciw->count, - device); + create_uid(private); + dasd_eckd_get_uid(device, &device_uid); - if (IS_ERR(cqr)) { - DBF_DEV_EVENT(DBF_WARNING, device, "%s", - "Could not allocate RCD request"); - return cqr; - } + return memcmp(&device_uid, &private->uid, sizeof(struct dasd_uid)); +} + +static void dasd_eckd_fill_rcd_cqr(struct dasd_device *device, + struct dasd_ccw_req *cqr, + __u8 *rcd_buffer, + __u8 lpm) +{ + struct ccw1 *ccw; + /* + * buffer has to start with EBCDIC "V1.0" to show + * support for virtual device SNEQ + */ + rcd_buffer[0] = 0xE5; + rcd_buffer[1] = 0xF1; + rcd_buffer[2] = 0x4B; + rcd_buffer[3] = 0xF0; ccw = cqr->cpaddr; - ccw->cmd_code = ciw->cmd; + ccw->cmd_code = DASD_ECKD_CCW_RCD; + ccw->flags = 0; ccw->cda = (__u32)(addr_t)rcd_buffer; - ccw->count = ciw->count; + ccw->count = DASD_ECKD_RCD_DATA_SIZE; + cqr->magic = DASD_ECKD_MAGIC; cqr->startdev = device; cqr->memdev = device; @@ -782,9 +864,57 @@ static struct dasd_ccw_req *dasd_eckd_build_rcd_lpm(struct dasd_device *device, cqr->expires = 10*HZ; cqr->lpm = lpm; cqr->retries = 256; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; - return cqr; + set_bit(DASD_CQR_VERIFY_PATH, &cqr->flags); +} + +/* + * Wakeup helper for read_conf + * if the cqr is not done and needs some error recovery + * the buffer has to be re-initialized with the EBCDIC "V1.0" + * to show support for virtual device SNEQ + */ +static void read_conf_cb(struct dasd_ccw_req *cqr, void *data) +{ + struct ccw1 *ccw; + __u8 *rcd_buffer; + + if (cqr->status != DASD_CQR_DONE) { + ccw = cqr->cpaddr; + rcd_buffer = (__u8 *)((addr_t) ccw->cda); + memset(rcd_buffer, 0, sizeof(*rcd_buffer)); + + rcd_buffer[0] = 0xE5; + rcd_buffer[1] = 0xF1; + rcd_buffer[2] = 0x4B; + rcd_buffer[3] = 0xF0; + } + dasd_wakeup_cb(cqr, data); +} + +static int dasd_eckd_read_conf_immediately(struct dasd_device *device, + struct dasd_ccw_req *cqr, + __u8 *rcd_buffer, + __u8 lpm) +{ + struct ciw *ciw; + int rc; + /* + * sanity check: scan for RCD command in extended SenseID data + * some devices do not support RCD + */ + ciw = ccw_device_get_ciw(device->cdev, CIW_TYPE_RCD); + if (!ciw || ciw->cmd != DASD_ECKD_CCW_RCD) + return -EOPNOTSUPP; + + dasd_eckd_fill_rcd_cqr(device, cqr, rcd_buffer, lpm); + clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags); + set_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags); + cqr->retries = 5; + cqr->callback = read_conf_cb; + rc = dasd_sleep_on_immediatly(cqr); + return rc; } static int dasd_eckd_read_conf_lpm(struct dasd_device *device, @@ -797,32 +927,30 @@ static int dasd_eckd_read_conf_lpm(struct dasd_device *device, struct dasd_ccw_req *cqr; /* - * scan for RCD command in extended SenseID data + * sanity check: scan for RCD command in extended SenseID data + * some devices do not support RCD */ ciw = ccw_device_get_ciw(device->cdev, CIW_TYPE_RCD); - if (!ciw || ciw->cmd == 0) { + if (!ciw || ciw->cmd != DASD_ECKD_CCW_RCD) { ret = -EOPNOTSUPP; goto out_error; } - rcd_buf = kzalloc(ciw->count, GFP_KERNEL | GFP_DMA); + rcd_buf = kzalloc(DASD_ECKD_RCD_DATA_SIZE, GFP_KERNEL | GFP_DMA); if (!rcd_buf) { ret = -ENOMEM; goto out_error; } - - /* - * buffer has to start with EBCDIC "V1.0" to show - * support for virtual device SNEQ - */ - rcd_buf[0] = 0xE5; - rcd_buf[1] = 0xF1; - rcd_buf[2] = 0x4B; - rcd_buf[3] = 0xF0; - cqr = dasd_eckd_build_rcd_lpm(device, rcd_buf, ciw, lpm); + cqr = dasd_smalloc_request(DASD_ECKD_MAGIC, 1 /* RCD */, + 0, /* use rcd_buf as data ara */ + device); if (IS_ERR(cqr)) { - ret = PTR_ERR(cqr); + DBF_DEV_EVENT(DBF_WARNING, device, "%s", + "Could not allocate RCD request"); + ret = -ENOMEM; goto out_error; } + dasd_eckd_fill_rcd_cqr(device, cqr, rcd_buf, lpm); + cqr->callback = read_conf_cb; ret = dasd_sleep_on(cqr); /* * on success we update the user input parms @@ -831,7 +959,7 @@ static int dasd_eckd_read_conf_lpm(struct dasd_device *device, if (ret) goto out_error; - *rcd_buffer_size = ciw->count; + *rcd_buffer_size = DASD_ECKD_RCD_DATA_SIZE; *rcd_buffer = rcd_buf; return 0; out_error: @@ -900,58 +1028,378 @@ static int dasd_eckd_read_conf(struct dasd_device *device) { void *conf_data; int conf_len, conf_data_saved; - int rc; - __u8 lpm; - struct dasd_eckd_private *private; - struct dasd_eckd_path *path_data; + int rc, path_err; + __u8 lpm, opm; + struct dasd_eckd_private *private, path_private; + struct dasd_path *path_data; + struct dasd_uid *uid; + char print_path_uid[60], print_device_uid[60]; private = (struct dasd_eckd_private *) device->private; - path_data = (struct dasd_eckd_path *) &private->path_data; - path_data->opm = ccw_device_get_path_mask(device->cdev); - lpm = 0x80; + path_data = &device->path_data; + opm = ccw_device_get_path_mask(device->cdev); conf_data_saved = 0; + path_err = 0; /* get configuration data per operational path */ for (lpm = 0x80; lpm; lpm>>= 1) { - if (lpm & path_data->opm){ - rc = dasd_eckd_read_conf_lpm(device, &conf_data, - &conf_len, lpm); - if (rc && rc != -EOPNOTSUPP) { /* -EOPNOTSUPP is ok */ - DBF_EVENT_DEVID(DBF_WARNING, device->cdev, - "Read configuration data returned " - "error %d", rc); - return rc; + if (!(lpm & opm)) + continue; + rc = dasd_eckd_read_conf_lpm(device, &conf_data, + &conf_len, lpm); + if (rc && rc != -EOPNOTSUPP) { /* -EOPNOTSUPP is ok */ + DBF_EVENT_DEVID(DBF_WARNING, device->cdev, + "Read configuration data returned " + "error %d", rc); + return rc; + } + if (conf_data == NULL) { + DBF_EVENT_DEVID(DBF_WARNING, device->cdev, "%s", + "No configuration data " + "retrieved"); + /* no further analysis possible */ + path_data->opm |= lpm; + continue; /* no error */ + } + /* save first valid configuration data */ + if (!conf_data_saved) { + kfree(private->conf_data); + private->conf_data = conf_data; + private->conf_len = conf_len; + if (dasd_eckd_identify_conf_parts(private)) { + private->conf_data = NULL; + private->conf_len = 0; + kfree(conf_data); + continue; } - if (conf_data == NULL) { - DBF_EVENT_DEVID(DBF_WARNING, device->cdev, "%s", - "No configuration data " - "retrieved"); - continue; /* no error */ + /* + * build device UID that other path data + * can be compared to it + */ + dasd_eckd_generate_uid(device); + conf_data_saved++; + } else { + path_private.conf_data = conf_data; + path_private.conf_len = DASD_ECKD_RCD_DATA_SIZE; + if (dasd_eckd_identify_conf_parts( + &path_private)) { + path_private.conf_data = NULL; + path_private.conf_len = 0; + kfree(conf_data); + continue; } - /* save first valid configuration data */ - if (!conf_data_saved) { - kfree(private->conf_data); - private->conf_data = conf_data; - private->conf_len = conf_len; - if (dasd_eckd_identify_conf_parts(private)) { - private->conf_data = NULL; - private->conf_len = 0; - kfree(conf_data); - continue; - } - conf_data_saved++; + + if (dasd_eckd_compare_path_uid( + device, &path_private)) { + uid = &path_private.uid; + if (strlen(uid->vduit) > 0) + snprintf(print_path_uid, + sizeof(print_path_uid), + "%s.%s.%04x.%02x.%s", + uid->vendor, uid->serial, + uid->ssid, uid->real_unit_addr, + uid->vduit); + else + snprintf(print_path_uid, + sizeof(print_path_uid), + "%s.%s.%04x.%02x", + uid->vendor, uid->serial, + uid->ssid, + uid->real_unit_addr); + uid = &private->uid; + if (strlen(uid->vduit) > 0) + snprintf(print_device_uid, + sizeof(print_device_uid), + "%s.%s.%04x.%02x.%s", + uid->vendor, uid->serial, + uid->ssid, uid->real_unit_addr, + uid->vduit); + else + snprintf(print_device_uid, + sizeof(print_device_uid), + "%s.%s.%04x.%02x", + uid->vendor, uid->serial, + uid->ssid, + uid->real_unit_addr); + dev_err(&device->cdev->dev, + "Not all channel paths lead to " + "the same device, path %02X leads to " + "device %s instead of %s\n", lpm, + print_path_uid, print_device_uid); + path_err = -EINVAL; + continue; } - switch (dasd_eckd_path_access(conf_data, conf_len)) { + + path_private.conf_data = NULL; + path_private.conf_len = 0; + } + switch (dasd_eckd_path_access(conf_data, conf_len)) { + case 0x02: + path_data->npm |= lpm; + break; + case 0x03: + path_data->ppm |= lpm; + break; + } + path_data->opm |= lpm; + + if (conf_data != private->conf_data) + kfree(conf_data); + } + + return path_err; +} + +static int verify_fcx_max_data(struct dasd_device *device, __u8 lpm) +{ + struct dasd_eckd_private *private; + int mdc; + u32 fcx_max_data; + + private = (struct dasd_eckd_private *) device->private; + if (private->fcx_max_data) { + mdc = ccw_device_get_mdc(device->cdev, lpm); + if ((mdc < 0)) { + dev_warn(&device->cdev->dev, + "Detecting the maximum data size for zHPF " + "requests failed (rc=%d) for a new path %x\n", + mdc, lpm); + return mdc; + } + fcx_max_data = mdc * FCX_MAX_DATA_FACTOR; + if (fcx_max_data < private->fcx_max_data) { + dev_warn(&device->cdev->dev, + "The maximum data size for zHPF requests %u " + "on a new path %x is below the active maximum " + "%u\n", fcx_max_data, lpm, + private->fcx_max_data); + return -EACCES; + } + } + return 0; +} + +static int rebuild_device_uid(struct dasd_device *device, + struct path_verification_work_data *data) +{ + struct dasd_eckd_private *private; + struct dasd_path *path_data; + __u8 lpm, opm; + int rc; + + rc = -ENODEV; + private = (struct dasd_eckd_private *) device->private; + path_data = &device->path_data; + opm = device->path_data.opm; + + for (lpm = 0x80; lpm; lpm >>= 1) { + if (!(lpm & opm)) + continue; + memset(&data->rcd_buffer, 0, sizeof(data->rcd_buffer)); + memset(&data->cqr, 0, sizeof(data->cqr)); + data->cqr.cpaddr = &data->ccw; + rc = dasd_eckd_read_conf_immediately(device, &data->cqr, + data->rcd_buffer, + lpm); + + if (rc) { + if (rc == -EOPNOTSUPP) /* -EOPNOTSUPP is ok */ + continue; + DBF_EVENT_DEVID(DBF_WARNING, device->cdev, + "Read configuration data " + "returned error %d", rc); + break; + } + memcpy(private->conf_data, data->rcd_buffer, + DASD_ECKD_RCD_DATA_SIZE); + if (dasd_eckd_identify_conf_parts(private)) { + rc = -ENODEV; + } else /* first valid path is enough */ + break; + } + + if (!rc) + rc = dasd_eckd_generate_uid(device); + + return rc; +} + +static void do_path_verification_work(struct work_struct *work) +{ + struct path_verification_work_data *data; + struct dasd_device *device; + struct dasd_eckd_private path_private; + struct dasd_uid *uid; + __u8 path_rcd_buf[DASD_ECKD_RCD_DATA_SIZE]; + __u8 lpm, opm, npm, ppm, epm; + unsigned long flags; + char print_uid[60]; + int rc; + + data = container_of(work, struct path_verification_work_data, worker); + device = data->device; + + /* delay path verification until device was resumed */ + if (test_bit(DASD_FLAG_SUSPENDED, &device->flags)) { + schedule_work(work); + return; + } + + opm = 0; + npm = 0; + ppm = 0; + epm = 0; + for (lpm = 0x80; lpm; lpm >>= 1) { + if (!(lpm & data->tbvpm)) + continue; + memset(&data->rcd_buffer, 0, sizeof(data->rcd_buffer)); + memset(&data->cqr, 0, sizeof(data->cqr)); + data->cqr.cpaddr = &data->ccw; + rc = dasd_eckd_read_conf_immediately(device, &data->cqr, + data->rcd_buffer, + lpm); + if (!rc) { + switch (dasd_eckd_path_access(data->rcd_buffer, + DASD_ECKD_RCD_DATA_SIZE) + ) { case 0x02: - path_data->npm |= lpm; + npm |= lpm; break; case 0x03: - path_data->ppm |= lpm; + ppm |= lpm; break; } - if (conf_data != private->conf_data) - kfree(conf_data); + opm |= lpm; + } else if (rc == -EOPNOTSUPP) { + DBF_EVENT_DEVID(DBF_WARNING, device->cdev, "%s", + "path verification: No configuration " + "data retrieved"); + opm |= lpm; + } else if (rc == -EAGAIN) { + DBF_EVENT_DEVID(DBF_WARNING, device->cdev, "%s", + "path verification: device is stopped," + " try again later"); + epm |= lpm; + } else { + dev_warn(&device->cdev->dev, + "Reading device feature codes failed " + "(rc=%d) for new path %x\n", rc, lpm); + continue; + } + if (verify_fcx_max_data(device, lpm)) { + opm &= ~lpm; + npm &= ~lpm; + ppm &= ~lpm; + continue; } + + /* + * save conf_data for comparison after + * rebuild_device_uid may have changed + * the original data + */ + memcpy(&path_rcd_buf, data->rcd_buffer, + DASD_ECKD_RCD_DATA_SIZE); + path_private.conf_data = (void *) &path_rcd_buf; + path_private.conf_len = DASD_ECKD_RCD_DATA_SIZE; + if (dasd_eckd_identify_conf_parts(&path_private)) { + path_private.conf_data = NULL; + path_private.conf_len = 0; + continue; + } + + /* + * compare path UID with device UID only if at least + * one valid path is left + * in other case the device UID may have changed and + * the first working path UID will be used as device UID + */ + if (device->path_data.opm && + dasd_eckd_compare_path_uid(device, &path_private)) { + /* + * the comparison was not successful + * rebuild the device UID with at least one + * known path in case a z/VM hyperswap command + * has changed the device + * + * after this compare again + * + * if either the rebuild or the recompare fails + * the path can not be used + */ + if (rebuild_device_uid(device, data) || + dasd_eckd_compare_path_uid( + device, &path_private)) { + uid = &path_private.uid; + if (strlen(uid->vduit) > 0) + snprintf(print_uid, sizeof(print_uid), + "%s.%s.%04x.%02x.%s", + uid->vendor, uid->serial, + uid->ssid, uid->real_unit_addr, + uid->vduit); + else + snprintf(print_uid, sizeof(print_uid), + "%s.%s.%04x.%02x", + uid->vendor, uid->serial, + uid->ssid, + uid->real_unit_addr); + dev_err(&device->cdev->dev, + "The newly added channel path %02X " + "will not be used because it leads " + "to a different device %s\n", + lpm, print_uid); + opm &= ~lpm; + npm &= ~lpm; + ppm &= ~lpm; + continue; + } + } + + /* + * There is a small chance that a path is lost again between + * above path verification and the following modification of + * the device opm mask. We could avoid that race here by using + * yet another path mask, but we rather deal with this unlikely + * situation in dasd_start_IO. + */ + spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags); + if (!device->path_data.opm && opm) { + device->path_data.opm = opm; + dasd_generic_path_operational(device); + } else + device->path_data.opm |= opm; + device->path_data.npm |= npm; + device->path_data.ppm |= ppm; + device->path_data.tbvpm |= epm; + spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags); } + + dasd_put_device(device); + if (data->isglobal) + mutex_unlock(&dasd_path_verification_mutex); + else + kfree(data); +} + +static int dasd_eckd_verify_path(struct dasd_device *device, __u8 lpm) +{ + struct path_verification_work_data *data; + + data = kmalloc(sizeof(*data), GFP_ATOMIC | GFP_DMA); + if (!data) { + if (mutex_trylock(&dasd_path_verification_mutex)) { + data = path_verification_worker; + data->isglobal = 1; + } else + return -ENOMEM; + } else { + memset(data, 0, sizeof(*data)); + data->isglobal = 0; + } + INIT_WORK(&data->worker, do_path_verification_work); + dasd_get_device(device); + data->device = device; + data->tbvpm = lpm; + schedule_work(&data->worker); return 0; } @@ -1003,7 +1451,7 @@ static int dasd_eckd_read_features(struct dasd_device *device) ccw->count = sizeof(struct dasd_rssd_features); ccw->cda = (__u32)(addr_t) features; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; rc = dasd_sleep_on(cqr); if (rc == 0) { @@ -1055,7 +1503,7 @@ static struct dasd_ccw_req *dasd_eckd_build_psf_ssc(struct dasd_device *device, cqr->block = NULL; cqr->retries = 256; cqr->expires = 10*HZ; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; return cqr; } @@ -1066,7 +1514,8 @@ static struct dasd_ccw_req *dasd_eckd_build_psf_ssc(struct dasd_device *device, * call might change behaviour of DASD devices. */ static int -dasd_eckd_psf_ssc(struct dasd_device *device, int enable_pav) +dasd_eckd_psf_ssc(struct dasd_device *device, int enable_pav, + unsigned long flags) { struct dasd_ccw_req *cqr; int rc; @@ -1075,10 +1524,19 @@ dasd_eckd_psf_ssc(struct dasd_device *device, int enable_pav) if (IS_ERR(cqr)) return PTR_ERR(cqr); + /* + * set flags e.g. turn on failfast, to prevent blocking + * the calling function should handle failed requests + */ + cqr->flags |= flags; + rc = dasd_sleep_on(cqr); if (!rc) /* trigger CIO to reprobe devices */ css_schedule_reprobe(); + else if (cqr->intrc == -EAGAIN) + rc = -EAGAIN; + dasd_sfree_request(cqr, cqr->memdev); return rc; } @@ -1086,23 +1544,92 @@ dasd_eckd_psf_ssc(struct dasd_device *device, int enable_pav) /* * Valide storage server of current device. */ -static void dasd_eckd_validate_server(struct dasd_device *device) +static int dasd_eckd_validate_server(struct dasd_device *device, + unsigned long flags) { int rc; struct dasd_eckd_private *private; int enable_pav; + private = (struct dasd_eckd_private *) device->private; + if (private->uid.type == UA_BASE_PAV_ALIAS || + private->uid.type == UA_HYPER_PAV_ALIAS) + return 0; if (dasd_nopav || MACHINE_IS_VM) enable_pav = 0; else enable_pav = 1; - rc = dasd_eckd_psf_ssc(device, enable_pav); + rc = dasd_eckd_psf_ssc(device, enable_pav, flags); /* may be requested feature is not available on server, * therefore just report error and go ahead */ - private = (struct dasd_eckd_private *) device->private; DBF_EVENT_DEVID(DBF_WARNING, device->cdev, "PSF-SSC for SSID %04x " "returned rc=%d", private->uid.ssid, rc); + return rc; +} + +/* + * worker to do a validate server in case of a lost pathgroup + */ +static void dasd_eckd_do_validate_server(struct work_struct *work) +{ + struct dasd_device *device = container_of(work, struct dasd_device, + kick_validate); + unsigned long flags = 0; + + set_bit(DASD_CQR_FLAGS_FAILFAST, &flags); + if (dasd_eckd_validate_server(device, flags) + == -EAGAIN) { + /* schedule worker again if failed */ + schedule_work(&device->kick_validate); + return; + } + + dasd_put_device(device); +} + +static void dasd_eckd_kick_validate_server(struct dasd_device *device) +{ + dasd_get_device(device); + /* exit if device not online or in offline processing */ + if (test_bit(DASD_FLAG_OFFLINE, &device->flags) || + device->state < DASD_STATE_ONLINE) { + dasd_put_device(device); + return; + } + /* queue call to do_validate_server to the kernel event daemon. */ + schedule_work(&device->kick_validate); +} + +static u32 get_fcx_max_data(struct dasd_device *device) +{ +#if defined(CONFIG_64BIT) + int tpm, mdc; + int fcx_in_css, fcx_in_gneq, fcx_in_features; + struct dasd_eckd_private *private; + + if (dasd_nofcx) + return 0; + /* is transport mode supported? */ + private = (struct dasd_eckd_private *) device->private; + fcx_in_css = css_general_characteristics.fcx; + fcx_in_gneq = private->gneq->reserved2[7] & 0x04; + fcx_in_features = private->features.feature[40] & 0x80; + tpm = fcx_in_css && fcx_in_gneq && fcx_in_features; + + if (!tpm) + return 0; + + mdc = ccw_device_get_mdc(device->cdev, 0); + if (mdc < 0) { + dev_warn(&device->cdev->dev, "Detecting the maximum supported" + " data size for zHPF requests failed\n"); + return 0; + } else + return mdc * FCX_MAX_DATA_FACTOR; +#else + return 0; +#endif } /* @@ -1115,10 +1642,13 @@ dasd_eckd_check_characteristics(struct dasd_device *device) struct dasd_eckd_private *private; struct dasd_block *block; struct dasd_uid temp_uid; - int is_known, rc, i; + int rc, i; int readonly; unsigned long value; + /* setup work queue for validate server*/ + INIT_WORK(&device->kick_validate, dasd_eckd_do_validate_server); + if (!ccw_device_is_pathgroup(device->cdev)) { dev_warn(&device->cdev->dev, "A channel path group could not be established\n"); @@ -1154,6 +1684,9 @@ dasd_eckd_check_characteristics(struct dasd_device *device) /* set default timeout */ device->default_expires = DASD_EXPIRES; + /* set default retry count */ + device->default_retries = DASD_RETRIES; + if (private->gneq) { value = 1; for (i = 0; i < private->gneq->timeout.value; i++) @@ -1164,11 +1697,6 @@ dasd_eckd_check_characteristics(struct dasd_device *device) device->default_expires = value; } - /* Generate device unique id */ - rc = dasd_eckd_generate_uid(device); - if (rc) - goto out_err1; - dasd_eckd_get_uid(device, &temp_uid); if (temp_uid.type == UA_BASE_DEVICE) { block = dasd_alloc_block(); @@ -1183,22 +1711,12 @@ dasd_eckd_check_characteristics(struct dasd_device *device) block->base = device; } - /* register lcu with alias handling, enable PAV if this is a new lcu */ - is_known = dasd_alias_make_device_known_to_lcu(device); - if (is_known < 0) { - rc = is_known; + /* register lcu with alias handling, enable PAV */ + rc = dasd_alias_make_device_known_to_lcu(device); + if (rc) goto out_err2; - } - /* - * dasd_eckd_validate_server is done on the first device that - * is found for an LCU. All later other devices have to wait - * for it, so they will read the correct feature codes. - */ - if (!is_known) { - dasd_eckd_validate_server(device); - dasd_alias_lcu_setup_complete(device); - } else - dasd_alias_wait_for_lcu_setup(device); + + dasd_eckd_validate_server(device, 0); /* device may report different configuration data after LCU setup */ rc = dasd_eckd_read_conf(device); @@ -1216,6 +1734,15 @@ dasd_eckd_check_characteristics(struct dasd_device *device) "Read device characteristic failed, rc=%d", rc); goto out_err3; } + + if ((device->features & DASD_FEATURE_USERAW) && + !(private->rdc_data.facilities.RT_in_LR)) { + dev_err(&device->cdev->dev, "The storage server does not " + "support raw-track access\n"); + rc = -EINVAL; + goto out_err3; + } + /* find the valid cylinder size */ if (private->rdc_data.no_cyl == LV_COMPAT_CYL && private->rdc_data.long_no_cyl) @@ -1223,6 +1750,8 @@ dasd_eckd_check_characteristics(struct dasd_device *device) else private->real_cyl = private->rdc_data.no_cyl; + private->fcx_max_data = get_fcx_max_data(device); + readonly = dasd_device_is_ro(device); if (readonly) set_bit(DASD_FLAG_DEVICE_RO, &device->flags); @@ -1320,7 +1849,7 @@ dasd_eckd_analysis_ccw(struct dasd_device *device) cqr->startdev = device; cqr->memdev = device; cqr->retries = 255; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; return cqr; } @@ -1364,10 +1893,8 @@ static void dasd_eckd_analysis_callback(struct dasd_ccw_req *init_cqr, static int dasd_eckd_start_analysis(struct dasd_block *block) { - struct dasd_eckd_private *private; struct dasd_ccw_req *init_cqr; - private = (struct dasd_eckd_private *) block->base->private; init_cqr = dasd_eckd_analysis_ccw(block->base); if (IS_ERR(init_cqr)) return PTR_ERR(init_cqr); @@ -1404,6 +1931,13 @@ static int dasd_eckd_end_analysis(struct dasd_block *block) dasd_sfree_request(init_cqr, device); } + if (device->features & DASD_FEATURE_USERAW) { + block->bp_block = DASD_RAW_BLOCKSIZE; + blk_per_trk = DASD_RAW_BLOCK_PER_TRACK; + block->s2b_shift = 3; + goto raw; + } + if (status == INIT_CQR_UNFORMATTED) { dev_warn(&device->cdev->dev, "The DASD is not formatted\n"); return -EMEDIUMTYPE; @@ -1419,7 +1953,10 @@ static int dasd_eckd_end_analysis(struct dasd_block *block) count_area = NULL; for (i = 0; i < 3; i++) { if (private->count_area[i].kl != 4 || - private->count_area[i].dl != dasd_eckd_cdl_reclen(i) - 4) { + private->count_area[i].dl != dasd_eckd_cdl_reclen(i) - 4 || + private->count_area[i].cyl != 0 || + private->count_area[i].head != count_area_head[i] || + private->count_area[i].record != count_area_rec[i]) { private->uses_cdl = 0; break; } @@ -1431,7 +1968,10 @@ static int dasd_eckd_end_analysis(struct dasd_block *block) for (i = 0; i < 5; i++) { if ((private->count_area[i].kl != 0) || (private->count_area[i].dl != - private->count_area[0].dl)) + private->count_area[0].dl) || + private->count_area[i].cyl != 0 || + private->count_area[i].head != count_area_head[i] || + private->count_area[i].record != count_area_rec[i]) break; } if (i == 5) @@ -1441,6 +1981,7 @@ static int dasd_eckd_end_analysis(struct dasd_block *block) dev_warn(&device->cdev->dev, "Track 0 has no records following the VTOC\n"); } + if (count_area != NULL && count_area->kl == 0) { /* we found notthing violating our disk layout */ if (dasd_check_blocksize(count_area->dl) == 0) @@ -1456,6 +1997,8 @@ static int dasd_eckd_end_analysis(struct dasd_block *block) block->s2b_shift++; blk_per_trk = recs_per_track(&private->rdc_data, 0, block->bp_block); + +raw: block->blocks = (private->real_cyl * private->rdc_data.trk_per_cyl * blk_per_trk); @@ -1484,7 +2027,7 @@ static int dasd_eckd_do_analysis(struct dasd_block *block) return dasd_eckd_end_analysis(block); } -static int dasd_eckd_ready_to_online(struct dasd_device *device) +static int dasd_eckd_basic_to_ready(struct dasd_device *device) { return dasd_alias_add_device(device); }; @@ -1492,6 +2035,12 @@ static int dasd_eckd_ready_to_online(struct dasd_device *device) static int dasd_eckd_online_to_ready(struct dasd_device *device) { cancel_work_sync(&device->reload_device); + cancel_work_sync(&device->kick_validate); + return 0; +}; + +static int dasd_eckd_ready_to_basic(struct dasd_device *device) +{ return dasd_alias_remove_device(device); }; @@ -1511,45 +2060,35 @@ dasd_eckd_fill_geometry(struct dasd_block *block, struct hd_geometry *geo) } static struct dasd_ccw_req * -dasd_eckd_format_device(struct dasd_device * device, - struct format_data_t * fdata) +dasd_eckd_build_format(struct dasd_device *base, + struct format_data_t *fdata) { - struct dasd_eckd_private *private; + struct dasd_eckd_private *base_priv; + struct dasd_eckd_private *start_priv; + struct dasd_device *startdev; struct dasd_ccw_req *fcp; struct eckd_count *ect; + struct ch_t address; struct ccw1 *ccw; void *data; int rpt; - struct ch_t address; int cplength, datasize; - int i; + int i, j; int intensity = 0; int r0_perm; + int nr_tracks; + int use_prefix; - private = (struct dasd_eckd_private *) device->private; - rpt = recs_per_track(&private->rdc_data, 0, fdata->blksize); - set_ch_t(&address, - fdata->start_unit / private->rdc_data.trk_per_cyl, - fdata->start_unit % private->rdc_data.trk_per_cyl); + startdev = dasd_alias_get_start_dev(base); + if (!startdev) + startdev = base; - /* Sanity checks. */ - if (fdata->start_unit >= - (private->real_cyl * private->rdc_data.trk_per_cyl)) { - dev_warn(&device->cdev->dev, "Start track number %d used in " - "formatting is too big\n", fdata->start_unit); - return ERR_PTR(-EINVAL); - } - if (fdata->start_unit > fdata->stop_unit) { - dev_warn(&device->cdev->dev, "Start track %d used in " - "formatting exceeds end track\n", fdata->start_unit); - return ERR_PTR(-EINVAL); - } - if (dasd_check_blocksize(fdata->blksize) != 0) { - dev_warn(&device->cdev->dev, - "The DASD cannot be formatted with block size %d\n", - fdata->blksize); - return ERR_PTR(-EINVAL); - } + start_priv = (struct dasd_eckd_private *) startdev->private; + base_priv = (struct dasd_eckd_private *) base->private; + + rpt = recs_per_track(&base_priv->rdc_data, 0, fdata->blksize); + + nr_tracks = fdata->stop_unit - fdata->start_unit + 1; /* * fdata->intensity is a bit string that tells us what to do: @@ -1567,155 +2106,336 @@ dasd_eckd_format_device(struct dasd_device * device, r0_perm = 1; intensity = fdata->intensity; } + + use_prefix = base_priv->features.feature[8] & 0x01; + switch (intensity) { case 0x00: /* Normal format */ case 0x08: /* Normal format, use cdl. */ - cplength = 2 + rpt; - datasize = sizeof(struct DE_eckd_data) + - sizeof(struct LO_eckd_data) + - rpt * sizeof(struct eckd_count); + cplength = 2 + (rpt*nr_tracks); + if (use_prefix) + datasize = sizeof(struct PFX_eckd_data) + + sizeof(struct LO_eckd_data) + + rpt * nr_tracks * sizeof(struct eckd_count); + else + datasize = sizeof(struct DE_eckd_data) + + sizeof(struct LO_eckd_data) + + rpt * nr_tracks * sizeof(struct eckd_count); break; case 0x01: /* Write record zero and format track. */ case 0x09: /* Write record zero and format track, use cdl. */ - cplength = 3 + rpt; - datasize = sizeof(struct DE_eckd_data) + - sizeof(struct LO_eckd_data) + - sizeof(struct eckd_count) + - rpt * sizeof(struct eckd_count); + cplength = 2 + rpt * nr_tracks; + if (use_prefix) + datasize = sizeof(struct PFX_eckd_data) + + sizeof(struct LO_eckd_data) + + sizeof(struct eckd_count) + + rpt * nr_tracks * sizeof(struct eckd_count); + else + datasize = sizeof(struct DE_eckd_data) + + sizeof(struct LO_eckd_data) + + sizeof(struct eckd_count) + + rpt * nr_tracks * sizeof(struct eckd_count); break; case 0x04: /* Invalidate track. */ case 0x0c: /* Invalidate track, use cdl. */ cplength = 3; - datasize = sizeof(struct DE_eckd_data) + - sizeof(struct LO_eckd_data) + - sizeof(struct eckd_count); + if (use_prefix) + datasize = sizeof(struct PFX_eckd_data) + + sizeof(struct LO_eckd_data) + + sizeof(struct eckd_count); + else + datasize = sizeof(struct DE_eckd_data) + + sizeof(struct LO_eckd_data) + + sizeof(struct eckd_count); break; default: - dev_warn(&device->cdev->dev, "An I/O control call used " - "incorrect flags 0x%x\n", fdata->intensity); + dev_warn(&startdev->cdev->dev, + "An I/O control call used incorrect flags 0x%x\n", + fdata->intensity); return ERR_PTR(-EINVAL); } /* Allocate the format ccw request. */ - fcp = dasd_smalloc_request(DASD_ECKD_MAGIC, cplength, datasize, device); + fcp = dasd_smalloc_request(DASD_ECKD_MAGIC, cplength, + datasize, startdev); if (IS_ERR(fcp)) return fcp; + start_priv->count++; data = fcp->data; ccw = fcp->cpaddr; switch (intensity & ~0x08) { case 0x00: /* Normal format. */ - define_extent(ccw++, (struct DE_eckd_data *) data, - fdata->start_unit, fdata->start_unit, - DASD_ECKD_CCW_WRITE_CKD, device); - /* grant subsystem permission to format R0 */ - if (r0_perm) - ((struct DE_eckd_data *)data)->ga_extended |= 0x04; - data += sizeof(struct DE_eckd_data); + if (use_prefix) { + prefix(ccw++, (struct PFX_eckd_data *) data, + fdata->start_unit, fdata->stop_unit, + DASD_ECKD_CCW_WRITE_CKD, base, startdev); + /* grant subsystem permission to format R0 */ + if (r0_perm) + ((struct PFX_eckd_data *)data) + ->define_extent.ga_extended |= 0x04; + data += sizeof(struct PFX_eckd_data); + } else { + define_extent(ccw++, (struct DE_eckd_data *) data, + fdata->start_unit, fdata->stop_unit, + DASD_ECKD_CCW_WRITE_CKD, startdev); + /* grant subsystem permission to format R0 */ + if (r0_perm) + ((struct DE_eckd_data *) data) + ->ga_extended |= 0x04; + data += sizeof(struct DE_eckd_data); + } ccw[-1].flags |= CCW_FLAG_CC; locate_record(ccw++, (struct LO_eckd_data *) data, - fdata->start_unit, 0, rpt, - DASD_ECKD_CCW_WRITE_CKD, device, + fdata->start_unit, 0, rpt*nr_tracks, + DASD_ECKD_CCW_WRITE_CKD, base, fdata->blksize); data += sizeof(struct LO_eckd_data); break; case 0x01: /* Write record zero + format track. */ - define_extent(ccw++, (struct DE_eckd_data *) data, - fdata->start_unit, fdata->start_unit, - DASD_ECKD_CCW_WRITE_RECORD_ZERO, - device); - data += sizeof(struct DE_eckd_data); + if (use_prefix) { + prefix(ccw++, (struct PFX_eckd_data *) data, + fdata->start_unit, fdata->stop_unit, + DASD_ECKD_CCW_WRITE_RECORD_ZERO, + base, startdev); + data += sizeof(struct PFX_eckd_data); + } else { + define_extent(ccw++, (struct DE_eckd_data *) data, + fdata->start_unit, fdata->stop_unit, + DASD_ECKD_CCW_WRITE_RECORD_ZERO, startdev); + data += sizeof(struct DE_eckd_data); + } ccw[-1].flags |= CCW_FLAG_CC; locate_record(ccw++, (struct LO_eckd_data *) data, - fdata->start_unit, 0, rpt + 1, - DASD_ECKD_CCW_WRITE_RECORD_ZERO, device, - device->block->bp_block); + fdata->start_unit, 0, rpt * nr_tracks + 1, + DASD_ECKD_CCW_WRITE_RECORD_ZERO, base, + base->block->bp_block); data += sizeof(struct LO_eckd_data); break; case 0x04: /* Invalidate track. */ - define_extent(ccw++, (struct DE_eckd_data *) data, - fdata->start_unit, fdata->start_unit, - DASD_ECKD_CCW_WRITE_CKD, device); - data += sizeof(struct DE_eckd_data); + if (use_prefix) { + prefix(ccw++, (struct PFX_eckd_data *) data, + fdata->start_unit, fdata->stop_unit, + DASD_ECKD_CCW_WRITE_CKD, base, startdev); + data += sizeof(struct PFX_eckd_data); + } else { + define_extent(ccw++, (struct DE_eckd_data *) data, + fdata->start_unit, fdata->stop_unit, + DASD_ECKD_CCW_WRITE_CKD, startdev); + data += sizeof(struct DE_eckd_data); + } ccw[-1].flags |= CCW_FLAG_CC; locate_record(ccw++, (struct LO_eckd_data *) data, fdata->start_unit, 0, 1, - DASD_ECKD_CCW_WRITE_CKD, device, 8); + DASD_ECKD_CCW_WRITE_CKD, base, 8); data += sizeof(struct LO_eckd_data); break; } - if (intensity & 0x01) { /* write record zero */ - ect = (struct eckd_count *) data; - data += sizeof(struct eckd_count); - ect->cyl = address.cyl; - ect->head = address.head; - ect->record = 0; - ect->kl = 0; - ect->dl = 8; - ccw[-1].flags |= CCW_FLAG_CC; - ccw->cmd_code = DASD_ECKD_CCW_WRITE_RECORD_ZERO; - ccw->flags = CCW_FLAG_SLI; - ccw->count = 8; - ccw->cda = (__u32)(addr_t) ect; - ccw++; - } - if ((intensity & ~0x08) & 0x04) { /* erase track */ - ect = (struct eckd_count *) data; - data += sizeof(struct eckd_count); - ect->cyl = address.cyl; - ect->head = address.head; - ect->record = 1; - ect->kl = 0; - ect->dl = 0; - ccw[-1].flags |= CCW_FLAG_CC; - ccw->cmd_code = DASD_ECKD_CCW_WRITE_CKD; - ccw->flags = CCW_FLAG_SLI; - ccw->count = 8; - ccw->cda = (__u32)(addr_t) ect; - } else { /* write remaining records */ - for (i = 0; i < rpt; i++) { + + for (j = 0; j < nr_tracks; j++) { + /* calculate cylinder and head for the current track */ + set_ch_t(&address, + (fdata->start_unit + j) / + base_priv->rdc_data.trk_per_cyl, + (fdata->start_unit + j) % + base_priv->rdc_data.trk_per_cyl); + if (intensity & 0x01) { /* write record zero */ ect = (struct eckd_count *) data; data += sizeof(struct eckd_count); ect->cyl = address.cyl; ect->head = address.head; - ect->record = i + 1; + ect->record = 0; ect->kl = 0; - ect->dl = fdata->blksize; - /* Check for special tracks 0-1 when formatting CDL */ - if ((intensity & 0x08) && - fdata->start_unit == 0) { - if (i < 3) { - ect->kl = 4; - ect->dl = sizes_trk0[i] - 4; - } - } - if ((intensity & 0x08) && - fdata->start_unit == 1) { - ect->kl = 44; - ect->dl = LABEL_SIZE - 44; - } + ect->dl = 8; ccw[-1].flags |= CCW_FLAG_CC; - ccw->cmd_code = DASD_ECKD_CCW_WRITE_CKD; + ccw->cmd_code = DASD_ECKD_CCW_WRITE_RECORD_ZERO; ccw->flags = CCW_FLAG_SLI; ccw->count = 8; ccw->cda = (__u32)(addr_t) ect; ccw++; } + if ((intensity & ~0x08) & 0x04) { /* erase track */ + ect = (struct eckd_count *) data; + data += sizeof(struct eckd_count); + ect->cyl = address.cyl; + ect->head = address.head; + ect->record = 1; + ect->kl = 0; + ect->dl = 0; + ccw[-1].flags |= CCW_FLAG_CC; + ccw->cmd_code = DASD_ECKD_CCW_WRITE_CKD; + ccw->flags = CCW_FLAG_SLI; + ccw->count = 8; + ccw->cda = (__u32)(addr_t) ect; + } else { /* write remaining records */ + for (i = 0; i < rpt; i++) { + ect = (struct eckd_count *) data; + data += sizeof(struct eckd_count); + ect->cyl = address.cyl; + ect->head = address.head; + ect->record = i + 1; + ect->kl = 0; + ect->dl = fdata->blksize; + /* + * Check for special tracks 0-1 + * when formatting CDL + */ + if ((intensity & 0x08) && + fdata->start_unit == 0) { + if (i < 3) { + ect->kl = 4; + ect->dl = sizes_trk0[i] - 4; + } + } + if ((intensity & 0x08) && + fdata->start_unit == 1) { + ect->kl = 44; + ect->dl = LABEL_SIZE - 44; + } + ccw[-1].flags |= CCW_FLAG_CC; + if (i != 0 || j == 0) + ccw->cmd_code = + DASD_ECKD_CCW_WRITE_CKD; + else + ccw->cmd_code = + DASD_ECKD_CCW_WRITE_CKD_MT; + ccw->flags = CCW_FLAG_SLI; + ccw->count = 8; + ccw->cda = (__u32)(addr_t) ect; + ccw++; + } + } } - fcp->startdev = device; - fcp->memdev = device; + + fcp->startdev = startdev; + fcp->memdev = startdev; fcp->retries = 256; - fcp->buildclk = get_clock(); + fcp->expires = startdev->default_expires * HZ; + fcp->buildclk = get_tod_clock(); fcp->status = DASD_CQR_FILLED; + return fcp; } +static int +dasd_eckd_format_device(struct dasd_device *base, + struct format_data_t *fdata) +{ + struct dasd_ccw_req *cqr, *n; + struct dasd_block *block; + struct dasd_eckd_private *private; + struct list_head format_queue; + struct dasd_device *device; + int old_stop, format_step; + int step, rc = 0; + + block = base->block; + private = (struct dasd_eckd_private *) base->private; + + /* Sanity checks. */ + if (fdata->start_unit >= + (private->real_cyl * private->rdc_data.trk_per_cyl)) { + dev_warn(&base->cdev->dev, + "Start track number %u used in formatting is too big\n", + fdata->start_unit); + return -EINVAL; + } + if (fdata->stop_unit >= + (private->real_cyl * private->rdc_data.trk_per_cyl)) { + dev_warn(&base->cdev->dev, + "Stop track number %u used in formatting is too big\n", + fdata->stop_unit); + return -EINVAL; + } + if (fdata->start_unit > fdata->stop_unit) { + dev_warn(&base->cdev->dev, + "Start track %u used in formatting exceeds end track\n", + fdata->start_unit); + return -EINVAL; + } + if (dasd_check_blocksize(fdata->blksize) != 0) { + dev_warn(&base->cdev->dev, + "The DASD cannot be formatted with block size %u\n", + fdata->blksize); + return -EINVAL; + } + + INIT_LIST_HEAD(&format_queue); + old_stop = fdata->stop_unit; + + while (fdata->start_unit <= 1) { + fdata->stop_unit = fdata->start_unit; + cqr = dasd_eckd_build_format(base, fdata); + list_add(&cqr->blocklist, &format_queue); + + fdata->stop_unit = old_stop; + fdata->start_unit++; + + if (fdata->start_unit > fdata->stop_unit) + goto sleep; + } + +retry: + format_step = 255 / recs_per_track(&private->rdc_data, 0, + fdata->blksize); + while (fdata->start_unit <= old_stop) { + step = fdata->stop_unit - fdata->start_unit + 1; + if (step > format_step) + fdata->stop_unit = fdata->start_unit + format_step - 1; + + cqr = dasd_eckd_build_format(base, fdata); + if (IS_ERR(cqr)) { + if (PTR_ERR(cqr) == -ENOMEM) { + /* + * not enough memory available + * go to out and start requests + * retry after first requests were finished + */ + fdata->stop_unit = old_stop; + goto sleep; + } else + return PTR_ERR(cqr); + } + list_add(&cqr->blocklist, &format_queue); + + fdata->start_unit = fdata->stop_unit + 1; + fdata->stop_unit = old_stop; + } + +sleep: + dasd_sleep_on_queue(&format_queue); + + list_for_each_entry_safe(cqr, n, &format_queue, blocklist) { + device = cqr->startdev; + private = (struct dasd_eckd_private *) device->private; + if (cqr->status == DASD_CQR_FAILED) + rc = -EIO; + list_del_init(&cqr->blocklist); + dasd_sfree_request(cqr, device); + private->count--; + } + + /* + * in case of ENOMEM we need to retry after + * first requests are finished + */ + if (fdata->start_unit <= fdata->stop_unit) + goto retry; + + return rc; +} + static void dasd_eckd_handle_terminated_request(struct dasd_ccw_req *cqr) { + if (cqr->retries < 0) { + cqr->status = DASD_CQR_FAILED; + return; + } cqr->status = DASD_CQR_FILLED; if (cqr->block && (cqr->startdev != cqr->block->base)) { dasd_eckd_reset_ccw_to_base_io(cqr); cqr->startdev = cqr->block->base; + cqr->lpm = cqr->block->base->path_data.opm; } }; @@ -1744,9 +2464,9 @@ dasd_eckd_erp_postaction(struct dasd_ccw_req * cqr) return dasd_default_erp_postaction; } - -static void dasd_eckd_handle_unsolicited_interrupt(struct dasd_device *device, - struct irb *irb) +static void dasd_eckd_check_for_device_change(struct dasd_device *device, + struct dasd_ccw_req *cqr, + struct irb *irb) { char mask; char *sense = NULL; @@ -1756,9 +2476,14 @@ static void dasd_eckd_handle_unsolicited_interrupt(struct dasd_device *device, /* first of all check for state change pending interrupt */ mask = DEV_STAT_ATTENTION | DEV_STAT_DEV_END | DEV_STAT_UNIT_EXCEP; if ((scsw_dstat(&irb->scsw) & mask) == mask) { - /* for alias only and not in offline processing*/ + /* + * for alias only, not in offline processing + * and only if not suspended + */ if (!device->block && private->lcu && - !test_bit(DASD_FLAG_OFFLINE, &device->flags)) { + device->state == DASD_STATE_ONLINE && + !test_bit(DASD_FLAG_OFFLINE, &device->flags) && + !test_bit(DASD_FLAG_SUSPENDED, &device->flags)) { /* * the state change could be caused by an alias * reassignment remove device from alias handling @@ -1770,40 +2495,42 @@ static void dasd_eckd_handle_unsolicited_interrupt(struct dasd_device *device, /* schedule worker to reload device */ dasd_reload_device(device); } - dasd_generic_handle_state_change(device); return; } - /* summary unit check */ sense = dasd_get_sense(irb); - if (sense && (sense[7] == 0x0D) && + if (!sense) + return; + + /* summary unit check */ + if ((sense[27] & DASD_SENSE_BIT_0) && (sense[7] == 0x0D) && (scsw_dstat(&irb->scsw) & DEV_STAT_UNIT_CHECK)) { dasd_alias_handle_summary_unit_check(device, irb); return; } /* service information message SIM */ - if (sense && !(sense[27] & DASD_SENSE_BIT_0) && + if (!cqr && !(sense[27] & DASD_SENSE_BIT_0) && ((sense[6] & DASD_SIM_SENSE) == DASD_SIM_SENSE)) { dasd_3990_erp_handle_sim(device, sense); - dasd_schedule_device_bh(device); return; } - if ((scsw_cc(&irb->scsw) == 1) && !sense && - (scsw_fctl(&irb->scsw) == SCSW_FCTL_START_FUNC) && - (scsw_actl(&irb->scsw) == SCSW_ACTL_START_PEND) && - (scsw_stctl(&irb->scsw) == SCSW_STCTL_STATUS_PEND)) { - /* fake irb do nothing, they are handled elsewhere */ - dasd_schedule_device_bh(device); - return; + /* loss of device reservation is handled via base devices only + * as alias devices may be used with several bases + */ + if (device->block && (sense[27] & DASD_SENSE_BIT_0) && + (sense[7] == 0x3F) && + (scsw_dstat(&irb->scsw) & DEV_STAT_UNIT_CHECK) && + test_bit(DASD_FLAG_IS_RESERVED, &device->flags)) { + if (device->features & DASD_FEATURE_FAILONSLCK) + set_bit(DASD_FLAG_LOCK_STOLEN, &device->flags); + clear_bit(DASD_FLAG_IS_RESERVED, &device->flags); + dev_err(&device->cdev->dev, + "The device reservation was lost\n"); } - - dasd_schedule_device_bh(device); - return; -}; - +} static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_single( struct dasd_device *startdev, @@ -1824,7 +2551,7 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_single( struct dasd_ccw_req *cqr; struct ccw1 *ccw; struct req_iterator iter; - struct bio_vec *bv; + struct bio_vec bv; char *dst; unsigned int off; int count, cidaw, cplength, datasize; @@ -1846,13 +2573,13 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_single( count = 0; cidaw = 0; rq_for_each_segment(bv, req, iter) { - if (bv->bv_len & (blksize - 1)) + if (bv.bv_len & (blksize - 1)) /* Eckd can only do full blocks. */ return ERR_PTR(-EINVAL); - count += bv->bv_len >> (block->s2b_shift + 9); + count += bv.bv_len >> (block->s2b_shift + 9); #if defined(CONFIG_64BIT) - if (idal_is_needed (page_address(bv->bv_page), bv->bv_len)) - cidaw += bv->bv_len >> (block->s2b_shift + 9); + if (idal_is_needed (page_address(bv.bv_page), bv.bv_len)) + cidaw += bv.bv_len >> (block->s2b_shift + 9); #endif } /* Paranoia. */ @@ -1903,7 +2630,7 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_single( sizeof(struct PFX_eckd_data)); } else { if (define_extent(ccw++, cqr->data, first_trk, - last_trk, cmd, startdev) == -EAGAIN) { + last_trk, cmd, basedev) == -EAGAIN) { /* Clock not in sync and XRC is enabled. * Try again later. */ @@ -1923,16 +2650,16 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_single( last_rec - recid + 1, cmd, basedev, blksize); } rq_for_each_segment(bv, req, iter) { - dst = page_address(bv->bv_page) + bv->bv_offset; + dst = page_address(bv.bv_page) + bv.bv_offset; if (dasd_page_cache) { char *copy = kmem_cache_alloc(dasd_page_cache, GFP_DMA | __GFP_NOWARN); if (copy && rq_data_dir(req) == WRITE) - memcpy(copy + bv->bv_offset, dst, bv->bv_len); + memcpy(copy + bv.bv_offset, dst, bv.bv_len); if (copy) - dst = copy + bv->bv_offset; + dst = copy + bv.bv_offset; } - for (off = 0; off < bv->bv_len; off += blksize) { + for (off = 0; off < bv.bv_len; off += blksize) { sector_t trkid = recid; unsigned int recoffs = sector_div(trkid, blk_per_trk); rcmd = cmd; @@ -1984,9 +2711,9 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_single( cqr->memdev = startdev; cqr->block = block; cqr->expires = startdev->default_expires * HZ; /* default 5 minutes */ - cqr->lpm = private->path_data.ppm; - cqr->retries = 256; - cqr->buildclk = get_clock(); + cqr->lpm = startdev->path_data.ppm; + cqr->retries = startdev->default_retries; + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; return cqr; } @@ -2004,12 +2731,11 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_track( unsigned int blk_per_trk, unsigned int blksize) { - struct dasd_eckd_private *private; unsigned long *idaws; struct dasd_ccw_req *cqr; struct ccw1 *ccw; struct req_iterator iter; - struct bio_vec *bv; + struct bio_vec bv; char *dst, *idaw_dst; unsigned int cidaw, cplength, datasize; unsigned int tlf; @@ -2023,7 +2749,6 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_track( unsigned int recoffs; basedev = block->base; - private = (struct dasd_eckd_private *) basedev->private; if (rq_data_dir(req) == READ) cmd = DASD_ECKD_CCW_READ_TRACK_DATA; else if (rq_data_dir(req) == WRITE) @@ -2085,11 +2810,11 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_track( new_track = 1; end_idaw = 0; len_to_track_end = 0; - idaw_dst = 0; + idaw_dst = NULL; idaw_len = 0; rq_for_each_segment(bv, req, iter) { - dst = page_address(bv->bv_page) + bv->bv_offset; - seg_len = bv->bv_len; + dst = page_address(bv.bv_page) + bv.bv_offset; + seg_len = bv.bv_len; while (seg_len) { if (new_track) { trkid = recid; @@ -2147,7 +2872,7 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_track( if (end_idaw) { idaws = idal_create_words(idaws, idaw_dst, idaw_len); - idaw_dst = 0; + idaw_dst = NULL; idaw_len = 0; end_idaw = 0; } @@ -2161,9 +2886,9 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_track( cqr->memdev = startdev; cqr->block = block; cqr->expires = startdev->default_expires * HZ; /* default 5 minutes */ - cqr->lpm = private->path_data.ppm; - cqr->retries = 256; - cqr->buildclk = get_clock(); + cqr->lpm = startdev->path_data.ppm; + cqr->retries = startdev->default_retries; + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; return cqr; } @@ -2296,8 +3021,7 @@ static int prepare_itcw(struct itcw *itcw, dcw = itcw_add_dcw(itcw, pfx_cmd, 0, &pfxdata, sizeof(pfxdata), total_data_size); - - return rc; + return PTR_RET(dcw); } static struct dasd_ccw_req *dasd_eckd_build_cp_tpm_track( @@ -2313,10 +3037,9 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_tpm_track( unsigned int blk_per_trk, unsigned int blksize) { - struct dasd_eckd_private *private; struct dasd_ccw_req *cqr; struct req_iterator iter; - struct bio_vec *bv; + struct bio_vec bv; char *dst; unsigned int trkcount, ctidaw; unsigned char cmd; @@ -2326,9 +3049,15 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_tpm_track( struct tidaw *last_tidaw = NULL; int itcw_op; size_t itcw_size; + u8 tidaw_flags; + unsigned int seg_len, part_len, len_to_track_end; + unsigned char new_track; + sector_t recid, trkid; + unsigned int offs; + unsigned int count, count_to_trk_end; + int ret; basedev = block->base; - private = (struct dasd_eckd_private *) basedev->private; if (rq_data_dir(req) == READ) { cmd = DASD_ECKD_CCW_READ_TRACK_DATA; itcw_op = ITCW_OP_READ; @@ -2341,12 +3070,16 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_tpm_track( /* trackbased I/O needs address all memory via TIDAWs, * not just for 64 bit addresses. This allows us to map * each segment directly to one tidaw. + * In the case of write requests, additional tidaws may + * be needed when a segment crosses a track boundary. */ trkcount = last_trk - first_trk + 1; ctidaw = 0; rq_for_each_segment(bv, req, iter) { ++ctidaw; } + if (rq_data_dir(req) == WRITE) + ctidaw += (last_trk - first_trk); /* Allocate the ccw request. */ itcw_size = itcw_calc_size(0, ctidaw, 0); @@ -2354,15 +3087,6 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_tpm_track( if (IS_ERR(cqr)) return cqr; - cqr->cpmode = 1; - cqr->startdev = startdev; - cqr->memdev = startdev; - cqr->block = block; - cqr->expires = 100*HZ; - cqr->buildclk = get_clock(); - cqr->status = DASD_CQR_FILLED; - cqr->retries = 10; - /* transfer length factor: how many bytes to read from the last track */ if (first_trk == last_trk) tlf = last_offs - first_offs + 1; @@ -2371,8 +3095,11 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_tpm_track( tlf *= blksize; itcw = itcw_init(cqr->data, itcw_size, itcw_op, 0, ctidaw, 0); + if (IS_ERR(itcw)) { + ret = -EINVAL; + goto out_error; + } cqr->cpaddr = itcw_get_tcw(itcw); - if (prepare_itcw(itcw, first_trk, last_trk, cmd, basedev, startdev, first_offs + 1, @@ -2382,49 +3109,93 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_tpm_track( /* Clock not in sync and XRC is enabled. * Try again later. */ - dasd_sfree_request(cqr, startdev); - return ERR_PTR(-EAGAIN); + ret = -EAGAIN; + goto out_error; } - + len_to_track_end = 0; /* * A tidaw can address 4k of memory, but must not cross page boundaries * We can let the block layer handle this by setting * blk_queue_segment_boundary to page boundaries and * blk_max_segment_size to page size when setting up the request queue. + * For write requests, a TIDAW must not cross track boundaries, because + * we have to set the CBC flag on the last tidaw for each track. */ - rq_for_each_segment(bv, req, iter) { - dst = page_address(bv->bv_page) + bv->bv_offset; - last_tidaw = itcw_add_tidaw(itcw, 0x00, dst, bv->bv_len); - if (IS_ERR(last_tidaw)) - return (struct dasd_ccw_req *)last_tidaw; + if (rq_data_dir(req) == WRITE) { + new_track = 1; + recid = first_rec; + rq_for_each_segment(bv, req, iter) { + dst = page_address(bv.bv_page) + bv.bv_offset; + seg_len = bv.bv_len; + while (seg_len) { + if (new_track) { + trkid = recid; + offs = sector_div(trkid, blk_per_trk); + count_to_trk_end = blk_per_trk - offs; + count = min((last_rec - recid + 1), + (sector_t)count_to_trk_end); + len_to_track_end = count * blksize; + recid += count; + new_track = 0; + } + part_len = min(seg_len, len_to_track_end); + seg_len -= part_len; + len_to_track_end -= part_len; + /* We need to end the tidaw at track end */ + if (!len_to_track_end) { + new_track = 1; + tidaw_flags = TIDAW_FLAGS_INSERT_CBC; + } else + tidaw_flags = 0; + last_tidaw = itcw_add_tidaw(itcw, tidaw_flags, + dst, part_len); + if (IS_ERR(last_tidaw)) { + ret = -EINVAL; + goto out_error; + } + dst += part_len; + } + } + } else { + rq_for_each_segment(bv, req, iter) { + dst = page_address(bv.bv_page) + bv.bv_offset; + last_tidaw = itcw_add_tidaw(itcw, 0x00, + dst, bv.bv_len); + if (IS_ERR(last_tidaw)) { + ret = -EINVAL; + goto out_error; + } + } } - - last_tidaw->flags |= 0x80; + last_tidaw->flags |= TIDAW_FLAGS_LAST; + last_tidaw->flags &= ~TIDAW_FLAGS_INSERT_CBC; itcw_finalize(itcw); if (blk_noretry_request(req) || block->base->features & DASD_FEATURE_FAILFAST) set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags); + cqr->cpmode = 1; cqr->startdev = startdev; cqr->memdev = startdev; cqr->block = block; cqr->expires = startdev->default_expires * HZ; /* default 5 minutes */ - cqr->lpm = private->path_data.ppm; - cqr->retries = 256; - cqr->buildclk = get_clock(); + cqr->lpm = startdev->path_data.ppm; + cqr->retries = startdev->default_retries; + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; return cqr; +out_error: + dasd_sfree_request(cqr, startdev); + return ERR_PTR(ret); } static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev, struct dasd_block *block, struct request *req) { - int tpm, cmdrtd, cmdwtd; + int cmdrtd, cmdwtd; int use_prefix; -#if defined(CONFIG_64BIT) - int fcx_in_css, fcx_in_gneq, fcx_in_features; -#endif + int fcx_multitrack; struct dasd_eckd_private *private; struct dasd_device *basedev; sector_t first_rec, last_rec; @@ -2432,6 +3203,7 @@ static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev, unsigned int first_offs, last_offs; unsigned int blk_per_trk, blksize; int cdlspecial; + unsigned int data_size; struct dasd_ccw_req *cqr; basedev = block->base; @@ -2450,15 +3222,13 @@ static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev, last_offs = sector_div(last_trk, blk_per_trk); cdlspecial = (private->uses_cdl && first_rec < 2*blk_per_trk); - /* is transport mode supported? */ -#if defined(CONFIG_64BIT) - fcx_in_css = css_general_characteristics.fcx; - fcx_in_gneq = private->gneq->reserved2[7] & 0x04; - fcx_in_features = private->features.feature[40] & 0x80; - tpm = fcx_in_css && fcx_in_gneq && fcx_in_features; -#else - tpm = 0; -#endif + fcx_multitrack = private->features.feature[40] & 0x20; + data_size = blk_rq_bytes(req); + if (data_size % blksize) + return ERR_PTR(-EINVAL); + /* tpm write request add CBC data on each track boundary */ + if (rq_data_dir(req) == WRITE) + data_size += (last_trk - first_trk) * 4; /* is read track data and write track data in command mode supported? */ cmdrtd = private->features.feature[9] & 0x20; @@ -2468,13 +3238,15 @@ static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev, cqr = NULL; if (cdlspecial || dasd_page_cache) { /* do nothing, just fall through to the cmd mode single case */ - } else if (!dasd_nofcx && tpm && (first_trk == last_trk)) { + } else if ((data_size <= private->fcx_max_data) + && (fcx_multitrack || (first_trk == last_trk))) { cqr = dasd_eckd_build_cp_tpm_track(startdev, block, req, first_rec, last_rec, first_trk, last_trk, first_offs, last_offs, blk_per_trk, blksize); - if (IS_ERR(cqr) && PTR_ERR(cqr) != -EAGAIN) + if (IS_ERR(cqr) && (PTR_ERR(cqr) != -EAGAIN) && + (PTR_ERR(cqr) != -ENOMEM)) cqr = NULL; } else if (use_prefix && (((rq_data_dir(req) == READ) && cmdrtd) || @@ -2484,7 +3256,8 @@ static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev, first_trk, last_trk, first_offs, last_offs, blk_per_trk, blksize); - if (IS_ERR(cqr) && PTR_ERR(cqr) != -EAGAIN) + if (IS_ERR(cqr) && (PTR_ERR(cqr) != -EAGAIN) && + (PTR_ERR(cqr) != -ENOMEM)) cqr = NULL; } if (!cqr) @@ -2496,13 +3269,162 @@ static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev, return cqr; } +static struct dasd_ccw_req *dasd_raw_build_cp(struct dasd_device *startdev, + struct dasd_block *block, + struct request *req) +{ + unsigned long *idaws; + struct dasd_device *basedev; + struct dasd_ccw_req *cqr; + struct ccw1 *ccw; + struct req_iterator iter; + struct bio_vec bv; + char *dst; + unsigned char cmd; + unsigned int trkcount; + unsigned int seg_len, len_to_track_end; + unsigned int first_offs; + unsigned int cidaw, cplength, datasize; + sector_t first_trk, last_trk, sectors; + sector_t start_padding_sectors, end_sector_offset, end_padding_sectors; + unsigned int pfx_datasize; + + /* + * raw track access needs to be mutiple of 64k and on 64k boundary + * For read requests we can fix an incorrect alignment by padding + * the request with dummy pages. + */ + start_padding_sectors = blk_rq_pos(req) % DASD_RAW_SECTORS_PER_TRACK; + end_sector_offset = (blk_rq_pos(req) + blk_rq_sectors(req)) % + DASD_RAW_SECTORS_PER_TRACK; + end_padding_sectors = (DASD_RAW_SECTORS_PER_TRACK - end_sector_offset) % + DASD_RAW_SECTORS_PER_TRACK; + basedev = block->base; + if ((start_padding_sectors || end_padding_sectors) && + (rq_data_dir(req) == WRITE)) { + DBF_DEV_EVENT(DBF_ERR, basedev, + "raw write not track aligned (%lu,%lu) req %p", + start_padding_sectors, end_padding_sectors, req); + cqr = ERR_PTR(-EINVAL); + goto out; + } + + first_trk = blk_rq_pos(req) / DASD_RAW_SECTORS_PER_TRACK; + last_trk = (blk_rq_pos(req) + blk_rq_sectors(req) - 1) / + DASD_RAW_SECTORS_PER_TRACK; + trkcount = last_trk - first_trk + 1; + first_offs = 0; + + if (rq_data_dir(req) == READ) + cmd = DASD_ECKD_CCW_READ_TRACK; + else if (rq_data_dir(req) == WRITE) + cmd = DASD_ECKD_CCW_WRITE_FULL_TRACK; + else { + cqr = ERR_PTR(-EINVAL); + goto out; + } + + /* + * Raw track based I/O needs IDAWs for each page, + * and not just for 64 bit addresses. + */ + cidaw = trkcount * DASD_RAW_BLOCK_PER_TRACK; + + /* 1x prefix + one read/write ccw per track */ + cplength = 1 + trkcount; + + /* + * struct PFX_eckd_data has up to 2 byte as extended parameter + * this is needed for write full track and has to be mentioned + * separately + * add 8 instead of 2 to keep 8 byte boundary + */ + pfx_datasize = sizeof(struct PFX_eckd_data) + 8; + + datasize = pfx_datasize + cidaw * sizeof(unsigned long long); + + /* Allocate the ccw request. */ + cqr = dasd_smalloc_request(DASD_ECKD_MAGIC, cplength, + datasize, startdev); + if (IS_ERR(cqr)) + goto out; + ccw = cqr->cpaddr; + + if (prefix_LRE(ccw++, cqr->data, first_trk, last_trk, cmd, + basedev, startdev, 1 /* format */, first_offs + 1, + trkcount, 0, 0) == -EAGAIN) { + /* Clock not in sync and XRC is enabled. + * Try again later. + */ + dasd_sfree_request(cqr, startdev); + cqr = ERR_PTR(-EAGAIN); + goto out; + } + + idaws = (unsigned long *)(cqr->data + pfx_datasize); + len_to_track_end = 0; + if (start_padding_sectors) { + ccw[-1].flags |= CCW_FLAG_CC; + ccw->cmd_code = cmd; + /* maximum 3390 track size */ + ccw->count = 57326; + /* 64k map to one track */ + len_to_track_end = 65536 - start_padding_sectors * 512; + ccw->cda = (__u32)(addr_t)idaws; + ccw->flags |= CCW_FLAG_IDA; + ccw->flags |= CCW_FLAG_SLI; + ccw++; + for (sectors = 0; sectors < start_padding_sectors; sectors += 8) + idaws = idal_create_words(idaws, rawpadpage, PAGE_SIZE); + } + rq_for_each_segment(bv, req, iter) { + dst = page_address(bv.bv_page) + bv.bv_offset; + seg_len = bv.bv_len; + if (cmd == DASD_ECKD_CCW_READ_TRACK) + memset(dst, 0, seg_len); + if (!len_to_track_end) { + ccw[-1].flags |= CCW_FLAG_CC; + ccw->cmd_code = cmd; + /* maximum 3390 track size */ + ccw->count = 57326; + /* 64k map to one track */ + len_to_track_end = 65536; + ccw->cda = (__u32)(addr_t)idaws; + ccw->flags |= CCW_FLAG_IDA; + ccw->flags |= CCW_FLAG_SLI; + ccw++; + } + len_to_track_end -= seg_len; + idaws = idal_create_words(idaws, dst, seg_len); + } + for (sectors = 0; sectors < end_padding_sectors; sectors += 8) + idaws = idal_create_words(idaws, rawpadpage, PAGE_SIZE); + if (blk_noretry_request(req) || + block->base->features & DASD_FEATURE_FAILFAST) + set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags); + cqr->startdev = startdev; + cqr->memdev = startdev; + cqr->block = block; + cqr->expires = startdev->default_expires * HZ; + cqr->lpm = startdev->path_data.ppm; + cqr->retries = startdev->default_retries; + cqr->buildclk = get_tod_clock(); + cqr->status = DASD_CQR_FILLED; + + if (IS_ERR(cqr) && PTR_ERR(cqr) != -EAGAIN) + cqr = NULL; +out: + return cqr; +} + + static int dasd_eckd_free_cp(struct dasd_ccw_req *cqr, struct request *req) { struct dasd_eckd_private *private; struct ccw1 *ccw; struct req_iterator iter; - struct bio_vec *bv; + struct bio_vec bv; char *dst, *cda; unsigned int blksize, blk_per_trk, off; sector_t recid; @@ -2520,8 +3442,8 @@ dasd_eckd_free_cp(struct dasd_ccw_req *cqr, struct request *req) if (private->uses_cdl == 0 || recid > 2*blk_per_trk) ccw++; rq_for_each_segment(bv, req, iter) { - dst = page_address(bv->bv_page) + bv->bv_offset; - for (off = 0; off < bv->bv_len; off += blksize) { + dst = page_address(bv.bv_page) + bv.bv_offset; + for (off = 0; off < bv.bv_len; off += blksize) { /* Skip locate record. */ if (private->uses_cdl && recid <= 2*blk_per_trk) ccw++; @@ -2532,7 +3454,7 @@ dasd_eckd_free_cp(struct dasd_ccw_req *cqr, struct request *req) cda = (char *)((addr_t) ccw->cda); if (dst != cda) { if (rq_data_dir(req) == READ) - memcpy(dst, cda, bv->bv_len); + memcpy(dst, cda, bv.bv_len); kmem_cache_free(dasd_page_cache, (void *)((addr_t)cda & PAGE_MASK)); } @@ -2600,7 +3522,10 @@ static struct dasd_ccw_req *dasd_eckd_build_alias_cp(struct dasd_device *base, spin_lock_irqsave(get_ccwdev_lock(startdev->cdev), flags); private->count++; - cqr = dasd_eckd_build_cp(startdev, block, req); + if ((base->features & DASD_FEATURE_USERAW)) + cqr = dasd_raw_build_cp(startdev, block, req); + else + cqr = dasd_eckd_build_cp(startdev, block, req); if (IS_ERR(cqr)) private->count--; spin_unlock_irqrestore(get_ccwdev_lock(startdev->cdev), flags); @@ -2684,10 +3609,12 @@ dasd_eckd_release(struct dasd_device *device) set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags); cqr->retries = 2; /* set retry counter to enable basic ERP */ cqr->expires = 2 * HZ; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; rc = dasd_sleep_on_immediatly(cqr); + if (!rc) + clear_bit(DASD_FLAG_IS_RESERVED, &device->flags); if (useglobal) mutex_unlock(&dasd_reserve_mutex); @@ -2737,10 +3664,12 @@ dasd_eckd_reserve(struct dasd_device *device) set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags); cqr->retries = 2; /* set retry counter to enable basic ERP */ cqr->expires = 2 * HZ; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; rc = dasd_sleep_on_immediatly(cqr); + if (!rc) + set_bit(DASD_FLAG_IS_RESERVED, &device->flags); if (useglobal) mutex_unlock(&dasd_reserve_mutex); @@ -2789,10 +3718,12 @@ dasd_eckd_steal_lock(struct dasd_device *device) set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags); cqr->retries = 2; /* set retry counter to enable basic ERP */ cqr->expires = 2 * HZ; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; rc = dasd_sleep_on_immediatly(cqr); + if (!rc) + set_bit(DASD_FLAG_IS_RESERVED, &device->flags); if (useglobal) mutex_unlock(&dasd_reserve_mutex); @@ -2845,9 +3776,10 @@ static int dasd_eckd_snid(struct dasd_device *device, cqr->memdev = device; clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags); set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags); + set_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags); cqr->retries = 5; cqr->expires = 10 * HZ; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; cqr->lpm = usrparm.path_mask; @@ -2917,7 +3849,7 @@ dasd_eckd_performance(struct dasd_device *device, void __user *argp) ccw->count = sizeof(struct dasd_rssd_perf_stats_t); ccw->cda = (__u32)(addr_t) stats; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; rc = dasd_sleep_on(cqr); if (rc == 0) { @@ -3043,7 +3975,7 @@ static int dasd_symm_io(struct dasd_device *device, void __user *argp) cqr->memdev = device; cqr->retries = 3; cqr->expires = 10 * HZ; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; /* Build the ccws */ @@ -3108,7 +4040,7 @@ dasd_eckd_ioctl(struct dasd_block *block, unsigned int cmd, void __user *argp) case BIODASDSYMMIO: return dasd_symm_io(device, argp); default: - return -ENOIOCTLCMD; + return -ENOTTY; } } @@ -3124,7 +4056,7 @@ dasd_eckd_dump_ccw_range(struct ccw1 *from, struct ccw1 *to, char *page) len = 0; while (from <= to) { - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " CCW %p: %08X %08X DAT:", from, ((int *) from)[0], ((int *) from)[1]); @@ -3185,23 +4117,23 @@ static void dasd_eckd_dump_sense_ccw(struct dasd_device *device, return; } /* dump the sense data */ - len = sprintf(page, KERN_ERR PRINTK_HEADER + len = sprintf(page, PRINTK_HEADER " I/O status report for device %s:\n", dev_name(&device->cdev->dev)); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " in req: %p CC:%02X FC:%02X AC:%02X SC:%02X DS:%02X " "CS:%02X RC:%d\n", req, scsw_cc(&irb->scsw), scsw_fctl(&irb->scsw), scsw_actl(&irb->scsw), scsw_stctl(&irb->scsw), scsw_dstat(&irb->scsw), scsw_cstat(&irb->scsw), req ? req->intrc : 0); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " device %s: Failing CCW: %p\n", dev_name(&device->cdev->dev), (void *) (addr_t) irb->scsw.cmd.cpa); if (irb->esw.esw0.erw.cons) { for (sl = 0; sl < 4; sl++) { - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " Sense(hex) %2d-%2d:", (8 * sl), ((8 * sl) + 7)); @@ -3214,23 +4146,23 @@ static void dasd_eckd_dump_sense_ccw(struct dasd_device *device, if (irb->ecw[27] & DASD_SENSE_BIT_0) { /* 24 Byte Sense Data */ - sprintf(page + len, KERN_ERR PRINTK_HEADER + sprintf(page + len, PRINTK_HEADER " 24 Byte: %x MSG %x, " "%s MSGb to SYSOP\n", irb->ecw[7] >> 4, irb->ecw[7] & 0x0f, irb->ecw[1] & 0x10 ? "" : "no"); } else { /* 32 Byte Sense Data */ - sprintf(page + len, KERN_ERR PRINTK_HEADER + sprintf(page + len, PRINTK_HEADER " 32 Byte: Format: %x " "Exception class %x\n", irb->ecw[6] & 0x0f, irb->ecw[22] >> 4); } } else { - sprintf(page + len, KERN_ERR PRINTK_HEADER + sprintf(page + len, PRINTK_HEADER " SORRY - NO VALID SENSE AVAILABLE\n"); } - printk("%s", page); + printk(KERN_ERR "%s", page); if (req) { /* req == NULL for unsolicited interrupts */ @@ -3239,10 +4171,10 @@ static void dasd_eckd_dump_sense_ccw(struct dasd_device *device, first = req->cpaddr; for (last = first; last->flags & (CCW_FLAG_CC | CCW_FLAG_DC); last++); to = min(first + 6, last); - len = sprintf(page, KERN_ERR PRINTK_HEADER + len = sprintf(page, PRINTK_HEADER " Related CP in req: %p\n", req); dasd_eckd_dump_ccw_range(first, to, page + len); - printk("%s", page); + printk(KERN_ERR "%s", page); /* print failing CCW area (maximum 4) */ /* scsw->cda is either valid or zero */ @@ -3252,7 +4184,7 @@ static void dasd_eckd_dump_sense_ccw(struct dasd_device *device, irb->scsw.cmd.cpa; /* failing CCW */ if (from < fail - 2) { from = fail - 2; /* there is a gap - print header */ - len += sprintf(page, KERN_ERR PRINTK_HEADER "......\n"); + len += sprintf(page, PRINTK_HEADER "......\n"); } to = min(fail + 1, last); len += dasd_eckd_dump_ccw_range(from, to, page + len); @@ -3261,11 +4193,11 @@ static void dasd_eckd_dump_sense_ccw(struct dasd_device *device, from = max(from, ++to); if (from < last - 1) { from = last - 1; /* there is a gap - print header */ - len += sprintf(page + len, KERN_ERR PRINTK_HEADER "......\n"); + len += sprintf(page + len, PRINTK_HEADER "......\n"); } len += dasd_eckd_dump_ccw_range(from, last, page + len); if (len > 0) - printk("%s", page); + printk(KERN_ERR "%s", page); } free_page((unsigned long) page); } @@ -3279,10 +4211,8 @@ static void dasd_eckd_dump_sense_tcw(struct dasd_device *device, { char *page; int len, sl, sct, residual; - struct tsb *tsb; - u8 *sense; - + u8 *sense, *rcq; page = (char *) get_zeroed_page(GFP_ATOMIC); if (page == NULL) { @@ -3291,10 +4221,10 @@ static void dasd_eckd_dump_sense_tcw(struct dasd_device *device, return; } /* dump the sense data */ - len = sprintf(page, KERN_ERR PRINTK_HEADER + len = sprintf(page, PRINTK_HEADER " I/O status report for device %s:\n", dev_name(&device->cdev->dev)); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " in req: %p CC:%02X FC:%02X AC:%02X SC:%02X DS:%02X " "CS:%02X fcxs:%02X schxs:%02X RC:%d\n", req, scsw_cc(&irb->scsw), scsw_fctl(&irb->scsw), @@ -3302,7 +4232,7 @@ static void dasd_eckd_dump_sense_tcw(struct dasd_device *device, scsw_dstat(&irb->scsw), scsw_cstat(&irb->scsw), irb->scsw.tm.fcxs, irb->scsw.tm.schxs, req ? req->intrc : 0); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " device %s: Failing TCW: %p\n", dev_name(&device->cdev->dev), (void *) (addr_t) irb->scsw.tm.tcw); @@ -3314,61 +4244,62 @@ static void dasd_eckd_dump_sense_tcw(struct dasd_device *device, (struct tcw *)(unsigned long)irb->scsw.tm.tcw); if (tsb) { - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " tsb->length %d\n", tsb->length); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " tsb->flags %x\n", tsb->flags); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " tsb->dcw_offset %d\n", tsb->dcw_offset); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " tsb->count %d\n", tsb->count); residual = tsb->count - 28; - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " residual %d\n", residual); switch (tsb->flags & 0x07) { case 1: /* tsa_iostat */ - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " tsb->tsa.iostat.dev_time %d\n", tsb->tsa.iostat.dev_time); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " tsb->tsa.iostat.def_time %d\n", tsb->tsa.iostat.def_time); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " tsb->tsa.iostat.queue_time %d\n", tsb->tsa.iostat.queue_time); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " tsb->tsa.iostat.dev_busy_time %d\n", tsb->tsa.iostat.dev_busy_time); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " tsb->tsa.iostat.dev_act_time %d\n", tsb->tsa.iostat.dev_act_time); sense = tsb->tsa.iostat.sense; break; case 2: /* ts_ddpc */ - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " tsb->tsa.ddpc.rc %d\n", tsb->tsa.ddpc.rc); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER - " tsb->tsa.ddpc.rcq: "); - for (sl = 0; sl < 16; sl++) { + for (sl = 0; sl < 2; sl++) { + len += sprintf(page + len, PRINTK_HEADER + " tsb->tsa.ddpc.rcq %2d-%2d: ", + (8 * sl), ((8 * sl) + 7)); + rcq = tsb->tsa.ddpc.rcq; for (sct = 0; sct < 8; sct++) { len += sprintf(page + len, " %02x", - tsb->tsa.ddpc.rcq[sl]); + rcq[8 * sl + sct]); } len += sprintf(page + len, "\n"); } sense = tsb->tsa.ddpc.sense; break; case 3: /* tsa_intrg */ - len += sprintf(page + len, KERN_ERR PRINTK_HEADER - " tsb->tsa.intrg.: not supportet yet \n"); + len += sprintf(page + len, PRINTK_HEADER + " tsb->tsa.intrg.: not supportet yet\n"); break; } if (sense) { for (sl = 0; sl < 4; sl++) { - len += sprintf(page + len, - KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " Sense(hex) %2d-%2d:", (8 * sl), ((8 * sl) + 7)); for (sct = 0; sct < 8; sct++) { @@ -3380,27 +4311,27 @@ static void dasd_eckd_dump_sense_tcw(struct dasd_device *device, if (sense[27] & DASD_SENSE_BIT_0) { /* 24 Byte Sense Data */ - sprintf(page + len, KERN_ERR PRINTK_HEADER + sprintf(page + len, PRINTK_HEADER " 24 Byte: %x MSG %x, " "%s MSGb to SYSOP\n", sense[7] >> 4, sense[7] & 0x0f, sense[1] & 0x10 ? "" : "no"); } else { /* 32 Byte Sense Data */ - sprintf(page + len, KERN_ERR PRINTK_HEADER + sprintf(page + len, PRINTK_HEADER " 32 Byte: Format: %x " "Exception class %x\n", sense[6] & 0x0f, sense[22] >> 4); } } else { - sprintf(page + len, KERN_ERR PRINTK_HEADER + sprintf(page + len, PRINTK_HEADER " SORRY - NO VALID SENSE AVAILABLE\n"); } } else { - sprintf(page + len, KERN_ERR PRINTK_HEADER + sprintf(page + len, PRINTK_HEADER " SORRY - NO TSB DATA AVAILABLE\n"); } - printk("%s", page); + printk(KERN_ERR "%s", page); free_page((unsigned long) page); } @@ -3430,16 +4361,15 @@ static int dasd_eckd_restore_device(struct dasd_device *device) { struct dasd_eckd_private *private; struct dasd_eckd_characteristics temp_rdc_data; - int is_known, rc; + int rc; struct dasd_uid temp_uid; unsigned long flags; + unsigned long cqr_flags = 0; private = (struct dasd_eckd_private *) device->private; /* Read Configuration Data */ - rc = dasd_eckd_read_conf(device); - if (rc) - goto out_err; + dasd_eckd_read_conf(device); dasd_eckd_get_uid(device, &temp_uid); /* Generate device unique id */ @@ -3453,19 +4383,15 @@ static int dasd_eckd_restore_device(struct dasd_device *device) goto out_err; /* register lcu with alias handling, enable PAV if this is a new lcu */ - is_known = dasd_alias_make_device_known_to_lcu(device); - if (is_known < 0) - return is_known; - if (!is_known) { - dasd_eckd_validate_server(device); - dasd_alias_lcu_setup_complete(device); - } else - dasd_alias_wait_for_lcu_setup(device); + rc = dasd_alias_make_device_known_to_lcu(device); + if (rc) + return rc; + + set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr_flags); + dasd_eckd_validate_server(device, cqr_flags); /* RE-Read Configuration Data */ - rc = dasd_eckd_read_conf(device); - if (rc) - goto out_err; + dasd_eckd_read_conf(device); /* Read Feature Codes */ dasd_eckd_read_features(device); @@ -3542,18 +4468,23 @@ out_err: } static struct ccw_driver dasd_eckd_driver = { - .name = "dasd-eckd", - .owner = THIS_MODULE, + .driver = { + .name = "dasd-eckd", + .owner = THIS_MODULE, + }, .ids = dasd_eckd_ids, .probe = dasd_eckd_probe, .remove = dasd_generic_remove, .set_offline = dasd_generic_set_offline, .set_online = dasd_eckd_set_online, .notify = dasd_generic_notify, + .path_event = dasd_generic_path_event, + .shutdown = dasd_generic_shutdown, .freeze = dasd_generic_pm_freeze, .thaw = dasd_generic_restore_device, .restore = dasd_generic_restore_device, .uc_handler = dasd_generic_uc_handler, + .int_class = IRQIO_DAS, }; /* @@ -3573,12 +4504,14 @@ static struct dasd_discipline dasd_eckd_discipline = { .owner = THIS_MODULE, .name = "ECKD", .ebcname = "ECKD", - .max_blocks = 240, + .max_blocks = 190, .check_device = dasd_eckd_check_characteristics, .uncheck_device = dasd_eckd_uncheck_device, .do_analysis = dasd_eckd_do_analysis, - .ready_to_online = dasd_eckd_ready_to_online, + .verify_path = dasd_eckd_verify_path, + .basic_to_ready = dasd_eckd_basic_to_ready, .online_to_ready = dasd_eckd_online_to_ready, + .ready_to_basic = dasd_eckd_ready_to_basic, .fill_geometry = dasd_eckd_fill_geometry, .start_IO = dasd_start_IO, .term_IO = dasd_term_IO, @@ -3586,7 +4519,7 @@ static struct dasd_discipline dasd_eckd_discipline = { .format_device = dasd_eckd_format_device, .erp_action = dasd_eckd_erp_action, .erp_postaction = dasd_eckd_erp_postaction, - .handle_unsolicited_interrupt = dasd_eckd_handle_unsolicited_interrupt, + .check_for_device_change = dasd_eckd_check_for_device_change, .build_cp = dasd_eckd_build_alias_cp, .free_cp = dasd_eckd_free_alias_cp, .dump_sense = dasd_eckd_dump_sense, @@ -3597,6 +4530,7 @@ static struct dasd_discipline dasd_eckd_discipline = { .restore = dasd_eckd_restore_device, .reload = dasd_eckd_reload_device, .get_uid = dasd_eckd_get_uid, + .kick_validate = dasd_eckd_kick_validate_server, }; static int __init @@ -3609,11 +4543,26 @@ dasd_eckd_init(void) GFP_KERNEL | GFP_DMA); if (!dasd_reserve_req) return -ENOMEM; + path_verification_worker = kmalloc(sizeof(*path_verification_worker), + GFP_KERNEL | GFP_DMA); + if (!path_verification_worker) { + kfree(dasd_reserve_req); + return -ENOMEM; + } + rawpadpage = (void *)__get_free_page(GFP_KERNEL); + if (!rawpadpage) { + kfree(path_verification_worker); + kfree(dasd_reserve_req); + return -ENOMEM; + } ret = ccw_driver_register(&dasd_eckd_driver); if (!ret) wait_for_device_probe(); - else + else { + kfree(path_verification_worker); kfree(dasd_reserve_req); + free_page((unsigned long)rawpadpage); + } return ret; } @@ -3621,7 +4570,9 @@ static void __exit dasd_eckd_cleanup(void) { ccw_driver_unregister(&dasd_eckd_driver); + kfree(path_verification_worker); kfree(dasd_reserve_req); + free_page((unsigned long)rawpadpage); } module_init(dasd_eckd_init); diff --git a/drivers/s390/block/dasd_eckd.h b/drivers/s390/block/dasd_eckd.h index 12097c24f2f..2555e494591 100644 --- a/drivers/s390/block/dasd_eckd.h +++ b/drivers/s390/block/dasd_eckd.h @@ -1,9 +1,8 @@ /* - * File...........: linux/drivers/s390/block/dasd_eckd.h * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Horst Hummel <Horst.Hummel@de.ibm.com> * Bugreports.to..: <Linux390@de.ibm.com> - * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 1999,2000 + * Copyright IBM Corp. 1999, 2000 * */ @@ -37,14 +36,17 @@ #define DASD_ECKD_CCW_WRITE_KD_MT 0x8d #define DASD_ECKD_CCW_READ_KD_MT 0x8e #define DASD_ECKD_CCW_RELEASE 0x94 +#define DASD_ECKD_CCW_WRITE_FULL_TRACK 0x95 #define DASD_ECKD_CCW_READ_CKD_MT 0x9e #define DASD_ECKD_CCW_WRITE_CKD_MT 0x9d #define DASD_ECKD_CCW_WRITE_TRACK_DATA 0xA5 #define DASD_ECKD_CCW_READ_TRACK_DATA 0xA6 #define DASD_ECKD_CCW_RESERVE 0xB4 +#define DASD_ECKD_CCW_READ_TRACK 0xDE #define DASD_ECKD_CCW_PFX 0xE7 #define DASD_ECKD_CCW_PFX_READ 0xEA #define DASD_ECKD_CCW_RSCK 0xF9 +#define DASD_ECKD_CCW_RCD 0xFA /* * Perform Subsystem Function / Sub-Orders @@ -57,6 +59,11 @@ */ #define LV_COMPAT_CYL 0xFFFE + +#define FCX_MAX_DATA_FACTOR 65536 +#define DASD_ECKD_RCD_DATA_SIZE 256 + + /***************************************************************************** * SECTION: Type Definitions ****************************************************************************/ @@ -331,12 +338,6 @@ struct dasd_gneq { __u8 reserved2[22]; } __attribute__ ((packed)); -struct dasd_eckd_path { - __u8 opm; - __u8 ppm; - __u8 npm; -}; - struct dasd_rssd_features { char feature[256]; } __attribute__((packed)); @@ -442,7 +443,6 @@ struct dasd_eckd_private { struct vd_sneq *vdsneq; struct dasd_gneq *gneq; - struct dasd_eckd_path path_data; struct eckd_count count_area[5]; int init_cqr_status; int uses_cdl; @@ -455,6 +455,8 @@ struct dasd_eckd_private { struct alias_pav_group *pavgroup; struct alias_lcu *lcu; int count; + + u32 fcx_max_data; }; diff --git a/drivers/s390/block/dasd_eer.c b/drivers/s390/block/dasd_eer.c index c71d89dba30..21ef63cf096 100644 --- a/drivers/s390/block/dasd_eer.c +++ b/drivers/s390/block/dasd_eer.c @@ -1,7 +1,7 @@ /* * Character device driver for extended error reporting. * - * Copyright (C) 2005 IBM Corporation + * Copyright IBM Corp. 2005 * extended error reporting for DASD ECKD devices * Author(s): Stefan Weinhuber <wein@de.ibm.com> */ @@ -17,12 +17,11 @@ #include <linux/device.h> #include <linux/poll.h> #include <linux/mutex.h> -#include <linux/smp_lock.h> #include <linux/err.h> #include <linux/slab.h> #include <asm/uaccess.h> -#include <asm/atomic.h> +#include <linux/atomic.h> #include <asm/ebcdic.h> #include "dasd_int.h" @@ -474,6 +473,7 @@ int dasd_eer_enable(struct dasd_device *device) cqr->retries = 255; cqr->expires = 10 * HZ; clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags); + set_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags); ccw = cqr->cpaddr; ccw->cmd_code = DASD_ECKD_CCW_SNSS; @@ -481,7 +481,7 @@ int dasd_eer_enable(struct dasd_device *device) ccw->flags = 0; ccw->cda = (__u32)(addr_t) cqr->data; - cqr->buildclk = get_clock(); + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; cqr->callback = dasd_eer_snss_cb; diff --git a/drivers/s390/block/dasd_erp.c b/drivers/s390/block/dasd_erp.c index 7656384a811..e1e88486b2b 100644 --- a/drivers/s390/block/dasd_erp.c +++ b/drivers/s390/block/dasd_erp.c @@ -1,11 +1,10 @@ /* - * File...........: linux/drivers/s390/block/dasd.c * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Horst Hummel <Horst.Hummel@de.ibm.com> * Carsten Otte <Cotte@de.ibm.com> * Martin Schwidefsky <schwidefsky@de.ibm.com> * Bugreports.to..: <Linux390@de.ibm.com> - * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 1999-2001 + * Copyright IBM Corp. 1999, 2001 * */ @@ -96,13 +95,14 @@ dasd_default_erp_action(struct dasd_ccw_req *cqr) DBF_DEV_EVENT(DBF_DEBUG, device, "default ERP called (%i retries left)", cqr->retries); - cqr->lpm = LPM_ANYPATH; + if (!test_bit(DASD_CQR_VERIFY_PATH, &cqr->flags)) + cqr->lpm = device->path_data.opm; cqr->status = DASD_CQR_FILLED; } else { pr_err("%s: default ERP has run out of retries and failed\n", dev_name(&device->cdev->dev)); cqr->status = DASD_CQR_FAILED; - cqr->stopclk = get_clock(); + cqr->stopclk = get_tod_clock(); } return cqr; } /* end dasd_default_erp_action */ @@ -124,10 +124,15 @@ dasd_default_erp_action(struct dasd_ccw_req *cqr) struct dasd_ccw_req *dasd_default_erp_postaction(struct dasd_ccw_req *cqr) { int success; + unsigned long long startclk, stopclk; + struct dasd_device *startdev; BUG_ON(cqr->refers == NULL || cqr->function == NULL); success = cqr->status == DASD_CQR_DONE; + startclk = cqr->startclk; + stopclk = cqr->stopclk; + startdev = cqr->startdev; /* free all ERPs - but NOT the original cqr */ while (cqr->refers != NULL) { @@ -142,11 +147,14 @@ struct dasd_ccw_req *dasd_default_erp_postaction(struct dasd_ccw_req *cqr) } /* set corresponding status to original cqr */ + cqr->startclk = startclk; + cqr->stopclk = stopclk; + cqr->startdev = startdev; if (success) cqr->status = DASD_CQR_DONE; else { cqr->status = DASD_CQR_FAILED; - cqr->stopclk = get_clock(); + cqr->stopclk = get_tod_clock(); } return cqr; @@ -159,6 +167,16 @@ dasd_log_sense(struct dasd_ccw_req *cqr, struct irb *irb) struct dasd_device *device; device = cqr->startdev; + if (cqr->intrc == -ETIMEDOUT) { + dev_err(&device->cdev->dev, + "A timeout error occurred for cqr %p", cqr); + return; + } + if (cqr->intrc == -ENOLINK) { + dev_err(&device->cdev->dev, + "A transport error occurred for cqr %p", cqr); + return; + } /* dump sense data */ if (device->discipline && device->discipline->dump_sense) device->discipline->dump_sense(device, cqr, irb); diff --git a/drivers/s390/block/dasd_fba.c b/drivers/s390/block/dasd_fba.c index bec5486e0e6..2c8e68bf9a1 100644 --- a/drivers/s390/block/dasd_fba.c +++ b/drivers/s390/block/dasd_fba.c @@ -1,5 +1,4 @@ /* - * File...........: linux/drivers/s390/block/dasd_fba.c * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Bugreports.to..: <Linux390@de.ibm.com> * Copyright IBM Corp. 1999, 2009 @@ -30,6 +29,8 @@ #endif /* PRINTK_HEADER */ #define PRINTK_HEADER "dasd(fba):" +#define FBA_DEFAULT_RETRIES 32 + #define DASD_FBA_CCW_WRITE 0x41 #define DASD_FBA_CCW_READ 0x42 #define DASD_FBA_CCW_LOCATE 0x43 @@ -65,17 +66,21 @@ dasd_fba_set_online(struct ccw_device *cdev) } static struct ccw_driver dasd_fba_driver = { - .name = "dasd-fba", - .owner = THIS_MODULE, + .driver = { + .name = "dasd-fba", + .owner = THIS_MODULE, + }, .ids = dasd_fba_ids, .probe = dasd_fba_probe, .remove = dasd_generic_remove, .set_offline = dasd_generic_set_offline, .set_online = dasd_fba_set_online, .notify = dasd_generic_notify, + .path_event = dasd_generic_path_event, .freeze = dasd_generic_pm_freeze, .thaw = dasd_generic_restore_device, .restore = dasd_generic_restore_device, + .int_class = IRQIO_DAS, }; static void @@ -164,6 +169,8 @@ dasd_fba_check_characteristics(struct dasd_device *device) } device->default_expires = DASD_EXPIRES; + device->default_retries = FBA_DEFAULT_RETRIES; + device->path_data.opm = LPM_ANYPATH; readonly = dasd_device_is_ro(device); if (readonly) @@ -231,24 +238,16 @@ dasd_fba_erp_postaction(struct dasd_ccw_req * cqr) return NULL; } -static void dasd_fba_handle_unsolicited_interrupt(struct dasd_device *device, - struct irb *irb) +static void dasd_fba_check_for_device_change(struct dasd_device *device, + struct dasd_ccw_req *cqr, + struct irb *irb) { char mask; /* first of all check for state change pending interrupt */ mask = DEV_STAT_ATTENTION | DEV_STAT_DEV_END | DEV_STAT_UNIT_EXCEP; - if ((irb->scsw.cmd.dstat & mask) == mask) { + if ((irb->scsw.cmd.dstat & mask) == mask) dasd_generic_handle_state_change(device); - return; - } - - /* check for unsolicited interrupts */ - DBF_DEV_EVENT(DBF_WARNING, device, "%s", - "unsolicited interrupt received"); - device->discipline->dump_sense_dbf(device, irb, "unsolicited"); - dasd_schedule_device_bh(device); - return; }; static struct dasd_ccw_req *dasd_fba_build_cp(struct dasd_device * memdev, @@ -261,7 +260,7 @@ static struct dasd_ccw_req *dasd_fba_build_cp(struct dasd_device * memdev, struct dasd_ccw_req *cqr; struct ccw1 *ccw; struct req_iterator iter; - struct bio_vec *bv; + struct bio_vec bv; char *dst; int count, cidaw, cplength, datasize; sector_t recid, first_rec, last_rec; @@ -284,13 +283,13 @@ static struct dasd_ccw_req *dasd_fba_build_cp(struct dasd_device * memdev, count = 0; cidaw = 0; rq_for_each_segment(bv, req, iter) { - if (bv->bv_len & (blksize - 1)) + if (bv.bv_len & (blksize - 1)) /* Fba can only do full blocks. */ return ERR_PTR(-EINVAL); - count += bv->bv_len >> (block->s2b_shift + 9); + count += bv.bv_len >> (block->s2b_shift + 9); #if defined(CONFIG_64BIT) - if (idal_is_needed (page_address(bv->bv_page), bv->bv_len)) - cidaw += bv->bv_len / blksize; + if (idal_is_needed (page_address(bv.bv_page), bv.bv_len)) + cidaw += bv.bv_len / blksize; #endif } /* Paranoia. */ @@ -327,16 +326,16 @@ static struct dasd_ccw_req *dasd_fba_build_cp(struct dasd_device * memdev, } recid = first_rec; rq_for_each_segment(bv, req, iter) { - dst = page_address(bv->bv_page) + bv->bv_offset; + dst = page_address(bv.bv_page) + bv.bv_offset; if (dasd_page_cache) { char *copy = kmem_cache_alloc(dasd_page_cache, GFP_DMA | __GFP_NOWARN); if (copy && rq_data_dir(req) == WRITE) - memcpy(copy + bv->bv_offset, dst, bv->bv_len); + memcpy(copy + bv.bv_offset, dst, bv.bv_len); if (copy) - dst = copy + bv->bv_offset; + dst = copy + bv.bv_offset; } - for (off = 0; off < bv->bv_len; off += blksize) { + for (off = 0; off < bv.bv_len; off += blksize) { /* Locate record for stupid devices. */ if (private->rdc_data.mode.bits.data_chain == 0) { ccw[-1].flags |= CCW_FLAG_CC; @@ -373,8 +372,8 @@ static struct dasd_ccw_req *dasd_fba_build_cp(struct dasd_device * memdev, cqr->memdev = memdev; cqr->block = block; cqr->expires = memdev->default_expires * HZ; /* default 5 minutes */ - cqr->retries = 32; - cqr->buildclk = get_clock(); + cqr->retries = memdev->default_retries; + cqr->buildclk = get_tod_clock(); cqr->status = DASD_CQR_FILLED; return cqr; } @@ -385,7 +384,7 @@ dasd_fba_free_cp(struct dasd_ccw_req *cqr, struct request *req) struct dasd_fba_private *private; struct ccw1 *ccw; struct req_iterator iter; - struct bio_vec *bv; + struct bio_vec bv; char *dst, *cda; unsigned int blksize, off; int status; @@ -400,8 +399,8 @@ dasd_fba_free_cp(struct dasd_ccw_req *cqr, struct request *req) if (private->rdc_data.mode.bits.data_chain != 0) ccw++; rq_for_each_segment(bv, req, iter) { - dst = page_address(bv->bv_page) + bv->bv_offset; - for (off = 0; off < bv->bv_len; off += blksize) { + dst = page_address(bv.bv_page) + bv.bv_offset; + for (off = 0; off < bv.bv_len; off += blksize) { /* Skip locate record. */ if (private->rdc_data.mode.bits.data_chain == 0) ccw++; @@ -412,7 +411,7 @@ dasd_fba_free_cp(struct dasd_ccw_req *cqr, struct request *req) cda = (char *)((addr_t) ccw->cda); if (dst != cda) { if (rq_data_dir(req) == READ) - memcpy(dst, cda, bv->bv_len); + memcpy(dst, cda, bv.bv_len); kmem_cache_free(dasd_page_cache, (void *)((addr_t)cda & PAGE_MASK)); } @@ -429,7 +428,10 @@ out: static void dasd_fba_handle_terminated_request(struct dasd_ccw_req *cqr) { - cqr->status = DASD_CQR_FILLED; + if (cqr->retries < 0) + cqr->status = DASD_CQR_FAILED; + else + cqr->status = DASD_CQR_FILLED; }; static int @@ -483,19 +485,19 @@ dasd_fba_dump_sense(struct dasd_device *device, struct dasd_ccw_req * req, "No memory to dump sense data"); return; } - len = sprintf(page, KERN_ERR PRINTK_HEADER + len = sprintf(page, PRINTK_HEADER " I/O status report for device %s:\n", dev_name(&device->cdev->dev)); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " in req: %p CS: 0x%02X DS: 0x%02X\n", req, irb->scsw.cmd.cstat, irb->scsw.cmd.dstat); - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " device %s: Failing CCW: %p\n", dev_name(&device->cdev->dev), (void *) (addr_t) irb->scsw.cmd.cpa); if (irb->esw.esw0.erw.cons) { for (sl = 0; sl < 4; sl++) { - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " Sense(hex) %2d-%2d:", (8 * sl), ((8 * sl) + 7)); @@ -506,7 +508,7 @@ dasd_fba_dump_sense(struct dasd_device *device, struct dasd_ccw_req * req, len += sprintf(page + len, "\n"); } } else { - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " SORRY - NO VALID SENSE AVAILABLE\n"); } printk(KERN_ERR "%s", page); @@ -516,10 +518,9 @@ dasd_fba_dump_sense(struct dasd_device *device, struct dasd_ccw_req * req, act = req->cpaddr; for (last = act; last->flags & (CCW_FLAG_CC | CCW_FLAG_DC); last++); end = min(act + 8, last); - len = sprintf(page, KERN_ERR PRINTK_HEADER - " Related CP in req: %p\n", req); + len = sprintf(page, PRINTK_HEADER " Related CP in req: %p\n", req); while (act <= end) { - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " CCW %p: %08X %08X DAT:", act, ((int *) act)[0], ((int *) act)[1]); for (count = 0; count < 32 && count < act->count; @@ -537,11 +538,11 @@ dasd_fba_dump_sense(struct dasd_device *device, struct dasd_ccw_req * req, len = 0; if (act < ((struct ccw1 *)(addr_t) irb->scsw.cmd.cpa) - 2) { act = ((struct ccw1 *)(addr_t) irb->scsw.cmd.cpa) - 2; - len += sprintf(page + len, KERN_ERR PRINTK_HEADER "......\n"); + len += sprintf(page + len, PRINTK_HEADER "......\n"); } end = min((struct ccw1 *)(addr_t) irb->scsw.cmd.cpa + 2, last); while (act <= end) { - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " CCW %p: %08X %08X DAT:", act, ((int *) act)[0], ((int *) act)[1]); for (count = 0; count < 32 && count < act->count; @@ -556,10 +557,10 @@ dasd_fba_dump_sense(struct dasd_device *device, struct dasd_ccw_req * req, /* print last CCWs */ if (act < last - 2) { act = last - 2; - len += sprintf(page + len, KERN_ERR PRINTK_HEADER "......\n"); + len += sprintf(page + len, PRINTK_HEADER "......\n"); } while (act <= last) { - len += sprintf(page + len, KERN_ERR PRINTK_HEADER + len += sprintf(page + len, PRINTK_HEADER " CCW %p: %08X %08X DAT:", act, ((int *) act)[0], ((int *) act)[1]); for (count = 0; count < 32 && count < act->count; @@ -596,13 +597,14 @@ static struct dasd_discipline dasd_fba_discipline = { .max_blocks = 96, .check_device = dasd_fba_check_characteristics, .do_analysis = dasd_fba_do_analysis, + .verify_path = dasd_generic_verify_path, .fill_geometry = dasd_fba_fill_geometry, .start_IO = dasd_start_IO, .term_IO = dasd_term_IO, .handle_terminated_request = dasd_fba_handle_terminated_request, .erp_action = dasd_fba_erp_action, .erp_postaction = dasd_fba_erp_postaction, - .handle_unsolicited_interrupt = dasd_fba_handle_unsolicited_interrupt, + .check_for_device_change = dasd_fba_check_for_device_change, .build_cp = dasd_fba_build_cp, .free_cp = dasd_fba_free_cp, .dump_sense = dasd_fba_dump_sense, diff --git a/drivers/s390/block/dasd_fba.h b/drivers/s390/block/dasd_fba.h index 14c910baa5f..b5d3db0e5ef 100644 --- a/drivers/s390/block/dasd_fba.h +++ b/drivers/s390/block/dasd_fba.h @@ -1,8 +1,7 @@ /* - * File...........: linux/drivers/s390/block/dasd_fba.h * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Bugreports.to..: <Linux390@de.ibm.com> - * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 1999,2000 + * Coypright IBM Corp. 1999, 2000 * */ diff --git a/drivers/s390/block/dasd_genhd.c b/drivers/s390/block/dasd_genhd.c index 30a1ca3d08b..f224d59c4b6 100644 --- a/drivers/s390/block/dasd_genhd.c +++ b/drivers/s390/block/dasd_genhd.c @@ -1,11 +1,10 @@ /* - * File...........: linux/drivers/s390/block/dasd_genhd.c * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Horst Hummel <Horst.Hummel@de.ibm.com> * Carsten Otte <Cotte@de.ibm.com> * Martin Schwidefsky <schwidefsky@de.ibm.com> * Bugreports.to..: <Linux390@de.ibm.com> - * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 1999-2001 + * Copyright IBM Corp. 1999, 2001 * * gendisk related functions for the dasd driver. * @@ -73,7 +72,7 @@ int dasd_gendisk_alloc(struct dasd_block *block) if (base->features & DASD_FEATURE_READONLY || test_bit(DASD_FLAG_DEVICE_RO, &base->flags)) set_disk_ro(gdp, 1); - gdp->private_data = block; + dasd_add_link_to_gendisk(gdp, base); gdp->queue = block->request_queue; block->gdp = gdp; set_capacity(block->gdp, 0); @@ -88,7 +87,6 @@ void dasd_gendisk_free(struct dasd_block *block) { if (block->gdp) { del_gendisk(block->gdp); - block->gdp->queue = NULL; block->gdp->private_data = NULL; put_disk(block->gdp); block->gdp = NULL; @@ -103,7 +101,7 @@ int dasd_scan_partitions(struct dasd_block *block) struct block_device *bdev; bdev = bdget_disk(block->gdp, 0); - if (!bdev || blkdev_get(bdev, FMODE_READ) < 0) + if (!bdev || blkdev_get(bdev, FMODE_READ, NULL) < 0) return -ENODEV; /* * See fs/partition/check.c:register_disk,rescan_partitions diff --git a/drivers/s390/block/dasd_int.h b/drivers/s390/block/dasd_int.h index 500678d7116..690001af0d0 100644 --- a/drivers/s390/block/dasd_int.h +++ b/drivers/s390/block/dasd_int.h @@ -1,5 +1,4 @@ /* - * File...........: linux/drivers/s390/block/dasd_int.h * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Horst Hummel <Horst.Hummel@de.ibm.com> * Martin Schwidefsky <schwidefsky@de.ibm.com> @@ -10,8 +9,6 @@ #ifndef DASD_INT_H #define DASD_INT_H -#ifdef __KERNEL__ - /* we keep old device allocation scheme; IOW, minors are still in 0..255 */ #define DASD_PER_MAJOR (1U << (MINORBITS - DASD_PARTN_BITS)) #define DASD_PARTN_MASK ((1 << DASD_PARTN_BITS) - 1) @@ -227,10 +224,17 @@ struct dasd_ccw_req { /* default expiration time*/ #define DASD_EXPIRES 300 #define DASD_EXPIRES_MAX 40000000 +#define DASD_RETRIES 256 +#define DASD_RETRIES_MAX 32768 /* per dasd_ccw_req flags */ #define DASD_CQR_FLAGS_USE_ERP 0 /* use ERP for this request */ #define DASD_CQR_FLAGS_FAILFAST 1 /* FAILFAST */ +#define DASD_CQR_VERIFY_PATH 2 /* path verification request */ +#define DASD_CQR_ALLOW_SLOCK 3 /* Try this request even when lock was + * stolen. Should not be combined with + * DASD_CQR_FLAGS_USE_ERP + */ /* Signature for error recovery functions. */ typedef struct dasd_ccw_req *(*dasd_erp_fn_t) (struct dasd_ccw_req *); @@ -287,13 +291,22 @@ struct dasd_discipline { int (*do_analysis) (struct dasd_block *); /* + * This function is called, when new paths become available. + * Disciplins may use this callback to do necessary setup work, + * e.g. verify that new path is compatible with the current + * configuration. + */ + int (*verify_path)(struct dasd_device *, __u8); + + /* * Last things to do when a device is set online, and first things * when it is set offline. */ - int (*ready_to_online) (struct dasd_device *); + int (*basic_to_ready) (struct dasd_device *); int (*online_to_ready) (struct dasd_device *); + int (*ready_to_basic) (struct dasd_device *); - /* + /* (struct dasd_device *); * Device operation functions. build_cp creates a ccw chain for * a block device request, start_io starts the request and * term_IO cancels it (e.g. in case of a timeout). format_device @@ -307,8 +320,8 @@ struct dasd_discipline { int (*start_IO) (struct dasd_ccw_req *); int (*term_IO) (struct dasd_ccw_req *); void (*handle_terminated_request) (struct dasd_ccw_req *); - struct dasd_ccw_req *(*format_device) (struct dasd_device *, - struct format_data_t *); + int (*format_device) (struct dasd_device *, + struct format_data_t *); int (*free_cp) (struct dasd_ccw_req *, struct request *); /* @@ -325,9 +338,9 @@ struct dasd_discipline { void (*dump_sense) (struct dasd_device *, struct dasd_ccw_req *, struct irb *); void (*dump_sense_dbf) (struct dasd_device *, struct irb *, char *); - - void (*handle_unsolicited_interrupt) (struct dasd_device *, - struct irb *); + void (*check_for_device_change) (struct dasd_device *, + struct dasd_ccw_req *, + struct irb *); /* i/o control functions. */ int (*fill_geometry) (struct dasd_block *, struct hd_geometry *); @@ -342,6 +355,7 @@ struct dasd_discipline { int (*reload) (struct dasd_device *); int (*get_uid) (struct dasd_device *, struct dasd_uid *); + void (*kick_validate) (struct dasd_device *); }; extern struct dasd_discipline *dasd_diag_discipline_pointer; @@ -362,6 +376,48 @@ extern struct dasd_discipline *dasd_diag_discipline_pointer; #define DASD_EER_STATECHANGE 3 #define DASD_EER_PPRCSUSPEND 4 +struct dasd_path { + __u8 opm; + __u8 tbvpm; + __u8 ppm; + __u8 npm; +}; + +struct dasd_profile_info { + /* legacy part of profile data, as in dasd_profile_info_t */ + unsigned int dasd_io_reqs; /* number of requests processed */ + unsigned int dasd_io_sects; /* number of sectors processed */ + unsigned int dasd_io_secs[32]; /* histogram of request's sizes */ + unsigned int dasd_io_times[32]; /* histogram of requests's times */ + unsigned int dasd_io_timps[32]; /* h. of requests's times per sector */ + unsigned int dasd_io_time1[32]; /* hist. of time from build to start */ + unsigned int dasd_io_time2[32]; /* hist. of time from start to irq */ + unsigned int dasd_io_time2ps[32]; /* hist. of time from start to irq */ + unsigned int dasd_io_time3[32]; /* hist. of time from irq to end */ + unsigned int dasd_io_nr_req[32]; /* hist. of # of requests in chanq */ + + /* new data */ + struct timespec starttod; /* time of start or last reset */ + unsigned int dasd_io_alias; /* requests using an alias */ + unsigned int dasd_io_tpm; /* requests using transport mode */ + unsigned int dasd_read_reqs; /* total number of read requests */ + unsigned int dasd_read_sects; /* total number read sectors */ + unsigned int dasd_read_alias; /* read request using an alias */ + unsigned int dasd_read_tpm; /* read requests in transport mode */ + unsigned int dasd_read_secs[32]; /* histogram of request's sizes */ + unsigned int dasd_read_times[32]; /* histogram of requests's times */ + unsigned int dasd_read_time1[32]; /* hist. time from build to start */ + unsigned int dasd_read_time2[32]; /* hist. of time from start to irq */ + unsigned int dasd_read_time3[32]; /* hist. of time from irq to end */ + unsigned int dasd_read_nr_req[32]; /* hist. of # of requests in chanq */ +}; + +struct dasd_profile { + struct dentry *dentry; + struct dasd_profile_info *data; + spinlock_t lock; +}; + struct dasd_device { /* Block device stuff. */ struct dasd_block *block; @@ -377,6 +433,7 @@ struct dasd_device { struct dasd_discipline *discipline; struct dasd_discipline *base_discipline; char *private; + struct dasd_path path_data; /* Device state and target state. */ int state, target; @@ -399,6 +456,7 @@ struct dasd_device { struct work_struct kick_work; struct work_struct restore_device; struct work_struct reload_device; + struct work_struct kick_validate; struct timer_list timer; debug_info_t *debug_area; @@ -410,6 +468,12 @@ struct dasd_device { /* default expiration time in s */ unsigned long default_expires; + unsigned long default_retries; + + unsigned long blk_timeout; + + struct dentry *debugfs_dentry; + struct dasd_profile profile; }; struct dasd_block { @@ -432,9 +496,8 @@ struct dasd_block { struct tasklet_struct tasklet; struct timer_list timer; -#ifdef CONFIG_DASD_PROFILE - struct dasd_profile_info_t profile; -#endif + struct dentry *debugfs_dentry; + struct dasd_profile profile; }; @@ -456,6 +519,15 @@ struct dasd_block { * confuse this with the user specified * read-only feature. */ +#define DASD_FLAG_IS_RESERVED 7 /* The device is reserved */ +#define DASD_FLAG_LOCK_STOLEN 8 /* The device lock was stolen */ +#define DASD_FLAG_SUSPENDED 9 /* The device was suspended */ +#define DASD_FLAG_SAFE_OFFLINE 10 /* safe offline processing requested*/ +#define DASD_FLAG_SAFE_OFFLINE_RUNNING 11 /* safe offline running */ +#define DASD_FLAG_ABORTALL 12 /* Abort all noretry requests */ + +#define DASD_SLEEPON_START_TAG ((void *) 1) +#define DASD_SLEEPON_END_TAG ((void *) 2) void dasd_put_device_wake(struct dasd_device *); @@ -565,12 +637,13 @@ dasd_check_blocksize(int bsize) } /* externals in dasd.c */ -#define DASD_PROFILE_ON 1 -#define DASD_PROFILE_OFF 0 +#define DASD_PROFILE_OFF 0 +#define DASD_PROFILE_ON 1 +#define DASD_PROFILE_GLOBAL_ONLY 2 extern debug_info_t *dasd_debug_area; -extern struct dasd_profile_info_t dasd_global_profile; -extern unsigned int dasd_profile_level; +extern struct dasd_profile_info dasd_global_profile_data; +extern unsigned int dasd_global_profile_level; extern const struct block_device_operations dasd_device_operations; extern struct kmem_cache *dasd_page_cache; @@ -581,6 +654,7 @@ struct dasd_ccw_req * dasd_smalloc_request(int , int, int, struct dasd_device *); void dasd_kfree_request(struct dasd_ccw_req *, struct dasd_device *); void dasd_sfree_request(struct dasd_ccw_req *, struct dasd_device *); +void dasd_wakeup_cb(struct dasd_ccw_req *, void *); static inline int dasd_kmalloc_set_cda(struct ccw1 *ccw, void *cda, struct dasd_device *device) @@ -594,6 +668,8 @@ void dasd_free_device(struct dasd_device *); struct dasd_block *dasd_alloc_block(void); void dasd_free_block(struct dasd_block *); +enum blk_eh_timer_return dasd_times_out(struct request *req); + void dasd_enable_device(struct dasd_device *); void dasd_set_target_state(struct dasd_device *, int); void dasd_kick_device(struct dasd_device *); @@ -607,6 +683,7 @@ int dasd_term_IO(struct dasd_ccw_req *); void dasd_schedule_device_bh(struct dasd_device *); void dasd_schedule_block_bh(struct dasd_block *); int dasd_sleep_on(struct dasd_ccw_req *); +int dasd_sleep_on_queue(struct list_head *); int dasd_sleep_on_immediatly(struct dasd_ccw_req *); int dasd_sleep_on_interruptible(struct dasd_ccw_req *); void dasd_device_set_timer(struct dasd_device *, int); @@ -620,10 +697,16 @@ void dasd_generic_remove (struct ccw_device *cdev); int dasd_generic_set_online(struct ccw_device *, struct dasd_discipline *); int dasd_generic_set_offline (struct ccw_device *cdev); int dasd_generic_notify(struct ccw_device *, int); +int dasd_generic_last_path_gone(struct dasd_device *); +int dasd_generic_path_operational(struct dasd_device *); +void dasd_generic_shutdown(struct ccw_device *); + void dasd_generic_handle_state_change(struct dasd_device *); int dasd_generic_pm_freeze(struct ccw_device *); int dasd_generic_restore_device(struct ccw_device *); enum uc_todo dasd_generic_uc_handler(struct ccw_device *, struct irb *); +void dasd_generic_path_event(struct ccw_device *, int *); +int dasd_generic_verify_path(struct dasd_device *, __u8); int dasd_generic_read_dev_chars(struct dasd_device *, int, void *, int); char *dasd_get_sense(struct irb *); @@ -633,6 +716,11 @@ void dasd_device_remove_stop_bits(struct dasd_device *, int); int dasd_device_is_ro(struct dasd_device *); +void dasd_profile_reset(struct dasd_profile *); +int dasd_profile_on(struct dasd_profile *); +void dasd_profile_off(struct dasd_profile *); +void dasd_global_profile_reset(void); +char *dasd_get_user_string(const char __user *, size_t); /* externals in dasd_devmap.c */ extern int dasd_max_devindex; @@ -657,6 +745,9 @@ struct dasd_device *dasd_device_from_cdev(struct ccw_device *); struct dasd_device *dasd_device_from_cdev_locked(struct ccw_device *); struct dasd_device *dasd_device_from_devindex(int); +void dasd_add_link_to_gendisk(struct gendisk *, struct dasd_device *); +struct dasd_device *dasd_device_from_gendisk(struct gendisk *); + int dasd_parse(void); int dasd_busid_known(const char *); @@ -712,6 +803,4 @@ static inline int dasd_eer_enabled(struct dasd_device *device) #define dasd_eer_enabled(d) (0) #endif /* CONFIG_DASD_ERR */ -#endif /* __KERNEL__ */ - #endif /* DASD_H */ diff --git a/drivers/s390/block/dasd_ioctl.c b/drivers/s390/block/dasd_ioctl.c index 26075e95b1b..25a0f2f8b0b 100644 --- a/drivers/s390/block/dasd_ioctl.c +++ b/drivers/s390/block/dasd_ioctl.c @@ -1,11 +1,10 @@ /* - * File...........: linux/drivers/s390/block/dasd_ioctl.c * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Horst Hummel <Horst.Hummel@de.ibm.com> * Carsten Otte <Cotte@de.ibm.com> * Martin Schwidefsky <schwidefsky@de.ibm.com> * Bugreports.to..: <Linux390@de.ibm.com> - * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 1999-2001 + * Copyright IBM Corp. 1999, 2001 * * i/o controls for the dasd driver. */ @@ -13,12 +12,14 @@ #define KMSG_COMPONENT "dasd" #include <linux/interrupt.h> +#include <linux/compat.h> #include <linux/major.h> #include <linux/fs.h> #include <linux/blkpg.h> #include <linux/slab.h> #include <asm/compat.h> #include <asm/ccwdev.h> +#include <asm/schid.h> #include <asm/cmb.h> #include <asm/uaccess.h> @@ -42,16 +43,22 @@ dasd_ioctl_api_version(void __user *argp) static int dasd_ioctl_enable(struct block_device *bdev) { - struct dasd_block *block = bdev->bd_disk->private_data; + struct dasd_device *base; if (!capable(CAP_SYS_ADMIN)) return -EACCES; - dasd_enable_device(block->base); + base = dasd_device_from_gendisk(bdev->bd_disk); + if (!base) + return -ENODEV; + + dasd_enable_device(base); /* Formatting the dasd device can change the capacity. */ mutex_lock(&bdev->bd_mutex); - i_size_write(bdev->bd_inode, (loff_t)get_capacity(block->gdp) << 9); + i_size_write(bdev->bd_inode, + (loff_t)get_capacity(base->block->gdp) << 9); mutex_unlock(&bdev->bd_mutex); + dasd_put_device(base); return 0; } @@ -62,11 +69,14 @@ dasd_ioctl_enable(struct block_device *bdev) static int dasd_ioctl_disable(struct block_device *bdev) { - struct dasd_block *block = bdev->bd_disk->private_data; + struct dasd_device *base; if (!capable(CAP_SYS_ADMIN)) return -EACCES; + base = dasd_device_from_gendisk(bdev->bd_disk); + if (!base) + return -ENODEV; /* * Man this is sick. We don't do a real disable but only downgrade * the device to DASD_STATE_BASIC. The reason is that dasdfmt uses @@ -75,7 +85,7 @@ dasd_ioctl_disable(struct block_device *bdev) * using the BIODASDFMT ioctl. Therefore the correct state for the * device is DASD_STATE_BASIC that allows to do basic i/o. */ - dasd_set_target_state(block->base, DASD_STATE_BASIC); + dasd_set_target_state(base, DASD_STATE_BASIC); /* * Set i_size to zero, since read, write, etc. check against this * value. @@ -83,6 +93,7 @@ dasd_ioctl_disable(struct block_device *bdev) mutex_lock(&bdev->bd_mutex); i_size_write(bdev->bd_inode, 0); mutex_unlock(&bdev->bd_mutex); + dasd_put_device(base); return 0; } @@ -130,14 +141,67 @@ static int dasd_ioctl_resume(struct dasd_block *block) } /* + * Abort all failfast I/O on a device. + */ +static int dasd_ioctl_abortio(struct dasd_block *block) +{ + unsigned long flags; + struct dasd_device *base; + struct dasd_ccw_req *cqr, *n; + + base = block->base; + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + + if (test_and_set_bit(DASD_FLAG_ABORTALL, &base->flags)) + return 0; + DBF_DEV_EVENT(DBF_NOTICE, base, "%s", "abortall flag set"); + + spin_lock_irqsave(&block->request_queue_lock, flags); + spin_lock(&block->queue_lock); + list_for_each_entry_safe(cqr, n, &block->ccw_queue, blocklist) { + if (test_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags) && + cqr->callback_data && + cqr->callback_data != DASD_SLEEPON_START_TAG && + cqr->callback_data != DASD_SLEEPON_END_TAG) { + spin_unlock(&block->queue_lock); + blk_abort_request(cqr->callback_data); + spin_lock(&block->queue_lock); + } + } + spin_unlock(&block->queue_lock); + spin_unlock_irqrestore(&block->request_queue_lock, flags); + + dasd_schedule_block_bh(block); + return 0; +} + +/* + * Allow I/O on a device + */ +static int dasd_ioctl_allowio(struct dasd_block *block) +{ + struct dasd_device *base; + + base = block->base; + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + + if (test_and_clear_bit(DASD_FLAG_ABORTALL, &base->flags)) + DBF_DEV_EVENT(DBF_NOTICE, base, "%s", "abortall flag unset"); + + return 0; +} + +/* * performs formatting of _device_ according to _fdata_ * Note: The discipline's format_function is assumed to deliver formatting - * commands to format a single unit of the device. In terms of the ECKD - * devices this means CCWs are generated to format a single track. + * commands to format multiple units of the device. In terms of the ECKD + * devices this means CCWs are generated to format multiple tracks. */ -static int dasd_format(struct dasd_block *block, struct format_data_t *fdata) +static int +dasd_format(struct dasd_block *block, struct format_data_t *fdata) { - struct dasd_ccw_req *cqr; struct dasd_device *base; int rc; @@ -146,8 +210,8 @@ static int dasd_format(struct dasd_block *block, struct format_data_t *fdata) return -EPERM; if (base->state != DASD_STATE_BASIC) { - pr_warning("%s: The DASD cannot be formatted while it is " - "enabled\n", dev_name(&base->cdev->dev)); + pr_warn("%s: The DASD cannot be formatted while it is enabled\n", + dev_name(&base->cdev->dev)); return -EBUSY; } @@ -167,21 +231,10 @@ static int dasd_format(struct dasd_block *block, struct format_data_t *fdata) bdput(bdev); } - while (fdata->start_unit <= fdata->stop_unit) { - cqr = base->discipline->format_device(base, fdata); - if (IS_ERR(cqr)) - return PTR_ERR(cqr); - rc = dasd_sleep_on_interruptible(cqr); - dasd_sfree_request(cqr, cqr->memdev); - if (rc) { - if (rc != -ERESTARTSYS) - pr_err("%s: Formatting unit %d failed with " - "rc=%d\n", dev_name(&base->cdev->dev), - fdata->start_unit, rc); - return rc; - } - fdata->start_unit++; - } + rc = base->discipline->format_device(base, fdata); + if (rc) + return rc; + return 0; } @@ -191,26 +244,36 @@ static int dasd_format(struct dasd_block *block, struct format_data_t *fdata) static int dasd_ioctl_format(struct block_device *bdev, void __user *argp) { - struct dasd_block *block = bdev->bd_disk->private_data; + struct dasd_device *base; struct format_data_t fdata; + int rc; if (!capable(CAP_SYS_ADMIN)) return -EACCES; if (!argp) return -EINVAL; - - if (block->base->features & DASD_FEATURE_READONLY || - test_bit(DASD_FLAG_DEVICE_RO, &block->base->flags)) + base = dasd_device_from_gendisk(bdev->bd_disk); + if (!base) + return -ENODEV; + if (base->features & DASD_FEATURE_READONLY || + test_bit(DASD_FLAG_DEVICE_RO, &base->flags)) { + dasd_put_device(base); return -EROFS; - if (copy_from_user(&fdata, argp, sizeof(struct format_data_t))) + } + if (copy_from_user(&fdata, argp, sizeof(struct format_data_t))) { + dasd_put_device(base); return -EFAULT; + } if (bdev != bdev->bd_contains) { pr_warning("%s: The specified DASD is a partition and cannot " "be formatted\n", - dev_name(&block->base->cdev->dev)); + dev_name(&base->cdev->dev)); + dasd_put_device(base); return -EINVAL; } - return dasd_format(block, &fdata); + rc = dasd_format(base->block, &fdata); + dasd_put_device(base); + return rc; } #ifdef CONFIG_DASD_PROFILE @@ -219,7 +282,7 @@ dasd_ioctl_format(struct block_device *bdev, void __user *argp) */ static int dasd_ioctl_reset_profile(struct dasd_block *block) { - memset(&block->profile, 0, sizeof(struct dasd_profile_info_t)); + dasd_profile_reset(&block->profile); return 0; } @@ -228,22 +291,56 @@ static int dasd_ioctl_reset_profile(struct dasd_block *block) */ static int dasd_ioctl_read_profile(struct dasd_block *block, void __user *argp) { - if (dasd_profile_level == DASD_PROFILE_OFF) - return -EIO; - if (copy_to_user(argp, &block->profile, - sizeof(struct dasd_profile_info_t))) - return -EFAULT; - return 0; + struct dasd_profile_info_t *data; + int rc = 0; + + data = kmalloc(sizeof(*data), GFP_KERNEL); + if (!data) + return -ENOMEM; + + spin_lock_bh(&block->profile.lock); + if (block->profile.data) { + data->dasd_io_reqs = block->profile.data->dasd_io_reqs; + data->dasd_io_sects = block->profile.data->dasd_io_sects; + memcpy(data->dasd_io_secs, block->profile.data->dasd_io_secs, + sizeof(data->dasd_io_secs)); + memcpy(data->dasd_io_times, block->profile.data->dasd_io_times, + sizeof(data->dasd_io_times)); + memcpy(data->dasd_io_timps, block->profile.data->dasd_io_timps, + sizeof(data->dasd_io_timps)); + memcpy(data->dasd_io_time1, block->profile.data->dasd_io_time1, + sizeof(data->dasd_io_time1)); + memcpy(data->dasd_io_time2, block->profile.data->dasd_io_time2, + sizeof(data->dasd_io_time2)); + memcpy(data->dasd_io_time2ps, + block->profile.data->dasd_io_time2ps, + sizeof(data->dasd_io_time2ps)); + memcpy(data->dasd_io_time3, block->profile.data->dasd_io_time3, + sizeof(data->dasd_io_time3)); + memcpy(data->dasd_io_nr_req, + block->profile.data->dasd_io_nr_req, + sizeof(data->dasd_io_nr_req)); + spin_unlock_bh(&block->profile.lock); + } else { + spin_unlock_bh(&block->profile.lock); + rc = -EIO; + goto out; + } + if (copy_to_user(argp, data, sizeof(*data))) + rc = -EFAULT; +out: + kfree(data); + return rc; } #else static int dasd_ioctl_reset_profile(struct dasd_block *block) { - return -ENOSYS; + return -ENOTTY; } static int dasd_ioctl_read_profile(struct dasd_block *block, void __user *argp) { - return -ENOSYS; + return -ENOTTY; } #endif @@ -254,11 +351,12 @@ static int dasd_ioctl_information(struct dasd_block *block, unsigned int cmd, void __user *argp) { struct dasd_information2_t *dasd_info; - unsigned long flags; - int rc; + struct subchannel_id sch_id; + struct ccw_dev_id dev_id; struct dasd_device *base; struct ccw_device *cdev; - struct ccw_dev_id dev_id; + unsigned long flags; + int rc; base = block->base; if (!base->discipline || !base->discipline->fill_info) @@ -276,9 +374,10 @@ static int dasd_ioctl_information(struct dasd_block *block, cdev = base->cdev; ccw_device_get_id(cdev, &dev_id); + ccw_device_get_schid(cdev, &sch_id); dasd_info->devno = dev_id.devno; - dasd_info->schid = _ccw_device_get_subchannel_number(base->cdev); + dasd_info->schid = sch_id.sch_no; dasd_info->cu_type = cdev->id.cu_type; dasd_info->cu_model = cdev->id.cu_model; dasd_info->dev_type = cdev->id.dev_type; @@ -340,8 +439,8 @@ static int dasd_ioctl_information(struct dasd_block *block, static int dasd_ioctl_set_ro(struct block_device *bdev, void __user *argp) { - struct dasd_block *block = bdev->bd_disk->private_data; - int intval; + struct dasd_device *base; + int intval, rc; if (!capable(CAP_SYS_ADMIN)) return -EACCES; @@ -350,10 +449,17 @@ dasd_ioctl_set_ro(struct block_device *bdev, void __user *argp) return -EINVAL; if (get_user(intval, (int __user *)argp)) return -EFAULT; - if (!intval && test_bit(DASD_FLAG_DEVICE_RO, &block->base->flags)) + base = dasd_device_from_gendisk(bdev->bd_disk); + if (!base) + return -ENODEV; + if (!intval && test_bit(DASD_FLAG_DEVICE_RO, &base->flags)) { + dasd_put_device(base); return -EROFS; + } set_disk_ro(bdev->bd_disk, intval); - return dasd_set_feature(block->base->cdev, DASD_FEATURE_READONLY, intval); + rc = dasd_set_feature(base->cdev, DASD_FEATURE_READONLY, intval); + dasd_put_device(base); + return rc; } static int dasd_ioctl_readall_cmb(struct dasd_block *block, unsigned int cmd, @@ -372,59 +478,81 @@ static int dasd_ioctl_readall_cmb(struct dasd_block *block, unsigned int cmd, int dasd_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, unsigned long arg) { - struct dasd_block *block = bdev->bd_disk->private_data; + struct dasd_block *block; + struct dasd_device *base; void __user *argp; + int rc; if (is_compat_task()) argp = compat_ptr(arg); else argp = (void __user *)arg; - if (!block) - return -ENODEV; - if ((_IOC_DIR(cmd) != _IOC_NONE) && !arg) { PRINT_DEBUG("empty data ptr"); return -EINVAL; } + base = dasd_device_from_gendisk(bdev->bd_disk); + if (!base) + return -ENODEV; + block = base->block; + rc = 0; switch (cmd) { case BIODASDDISABLE: - return dasd_ioctl_disable(bdev); + rc = dasd_ioctl_disable(bdev); + break; case BIODASDENABLE: - return dasd_ioctl_enable(bdev); + rc = dasd_ioctl_enable(bdev); + break; case BIODASDQUIESCE: - return dasd_ioctl_quiesce(block); + rc = dasd_ioctl_quiesce(block); + break; case BIODASDRESUME: - return dasd_ioctl_resume(block); + rc = dasd_ioctl_resume(block); + break; + case BIODASDABORTIO: + rc = dasd_ioctl_abortio(block); + break; + case BIODASDALLOWIO: + rc = dasd_ioctl_allowio(block); + break; case BIODASDFMT: - return dasd_ioctl_format(bdev, argp); + rc = dasd_ioctl_format(bdev, argp); + break; case BIODASDINFO: - return dasd_ioctl_information(block, cmd, argp); + rc = dasd_ioctl_information(block, cmd, argp); + break; case BIODASDINFO2: - return dasd_ioctl_information(block, cmd, argp); + rc = dasd_ioctl_information(block, cmd, argp); + break; case BIODASDPRRD: - return dasd_ioctl_read_profile(block, argp); + rc = dasd_ioctl_read_profile(block, argp); + break; case BIODASDPRRST: - return dasd_ioctl_reset_profile(block); + rc = dasd_ioctl_reset_profile(block); + break; case BLKROSET: - return dasd_ioctl_set_ro(bdev, argp); + rc = dasd_ioctl_set_ro(bdev, argp); + break; case DASDAPIVER: - return dasd_ioctl_api_version(argp); + rc = dasd_ioctl_api_version(argp); + break; case BIODASDCMFENABLE: - return enable_cmf(block->base->cdev); + rc = enable_cmf(base->cdev); + break; case BIODASDCMFDISABLE: - return disable_cmf(block->base->cdev); + rc = disable_cmf(base->cdev); + break; case BIODASDREADALLCMB: - return dasd_ioctl_readall_cmb(block, cmd, argp); + rc = dasd_ioctl_readall_cmb(block, cmd, argp); + break; default: /* if the discipline has an ioctl method try it. */ - if (block->base->discipline->ioctl) { - int rval = block->base->discipline->ioctl(block, cmd, argp); - if (rval != -ENOIOCTLCMD) - return rval; - } - - return -EINVAL; + rc = -ENOTTY; + if (base->discipline->ioctl) + rc = base->discipline->ioctl(block, cmd, argp); } + dasd_put_device(base); + return rc; } diff --git a/drivers/s390/block/dasd_proc.c b/drivers/s390/block/dasd_proc.c index c4a6a31bd9c..78ac905a5b7 100644 --- a/drivers/s390/block/dasd_proc.c +++ b/drivers/s390/block/dasd_proc.c @@ -1,11 +1,10 @@ /* - * File...........: linux/drivers/s390/block/dasd_proc.c * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com> * Horst Hummel <Horst.Hummel@de.ibm.com> * Carsten Otte <Cotte@de.ibm.com> * Martin Schwidefsky <schwidefsky@de.ibm.com> * Bugreports.to..: <Linux390@de.ibm.com> - * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 1999-2002 + * Coypright IBM Corp. 1999, 2002 * * /proc interface for the dasd driver. * @@ -32,28 +31,6 @@ static struct proc_dir_entry *dasd_proc_root_entry = NULL; static struct proc_dir_entry *dasd_devices_entry = NULL; static struct proc_dir_entry *dasd_statistics_entry = NULL; -#ifdef CONFIG_DASD_PROFILE -static char * -dasd_get_user_string(const char __user *user_buf, size_t user_len) -{ - char *buffer; - - buffer = kmalloc(user_len + 1, GFP_KERNEL); - if (buffer == NULL) - return ERR_PTR(-ENOMEM); - if (copy_from_user(buffer, user_buf, user_len) != 0) { - kfree(buffer); - return ERR_PTR(-EFAULT); - } - /* got the string, now strip linefeed. */ - if (buffer[user_len - 1] == '\n') - buffer[user_len - 1] = 0; - else - buffer[user_len] = 0; - return buffer; -} -#endif /* CONFIG_DASD_PROFILE */ - static int dasd_devices_show(struct seq_file *m, void *v) { @@ -167,6 +144,55 @@ static const struct file_operations dasd_devices_file_ops = { }; #ifdef CONFIG_DASD_PROFILE +static int dasd_stats_all_block_on(void) +{ + int i, rc; + struct dasd_device *device; + + rc = 0; + for (i = 0; i < dasd_max_devindex; ++i) { + device = dasd_device_from_devindex(i); + if (IS_ERR(device)) + continue; + if (device->block) + rc = dasd_profile_on(&device->block->profile); + dasd_put_device(device); + if (rc) + return rc; + } + return 0; +} + +static void dasd_stats_all_block_off(void) +{ + int i; + struct dasd_device *device; + + for (i = 0; i < dasd_max_devindex; ++i) { + device = dasd_device_from_devindex(i); + if (IS_ERR(device)) + continue; + if (device->block) + dasd_profile_off(&device->block->profile); + dasd_put_device(device); + } +} + +static void dasd_stats_all_block_reset(void) +{ + int i; + struct dasd_device *device; + + for (i = 0; i < dasd_max_devindex; ++i) { + device = dasd_device_from_devindex(i); + if (IS_ERR(device)) + continue; + if (device->block) + dasd_profile_reset(&device->block->profile); + dasd_put_device(device); + } +} + static void dasd_statistics_array(struct seq_file *m, unsigned int *array, int factor) { int i; @@ -183,18 +209,18 @@ static void dasd_statistics_array(struct seq_file *m, unsigned int *array, int f static int dasd_stats_proc_show(struct seq_file *m, void *v) { #ifdef CONFIG_DASD_PROFILE - struct dasd_profile_info_t *prof; + struct dasd_profile_info *prof; int factor; /* check for active profiling */ - if (dasd_profile_level == DASD_PROFILE_OFF) { + if (!dasd_global_profile_level) { seq_printf(m, "Statistics are off - they might be " "switched on using 'echo set on > " "/proc/dasd/statistics'\n"); return 0; } + prof = &dasd_global_profile_data; - prof = &dasd_global_profile; /* prevent counter 'overflow' on output */ for (factor = 1; (prof->dasd_io_reqs / factor) > 9999999; factor *= 10); @@ -245,6 +271,7 @@ static ssize_t dasd_stats_proc_write(struct file *file, { #ifdef CONFIG_DASD_PROFILE char *buffer, *str; + int rc; if (user_len > 65536) user_len = 65536; @@ -259,32 +286,40 @@ static ssize_t dasd_stats_proc_write(struct file *file, str = skip_spaces(str + 4); if (strcmp(str, "on") == 0) { /* switch on statistics profiling */ - dasd_profile_level = DASD_PROFILE_ON; + rc = dasd_stats_all_block_on(); + if (rc) { + dasd_stats_all_block_off(); + goto out_error; + } + dasd_global_profile_reset(); + dasd_global_profile_level = DASD_PROFILE_ON; pr_info("The statistics feature has been switched " "on\n"); } else if (strcmp(str, "off") == 0) { /* switch off and reset statistics profiling */ - memset(&dasd_global_profile, - 0, sizeof (struct dasd_profile_info_t)); - dasd_profile_level = DASD_PROFILE_OFF; + dasd_global_profile_level = DASD_PROFILE_OFF; + dasd_global_profile_reset(); + dasd_stats_all_block_off(); pr_info("The statistics feature has been switched " "off\n"); } else - goto out_error; + goto out_parse_error; } else if (strncmp(str, "reset", 5) == 0) { /* reset the statistics */ - memset(&dasd_global_profile, 0, - sizeof (struct dasd_profile_info_t)); + dasd_global_profile_reset(); + dasd_stats_all_block_reset(); pr_info("The statistics have been reset\n"); } else - goto out_error; - kfree(buffer); + goto out_parse_error; + vfree(buffer); return user_len; -out_error: +out_parse_error: + rc = -EINVAL; pr_warning("%s is not a supported value for /proc/dasd/statistics\n", str); - kfree(buffer); - return -EINVAL; +out_error: + vfree(buffer); + return rc; #else pr_warning("/proc/dasd/statistics: is not activated in this kernel\n"); return user_len; diff --git a/drivers/s390/block/dcssblk.c b/drivers/s390/block/dcssblk.c index 9b43ae94beb..0f471750327 100644 --- a/drivers/s390/block/dcssblk.c +++ b/drivers/s390/block/dcssblk.c @@ -26,8 +26,8 @@ #define DCSS_BUS_ID_SIZE 20 static int dcssblk_open(struct block_device *bdev, fmode_t mode); -static int dcssblk_release(struct gendisk *disk, fmode_t mode); -static int dcssblk_make_request(struct request_queue *q, struct bio *bio); +static void dcssblk_release(struct gendisk *disk, fmode_t mode); +static void dcssblk_make_request(struct request_queue *q, struct bio *bio); static int dcssblk_direct_access(struct block_device *bdev, sector_t secnum, void **kaddr, unsigned long *pfn); @@ -69,23 +69,9 @@ static ssize_t dcssblk_add_store(struct device * dev, struct device_attribute *a size_t count); static ssize_t dcssblk_remove_store(struct device * dev, struct device_attribute *attr, const char * buf, size_t count); -static ssize_t dcssblk_save_store(struct device * dev, struct device_attribute *attr, const char * buf, - size_t count); -static ssize_t dcssblk_save_show(struct device *dev, struct device_attribute *attr, char *buf); -static ssize_t dcssblk_shared_store(struct device * dev, struct device_attribute *attr, const char * buf, - size_t count); -static ssize_t dcssblk_shared_show(struct device *dev, struct device_attribute *attr, char *buf); -static ssize_t dcssblk_seglist_show(struct device *dev, - struct device_attribute *attr, - char *buf); static DEVICE_ATTR(add, S_IWUSR, NULL, dcssblk_add_store); static DEVICE_ATTR(remove, S_IWUSR, NULL, dcssblk_remove_store); -static DEVICE_ATTR(save, S_IWUSR | S_IRUSR, dcssblk_save_show, - dcssblk_save_store); -static DEVICE_ATTR(shared, S_IWUSR | S_IRUSR, dcssblk_shared_show, - dcssblk_shared_store); -static DEVICE_ATTR(seglist, S_IRUSR, dcssblk_seglist_show, NULL); static struct device *dcssblk_root_dev; @@ -318,12 +304,6 @@ dcssblk_load_segment(char *name, struct segment_info **seg_info) return rc; } -static void dcssblk_unregister_callback(struct device *dev) -{ - device_unregister(dev); - put_device(dev); -} - /* * device attribute for switching shared/nonshared (exclusive) * operation (show + store) @@ -411,11 +391,19 @@ removeseg: blk_cleanup_queue(dev_info->dcssblk_queue); dev_info->gd->queue = NULL; put_disk(dev_info->gd); - rc = device_schedule_callback(dev, dcssblk_unregister_callback); + up_write(&dcssblk_devices_sem); + + if (device_remove_file_self(dev, attr)) { + device_unregister(dev); + put_device(dev); + } + return rc; out: up_write(&dcssblk_devices_sem); return rc; } +static DEVICE_ATTR(shared, S_IWUSR | S_IRUSR, dcssblk_shared_show, + dcssblk_shared_store); /* * device attribute for save operation on current copy @@ -476,6 +464,8 @@ dcssblk_save_store(struct device *dev, struct device_attribute *attr, const char up_write(&dcssblk_devices_sem); return count; } +static DEVICE_ATTR(save, S_IWUSR | S_IRUSR, dcssblk_save_show, + dcssblk_save_store); /* * device attribute for showing all segments in a device @@ -502,6 +492,21 @@ dcssblk_seglist_show(struct device *dev, struct device_attribute *attr, up_read(&dcssblk_devices_sem); return i; } +static DEVICE_ATTR(seglist, S_IRUSR, dcssblk_seglist_show, NULL); + +static struct attribute *dcssblk_dev_attrs[] = { + &dev_attr_shared.attr, + &dev_attr_save.attr, + &dev_attr_seglist.attr, + NULL, +}; +static struct attribute_group dcssblk_dev_attr_group = { + .attrs = dcssblk_dev_attrs, +}; +static const struct attribute_group *dcssblk_dev_attr_groups[] = { + &dcssblk_dev_attr_group, + NULL, +}; /* * device attribute for adding devices @@ -588,8 +593,9 @@ dcssblk_add_store(struct device *dev, struct device_attribute *attr, const char dev_info->start = dcssblk_find_lowest_addr(dev_info); dev_info->end = dcssblk_find_highest_addr(dev_info); - dev_set_name(&dev_info->dev, dev_info->segment_name); + dev_set_name(&dev_info->dev, "%s", dev_info->segment_name); dev_info->dev.release = dcssblk_release_segment; + dev_info->dev.groups = dcssblk_dev_attr_groups; INIT_LIST_HEAD(&dev_info->lh); dev_info->gd = alloc_disk(DCSSBLK_MINORS_PER_DISK); if (dev_info->gd == NULL) { @@ -637,21 +643,10 @@ dcssblk_add_store(struct device *dev, struct device_attribute *attr, const char * register the device */ rc = device_register(&dev_info->dev); - if (rc) { - module_put(THIS_MODULE); - goto dev_list_del; - } - get_device(&dev_info->dev); - rc = device_create_file(&dev_info->dev, &dev_attr_shared); - if (rc) - goto unregister_dev; - rc = device_create_file(&dev_info->dev, &dev_attr_save); - if (rc) - goto unregister_dev; - rc = device_create_file(&dev_info->dev, &dev_attr_seglist); if (rc) - goto unregister_dev; + goto put_dev; + get_device(&dev_info->dev); add_disk(dev_info->gd); switch (dev_info->segment_type) { @@ -668,12 +663,11 @@ dcssblk_add_store(struct device *dev, struct device_attribute *attr, const char rc = count; goto out; -unregister_dev: +put_dev: list_del(&dev_info->lh); blk_cleanup_queue(dev_info->dcssblk_queue); dev_info->gd->queue = NULL; put_disk(dev_info->gd); - device_unregister(&dev_info->dev); list_for_each_entry(seg_info, &dev_info->seg_list, lh) { segment_unload(seg_info->segment_name); } @@ -787,16 +781,15 @@ out: return rc; } -static int +static void dcssblk_release(struct gendisk *disk, fmode_t mode) { struct dcssblk_dev_info *dev_info = disk->private_data; struct segment_info *entry; - int rc; if (!dev_info) { - rc = -ENODEV; - goto out; + WARN_ON(1); + return; } down_write(&dcssblk_devices_sem); if (atomic_dec_and_test(&dev_info->use_count) @@ -809,31 +802,28 @@ dcssblk_release(struct gendisk *disk, fmode_t mode) dev_info->save_pending = 0; } up_write(&dcssblk_devices_sem); - rc = 0; -out: - return rc; } -static int +static void dcssblk_make_request(struct request_queue *q, struct bio *bio) { struct dcssblk_dev_info *dev_info; - struct bio_vec *bvec; + struct bio_vec bvec; + struct bvec_iter iter; unsigned long index; unsigned long page_addr; unsigned long source_addr; unsigned long bytes_done; - int i; bytes_done = 0; dev_info = bio->bi_bdev->bd_disk->private_data; if (dev_info == NULL) goto fail; - if ((bio->bi_sector & 7) != 0 || (bio->bi_size & 4095) != 0) + if ((bio->bi_iter.bi_sector & 7) != 0 || + (bio->bi_iter.bi_size & 4095) != 0) /* Request is not page-aligned. */ goto fail; - if (((bio->bi_size >> 9) + bio->bi_sector) - > get_capacity(bio->bi_bdev->bd_disk)) { + if (bio_end_sector(bio) > get_capacity(bio->bi_bdev->bd_disk)) { /* Request beyond end of DCSS segment. */ goto fail; } @@ -853,28 +843,27 @@ dcssblk_make_request(struct request_queue *q, struct bio *bio) } } - index = (bio->bi_sector >> 3); - bio_for_each_segment(bvec, bio, i) { + index = (bio->bi_iter.bi_sector >> 3); + bio_for_each_segment(bvec, bio, iter) { page_addr = (unsigned long) - page_address(bvec->bv_page) + bvec->bv_offset; + page_address(bvec.bv_page) + bvec.bv_offset; source_addr = dev_info->start + (index<<12) + bytes_done; - if (unlikely((page_addr & 4095) != 0) || (bvec->bv_len & 4095) != 0) + if (unlikely((page_addr & 4095) != 0) || (bvec.bv_len & 4095) != 0) // More paranoia. goto fail; if (bio_data_dir(bio) == READ) { memcpy((void*)page_addr, (void*)source_addr, - bvec->bv_len); + bvec.bv_len); } else { memcpy((void*)source_addr, (void*)page_addr, - bvec->bv_len); + bvec.bv_len); } - bytes_done += bvec->bv_len; + bytes_done += bvec.bv_len; } bio_endio(bio, 0); - return 0; + return; fail: bio_io_error(bio); - return 0; } static int diff --git a/drivers/s390/block/scm_blk.c b/drivers/s390/block/scm_blk.c new file mode 100644 index 00000000000..76bed1743db --- /dev/null +++ b/drivers/s390/block/scm_blk.c @@ -0,0 +1,493 @@ +/* + * Block driver for s390 storage class memory. + * + * Copyright IBM Corp. 2012 + * Author(s): Sebastian Ott <sebott@linux.vnet.ibm.com> + */ + +#define KMSG_COMPONENT "scm_block" +#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt + +#include <linux/interrupt.h> +#include <linux/spinlock.h> +#include <linux/module.h> +#include <linux/blkdev.h> +#include <linux/genhd.h> +#include <linux/slab.h> +#include <linux/list.h> +#include <asm/eadm.h> +#include "scm_blk.h" + +debug_info_t *scm_debug; +static int scm_major; +static DEFINE_SPINLOCK(list_lock); +static LIST_HEAD(inactive_requests); +static unsigned int nr_requests = 64; +static atomic_t nr_devices = ATOMIC_INIT(0); +module_param(nr_requests, uint, S_IRUGO); +MODULE_PARM_DESC(nr_requests, "Number of parallel requests."); + +MODULE_DESCRIPTION("Block driver for s390 storage class memory."); +MODULE_LICENSE("GPL"); +MODULE_ALIAS("scm:scmdev*"); + +static void __scm_free_rq(struct scm_request *scmrq) +{ + struct aob_rq_header *aobrq = to_aobrq(scmrq); + + free_page((unsigned long) scmrq->aob); + free_page((unsigned long) scmrq->aidaw); + __scm_free_rq_cluster(scmrq); + kfree(aobrq); +} + +static void scm_free_rqs(void) +{ + struct list_head *iter, *safe; + struct scm_request *scmrq; + + spin_lock_irq(&list_lock); + list_for_each_safe(iter, safe, &inactive_requests) { + scmrq = list_entry(iter, struct scm_request, list); + list_del(&scmrq->list); + __scm_free_rq(scmrq); + } + spin_unlock_irq(&list_lock); +} + +static int __scm_alloc_rq(void) +{ + struct aob_rq_header *aobrq; + struct scm_request *scmrq; + + aobrq = kzalloc(sizeof(*aobrq) + sizeof(*scmrq), GFP_KERNEL); + if (!aobrq) + return -ENOMEM; + + scmrq = (void *) aobrq->data; + scmrq->aidaw = (void *) get_zeroed_page(GFP_DMA); + scmrq->aob = (void *) get_zeroed_page(GFP_DMA); + if (!scmrq->aob || !scmrq->aidaw) { + __scm_free_rq(scmrq); + return -ENOMEM; + } + + if (__scm_alloc_rq_cluster(scmrq)) { + __scm_free_rq(scmrq); + return -ENOMEM; + } + + INIT_LIST_HEAD(&scmrq->list); + spin_lock_irq(&list_lock); + list_add(&scmrq->list, &inactive_requests); + spin_unlock_irq(&list_lock); + + return 0; +} + +static int scm_alloc_rqs(unsigned int nrqs) +{ + int ret = 0; + + while (nrqs-- && !ret) + ret = __scm_alloc_rq(); + + return ret; +} + +static struct scm_request *scm_request_fetch(void) +{ + struct scm_request *scmrq = NULL; + + spin_lock(&list_lock); + if (list_empty(&inactive_requests)) + goto out; + scmrq = list_first_entry(&inactive_requests, struct scm_request, list); + list_del(&scmrq->list); +out: + spin_unlock(&list_lock); + return scmrq; +} + +static void scm_request_done(struct scm_request *scmrq) +{ + unsigned long flags; + + spin_lock_irqsave(&list_lock, flags); + list_add(&scmrq->list, &inactive_requests); + spin_unlock_irqrestore(&list_lock, flags); +} + +static bool scm_permit_request(struct scm_blk_dev *bdev, struct request *req) +{ + return rq_data_dir(req) != WRITE || bdev->state != SCM_WR_PROHIBIT; +} + +static void scm_request_prepare(struct scm_request *scmrq) +{ + struct scm_blk_dev *bdev = scmrq->bdev; + struct scm_device *scmdev = bdev->gendisk->private_data; + struct aidaw *aidaw = scmrq->aidaw; + struct msb *msb = &scmrq->aob->msb[0]; + struct req_iterator iter; + struct bio_vec bv; + + msb->bs = MSB_BS_4K; + scmrq->aob->request.msb_count = 1; + msb->scm_addr = scmdev->address + + ((u64) blk_rq_pos(scmrq->request) << 9); + msb->oc = (rq_data_dir(scmrq->request) == READ) ? + MSB_OC_READ : MSB_OC_WRITE; + msb->flags |= MSB_FLAG_IDA; + msb->data_addr = (u64) aidaw; + + rq_for_each_segment(bv, scmrq->request, iter) { + WARN_ON(bv.bv_offset); + msb->blk_count += bv.bv_len >> 12; + aidaw->data_addr = (u64) page_address(bv.bv_page); + aidaw++; + } +} + +static inline void scm_request_init(struct scm_blk_dev *bdev, + struct scm_request *scmrq, + struct request *req) +{ + struct aob_rq_header *aobrq = to_aobrq(scmrq); + struct aob *aob = scmrq->aob; + + memset(aob, 0, sizeof(*aob)); + memset(scmrq->aidaw, 0, PAGE_SIZE); + aobrq->scmdev = bdev->scmdev; + aob->request.cmd_code = ARQB_CMD_MOVE; + aob->request.data = (u64) aobrq; + scmrq->request = req; + scmrq->bdev = bdev; + scmrq->retries = 4; + scmrq->error = 0; + scm_request_cluster_init(scmrq); +} + +static void scm_ensure_queue_restart(struct scm_blk_dev *bdev) +{ + if (atomic_read(&bdev->queued_reqs)) { + /* Queue restart is triggered by the next interrupt. */ + return; + } + blk_delay_queue(bdev->rq, SCM_QUEUE_DELAY); +} + +void scm_request_requeue(struct scm_request *scmrq) +{ + struct scm_blk_dev *bdev = scmrq->bdev; + + scm_release_cluster(scmrq); + blk_requeue_request(bdev->rq, scmrq->request); + atomic_dec(&bdev->queued_reqs); + scm_request_done(scmrq); + scm_ensure_queue_restart(bdev); +} + +void scm_request_finish(struct scm_request *scmrq) +{ + struct scm_blk_dev *bdev = scmrq->bdev; + + scm_release_cluster(scmrq); + blk_end_request_all(scmrq->request, scmrq->error); + atomic_dec(&bdev->queued_reqs); + scm_request_done(scmrq); +} + +static void scm_blk_request(struct request_queue *rq) +{ + struct scm_device *scmdev = rq->queuedata; + struct scm_blk_dev *bdev = dev_get_drvdata(&scmdev->dev); + struct scm_request *scmrq; + struct request *req; + int ret; + + while ((req = blk_peek_request(rq))) { + if (req->cmd_type != REQ_TYPE_FS) { + blk_start_request(req); + blk_dump_rq_flags(req, KMSG_COMPONENT " bad request"); + blk_end_request_all(req, -EIO); + continue; + } + + if (!scm_permit_request(bdev, req)) { + scm_ensure_queue_restart(bdev); + return; + } + scmrq = scm_request_fetch(); + if (!scmrq) { + SCM_LOG(5, "no request"); + scm_ensure_queue_restart(bdev); + return; + } + scm_request_init(bdev, scmrq, req); + if (!scm_reserve_cluster(scmrq)) { + SCM_LOG(5, "cluster busy"); + scm_request_done(scmrq); + return; + } + if (scm_need_cluster_request(scmrq)) { + atomic_inc(&bdev->queued_reqs); + blk_start_request(req); + scm_initiate_cluster_request(scmrq); + return; + } + scm_request_prepare(scmrq); + atomic_inc(&bdev->queued_reqs); + blk_start_request(req); + + ret = eadm_start_aob(scmrq->aob); + if (ret) { + SCM_LOG(5, "no subchannel"); + scm_request_requeue(scmrq); + return; + } + } +} + +static void __scmrq_log_error(struct scm_request *scmrq) +{ + struct aob *aob = scmrq->aob; + + if (scmrq->error == -ETIMEDOUT) + SCM_LOG(1, "Request timeout"); + else { + SCM_LOG(1, "Request error"); + SCM_LOG_HEX(1, &aob->response, sizeof(aob->response)); + } + if (scmrq->retries) + SCM_LOG(1, "Retry request"); + else + pr_err("An I/O operation to SCM failed with rc=%d\n", + scmrq->error); +} + +void scm_blk_irq(struct scm_device *scmdev, void *data, int error) +{ + struct scm_request *scmrq = data; + struct scm_blk_dev *bdev = scmrq->bdev; + + scmrq->error = error; + if (error) + __scmrq_log_error(scmrq); + + spin_lock(&bdev->lock); + list_add_tail(&scmrq->list, &bdev->finished_requests); + spin_unlock(&bdev->lock); + tasklet_hi_schedule(&bdev->tasklet); +} + +static void scm_blk_handle_error(struct scm_request *scmrq) +{ + struct scm_blk_dev *bdev = scmrq->bdev; + unsigned long flags; + + if (scmrq->error != -EIO) + goto restart; + + /* For -EIO the response block is valid. */ + switch (scmrq->aob->response.eqc) { + case EQC_WR_PROHIBIT: + spin_lock_irqsave(&bdev->lock, flags); + if (bdev->state != SCM_WR_PROHIBIT) + pr_info("%lx: Write access to the SCM increment is suspended\n", + (unsigned long) bdev->scmdev->address); + bdev->state = SCM_WR_PROHIBIT; + spin_unlock_irqrestore(&bdev->lock, flags); + goto requeue; + default: + break; + } + +restart: + if (!eadm_start_aob(scmrq->aob)) + return; + +requeue: + spin_lock_irqsave(&bdev->rq_lock, flags); + scm_request_requeue(scmrq); + spin_unlock_irqrestore(&bdev->rq_lock, flags); +} + +static void scm_blk_tasklet(struct scm_blk_dev *bdev) +{ + struct scm_request *scmrq; + unsigned long flags; + + spin_lock_irqsave(&bdev->lock, flags); + while (!list_empty(&bdev->finished_requests)) { + scmrq = list_first_entry(&bdev->finished_requests, + struct scm_request, list); + list_del(&scmrq->list); + spin_unlock_irqrestore(&bdev->lock, flags); + + if (scmrq->error && scmrq->retries-- > 0) { + scm_blk_handle_error(scmrq); + + /* Request restarted or requeued, handle next. */ + spin_lock_irqsave(&bdev->lock, flags); + continue; + } + + if (scm_test_cluster_request(scmrq)) { + scm_cluster_request_irq(scmrq); + spin_lock_irqsave(&bdev->lock, flags); + continue; + } + + scm_request_finish(scmrq); + spin_lock_irqsave(&bdev->lock, flags); + } + spin_unlock_irqrestore(&bdev->lock, flags); + /* Look out for more requests. */ + blk_run_queue(bdev->rq); +} + +static const struct block_device_operations scm_blk_devops = { + .owner = THIS_MODULE, +}; + +int scm_blk_dev_setup(struct scm_blk_dev *bdev, struct scm_device *scmdev) +{ + struct request_queue *rq; + int len, ret = -ENOMEM; + unsigned int devindex, nr_max_blk; + + devindex = atomic_inc_return(&nr_devices) - 1; + /* scma..scmz + scmaa..scmzz */ + if (devindex > 701) { + ret = -ENODEV; + goto out; + } + + bdev->scmdev = scmdev; + bdev->state = SCM_OPER; + spin_lock_init(&bdev->rq_lock); + spin_lock_init(&bdev->lock); + INIT_LIST_HEAD(&bdev->finished_requests); + atomic_set(&bdev->queued_reqs, 0); + tasklet_init(&bdev->tasklet, + (void (*)(unsigned long)) scm_blk_tasklet, + (unsigned long) bdev); + + rq = blk_init_queue(scm_blk_request, &bdev->rq_lock); + if (!rq) + goto out; + + bdev->rq = rq; + nr_max_blk = min(scmdev->nr_max_block, + (unsigned int) (PAGE_SIZE / sizeof(struct aidaw))); + + blk_queue_logical_block_size(rq, 1 << 12); + blk_queue_max_hw_sectors(rq, nr_max_blk << 3); /* 8 * 512 = blk_size */ + blk_queue_max_segments(rq, nr_max_blk); + queue_flag_set_unlocked(QUEUE_FLAG_NONROT, rq); + scm_blk_dev_cluster_setup(bdev); + + bdev->gendisk = alloc_disk(SCM_NR_PARTS); + if (!bdev->gendisk) + goto out_queue; + + rq->queuedata = scmdev; + bdev->gendisk->driverfs_dev = &scmdev->dev; + bdev->gendisk->private_data = scmdev; + bdev->gendisk->fops = &scm_blk_devops; + bdev->gendisk->queue = rq; + bdev->gendisk->major = scm_major; + bdev->gendisk->first_minor = devindex * SCM_NR_PARTS; + + len = snprintf(bdev->gendisk->disk_name, DISK_NAME_LEN, "scm"); + if (devindex > 25) { + len += snprintf(bdev->gendisk->disk_name + len, + DISK_NAME_LEN - len, "%c", + 'a' + (devindex / 26) - 1); + devindex = devindex % 26; + } + snprintf(bdev->gendisk->disk_name + len, DISK_NAME_LEN - len, "%c", + 'a' + devindex); + + /* 512 byte sectors */ + set_capacity(bdev->gendisk, scmdev->size >> 9); + add_disk(bdev->gendisk); + return 0; + +out_queue: + blk_cleanup_queue(rq); +out: + atomic_dec(&nr_devices); + return ret; +} + +void scm_blk_dev_cleanup(struct scm_blk_dev *bdev) +{ + tasklet_kill(&bdev->tasklet); + del_gendisk(bdev->gendisk); + blk_cleanup_queue(bdev->gendisk->queue); + put_disk(bdev->gendisk); +} + +void scm_blk_set_available(struct scm_blk_dev *bdev) +{ + unsigned long flags; + + spin_lock_irqsave(&bdev->lock, flags); + if (bdev->state == SCM_WR_PROHIBIT) + pr_info("%lx: Write access to the SCM increment is restored\n", + (unsigned long) bdev->scmdev->address); + bdev->state = SCM_OPER; + spin_unlock_irqrestore(&bdev->lock, flags); +} + +static int __init scm_blk_init(void) +{ + int ret = -EINVAL; + + if (!scm_cluster_size_valid()) + goto out; + + ret = register_blkdev(0, "scm"); + if (ret < 0) + goto out; + + scm_major = ret; + ret = scm_alloc_rqs(nr_requests); + if (ret) + goto out_free; + + scm_debug = debug_register("scm_log", 16, 1, 16); + if (!scm_debug) { + ret = -ENOMEM; + goto out_free; + } + + debug_register_view(scm_debug, &debug_hex_ascii_view); + debug_set_level(scm_debug, 2); + + ret = scm_drv_init(); + if (ret) + goto out_dbf; + + return ret; + +out_dbf: + debug_unregister(scm_debug); +out_free: + scm_free_rqs(); + unregister_blkdev(scm_major, "scm"); +out: + return ret; +} +module_init(scm_blk_init); + +static void __exit scm_blk_cleanup(void) +{ + scm_drv_cleanup(); + debug_unregister(scm_debug); + scm_free_rqs(); + unregister_blkdev(scm_major, "scm"); +} +module_exit(scm_blk_cleanup); diff --git a/drivers/s390/block/scm_blk.h b/drivers/s390/block/scm_blk.h new file mode 100644 index 00000000000..e59331e6c2e --- /dev/null +++ b/drivers/s390/block/scm_blk.h @@ -0,0 +1,134 @@ +#ifndef SCM_BLK_H +#define SCM_BLK_H + +#include <linux/interrupt.h> +#include <linux/spinlock.h> +#include <linux/blkdev.h> +#include <linux/genhd.h> +#include <linux/list.h> + +#include <asm/debug.h> +#include <asm/eadm.h> + +#define SCM_NR_PARTS 8 +#define SCM_QUEUE_DELAY 5 + +struct scm_blk_dev { + struct tasklet_struct tasklet; + struct request_queue *rq; + struct gendisk *gendisk; + struct scm_device *scmdev; + spinlock_t rq_lock; /* guard the request queue */ + spinlock_t lock; /* guard the rest of the blockdev */ + atomic_t queued_reqs; + enum {SCM_OPER, SCM_WR_PROHIBIT} state; + struct list_head finished_requests; +#ifdef CONFIG_SCM_BLOCK_CLUSTER_WRITE + struct list_head cluster_list; +#endif +}; + +struct scm_request { + struct scm_blk_dev *bdev; + struct request *request; + struct aidaw *aidaw; + struct aob *aob; + struct list_head list; + u8 retries; + int error; +#ifdef CONFIG_SCM_BLOCK_CLUSTER_WRITE + struct { + enum {CLUSTER_NONE, CLUSTER_READ, CLUSTER_WRITE} state; + struct list_head list; + void **buf; + } cluster; +#endif +}; + +#define to_aobrq(rq) container_of((void *) rq, struct aob_rq_header, data) + +int scm_blk_dev_setup(struct scm_blk_dev *, struct scm_device *); +void scm_blk_dev_cleanup(struct scm_blk_dev *); +void scm_blk_set_available(struct scm_blk_dev *); +void scm_blk_irq(struct scm_device *, void *, int); + +void scm_request_finish(struct scm_request *); +void scm_request_requeue(struct scm_request *); + +int scm_drv_init(void); +void scm_drv_cleanup(void); + +#ifdef CONFIG_SCM_BLOCK_CLUSTER_WRITE +void __scm_free_rq_cluster(struct scm_request *); +int __scm_alloc_rq_cluster(struct scm_request *); +void scm_request_cluster_init(struct scm_request *); +bool scm_reserve_cluster(struct scm_request *); +void scm_release_cluster(struct scm_request *); +void scm_blk_dev_cluster_setup(struct scm_blk_dev *); +bool scm_need_cluster_request(struct scm_request *); +void scm_initiate_cluster_request(struct scm_request *); +void scm_cluster_request_irq(struct scm_request *); +bool scm_test_cluster_request(struct scm_request *); +bool scm_cluster_size_valid(void); +#else /* CONFIG_SCM_BLOCK_CLUSTER_WRITE */ +static inline void __scm_free_rq_cluster(struct scm_request *scmrq) {} +static inline int __scm_alloc_rq_cluster(struct scm_request *scmrq) +{ + return 0; +} +static inline void scm_request_cluster_init(struct scm_request *scmrq) {} +static inline bool scm_reserve_cluster(struct scm_request *scmrq) +{ + return true; +} +static inline void scm_release_cluster(struct scm_request *scmrq) {} +static inline void scm_blk_dev_cluster_setup(struct scm_blk_dev *bdev) {} +static inline bool scm_need_cluster_request(struct scm_request *scmrq) +{ + return false; +} +static inline void scm_initiate_cluster_request(struct scm_request *scmrq) {} +static inline void scm_cluster_request_irq(struct scm_request *scmrq) {} +static inline bool scm_test_cluster_request(struct scm_request *scmrq) +{ + return false; +} +static inline bool scm_cluster_size_valid(void) +{ + return true; +} +#endif /* CONFIG_SCM_BLOCK_CLUSTER_WRITE */ + +extern debug_info_t *scm_debug; + +#define SCM_LOG(imp, txt) do { \ + debug_text_event(scm_debug, imp, txt); \ + } while (0) + +static inline void SCM_LOG_HEX(int level, void *data, int length) +{ + if (!debug_level_enabled(scm_debug, level)) + return; + while (length > 0) { + debug_event(scm_debug, level, data, length); + length -= scm_debug->buf_size; + data += scm_debug->buf_size; + } +} + +static inline void SCM_LOG_STATE(int level, struct scm_device *scmdev) +{ + struct { + u64 address; + u8 oper_state; + u8 rank; + } __packed data = { + .address = scmdev->address, + .oper_state = scmdev->attrs.oper_state, + .rank = scmdev->attrs.rank, + }; + + SCM_LOG_HEX(level, &data, sizeof(data)); +} + +#endif /* SCM_BLK_H */ diff --git a/drivers/s390/block/scm_blk_cluster.c b/drivers/s390/block/scm_blk_cluster.c new file mode 100644 index 00000000000..9aae909d47a --- /dev/null +++ b/drivers/s390/block/scm_blk_cluster.c @@ -0,0 +1,230 @@ +/* + * Block driver for s390 storage class memory. + * + * Copyright IBM Corp. 2012 + * Author(s): Sebastian Ott <sebott@linux.vnet.ibm.com> + */ + +#include <linux/spinlock.h> +#include <linux/module.h> +#include <linux/blkdev.h> +#include <linux/genhd.h> +#include <linux/slab.h> +#include <linux/list.h> +#include <asm/eadm.h> +#include "scm_blk.h" + +static unsigned int write_cluster_size = 64; +module_param(write_cluster_size, uint, S_IRUGO); +MODULE_PARM_DESC(write_cluster_size, + "Number of pages used for contiguous writes."); + +#define CLUSTER_SIZE (write_cluster_size * PAGE_SIZE) + +void __scm_free_rq_cluster(struct scm_request *scmrq) +{ + int i; + + if (!scmrq->cluster.buf) + return; + + for (i = 0; i < 2 * write_cluster_size; i++) + free_page((unsigned long) scmrq->cluster.buf[i]); + + kfree(scmrq->cluster.buf); +} + +int __scm_alloc_rq_cluster(struct scm_request *scmrq) +{ + int i; + + scmrq->cluster.buf = kzalloc(sizeof(void *) * 2 * write_cluster_size, + GFP_KERNEL); + if (!scmrq->cluster.buf) + return -ENOMEM; + + for (i = 0; i < 2 * write_cluster_size; i++) { + scmrq->cluster.buf[i] = (void *) get_zeroed_page(GFP_DMA); + if (!scmrq->cluster.buf[i]) + return -ENOMEM; + } + INIT_LIST_HEAD(&scmrq->cluster.list); + return 0; +} + +void scm_request_cluster_init(struct scm_request *scmrq) +{ + scmrq->cluster.state = CLUSTER_NONE; +} + +static bool clusters_intersect(struct scm_request *A, struct scm_request *B) +{ + unsigned long firstA, lastA, firstB, lastB; + + firstA = ((u64) blk_rq_pos(A->request) << 9) / CLUSTER_SIZE; + lastA = (((u64) blk_rq_pos(A->request) << 9) + + blk_rq_bytes(A->request) - 1) / CLUSTER_SIZE; + + firstB = ((u64) blk_rq_pos(B->request) << 9) / CLUSTER_SIZE; + lastB = (((u64) blk_rq_pos(B->request) << 9) + + blk_rq_bytes(B->request) - 1) / CLUSTER_SIZE; + + return (firstB <= lastA && firstA <= lastB); +} + +bool scm_reserve_cluster(struct scm_request *scmrq) +{ + struct scm_blk_dev *bdev = scmrq->bdev; + struct scm_request *iter; + + if (write_cluster_size == 0) + return true; + + spin_lock(&bdev->lock); + list_for_each_entry(iter, &bdev->cluster_list, cluster.list) { + if (clusters_intersect(scmrq, iter) && + (rq_data_dir(scmrq->request) == WRITE || + rq_data_dir(iter->request) == WRITE)) { + spin_unlock(&bdev->lock); + return false; + } + } + list_add(&scmrq->cluster.list, &bdev->cluster_list); + spin_unlock(&bdev->lock); + + return true; +} + +void scm_release_cluster(struct scm_request *scmrq) +{ + struct scm_blk_dev *bdev = scmrq->bdev; + unsigned long flags; + + if (write_cluster_size == 0) + return; + + spin_lock_irqsave(&bdev->lock, flags); + list_del(&scmrq->cluster.list); + spin_unlock_irqrestore(&bdev->lock, flags); +} + +void scm_blk_dev_cluster_setup(struct scm_blk_dev *bdev) +{ + INIT_LIST_HEAD(&bdev->cluster_list); + blk_queue_io_opt(bdev->rq, CLUSTER_SIZE); +} + +static void scm_prepare_cluster_request(struct scm_request *scmrq) +{ + struct scm_blk_dev *bdev = scmrq->bdev; + struct scm_device *scmdev = bdev->gendisk->private_data; + struct request *req = scmrq->request; + struct aidaw *aidaw = scmrq->aidaw; + struct msb *msb = &scmrq->aob->msb[0]; + struct req_iterator iter; + struct bio_vec bv; + int i = 0; + u64 addr; + + switch (scmrq->cluster.state) { + case CLUSTER_NONE: + scmrq->cluster.state = CLUSTER_READ; + /* fall through */ + case CLUSTER_READ: + scmrq->aob->request.msb_count = 1; + msb->bs = MSB_BS_4K; + msb->oc = MSB_OC_READ; + msb->flags = MSB_FLAG_IDA; + msb->data_addr = (u64) aidaw; + msb->blk_count = write_cluster_size; + + addr = scmdev->address + ((u64) blk_rq_pos(req) << 9); + msb->scm_addr = round_down(addr, CLUSTER_SIZE); + + if (msb->scm_addr != + round_down(addr + (u64) blk_rq_bytes(req) - 1, + CLUSTER_SIZE)) + msb->blk_count = 2 * write_cluster_size; + + for (i = 0; i < msb->blk_count; i++) { + aidaw->data_addr = (u64) scmrq->cluster.buf[i]; + aidaw++; + } + + break; + case CLUSTER_WRITE: + msb->oc = MSB_OC_WRITE; + + for (addr = msb->scm_addr; + addr < scmdev->address + ((u64) blk_rq_pos(req) << 9); + addr += PAGE_SIZE) { + aidaw->data_addr = (u64) scmrq->cluster.buf[i]; + aidaw++; + i++; + } + rq_for_each_segment(bv, req, iter) { + aidaw->data_addr = (u64) page_address(bv.bv_page); + aidaw++; + i++; + } + for (; i < msb->blk_count; i++) { + aidaw->data_addr = (u64) scmrq->cluster.buf[i]; + aidaw++; + } + break; + } +} + +bool scm_need_cluster_request(struct scm_request *scmrq) +{ + if (rq_data_dir(scmrq->request) == READ) + return false; + + return blk_rq_bytes(scmrq->request) < CLUSTER_SIZE; +} + +/* Called with queue lock held. */ +void scm_initiate_cluster_request(struct scm_request *scmrq) +{ + scm_prepare_cluster_request(scmrq); + if (eadm_start_aob(scmrq->aob)) + scm_request_requeue(scmrq); +} + +bool scm_test_cluster_request(struct scm_request *scmrq) +{ + return scmrq->cluster.state != CLUSTER_NONE; +} + +void scm_cluster_request_irq(struct scm_request *scmrq) +{ + struct scm_blk_dev *bdev = scmrq->bdev; + unsigned long flags; + + switch (scmrq->cluster.state) { + case CLUSTER_NONE: + BUG(); + break; + case CLUSTER_READ: + if (scmrq->error) { + scm_request_finish(scmrq); + break; + } + scmrq->cluster.state = CLUSTER_WRITE; + spin_lock_irqsave(&bdev->rq_lock, flags); + scm_initiate_cluster_request(scmrq); + spin_unlock_irqrestore(&bdev->rq_lock, flags); + break; + case CLUSTER_WRITE: + scm_request_finish(scmrq); + break; + } +} + +bool scm_cluster_size_valid(void) +{ + if (write_cluster_size == 1 || write_cluster_size > 128) + return false; + + return !(write_cluster_size & (write_cluster_size - 1)); +} diff --git a/drivers/s390/block/scm_drv.c b/drivers/s390/block/scm_drv.c new file mode 100644 index 00000000000..c98cf52d78d --- /dev/null +++ b/drivers/s390/block/scm_drv.c @@ -0,0 +1,92 @@ +/* + * Device driver for s390 storage class memory. + * + * Copyright IBM Corp. 2012 + * Author(s): Sebastian Ott <sebott@linux.vnet.ibm.com> + */ + +#define KMSG_COMPONENT "scm_block" +#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt + +#include <linux/module.h> +#include <linux/slab.h> +#include <asm/eadm.h> +#include "scm_blk.h" + +static void scm_notify(struct scm_device *scmdev, enum scm_event event) +{ + struct scm_blk_dev *bdev = dev_get_drvdata(&scmdev->dev); + + switch (event) { + case SCM_CHANGE: + pr_info("%lx: The capabilities of the SCM increment changed\n", + (unsigned long) scmdev->address); + SCM_LOG(2, "State changed"); + SCM_LOG_STATE(2, scmdev); + break; + case SCM_AVAIL: + SCM_LOG(2, "Increment available"); + SCM_LOG_STATE(2, scmdev); + scm_blk_set_available(bdev); + break; + } +} + +static int scm_probe(struct scm_device *scmdev) +{ + struct scm_blk_dev *bdev; + int ret; + + SCM_LOG(2, "probe"); + SCM_LOG_STATE(2, scmdev); + + if (scmdev->attrs.oper_state != OP_STATE_GOOD) + return -EINVAL; + + bdev = kzalloc(sizeof(*bdev), GFP_KERNEL); + if (!bdev) + return -ENOMEM; + + dev_set_drvdata(&scmdev->dev, bdev); + ret = scm_blk_dev_setup(bdev, scmdev); + if (ret) { + dev_set_drvdata(&scmdev->dev, NULL); + kfree(bdev); + goto out; + } + +out: + return ret; +} + +static int scm_remove(struct scm_device *scmdev) +{ + struct scm_blk_dev *bdev = dev_get_drvdata(&scmdev->dev); + + scm_blk_dev_cleanup(bdev); + dev_set_drvdata(&scmdev->dev, NULL); + kfree(bdev); + + return 0; +} + +static struct scm_driver scm_drv = { + .drv = { + .name = "scm_block", + .owner = THIS_MODULE, + }, + .notify = scm_notify, + .probe = scm_probe, + .remove = scm_remove, + .handler = scm_blk_irq, +}; + +int __init scm_drv_init(void) +{ + return scm_driver_register(&scm_drv); +} + +void scm_drv_cleanup(void) +{ + scm_driver_unregister(&scm_drv); +} diff --git a/drivers/s390/block/xpram.c b/drivers/s390/block/xpram.c index c881a14fa5d..6969d39f1e2 100644 --- a/drivers/s390/block/xpram.c +++ b/drivers/s390/block/xpram.c @@ -36,7 +36,7 @@ #include <linux/blkdev.h> #include <linux/blkpg.h> #include <linux/hdreg.h> /* HDIO_GETGEO */ -#include <linux/sysdev.h> +#include <linux/device.h> #include <linux/bio.h> #include <linux/suspend.h> #include <linux/platform_device.h> @@ -62,8 +62,8 @@ static int xpram_devs; /* * Parameter parsing functions. */ -static int __initdata devs = XPRAM_DEVS; -static char __initdata *sizes[XPRAM_MAX_DEVS]; +static int devs = XPRAM_DEVS; +static char *sizes[XPRAM_MAX_DEVS]; module_param(devs, int, 0); module_param_array(sizes, charp, NULL, 0); @@ -181,28 +181,29 @@ static unsigned long xpram_highest_page_index(void) /* * Block device make request function. */ -static int xpram_make_request(struct request_queue *q, struct bio *bio) +static void xpram_make_request(struct request_queue *q, struct bio *bio) { xpram_device_t *xdev = bio->bi_bdev->bd_disk->private_data; - struct bio_vec *bvec; + struct bio_vec bvec; + struct bvec_iter iter; unsigned int index; unsigned long page_addr; unsigned long bytes; - int i; - if ((bio->bi_sector & 7) != 0 || (bio->bi_size & 4095) != 0) + if ((bio->bi_iter.bi_sector & 7) != 0 || + (bio->bi_iter.bi_size & 4095) != 0) /* Request is not page-aligned. */ goto fail; - if ((bio->bi_size >> 12) > xdev->size) + if ((bio->bi_iter.bi_size >> 12) > xdev->size) /* Request size is no page-aligned. */ goto fail; - if ((bio->bi_sector >> 3) > 0xffffffffU - xdev->offset) + if ((bio->bi_iter.bi_sector >> 3) > 0xffffffffU - xdev->offset) goto fail; - index = (bio->bi_sector >> 3) + xdev->offset; - bio_for_each_segment(bvec, bio, i) { + index = (bio->bi_iter.bi_sector >> 3) + xdev->offset; + bio_for_each_segment(bvec, bio, iter) { page_addr = (unsigned long) - kmap(bvec->bv_page) + bvec->bv_offset; - bytes = bvec->bv_len; + kmap(bvec.bv_page) + bvec.bv_offset; + bytes = bvec.bv_len; if ((page_addr & 4095) != 0 || (bytes & 4095) != 0) /* More paranoia. */ goto fail; @@ -221,10 +222,9 @@ static int xpram_make_request(struct request_queue *q, struct bio *bio) } set_bit(BIO_UPTODATE, &bio->bi_flags); bio_endio(bio, 0); - return 0; + return; fail: bio_io_error(bio); - return 0; } static int xpram_getgeo(struct block_device *bdev, struct hd_geometry *geo) @@ -258,6 +258,7 @@ static int __init xpram_setup_sizes(unsigned long pages) unsigned long mem_needed; unsigned long mem_auto; unsigned long long size; + char *sizes_end; int mem_auto_no; int i; @@ -276,8 +277,8 @@ static int __init xpram_setup_sizes(unsigned long pages) mem_auto_no = 0; for (i = 0; i < xpram_devs; i++) { if (sizes[i]) { - size = simple_strtoull(sizes[i], &sizes[i], 0); - switch (sizes[i][0]) { + size = simple_strtoull(sizes[i], &sizes_end, 0); + switch (*sizes_end) { case 'g': case 'G': size <<= 20; @@ -344,6 +345,7 @@ static int __init xpram_setup_blkdev(void) put_disk(xpram_disks[i]); goto out; } + queue_flag_set_unlocked(QUEUE_FLAG_NONROT, xpram_queues[i]); blk_queue_make_request(xpram_queues[i], xpram_make_request); blk_queue_logical_block_size(xpram_queues[i], 4096); } |