aboutsummaryrefslogtreecommitdiff
path: root/drivers/block/aoe
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/block/aoe')
-rw-r--r--drivers/block/aoe/aoe.h78
-rw-r--r--drivers/block/aoe/aoeblk.c210
-rw-r--r--drivers/block/aoe/aoechr.c10
-rw-r--r--drivers/block/aoe/aoecmd.c986
-rw-r--r--drivers/block/aoe/aoedev.c244
-rw-r--r--drivers/block/aoe/aoemain.c2
-rw-r--r--drivers/block/aoe/aoenet.c20
7 files changed, 1100 insertions, 450 deletions
diff --git a/drivers/block/aoe/aoe.h b/drivers/block/aoe/aoe.h
index d2ed7f18d1a..9220f8e833d 100644
--- a/drivers/block/aoe/aoe.h
+++ b/drivers/block/aoe/aoe.h
@@ -1,5 +1,5 @@
-/* Copyright (c) 2012 Coraid, Inc. See COPYING for GPL terms. */
-#define VERSION "50"
+/* Copyright (c) 2013 Coraid, Inc. See COPYING for GPL terms. */
+#define VERSION "85"
#define AOE_MAJOR 152
#define DEVICE_NAME "aoe"
@@ -10,7 +10,7 @@
#define AOE_PARTITIONS (16)
#endif
-#define WHITESPACE " \t\v\f\n"
+#define WHITESPACE " \t\v\f\n,"
enum {
AOECMD_ATA,
@@ -73,45 +73,55 @@ enum {
DEVFL_TKILL = (1<<1), /* flag for timer to know when to kill self */
DEVFL_EXT = (1<<2), /* device accepts lba48 commands */
DEVFL_GDALLOC = (1<<3), /* need to alloc gendisk */
- DEVFL_KICKME = (1<<4), /* slow polling network card catch */
- DEVFL_NEWSIZE = (1<<5), /* need to update dev size in block layer */
+ DEVFL_GD_NOW = (1<<4), /* allocating gendisk */
+ DEVFL_KICKME = (1<<5), /* slow polling network card catch */
+ DEVFL_NEWSIZE = (1<<6), /* need to update dev size in block layer */
+ DEVFL_FREEING = (1<<7), /* set when device is being cleaned up */
+ DEVFL_FREED = (1<<8), /* device has been cleaned up */
};
enum {
DEFAULTBCNT = 2 * 512, /* 2 sectors */
MIN_BUFS = 16,
- NTARGETS = 8,
+ NTARGETS = 4,
NAOEIFS = 8,
NSKBPOOLMAX = 256,
NFACTIVE = 61,
TIMERTICK = HZ / 10,
- MINTIMER = HZ >> 2,
- MAXTIMER = HZ << 1,
+ RTTSCALE = 8,
+ RTTDSCALE = 3,
+ RTTAVG_INIT = USEC_PER_SEC / 4 << RTTSCALE,
+ RTTDEV_INIT = RTTAVG_INIT / 4,
+
+ HARD_SCORN_SECS = 10, /* try another remote port after this */
+ MAX_TAINT = 1000, /* cap on aoetgt taint */
};
struct buf {
ulong nframesout;
- ulong resid;
- ulong bv_resid;
- sector_t sector;
struct bio *bio;
- struct bio_vec *bv;
+ struct bvec_iter iter;
struct request *rq;
};
+enum frame_flags {
+ FFL_PROBE = 1,
+};
+
struct frame {
struct list_head head;
u32 tag;
+ struct timeval sent; /* high-res time packet was sent */
+ u32 sent_jiffs; /* low-res jiffies-based sent time */
ulong waited;
+ ulong waited_total;
struct aoetgt *t; /* parent target I belong to */
- sector_t lba;
struct sk_buff *skb; /* command skb freed on module exit */
struct sk_buff *r_skb; /* response skb for async processing */
struct buf *buf;
- struct bio_vec *bv;
- ulong bcnt;
- ulong bv_off;
+ struct bvec_iter iter;
+ char flags;
};
struct aoeif {
@@ -122,36 +132,40 @@ struct aoeif {
struct aoetgt {
unsigned char addr[6];
- ushort nframes;
+ ushort nframes; /* cap on frames to use */
struct aoedev *d; /* parent device I belong to */
struct list_head ffree; /* list of free frames */
struct aoeif ifs[NAOEIFS];
struct aoeif *ifp; /* current aoeif in use */
- ushort nout;
- ushort maxout;
- ulong falloc;
- ulong lastwadj; /* last window adjustment */
+ ushort nout; /* number of AoE commands outstanding */
+ ushort maxout; /* current value for max outstanding */
+ ushort next_cwnd; /* incr maxout after decrementing to zero */
+ ushort ssthresh; /* slow start threshold */
+ ulong falloc; /* number of allocated frames */
+ int taint; /* how much we want to avoid this aoetgt */
int minbcnt;
int wpkts, rpkts;
+ char nout_probes;
};
struct aoedev {
struct aoedev *next;
ulong sysminor;
ulong aoemajor;
+ u32 rttavg; /* scaled AoE round trip time average */
+ u32 rttdev; /* scaled round trip time mean deviation */
u16 aoeminor;
u16 flags;
u16 nopen; /* (bd_openers isn't available without sleeping) */
- u16 rttavg; /* round trip average of requests/responses */
- u16 mintimer;
u16 fw_ver; /* version of blade's firmware */
u16 lasttag; /* last tag sent */
u16 useme;
ulong ref;
struct work_struct work;/* disk create work struct */
struct gendisk *gd;
+ struct dentry *debugfs;
struct request_queue *blkq;
- struct hd_geometry geo;
+ struct hd_geometry geo;
sector_t ssize;
struct timer_list timer;
spinlock_t lock;
@@ -164,11 +178,12 @@ struct aoedev {
} ip;
ulong maxbcnt;
struct list_head factive[NFACTIVE]; /* hash of active frames */
- struct aoetgt *targets[NTARGETS];
+ struct list_head rexmitq; /* deferred retransmissions */
+ struct aoetgt **targets;
+ ulong ntargets; /* number of allocated aoetgt pointers */
struct aoetgt **tgt; /* target in use when working */
- struct aoetgt *htgt; /* target needing rexmit assistance */
- ulong ntargets;
ulong kicked;
+ char ident[512];
};
/* kthread tracking */
@@ -176,14 +191,17 @@ struct ktstate {
struct completion rendez;
struct task_struct *task;
wait_queue_head_t *waitq;
- int (*fn) (void);
- char *name;
+ int (*fn) (int);
+ char name[12];
spinlock_t *lock;
+ int id;
+ int active;
};
int aoeblk_init(void);
void aoeblk_exit(void);
void aoeblk_gdalloc(void *);
+void aoedisk_rm_debugfs(struct aoedev *d);
void aoedisk_rm_sysfs(struct aoedev *d);
int aoechr_init(void);
@@ -195,12 +213,14 @@ void aoecmd_cfg(ushort aoemajor, unsigned char aoeminor);
struct sk_buff *aoecmd_ata_rsp(struct sk_buff *);
void aoecmd_cfg_rsp(struct sk_buff *);
void aoecmd_sleepwork(struct work_struct *);
+void aoecmd_wreset(struct aoetgt *t);
void aoecmd_cleanslate(struct aoedev *);
void aoecmd_exit(void);
int aoecmd_init(void);
struct sk_buff *aoecmd_ata_id(struct aoedev *);
void aoe_freetframe(struct frame *);
void aoe_flush_iocq(void);
+void aoe_flush_iocq_by_index(int);
void aoe_end_request(struct aoedev *, struct request *, int);
int aoe_ktstart(struct ktstate *k);
void aoe_ktstop(struct ktstate *k);
diff --git a/drivers/block/aoe/aoeblk.c b/drivers/block/aoe/aoeblk.c
index 00dfc5008ad..dd73e1ff175 100644
--- a/drivers/block/aoe/aoeblk.c
+++ b/drivers/block/aoe/aoeblk.c
@@ -1,4 +1,4 @@
-/* Copyright (c) 2012 Coraid, Inc. See COPYING for GPL terms. */
+/* Copyright (c) 2013 Coraid, Inc. See COPYING for GPL terms. */
/*
* aoeblk.c
* block device routines
@@ -16,10 +16,20 @@
#include <linux/netdevice.h>
#include <linux/mutex.h>
#include <linux/export.h>
+#include <linux/moduleparam.h>
+#include <linux/debugfs.h>
+#include <scsi/sg.h>
#include "aoe.h"
static DEFINE_MUTEX(aoeblk_mutex);
static struct kmem_cache *buf_pool_cache;
+static struct dentry *aoe_debugfs_dir;
+
+/* GPFS needs a larger value than the default. */
+static int aoe_maxsectors;
+module_param(aoe_maxsectors, int, 0644);
+MODULE_PARM_DESC(aoe_maxsectors,
+ "When nonzero, set the maximum number of sectors per I/O request");
static ssize_t aoedisk_show_state(struct device *dev,
struct device_attribute *attr, char *page)
@@ -59,7 +69,7 @@ static ssize_t aoedisk_show_netif(struct device *dev,
nd = nds;
ne = nd + ARRAY_SIZE(nds);
t = d->targets;
- te = t + NTARGETS;
+ te = t + d->ntargets;
for (; t < te && *t; t++) {
ifp = (*t)->ifs;
e = ifp + NAOEIFS;
@@ -91,6 +101,63 @@ static ssize_t aoedisk_show_fwver(struct device *dev,
return snprintf(page, PAGE_SIZE, "0x%04x\n", (unsigned int) d->fw_ver);
}
+static ssize_t aoedisk_show_payload(struct device *dev,
+ struct device_attribute *attr, char *page)
+{
+ struct gendisk *disk = dev_to_disk(dev);
+ struct aoedev *d = disk->private_data;
+
+ return snprintf(page, PAGE_SIZE, "%lu\n", d->maxbcnt);
+}
+
+static int aoedisk_debugfs_show(struct seq_file *s, void *ignored)
+{
+ struct aoedev *d;
+ struct aoetgt **t, **te;
+ struct aoeif *ifp, *ife;
+ unsigned long flags;
+ char c;
+
+ d = s->private;
+ seq_printf(s, "rttavg: %d rttdev: %d\n",
+ d->rttavg >> RTTSCALE,
+ d->rttdev >> RTTDSCALE);
+ seq_printf(s, "nskbpool: %d\n", skb_queue_len(&d->skbpool));
+ seq_printf(s, "kicked: %ld\n", d->kicked);
+ seq_printf(s, "maxbcnt: %ld\n", d->maxbcnt);
+ seq_printf(s, "ref: %ld\n", d->ref);
+
+ spin_lock_irqsave(&d->lock, flags);
+ t = d->targets;
+ te = t + d->ntargets;
+ for (; t < te && *t; t++) {
+ c = '\t';
+ seq_printf(s, "falloc: %ld\n", (*t)->falloc);
+ seq_printf(s, "ffree: %p\n",
+ list_empty(&(*t)->ffree) ? NULL : (*t)->ffree.next);
+ seq_printf(s, "%pm:%d:%d:%d\n", (*t)->addr, (*t)->nout,
+ (*t)->maxout, (*t)->nframes);
+ seq_printf(s, "\tssthresh:%d\n", (*t)->ssthresh);
+ seq_printf(s, "\ttaint:%d\n", (*t)->taint);
+ seq_printf(s, "\tr:%d\n", (*t)->rpkts);
+ seq_printf(s, "\tw:%d\n", (*t)->wpkts);
+ ifp = (*t)->ifs;
+ ife = ifp + ARRAY_SIZE((*t)->ifs);
+ for (; ifp->nd && ifp < ife; ifp++) {
+ seq_printf(s, "%c%s", c, ifp->nd->name);
+ c = ',';
+ }
+ seq_puts(s, "\n");
+ }
+ spin_unlock_irqrestore(&d->lock, flags);
+
+ return 0;
+}
+
+static int aoe_debugfs_open(struct inode *inode, struct file *file)
+{
+ return single_open(file, aoedisk_debugfs_show, inode->i_private);
+}
static DEVICE_ATTR(state, S_IRUGO, aoedisk_show_state, NULL);
static DEVICE_ATTR(mac, S_IRUGO, aoedisk_show_mac, NULL);
@@ -99,12 +166,14 @@ static struct device_attribute dev_attr_firmware_version = {
.attr = { .name = "firmware-version", .mode = S_IRUGO },
.show = aoedisk_show_fwver,
};
+static DEVICE_ATTR(payload, S_IRUGO, aoedisk_show_payload, NULL);
static struct attribute *aoe_attrs[] = {
&dev_attr_state.attr,
&dev_attr_mac.attr,
&dev_attr_netif.attr,
&dev_attr_firmware_version.attr,
+ &dev_attr_payload.attr,
NULL,
};
@@ -112,6 +181,44 @@ static const struct attribute_group attr_group = {
.attrs = aoe_attrs,
};
+static const struct file_operations aoe_debugfs_fops = {
+ .open = aoe_debugfs_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = single_release,
+};
+
+static void
+aoedisk_add_debugfs(struct aoedev *d)
+{
+ struct dentry *entry;
+ char *p;
+
+ if (aoe_debugfs_dir == NULL)
+ return;
+ p = strchr(d->gd->disk_name, '/');
+ if (p == NULL)
+ p = d->gd->disk_name;
+ else
+ p++;
+ BUG_ON(*p == '\0');
+ entry = debugfs_create_file(p, 0444, aoe_debugfs_dir, d,
+ &aoe_debugfs_fops);
+ if (IS_ERR_OR_NULL(entry)) {
+ pr_info("aoe: cannot create debugfs file for %s\n",
+ d->gd->disk_name);
+ return;
+ }
+ BUG_ON(d->debugfs);
+ d->debugfs = entry;
+}
+void
+aoedisk_rm_debugfs(struct aoedev *d)
+{
+ debugfs_remove(d->debugfs);
+ d->debugfs = NULL;
+}
+
static int
aoedisk_add_sysfs(struct aoedev *d)
{
@@ -129,9 +236,18 @@ aoeblk_open(struct block_device *bdev, fmode_t mode)
struct aoedev *d = bdev->bd_disk->private_data;
ulong flags;
+ if (!virt_addr_valid(d)) {
+ pr_crit("aoe: invalid device pointer in %s\n",
+ __func__);
+ WARN_ON(1);
+ return -ENODEV;
+ }
+ if (!(d->flags & DEVFL_UP) || d->flags & DEVFL_TKILL)
+ return -ENODEV;
+
mutex_lock(&aoeblk_mutex);
spin_lock_irqsave(&d->lock, flags);
- if (d->flags & DEVFL_UP) {
+ if (d->flags & DEVFL_UP && !(d->flags & DEVFL_TKILL)) {
d->nopen++;
spin_unlock_irqrestore(&d->lock, flags);
mutex_unlock(&aoeblk_mutex);
@@ -142,7 +258,7 @@ aoeblk_open(struct block_device *bdev, fmode_t mode)
return -ENODEV;
}
-static int
+static void
aoeblk_release(struct gendisk *disk, fmode_t mode)
{
struct aoedev *d = disk->private_data;
@@ -153,11 +269,9 @@ aoeblk_release(struct gendisk *disk, fmode_t mode)
if (--d->nopen == 0) {
spin_unlock_irqrestore(&d->lock, flags);
aoecmd_cfg(d->aoemajor, d->aoeminor);
- return 0;
+ return;
}
spin_unlock_irqrestore(&d->lock, flags);
-
- return 0;
}
static void
@@ -195,9 +309,38 @@ aoeblk_getgeo(struct block_device *bdev, struct hd_geometry *geo)
return 0;
}
+static int
+aoeblk_ioctl(struct block_device *bdev, fmode_t mode, uint cmd, ulong arg)
+{
+ struct aoedev *d;
+
+ if (!arg)
+ return -EINVAL;
+
+ d = bdev->bd_disk->private_data;
+ if ((d->flags & DEVFL_UP) == 0) {
+ pr_err("aoe: disk not up\n");
+ return -ENODEV;
+ }
+
+ if (cmd == HDIO_GET_IDENTITY) {
+ if (!copy_to_user((void __user *) arg, &d->ident,
+ sizeof(d->ident)))
+ return 0;
+ return -EFAULT;
+ }
+
+ /* udev calls scsi_id, which uses SG_IO, resulting in noise */
+ if (cmd != SG_IO)
+ pr_info("aoe: unknown ioctl 0x%x\n", cmd);
+
+ return -ENOTTY;
+}
+
static const struct block_device_operations aoe_bdops = {
.open = aoeblk_open,
.release = aoeblk_release,
+ .ioctl = aoeblk_ioctl,
.getgeo = aoeblk_getgeo,
.owner = THIS_MODULE,
};
@@ -212,6 +355,18 @@ aoeblk_gdalloc(void *vp)
struct request_queue *q;
enum { KB = 1024, MB = KB * KB, READ_AHEAD = 2 * MB, };
ulong flags;
+ int late = 0;
+
+ spin_lock_irqsave(&d->lock, flags);
+ if (d->flags & DEVFL_GDALLOC
+ && !(d->flags & DEVFL_TKILL)
+ && !(d->flags & DEVFL_GD_NOW))
+ d->flags |= DEVFL_GD_NOW;
+ else
+ late = 1;
+ spin_unlock_irqrestore(&d->lock, flags);
+ if (late)
+ return;
gd = alloc_disk(AOE_PARTITIONS);
if (gd == NULL) {
@@ -231,23 +386,24 @@ aoeblk_gdalloc(void *vp)
if (q == NULL) {
pr_err("aoe: cannot allocate block queue for %ld.%d\n",
d->aoemajor, d->aoeminor);
- mempool_destroy(mp);
- goto err_disk;
+ goto err_mempool;
}
- d->blkq = blk_alloc_queue(GFP_KERNEL);
- if (!d->blkq)
- goto err_mempool;
- d->blkq->backing_dev_info.name = "aoe";
- if (bdi_init(&d->blkq->backing_dev_info))
- goto err_blkq;
spin_lock_irqsave(&d->lock, flags);
- blk_queue_max_hw_sectors(d->blkq, BLK_DEF_MAX_SECTORS);
+ WARN_ON(!(d->flags & DEVFL_GD_NOW));
+ WARN_ON(!(d->flags & DEVFL_GDALLOC));
+ WARN_ON(d->flags & DEVFL_TKILL);
+ WARN_ON(d->gd);
+ WARN_ON(d->flags & DEVFL_UP);
+ blk_queue_max_hw_sectors(q, BLK_DEF_MAX_SECTORS);
+ q->backing_dev_info.name = "aoe";
q->backing_dev_info.ra_pages = READ_AHEAD / PAGE_CACHE_SIZE;
d->bufpool = mp;
d->blkq = gd->queue = q;
q->queuedata = d;
d->gd = gd;
+ if (aoe_maxsectors)
+ blk_queue_max_hw_sectors(q, aoe_maxsectors);
gd->major = AOE_MAJOR;
gd->first_minor = d->sysminor;
gd->fops = &aoe_bdops;
@@ -263,24 +419,30 @@ aoeblk_gdalloc(void *vp)
add_disk(gd);
aoedisk_add_sysfs(d);
+ aoedisk_add_debugfs(d);
+
+ spin_lock_irqsave(&d->lock, flags);
+ WARN_ON(!(d->flags & DEVFL_GD_NOW));
+ d->flags &= ~DEVFL_GD_NOW;
+ spin_unlock_irqrestore(&d->lock, flags);
return;
-err_blkq:
- blk_cleanup_queue(d->blkq);
- d->blkq = NULL;
err_mempool:
- mempool_destroy(d->bufpool);
+ mempool_destroy(mp);
err_disk:
put_disk(gd);
err:
spin_lock_irqsave(&d->lock, flags);
- d->flags &= ~DEVFL_GDALLOC;
+ d->flags &= ~DEVFL_GD_NOW;
+ schedule_work(&d->work);
spin_unlock_irqrestore(&d->lock, flags);
}
void
aoeblk_exit(void)
{
+ debugfs_remove_recursive(aoe_debugfs_dir);
+ aoe_debugfs_dir = NULL;
kmem_cache_destroy(buf_pool_cache);
}
@@ -292,7 +454,11 @@ aoeblk_init(void)
0, 0, NULL);
if (buf_pool_cache == NULL)
return -ENOMEM;
-
+ aoe_debugfs_dir = debugfs_create_dir("aoe", NULL);
+ if (IS_ERR_OR_NULL(aoe_debugfs_dir)) {
+ pr_info("aoe: cannot create debugfs directory\n");
+ aoe_debugfs_dir = NULL;
+ }
return 0;
}
diff --git a/drivers/block/aoe/aoechr.c b/drivers/block/aoe/aoechr.c
index ed57a890c64..ab41be625a5 100644
--- a/drivers/block/aoe/aoechr.c
+++ b/drivers/block/aoe/aoechr.c
@@ -39,6 +39,11 @@ struct ErrMsg {
};
static DEFINE_MUTEX(aoechr_mutex);
+
+/* A ring buffer of error messages, to be read through
+ * "/dev/etherd/err". When no messages are present,
+ * readers will block waiting for messages to appear.
+ */
static struct ErrMsg emsgs[NMSG];
static int emsgs_head_idx, emsgs_tail_idx;
static struct completion emsgs_comp;
@@ -134,13 +139,12 @@ bail: spin_unlock_irqrestore(&emsgs_lock, flags);
return;
}
- mp = kmalloc(n, GFP_ATOMIC);
+ mp = kmemdup(msg, n, GFP_ATOMIC);
if (mp == NULL) {
printk(KERN_ERR "aoe: allocation failure, len=%ld\n", n);
goto bail;
}
- memcpy(mp, msg, n);
em->msg = mp;
em->flags |= EMFL_VALID;
em->len = n;
@@ -282,7 +286,7 @@ aoechr_init(void)
int n, i;
n = register_chrdev(AOE_MAJOR, "aoechr", &aoe_fops);
- if (n < 0) {
+ if (n < 0) {
printk(KERN_ERR "aoe: can't register char device\n");
return n;
}
diff --git a/drivers/block/aoe/aoecmd.c b/drivers/block/aoe/aoecmd.c
index 3804a0af3ef..422b7d84f68 100644
--- a/drivers/block/aoe/aoecmd.c
+++ b/drivers/block/aoe/aoecmd.c
@@ -1,4 +1,4 @@
-/* Copyright (c) 2012 Coraid, Inc. See COPYING for GPL terms. */
+/* Copyright (c) 2013 Coraid, Inc. See COPYING for GPL terms. */
/*
* aoecmd.c
* Filesystem request handling methods
@@ -22,6 +22,7 @@
#define MAXIOC (8192) /* default meant to avoid most soft lockups */
static void ktcomplete(struct frame *, struct sk_buff *);
+static int count_targets(struct aoedev *d, int *untainted);
static struct buf *nextbuf(struct aoedev *);
@@ -29,27 +30,43 @@ static int aoe_deadsecs = 60 * 3;
module_param(aoe_deadsecs, int, 0644);
MODULE_PARM_DESC(aoe_deadsecs, "After aoe_deadsecs seconds, give up and fail dev.");
-static int aoe_maxout = 16;
+static int aoe_maxout = 64;
module_param(aoe_maxout, int, 0644);
MODULE_PARM_DESC(aoe_maxout,
"Only aoe_maxout outstanding packets for every MAC on eX.Y.");
-static wait_queue_head_t ktiowq;
-static struct ktstate kts;
+/* The number of online cpus during module initialization gives us a
+ * convenient heuristic cap on the parallelism used for ktio threads
+ * doing I/O completion. It is not important that the cap equal the
+ * actual number of running CPUs at any given time, but because of CPU
+ * hotplug, we take care to use ncpus instead of using
+ * num_online_cpus() after module initialization.
+ */
+static int ncpus;
+
+/* mutex lock used for synchronization while thread spawning */
+static DEFINE_MUTEX(ktio_spawn_lock);
+
+static wait_queue_head_t *ktiowq;
+static struct ktstate *kts;
/* io completion queue */
-static struct {
+struct iocq_ktio {
struct list_head head;
spinlock_t lock;
-} iocq;
+};
+static struct iocq_ktio *iocq;
+
+static struct page *empty_page;
static struct sk_buff *
new_skb(ulong len)
{
struct sk_buff *skb;
- skb = alloc_skb(len, GFP_ATOMIC);
+ skb = alloc_skb(len + MAX_HEADER, GFP_ATOMIC);
if (skb) {
+ skb_reserve(skb, MAX_HEADER);
skb_reset_mac_header(skb);
skb_reset_network_header(skb);
skb->protocol = __constant_htons(ETH_P_AOE);
@@ -59,6 +76,23 @@ new_skb(ulong len)
}
static struct frame *
+getframe_deferred(struct aoedev *d, u32 tag)
+{
+ struct list_head *head, *pos, *nx;
+ struct frame *f;
+
+ head = &d->rexmitq;
+ list_for_each_safe(pos, nx, head) {
+ f = list_entry(pos, struct frame, head);
+ if (f->tag == tag) {
+ list_del(pos);
+ return f;
+ }
+ }
+ return NULL;
+}
+
+static struct frame *
getframe(struct aoedev *d, u32 tag)
{
struct frame *f;
@@ -162,8 +196,9 @@ aoe_freetframe(struct frame *f)
t = f->t;
f->buf = NULL;
- f->bv = NULL;
+ memset(&f->iter, 0, sizeof(f->iter));
f->r_skb = NULL;
+ f->flags = 0;
list_add(&f->head, &t->ffree);
}
@@ -217,20 +252,25 @@ newframe(struct aoedev *d)
struct frame *f;
struct aoetgt *t, **tt;
int totout = 0;
+ int use_tainted;
+ int has_untainted;
- if (d->targets[0] == NULL) { /* shouldn't happen, but I'm paranoid */
+ if (!d->targets || !d->targets[0]) {
printk(KERN_ERR "aoe: NULL TARGETS!\n");
return NULL;
}
tt = d->tgt; /* last used target */
- for (;;) {
+ for (use_tainted = 0, has_untainted = 0;;) {
tt++;
- if (tt >= &d->targets[NTARGETS] || !*tt)
+ if (tt >= &d->targets[d->ntargets] || !*tt)
tt = d->targets;
t = *tt;
- totout += t->nout;
+ if (!t->taint) {
+ has_untainted = 1;
+ totout += t->nout;
+ }
if (t->nout < t->maxout
- && t != d->htgt
+ && (use_tainted || !t->taint)
&& t->ifp->nd) {
f = newtframe(d, t);
if (f) {
@@ -239,8 +279,12 @@ newframe(struct aoedev *d)
return f;
}
}
- if (tt == d->tgt) /* we've looped and found nada */
- break;
+ if (tt == d->tgt) { /* we've looped and found nada */
+ if (!use_tainted && !has_untainted)
+ use_tainted = 1;
+ else
+ break;
+ }
}
if (totout == 0) {
d->kicked++;
@@ -250,21 +294,14 @@ newframe(struct aoedev *d)
}
static void
-skb_fillup(struct sk_buff *skb, struct bio_vec *bv, ulong off, ulong cnt)
+skb_fillup(struct sk_buff *skb, struct bio *bio, struct bvec_iter iter)
{
int frag = 0;
- ulong fcnt;
-loop:
- fcnt = bv->bv_len - (off - bv->bv_offset);
- if (fcnt > cnt)
- fcnt = cnt;
- skb_fill_page_desc(skb, frag++, bv->bv_page, off, fcnt);
- cnt -= fcnt;
- if (cnt <= 0)
- return;
- bv++;
- off = bv->bv_offset;
- goto loop;
+ struct bio_vec bv;
+
+ __bio_for_each_segment(bv, bio, iter, iter)
+ skb_fill_page_desc(skb, frag++, bv.bv_page,
+ bv.bv_offset, bv.bv_len);
}
static void
@@ -277,84 +314,47 @@ fhash(struct frame *f)
list_add_tail(&f->head, &d->factive[n]);
}
-static int
-aoecmd_ata_rw(struct aoedev *d)
+static void
+ata_rw_frameinit(struct frame *f)
{
- struct frame *f;
+ struct aoetgt *t;
struct aoe_hdr *h;
struct aoe_atahdr *ah;
- struct buf *buf;
- struct aoetgt *t;
struct sk_buff *skb;
- struct sk_buff_head queue;
- ulong bcnt, fbcnt;
char writebit, extbit;
- writebit = 0x10;
- extbit = 0x4;
-
- buf = nextbuf(d);
- if (buf == NULL)
- return 0;
- f = newframe(d);
- if (f == NULL)
- return 0;
- t = *d->tgt;
- bcnt = d->maxbcnt;
- if (bcnt == 0)
- bcnt = DEFAULTBCNT;
- if (bcnt > buf->resid)
- bcnt = buf->resid;
- fbcnt = bcnt;
- f->bv = buf->bv;
- f->bv_off = f->bv->bv_offset + (f->bv->bv_len - buf->bv_resid);
- do {
- if (fbcnt < buf->bv_resid) {
- buf->bv_resid -= fbcnt;
- buf->resid -= fbcnt;
- break;
- }
- fbcnt -= buf->bv_resid;
- buf->resid -= buf->bv_resid;
- if (buf->resid == 0) {
- d->ip.buf = NULL;
- break;
- }
- buf->bv++;
- buf->bv_resid = buf->bv->bv_len;
- WARN_ON(buf->bv_resid == 0);
- } while (fbcnt);
-
- /* initialize the headers & frame */
skb = f->skb;
h = (struct aoe_hdr *) skb_mac_header(skb);
- ah = (struct aoe_atahdr *) (h+1);
- skb_put(skb, sizeof *h + sizeof *ah);
+ ah = (struct aoe_atahdr *) (h + 1);
+ skb_put(skb, sizeof(*h) + sizeof(*ah));
memset(h, 0, skb->len);
- f->tag = aoehdr_atainit(d, t, h);
+
+ writebit = 0x10;
+ extbit = 0x4;
+
+ t = f->t;
+ f->tag = aoehdr_atainit(t->d, t, h);
fhash(f);
t->nout++;
f->waited = 0;
- f->buf = buf;
- f->bcnt = bcnt;
- f->lba = buf->sector;
+ f->waited_total = 0;
/* set up ata header */
- ah->scnt = bcnt >> 9;
- put_lba(ah, buf->sector);
- if (d->flags & DEVFL_EXT) {
+ ah->scnt = f->iter.bi_size >> 9;
+ put_lba(ah, f->iter.bi_sector);
+ if (t->d->flags & DEVFL_EXT) {
ah->aflags |= AOEAFL_EXT;
} else {
extbit = 0;
ah->lba3 &= 0x0f;
ah->lba3 |= 0xe0; /* LBA bit + obsolete 0xa0 */
}
- if (bio_data_dir(buf->bio) == WRITE) {
- skb_fillup(skb, f->bv, f->bv_off, bcnt);
+ if (f->buf && bio_data_dir(f->buf->bio) == WRITE) {
+ skb_fillup(skb, f->buf->bio, f->iter);
ah->aflags |= AOEAFL_WRITE;
- skb->len += bcnt;
- skb->data_len = bcnt;
- skb->truesize += bcnt;
+ skb->len += f->iter.bi_size;
+ skb->data_len = f->iter.bi_size;
+ skb->truesize += f->iter.bi_size;
t->wpkts++;
} else {
t->rpkts++;
@@ -362,14 +362,44 @@ aoecmd_ata_rw(struct aoedev *d)
}
ah->cmdstat = ATA_CMD_PIO_READ | writebit | extbit;
+ skb->dev = t->ifp->nd;
+}
+
+static int
+aoecmd_ata_rw(struct aoedev *d)
+{
+ struct frame *f;
+ struct buf *buf;
+ struct sk_buff *skb;
+ struct sk_buff_head queue;
+
+ buf = nextbuf(d);
+ if (buf == NULL)
+ return 0;
+ f = newframe(d);
+ if (f == NULL)
+ return 0;
+
+ /* initialize the headers & frame */
+ f->buf = buf;
+ f->iter = buf->iter;
+ f->iter.bi_size = min_t(unsigned long,
+ d->maxbcnt ?: DEFAULTBCNT,
+ f->iter.bi_size);
+ bio_advance_iter(buf->bio, &buf->iter, f->iter.bi_size);
+
+ if (!buf->iter.bi_size)
+ d->ip.buf = NULL;
/* mark all tracking fields and load out */
buf->nframesout += 1;
- buf->sector += bcnt >> 9;
- skb->dev = t->ifp->nd;
- skb = skb_clone(skb, GFP_ATOMIC);
+ ata_rw_frameinit(f);
+
+ skb = skb_clone(f->skb, GFP_ATOMIC);
if (skb) {
+ do_gettimeofday(&f->sent);
+ f->sent_jiffs = (u32) jiffies;
__skb_queue_head_init(&queue);
__skb_queue_tail(&queue, skb);
aoenet_xmit(&queue);
@@ -425,7 +455,6 @@ resend(struct aoedev *d, struct frame *f)
struct sk_buff *skb;
struct sk_buff_head queue;
struct aoe_hdr *h;
- struct aoe_atahdr *ah;
struct aoetgt *t;
char buf[128];
u32 n;
@@ -440,13 +469,15 @@ resend(struct aoedev *d, struct frame *f)
return;
}
h = (struct aoe_hdr *) skb_mac_header(skb);
- ah = (struct aoe_atahdr *) (h+1);
- snprintf(buf, sizeof buf,
- "%15s e%ld.%d oldtag=%08x@%08lx newtag=%08x s=%pm d=%pm nout=%d\n",
- "retransmit", d->aoemajor, d->aoeminor, f->tag, jiffies, n,
- h->src, h->dst, t->nout);
- aoechr_error(buf);
+ if (!(f->flags & FFL_PROBE)) {
+ snprintf(buf, sizeof(buf),
+ "%15s e%ld.%d oldtag=%08x@%08lx newtag=%08x s=%pm d=%pm nout=%d\n",
+ "retransmit", d->aoemajor, d->aoeminor,
+ f->tag, jiffies, n,
+ h->src, h->dst, t->nout);
+ aoechr_error(buf);
+ }
f->tag = n;
fhash(f);
@@ -458,12 +489,46 @@ resend(struct aoedev *d, struct frame *f)
skb = skb_clone(skb, GFP_ATOMIC);
if (skb == NULL)
return;
+ do_gettimeofday(&f->sent);
+ f->sent_jiffs = (u32) jiffies;
__skb_queue_head_init(&queue);
__skb_queue_tail(&queue, skb);
aoenet_xmit(&queue);
}
static int
+tsince_hr(struct frame *f)
+{
+ struct timeval now;
+ int n;
+
+ do_gettimeofday(&now);
+ n = now.tv_usec - f->sent.tv_usec;
+ n += (now.tv_sec - f->sent.tv_sec) * USEC_PER_SEC;
+
+ if (n < 0)
+ n = -n;
+
+ /* For relatively long periods, use jiffies to avoid
+ * discrepancies caused by updates to the system time.
+ *
+ * On system with HZ of 1000, 32-bits is over 49 days
+ * worth of jiffies, or over 71 minutes worth of usecs.
+ *
+ * Jiffies overflow is handled by subtraction of unsigned ints:
+ * (gdb) print (unsigned) 2 - (unsigned) 0xfffffffe
+ * $3 = 4
+ * (gdb)
+ */
+ if (n > USEC_PER_SEC / 4) {
+ n = ((u32) jiffies) - f->sent_jiffs;
+ n *= USEC_PER_SEC / HZ;
+ }
+
+ return n;
+}
+
+static int
tsince(u32 tag)
{
int n;
@@ -472,7 +537,7 @@ tsince(u32 tag)
n -= tag & 0xffff;
if (n < 0)
n += 1<<16;
- return n;
+ return jiffies_to_usecs(n + 1);
}
static struct aoeif *
@@ -503,70 +568,186 @@ ejectif(struct aoetgt *t, struct aoeif *ifp)
dev_put(nd);
}
-static int
-sthtith(struct aoedev *d)
+static struct frame *
+reassign_frame(struct frame *f)
{
- struct frame *f, *nf;
- struct list_head *nx, *pos, *head;
+ struct frame *nf;
struct sk_buff *skb;
- struct aoetgt *ht = d->htgt;
- int i;
- for (i = 0; i < NFACTIVE; i++) {
- head = &d->factive[i];
- list_for_each_safe(pos, nx, head) {
- f = list_entry(pos, struct frame, head);
- if (f->t != ht)
- continue;
+ nf = newframe(f->t->d);
+ if (!nf)
+ return NULL;
+ if (nf->t == f->t) {
+ aoe_freetframe(nf);
+ return NULL;
+ }
- nf = newframe(d);
- if (!nf)
- return 0;
+ skb = nf->skb;
+ nf->skb = f->skb;
+ nf->buf = f->buf;
+ nf->iter = f->iter;
+ nf->waited = 0;
+ nf->waited_total = f->waited_total;
+ nf->sent = f->sent;
+ nf->sent_jiffs = f->sent_jiffs;
+ f->skb = skb;
+
+ return nf;
+}
- /* remove frame from active list */
- list_del(pos);
+static void
+probe(struct aoetgt *t)
+{
+ struct aoedev *d;
+ struct frame *f;
+ struct sk_buff *skb;
+ struct sk_buff_head queue;
+ size_t n, m;
+ int frag;
+
+ d = t->d;
+ f = newtframe(d, t);
+ if (!f) {
+ pr_err("%s %pm for e%ld.%d: %s\n",
+ "aoe: cannot probe remote address",
+ t->addr,
+ (long) d->aoemajor, d->aoeminor,
+ "no frame available");
+ return;
+ }
+ f->flags |= FFL_PROBE;
+ ifrotate(t);
+ f->iter.bi_size = t->d->maxbcnt ? t->d->maxbcnt : DEFAULTBCNT;
+ ata_rw_frameinit(f);
+ skb = f->skb;
+ for (frag = 0, n = f->iter.bi_size; n > 0; ++frag, n -= m) {
+ if (n < PAGE_SIZE)
+ m = n;
+ else
+ m = PAGE_SIZE;
+ skb_fill_page_desc(skb, frag, empty_page, 0, m);
+ }
+ skb->len += f->iter.bi_size;
+ skb->data_len = f->iter.bi_size;
+ skb->truesize += f->iter.bi_size;
+
+ skb = skb_clone(f->skb, GFP_ATOMIC);
+ if (skb) {
+ do_gettimeofday(&f->sent);
+ f->sent_jiffs = (u32) jiffies;
+ __skb_queue_head_init(&queue);
+ __skb_queue_tail(&queue, skb);
+ aoenet_xmit(&queue);
+ }
+}
+
+static long
+rto(struct aoedev *d)
+{
+ long t;
+
+ t = 2 * d->rttavg >> RTTSCALE;
+ t += 8 * d->rttdev >> RTTDSCALE;
+ if (t == 0)
+ t = 1;
+
+ return t;
+}
+
+static void
+rexmit_deferred(struct aoedev *d)
+{
+ struct aoetgt *t;
+ struct frame *f;
+ struct frame *nf;
+ struct list_head *pos, *nx, *head;
+ int since;
+ int untainted;
- /* reassign all pertinent bits to new outbound frame */
- skb = nf->skb;
- nf->skb = f->skb;
- nf->buf = f->buf;
- nf->bcnt = f->bcnt;
- nf->lba = f->lba;
- nf->bv = f->bv;
- nf->bv_off = f->bv_off;
- nf->waited = 0;
- f->skb = skb;
+ count_targets(d, &untainted);
+
+ head = &d->rexmitq;
+ list_for_each_safe(pos, nx, head) {
+ f = list_entry(pos, struct frame, head);
+ t = f->t;
+ if (t->taint) {
+ if (!(f->flags & FFL_PROBE)) {
+ nf = reassign_frame(f);
+ if (nf) {
+ if (t->nout_probes == 0
+ && untainted > 0) {
+ probe(t);
+ t->nout_probes++;
+ }
+ list_replace(&f->head, &nf->head);
+ pos = &nf->head;
+ aoe_freetframe(f);
+ f = nf;
+ t = f->t;
+ }
+ } else if (untainted < 1) {
+ /* don't probe w/o other untainted aoetgts */
+ goto stop_probe;
+ } else if (tsince_hr(f) < t->taint * rto(d)) {
+ /* reprobe slowly when taint is high */
+ continue;
+ }
+ } else if (f->flags & FFL_PROBE) {
+stop_probe: /* don't probe untainted aoetgts */
+ list_del(pos);
aoe_freetframe(f);
- ht->nout--;
- nf->t->nout++;
- resend(d, nf);
+ /* leaving d->kicked, because this is routine */
+ f->t->d->flags |= DEVFL_KICKME;
+ continue;
}
+ if (t->nout >= t->maxout)
+ continue;
+ list_del(pos);
+ t->nout++;
+ if (f->flags & FFL_PROBE)
+ t->nout_probes++;
+ since = tsince_hr(f);
+ f->waited += since;
+ f->waited_total += since;
+ resend(d, f);
}
- /* We've cleaned up the outstanding so take away his
- * interfaces so he won't be used. We should remove him from
- * the target array here, but cleaning up a target is
- * involved. PUNT!
- */
- memset(ht->ifs, 0, sizeof ht->ifs);
- d->htgt = NULL;
- return 1;
}
-static inline unsigned char
-ata_scnt(unsigned char *packet) {
- struct aoe_hdr *h;
- struct aoe_atahdr *ah;
+/* An aoetgt accumulates demerits quickly, and successful
+ * probing redeems the aoetgt slowly.
+ */
+static void
+scorn(struct aoetgt *t)
+{
+ int n;
- h = (struct aoe_hdr *) packet;
- ah = (struct aoe_atahdr *) (h+1);
- return ah->scnt;
+ n = t->taint++;
+ t->taint += t->taint * 2;
+ if (n > t->taint)
+ t->taint = n;
+ if (t->taint > MAX_TAINT)
+ t->taint = MAX_TAINT;
+}
+
+static int
+count_targets(struct aoedev *d, int *untainted)
+{
+ int i, good;
+
+ for (i = good = 0; i < d->ntargets && d->targets[i]; ++i)
+ if (d->targets[i]->taint == 0)
+ good++;
+
+ if (untainted)
+ *untainted = good;
+ return i;
}
static void
rexmit_timer(ulong vp)
{
struct aoedev *d;
- struct aoetgt *t, **tt, **te;
+ struct aoetgt *t;
struct aoeif *ifp;
struct frame *f;
struct list_head *head, *pos, *nx;
@@ -574,15 +755,18 @@ rexmit_timer(ulong vp)
register long timeout;
ulong flags, n;
int i;
+ int utgts; /* number of aoetgt descriptors (not slots) */
+ int since;
d = (struct aoedev *) vp;
- /* timeout is always ~150% of the moving average */
- timeout = d->rttavg;
- timeout += timeout >> 1;
-
spin_lock_irqsave(&d->lock, flags);
+ /* timeout based on observed timings and variations */
+ timeout = rto(d);
+
+ utgts = count_targets(d, NULL);
+
if (d->flags & DEVFL_TKILL) {
spin_unlock_irqrestore(&d->lock, flags);
return;
@@ -593,67 +777,61 @@ rexmit_timer(ulong vp)
head = &d->factive[i];
list_for_each_safe(pos, nx, head) {
f = list_entry(pos, struct frame, head);
- if (tsince(f->tag) < timeout)
+ if (tsince_hr(f) < timeout)
break; /* end of expired frames */
/* move to flist for later processing */
list_move_tail(pos, &flist);
}
}
- /* window check */
- tt = d->targets;
- te = tt + d->ntargets;
- for (; tt < te && (t = *tt); tt++) {
- if (t->nout == t->maxout
- && t->maxout < t->nframes
- && (jiffies - t->lastwadj)/HZ > 10) {
- t->maxout++;
- t->lastwadj = jiffies;
- }
- }
-
- if (!list_empty(&flist)) { /* retransmissions necessary */
- n = d->rttavg <<= 1;
- if (n > MAXTIMER)
- d->rttavg = MAXTIMER;
- }
/* process expired frames */
while (!list_empty(&flist)) {
pos = flist.next;
f = list_entry(pos, struct frame, head);
- n = f->waited += timeout;
- n /= HZ;
- if (n > aoe_deadsecs) {
+ since = tsince_hr(f);
+ n = f->waited_total + since;
+ n /= USEC_PER_SEC;
+ if (aoe_deadsecs
+ && n > aoe_deadsecs
+ && !(f->flags & FFL_PROBE)) {
/* Waited too long. Device failure.
* Hang all frames on first hash bucket for downdev
* to clean up.
*/
list_splice(&flist, &d->factive[0]);
aoedev_downdev(d);
- break;
+ goto out;
}
- list_del(pos);
t = f->t;
- if (n > aoe_deadsecs/2)
- d->htgt = t; /* see if another target can help */
-
- if (t->nout == t->maxout) {
- if (t->maxout > 1)
- t->maxout--;
- t->lastwadj = jiffies;
+ n = f->waited + since;
+ n /= USEC_PER_SEC;
+ if (aoe_deadsecs && utgts > 0
+ && (n > aoe_deadsecs / utgts || n > HARD_SCORN_SECS))
+ scorn(t); /* avoid this target */
+
+ if (t->maxout != 1) {
+ t->ssthresh = t->maxout / 2;
+ t->maxout = 1;
}
- ifp = getif(t, f->skb->dev);
- if (ifp && ++ifp->lost > (t->nframes << 1)
- && (ifp != t->ifs || t->ifs[1].nd)) {
- ejectif(t, ifp);
- ifp = NULL;
+ if (f->flags & FFL_PROBE) {
+ t->nout_probes--;
+ } else {
+ ifp = getif(t, f->skb->dev);
+ if (ifp && ++ifp->lost > (t->nframes << 1)
+ && (ifp != t->ifs || t->ifs[1].nd)) {
+ ejectif(t, ifp);
+ ifp = NULL;
+ }
}
- resend(d, f);
+ list_move_tail(pos, &d->rexmitq);
+ t->nout--;
}
+ rexmit_deferred(d);
- if ((d->flags & DEVFL_KICKME || d->htgt) && d->blkq) {
+out:
+ if ((d->flags & DEVFL_KICKME) && d->blkq) {
d->flags &= ~DEVFL_KICKME;
d->blkq->request_fn(d->blkq);
}
@@ -688,21 +866,15 @@ rqbiocnt(struct request *r)
static void
bio_pageinc(struct bio *bio)
{
- struct bio_vec *bv;
+ struct bio_vec bv;
struct page *page;
- int i;
+ struct bvec_iter iter;
- bio_for_each_segment(bv, bio, i) {
- page = bv->bv_page;
+ bio_for_each_segment(bv, bio, iter) {
/* Non-zero page count for non-head members of
- * compound pages is no longer allowed by the kernel,
- * but this has never been seen here.
+ * compound pages is no longer allowed by the kernel.
*/
- if (unlikely(PageCompound(page)))
- if (compound_trans_head(page) != page) {
- pr_crit("page tail used for block I/O\n");
- BUG();
- }
+ page = compound_head(bv.bv_page);
atomic_inc(&page->_count);
}
}
@@ -710,27 +882,24 @@ bio_pageinc(struct bio *bio)
static void
bio_pagedec(struct bio *bio)
{
- struct bio_vec *bv;
- int i;
+ struct page *page;
+ struct bio_vec bv;
+ struct bvec_iter iter;
- bio_for_each_segment(bv, bio, i)
- atomic_dec(&bv->bv_page->_count);
+ bio_for_each_segment(bv, bio, iter) {
+ page = compound_head(bv.bv_page);
+ atomic_dec(&page->_count);
+ }
}
static void
bufinit(struct buf *buf, struct request *rq, struct bio *bio)
{
- struct bio_vec *bv;
-
memset(buf, 0, sizeof(*buf));
buf->rq = rq;
buf->bio = bio;
- buf->resid = bio->bi_size;
- buf->sector = bio->bi_sector;
+ buf->iter = bio->bi_iter;
bio_pageinc(bio);
- buf->bv = bv = &bio->bi_io_vec[bio->bi_idx];
- buf->bv_resid = bv->bv_len;
- WARN_ON(buf->bv_resid == 0);
}
static struct buf *
@@ -774,8 +943,7 @@ nextbuf(struct aoedev *d)
void
aoecmd_work(struct aoedev *d)
{
- if (d->htgt && !sthtith(d))
- return;
+ rexmit_deferred(d);
while (aoecmd_ata_rw(d))
;
}
@@ -809,6 +977,17 @@ aoecmd_sleepwork(struct work_struct *work)
}
static void
+ata_ident_fixstring(u16 *id, int ns)
+{
+ u16 s;
+
+ while (ns-- > 0) {
+ s = *id;
+ *id++ = s >> 8 | s << 8;
+ }
+}
+
+static void
ataid_complete(struct aoedev *d, struct aoetgt *t, unsigned char *id)
{
u64 ssize;
@@ -843,6 +1022,11 @@ ataid_complete(struct aoedev *d, struct aoetgt *t, unsigned char *id)
d->geo.sectors = get_unaligned_le16(&id[56 << 1]);
}
+ ata_ident_fixstring((u16 *) &id[10<<1], 10); /* serial */
+ ata_ident_fixstring((u16 *) &id[23<<1], 4); /* firmware */
+ ata_ident_fixstring((u16 *) &id[27<<1], 20); /* model */
+ memcpy(d->ident, id, sizeof(d->ident));
+
if (d->ssize != ssize)
printk(KERN_INFO
"aoe: %pm e%ld.%d v%04x has %llu sectors\n",
@@ -862,26 +1046,28 @@ ataid_complete(struct aoedev *d, struct aoetgt *t, unsigned char *id)
}
static void
-calc_rttavg(struct aoedev *d, int rtt)
+calc_rttavg(struct aoedev *d, struct aoetgt *t, int rtt)
{
register long n;
n = rtt;
- if (n < 0) {
- n = -rtt;
- if (n < MINTIMER)
- n = MINTIMER;
- else if (n > MAXTIMER)
- n = MAXTIMER;
- d->mintimer += (n - d->mintimer) >> 1;
- } else if (n < d->mintimer)
- n = d->mintimer;
- else if (n > MAXTIMER)
- n = MAXTIMER;
-
- /* g == .25; cf. Congestion Avoidance and Control, Jacobson & Karels; 1988 */
- n -= d->rttavg;
- d->rttavg += n >> 2;
+
+ /* cf. Congestion Avoidance and Control, Jacobson & Karels, 1988 */
+ n -= d->rttavg >> RTTSCALE;
+ d->rttavg += n;
+ if (n < 0)
+ n = -n;
+ n -= d->rttdev >> RTTDSCALE;
+ d->rttdev += n;
+
+ if (!t || t->maxout >= t->nframes)
+ return;
+ if (t->maxout < t->ssthresh)
+ t->maxout += 1;
+ else if (t->nout == t->maxout && t->next_cwnd-- == 0) {
+ t->maxout += 1;
+ t->next_cwnd = t->maxout;
+ }
}
static struct aoetgt *
@@ -890,7 +1076,7 @@ gettgt(struct aoedev *d, char *addr)
struct aoetgt **t, **e;
t = d->targets;
- e = t + NTARGETS;
+ e = t + d->ntargets;
for (; t < e && *t; t++)
if (memcmp((*t)->addr, addr, sizeof((*t)->addr)) == 0)
return *t;
@@ -898,24 +1084,18 @@ gettgt(struct aoedev *d, char *addr)
}
static void
-bvcpy(struct bio_vec *bv, ulong off, struct sk_buff *skb, long cnt)
+bvcpy(struct sk_buff *skb, struct bio *bio, struct bvec_iter iter, long cnt)
{
- ulong fcnt;
- char *p;
int soff = 0;
-loop:
- fcnt = bv->bv_len - (off - bv->bv_offset);
- if (fcnt > cnt)
- fcnt = cnt;
- p = page_address(bv->bv_page) + off;
- skb_copy_bits(skb, soff, p, fcnt);
- soff += fcnt;
- cnt -= fcnt;
- if (cnt <= 0)
- return;
- bv++;
- off = bv->bv_offset;
- goto loop;
+ struct bio_vec bv;
+
+ iter.bi_size = cnt;
+
+ __bio_for_each_segment(bv, bio, iter, iter) {
+ char *p = page_address(bv.bv_page) + bv.bv_offset;
+ skb_copy_bits(skb, soff, p, bv.bv_len);
+ soff += bv.bv_len;
+ }
}
void
@@ -931,11 +1111,11 @@ aoe_end_request(struct aoedev *d, struct request *rq, int fastfail)
do {
bio = rq->bio;
bok = !fastfail && test_bit(BIO_UPTODATE, &bio->bi_flags);
- } while (__blk_end_request(rq, bok ? 0 : -EIO, bio->bi_size));
+ } while (__blk_end_request(rq, bok ? 0 : -EIO, bio->bi_iter.bi_size));
/* cf. http://lkml.org/lkml/2006/10/31/28 */
if (!fastfail)
- q->request_fn(q);
+ __blk_run_queue(q);
}
static void
@@ -966,19 +1146,22 @@ ktiocomplete(struct frame *f)
struct aoeif *ifp;
struct aoedev *d;
long n;
+ int untainted;
if (f == NULL)
return;
t = f->t;
d = t->d;
+ skb = f->r_skb;
+ buf = f->buf;
+ if (f->flags & FFL_PROBE)
+ goto out;
+ if (!skb) /* just fail the buf. */
+ goto noskb;
hout = (struct aoe_hdr *) skb_mac_header(f->skb);
ahout = (struct aoe_atahdr *) (hout+1);
- buf = f->buf;
- skb = f->r_skb;
- if (skb == NULL)
- goto noskb; /* just fail the buf. */
hin = (struct aoe_hdr *) skb->data;
skb_pull(skb, sizeof(*hin));
@@ -988,9 +1171,9 @@ ktiocomplete(struct frame *f)
pr_err("aoe: ata error cmd=%2.2Xh stat=%2.2Xh from e%ld.%d\n",
ahout->cmdstat, ahin->cmdstat,
d->aoemajor, d->aoeminor);
-noskb: if (buf)
+noskb: if (buf)
clear_bit(BIO_UPTODATE, &buf->bio->bi_flags);
- goto badrsp;
+ goto out;
}
n = ahout->scnt << 9;
@@ -998,25 +1181,35 @@ noskb: if (buf)
case ATA_CMD_PIO_READ:
case ATA_CMD_PIO_READ_EXT:
if (skb->len < n) {
- pr_err("aoe: runt data size in read. skb->len=%d need=%ld\n",
- skb->len, n);
+ pr_err("%s e%ld.%d. skb->len=%d need=%ld\n",
+ "aoe: runt data size in read from",
+ (long) d->aoemajor, d->aoeminor,
+ skb->len, n);
+ clear_bit(BIO_UPTODATE, &buf->bio->bi_flags);
+ break;
+ }
+ if (n > f->iter.bi_size) {
+ pr_err_ratelimited("%s e%ld.%d. bytes=%ld need=%u\n",
+ "aoe: too-large data size in read from",
+ (long) d->aoemajor, d->aoeminor,
+ n, f->iter.bi_size);
clear_bit(BIO_UPTODATE, &buf->bio->bi_flags);
break;
}
- bvcpy(f->bv, f->bv_off, skb, n);
+ bvcpy(skb, f->buf->bio, f->iter, n);
case ATA_CMD_PIO_WRITE:
case ATA_CMD_PIO_WRITE_EXT:
spin_lock_irq(&d->lock);
ifp = getif(t, skb->dev);
if (ifp)
ifp->lost = 0;
- if (d->htgt == t) /* I'll help myself, thank you. */
- d->htgt = NULL;
spin_unlock_irq(&d->lock);
break;
case ATA_CMD_ID_ATA:
if (skb->len < 512) {
- pr_info("aoe: runt data size in ataid. skb->len=%d\n",
+ pr_info("%s e%ld.%d. skb->len=%d need=512\n",
+ "aoe: runt data size in ataid from",
+ (long) d->aoemajor, d->aoeminor,
skb->len);
break;
}
@@ -1032,16 +1225,23 @@ noskb: if (buf)
be16_to_cpu(get_unaligned(&hin->major)),
hin->minor);
}
-badrsp:
+out:
spin_lock_irq(&d->lock);
+ if (t->taint > 0
+ && --t->taint > 0
+ && t->nout_probes == 0) {
+ count_targets(d, &untainted);
+ if (untainted > 0) {
+ probe(t);
+ t->nout_probes++;
+ }
+ }
aoe_freetframe(f);
- if (buf && --buf->nframesout == 0 && buf->resid == 0)
+ if (buf && --buf->nframesout == 0 && buf->iter.bi_size == 0)
aoe_end_buf(d, buf);
- aoecmd_work(d);
-
spin_unlock_irq(&d->lock);
aoedev_put(d);
dev_kfree_skb(skb);
@@ -1051,23 +1251,36 @@ badrsp:
* Returns true iff responses needing processing remain.
*/
static int
-ktio(void)
+ktio(int id)
{
struct frame *f;
struct list_head *pos;
int i;
+ int actual_id;
for (i = 0; ; ++i) {
if (i == MAXIOC)
return 1;
- if (list_empty(&iocq.head))
+ if (list_empty(&iocq[id].head))
return 0;
- pos = iocq.head.next;
+ pos = iocq[id].head.next;
list_del(pos);
- spin_unlock_irq(&iocq.lock);
f = list_entry(pos, struct frame, head);
+ spin_unlock_irq(&iocq[id].lock);
ktiocomplete(f);
- spin_lock_irq(&iocq.lock);
+
+ /* Figure out if extra threads are required. */
+ actual_id = f->t->d->aoeminor % ncpus;
+
+ if (!kts[actual_id].active) {
+ BUG_ON(id != 0);
+ mutex_lock(&ktio_spawn_lock);
+ if (!kts[actual_id].active
+ && aoe_ktstart(&kts[actual_id]) == 0)
+ kts[actual_id].active = 1;
+ mutex_unlock(&ktio_spawn_lock);
+ }
+ spin_lock_irq(&iocq[id].lock);
}
}
@@ -1084,7 +1297,7 @@ kthread(void *vp)
complete(&k->rendez); /* tell spawner we're running */
do {
spin_lock_irq(k->lock);
- more = k->fn();
+ more = k->fn(k->id);
if (!more) {
add_wait_queue(k->waitq, &wait);
__set_current_state(TASK_INTERRUPTIBLE);
@@ -1113,7 +1326,7 @@ aoe_ktstart(struct ktstate *k)
struct task_struct *task;
init_completion(&k->rendez);
- task = kthread_run(kthread, k, k->name);
+ task = kthread_run(kthread, k, "%s", k->name);
if (task == NULL || IS_ERR(task))
return -ENOMEM;
k->task = task;
@@ -1126,13 +1339,24 @@ aoe_ktstart(struct ktstate *k)
static void
ktcomplete(struct frame *f, struct sk_buff *skb)
{
+ int id;
ulong flags;
f->r_skb = skb;
- spin_lock_irqsave(&iocq.lock, flags);
- list_add_tail(&f->head, &iocq.head);
- spin_unlock_irqrestore(&iocq.lock, flags);
- wake_up(&ktiowq);
+ id = f->t->d->aoeminor % ncpus;
+ spin_lock_irqsave(&iocq[id].lock, flags);
+ if (!kts[id].active) {
+ spin_unlock_irqrestore(&iocq[id].lock, flags);
+ /* The thread with id has not been spawned yet,
+ * so delegate the work to the main thread and
+ * try spawning a new thread.
+ */
+ id = 0;
+ spin_lock_irqsave(&iocq[id].lock, flags);
+ }
+ list_add_tail(&f->head, &iocq[id].head);
+ spin_unlock_irqrestore(&iocq[id].lock, flags);
+ wake_up(&ktiowq[id]);
}
struct sk_buff *
@@ -1141,7 +1365,6 @@ aoecmd_ata_rsp(struct sk_buff *skb)
struct aoedev *d;
struct aoe_hdr *h;
struct frame *f;
- struct aoetgt *t;
u32 n;
ulong flags;
char ebuf[128];
@@ -1162,23 +1385,32 @@ aoecmd_ata_rsp(struct sk_buff *skb)
n = be32_to_cpu(get_unaligned(&h->tag));
f = getframe(d, n);
- if (f == NULL) {
- calc_rttavg(d, -tsince(n));
- spin_unlock_irqrestore(&d->lock, flags);
- aoedev_put(d);
- snprintf(ebuf, sizeof ebuf,
- "%15s e%d.%d tag=%08x@%08lx\n",
- "unexpected rsp",
- get_unaligned_be16(&h->major),
- h->minor,
- get_unaligned_be32(&h->tag),
- jiffies);
- aoechr_error(ebuf);
- return skb;
+ if (f) {
+ calc_rttavg(d, f->t, tsince_hr(f));
+ f->t->nout--;
+ if (f->flags & FFL_PROBE)
+ f->t->nout_probes--;
+ } else {
+ f = getframe_deferred(d, n);
+ if (f) {
+ calc_rttavg(d, NULL, tsince_hr(f));
+ } else {
+ calc_rttavg(d, NULL, tsince(n));
+ spin_unlock_irqrestore(&d->lock, flags);
+ aoedev_put(d);
+ snprintf(ebuf, sizeof(ebuf),
+ "%15s e%d.%d tag=%08x@%08lx s=%pm d=%pm\n",
+ "unexpected rsp",
+ get_unaligned_be16(&h->major),
+ h->minor,
+ get_unaligned_be32(&h->tag),
+ jiffies,
+ h->src,
+ h->dst);
+ aoechr_error(ebuf);
+ return skb;
+ }
}
- t = f->t;
- calc_rttavg(d, tsince(f->tag));
- t->nout--;
aoecmd_work(d);
spin_unlock_irqrestore(&d->lock, flags);
@@ -1201,7 +1433,7 @@ aoecmd_cfg(ushort aoemajor, unsigned char aoeminor)
aoecmd_cfg_pkts(aoemajor, aoeminor, &queue);
aoenet_xmit(&queue);
}
-
+
struct sk_buff *
aoecmd_ata_id(struct aoedev *d)
{
@@ -1227,6 +1459,7 @@ aoecmd_ata_id(struct aoedev *d)
fhash(f);
t->nout++;
f->waited = 0;
+ f->waited_total = 0;
/* set up ata header */
ah->scnt = 1;
@@ -1235,41 +1468,69 @@ aoecmd_ata_id(struct aoedev *d)
skb->dev = t->ifp->nd;
- d->rttavg = MAXTIMER;
+ d->rttavg = RTTAVG_INIT;
+ d->rttdev = RTTDEV_INIT;
d->timer.function = rexmit_timer;
- return skb_clone(skb, GFP_ATOMIC);
+ skb = skb_clone(skb, GFP_ATOMIC);
+ if (skb) {
+ do_gettimeofday(&f->sent);
+ f->sent_jiffs = (u32) jiffies;
+ }
+
+ return skb;
+}
+
+static struct aoetgt **
+grow_targets(struct aoedev *d)
+{
+ ulong oldn, newn;
+ struct aoetgt **tt;
+
+ oldn = d->ntargets;
+ newn = oldn * 2;
+ tt = kcalloc(newn, sizeof(*d->targets), GFP_ATOMIC);
+ if (!tt)
+ return NULL;
+ memmove(tt, d->targets, sizeof(*d->targets) * oldn);
+ d->tgt = tt + (d->tgt - d->targets);
+ kfree(d->targets);
+ d->targets = tt;
+ d->ntargets = newn;
+
+ return &d->targets[oldn];
}
-
+
static struct aoetgt *
addtgt(struct aoedev *d, char *addr, ulong nframes)
{
struct aoetgt *t, **tt, **te;
tt = d->targets;
- te = tt + NTARGETS;
+ te = tt + d->ntargets;
for (; tt < te && *tt; tt++)
;
if (tt == te) {
- printk(KERN_INFO
- "aoe: device addtgt failure; too many targets\n");
- return NULL;
+ tt = grow_targets(d);
+ if (!tt)
+ goto nomem;
}
t = kzalloc(sizeof(*t), GFP_ATOMIC);
- if (!t) {
- printk(KERN_INFO "aoe: cannot allocate memory to add target\n");
- return NULL;
- }
-
- d->ntargets++;
+ if (!t)
+ goto nomem;
t->nframes = nframes;
t->d = d;
memcpy(t->addr, addr, sizeof t->addr);
t->ifp = t->ifs;
- t->maxout = t->nframes;
+ aoecmd_wreset(t);
+ t->maxout = t->nframes / 2;
INIT_LIST_HEAD(&t->ffree);
return *tt = t;
+
+ nomem:
+ pr_info("aoe: cannot allocate memory to add target\n");
+ return NULL;
}
static void
@@ -1279,7 +1540,7 @@ setdbcnt(struct aoedev *d)
int bcnt = 0;
t = d->targets;
- e = t + NTARGETS;
+ e = t + d->ntargets;
for (; t < e && *t; t++)
if (bcnt == 0 || bcnt > (*t)->minbcnt)
bcnt = (*t)->minbcnt;
@@ -1373,7 +1634,11 @@ aoecmd_cfg_rsp(struct sk_buff *skb)
spin_lock_irqsave(&d->lock, flags);
t = gettgt(d, h->src);
- if (!t) {
+ if (t) {
+ t->nframes = n;
+ if (n < t->maxout)
+ aoecmd_wreset(t);
+ } else {
t = addtgt(d, h->src, n);
if (!t)
goto bail;
@@ -1402,17 +1667,26 @@ bail:
}
void
+aoecmd_wreset(struct aoetgt *t)
+{
+ t->maxout = 1;
+ t->ssthresh = t->nframes / 2;
+ t->next_cwnd = t->nframes;
+}
+
+void
aoecmd_cleanslate(struct aoedev *d)
{
struct aoetgt **t, **te;
- d->mintimer = MINTIMER;
+ d->rttavg = RTTAVG_INIT;
+ d->rttdev = RTTDEV_INIT;
d->maxbcnt = 0;
t = d->targets;
- te = t + NTARGETS;
+ te = t + d->ntargets;
for (; t < te && *t; t++)
- (*t)->maxout = (*t)->nframes;
+ aoecmd_wreset(*t);
}
void
@@ -1420,7 +1694,7 @@ aoe_failbuf(struct aoedev *d, struct buf *buf)
{
if (buf == NULL)
return;
- buf->resid = 0;
+ buf->iter.bi_size = 0;
clear_bit(BIO_UPTODATE, &buf->bio->bi_flags);
if (buf->nframesout == 0)
aoe_end_buf(d, buf);
@@ -1429,6 +1703,17 @@ aoe_failbuf(struct aoedev *d, struct buf *buf)
void
aoe_flush_iocq(void)
{
+ int i;
+
+ for (i = 0; i < ncpus; i++) {
+ if (kts[i].active)
+ aoe_flush_iocq_by_index(i);
+ }
+}
+
+void
+aoe_flush_iocq_by_index(int id)
+{
struct frame *f;
struct aoedev *d;
LIST_HEAD(flist);
@@ -1436,9 +1721,9 @@ aoe_flush_iocq(void)
struct sk_buff *skb;
ulong flags;
- spin_lock_irqsave(&iocq.lock, flags);
- list_splice_init(&iocq.head, &flist);
- spin_unlock_irqrestore(&iocq.lock, flags);
+ spin_lock_irqsave(&iocq[id].lock, flags);
+ list_splice_init(&iocq[id].head, &flist);
+ spin_unlock_irqrestore(&iocq[id].lock, flags);
while (!list_empty(&flist)) {
pos = flist.next;
list_del(pos);
@@ -1460,19 +1745,82 @@ aoe_flush_iocq(void)
int __init
aoecmd_init(void)
{
- INIT_LIST_HEAD(&iocq.head);
- spin_lock_init(&iocq.lock);
- init_waitqueue_head(&ktiowq);
- kts.name = "aoe_ktio";
- kts.fn = ktio;
- kts.waitq = &ktiowq;
- kts.lock = &iocq.lock;
- return aoe_ktstart(&kts);
+ void *p;
+ int i;
+ int ret;
+
+ /* get_zeroed_page returns page with ref count 1 */
+ p = (void *) get_zeroed_page(GFP_KERNEL | __GFP_REPEAT);
+ if (!p)
+ return -ENOMEM;
+ empty_page = virt_to_page(p);
+
+ ncpus = num_online_cpus();
+
+ iocq = kcalloc(ncpus, sizeof(struct iocq_ktio), GFP_KERNEL);
+ if (!iocq)
+ return -ENOMEM;
+
+ kts = kcalloc(ncpus, sizeof(struct ktstate), GFP_KERNEL);
+ if (!kts) {
+ ret = -ENOMEM;
+ goto kts_fail;
+ }
+
+ ktiowq = kcalloc(ncpus, sizeof(wait_queue_head_t), GFP_KERNEL);
+ if (!ktiowq) {
+ ret = -ENOMEM;
+ goto ktiowq_fail;
+ }
+
+ mutex_init(&ktio_spawn_lock);
+
+ for (i = 0; i < ncpus; i++) {
+ INIT_LIST_HEAD(&iocq[i].head);
+ spin_lock_init(&iocq[i].lock);
+ init_waitqueue_head(&ktiowq[i]);
+ snprintf(kts[i].name, sizeof(kts[i].name), "aoe_ktio%d", i);
+ kts[i].fn = ktio;
+ kts[i].waitq = &ktiowq[i];
+ kts[i].lock = &iocq[i].lock;
+ kts[i].id = i;
+ kts[i].active = 0;
+ }
+ kts[0].active = 1;
+ if (aoe_ktstart(&kts[0])) {
+ ret = -ENOMEM;
+ goto ktstart_fail;
+ }
+ return 0;
+
+ktstart_fail:
+ kfree(ktiowq);
+ktiowq_fail:
+ kfree(kts);
+kts_fail:
+ kfree(iocq);
+
+ return ret;
}
void
aoecmd_exit(void)
{
- aoe_ktstop(&kts);
+ int i;
+
+ for (i = 0; i < ncpus; i++)
+ if (kts[i].active)
+ aoe_ktstop(&kts[i]);
+
aoe_flush_iocq();
+
+ /* Free up the iocq and thread speicific configuration
+ * allocated during startup.
+ */
+ kfree(iocq);
+ kfree(kts);
+ kfree(ktiowq);
+
+ free_page((unsigned long) page_address(empty_page));
+ empty_page = NULL;
}
diff --git a/drivers/block/aoe/aoedev.c b/drivers/block/aoe/aoedev.c
index 90e5b537f94..e774c50b684 100644
--- a/drivers/block/aoe/aoedev.c
+++ b/drivers/block/aoe/aoedev.c
@@ -1,4 +1,4 @@
-/* Copyright (c) 2012 Coraid, Inc. See COPYING for GPL terms. */
+/* Copyright (c) 2013 Coraid, Inc. See COPYING for GPL terms. */
/*
* aoedev.c
* AoE device utility functions; maintains device list.
@@ -12,10 +12,10 @@
#include <linux/bitmap.h>
#include <linux/kdev_t.h>
#include <linux/moduleparam.h>
+#include <linux/string.h>
#include "aoe.h"
static void dummy_timer(ulong);
-static void aoedev_freedev(struct aoedev *);
static void freetgt(struct aoedev *d, struct aoetgt *t);
static void skbpoolfree(struct aoedev *d);
@@ -69,25 +69,34 @@ minor_get_static(ulong *sysminor, ulong aoemaj, int aoemin)
NPERSHELF = 16,
};
+ if (aoemin >= NPERSHELF) {
+ pr_err("aoe: %s %d slots per shelf\n",
+ "static minor device numbers support only",
+ NPERSHELF);
+ error = -1;
+ goto out;
+ }
+
n = aoemaj * NPERSHELF + aoemin;
- if (aoemin >= NPERSHELF || n >= N_DEVS) {
+ if (n >= N_DEVS) {
pr_err("aoe: %s with e%ld.%d\n",
"cannot use static minor device numbers",
aoemaj, aoemin);
error = -1;
- } else {
- spin_lock_irqsave(&used_minors_lock, flags);
- if (test_bit(n, used_minors)) {
- pr_err("aoe: %s %lu\n",
- "existing device already has static minor number",
- n);
- error = -1;
- } else
- set_bit(n, used_minors);
- spin_unlock_irqrestore(&used_minors_lock, flags);
+ goto out;
}
- *sysminor = n;
+ spin_lock_irqsave(&used_minors_lock, flags);
+ if (test_bit(n, used_minors)) {
+ pr_err("aoe: %s %lu\n",
+ "existing device already has static minor number",
+ n);
+ error = -1;
+ } else
+ set_bit(n, used_minors);
+ spin_unlock_irqrestore(&used_minors_lock, flags);
+ *sysminor = n * AOE_PARTITIONS;
+out:
return error;
}
@@ -170,41 +179,50 @@ aoe_failip(struct aoedev *d)
aoe_end_request(d, rq, 0);
}
+static void
+downdev_frame(struct list_head *pos)
+{
+ struct frame *f;
+
+ f = list_entry(pos, struct frame, head);
+ list_del(pos);
+ if (f->buf) {
+ f->buf->nframesout--;
+ aoe_failbuf(f->t->d, f->buf);
+ }
+ aoe_freetframe(f);
+}
+
void
aoedev_downdev(struct aoedev *d)
{
struct aoetgt *t, **tt, **te;
- struct frame *f;
struct list_head *head, *pos, *nx;
struct request *rq;
int i;
d->flags &= ~DEVFL_UP;
- /* clean out active buffers */
+ /* clean out active and to-be-retransmitted buffers */
for (i = 0; i < NFACTIVE; i++) {
head = &d->factive[i];
- list_for_each_safe(pos, nx, head) {
- f = list_entry(pos, struct frame, head);
- list_del(pos);
- if (f->buf) {
- f->buf->nframesout--;
- aoe_failbuf(d, f->buf);
- }
- aoe_freetframe(f);
- }
+ list_for_each_safe(pos, nx, head)
+ downdev_frame(pos);
}
+ head = &d->rexmitq;
+ list_for_each_safe(pos, nx, head)
+ downdev_frame(pos);
+
/* reset window dressings */
tt = d->targets;
- te = tt + NTARGETS;
+ te = tt + d->ntargets;
for (; tt < te && (t = *tt); tt++) {
- t->maxout = t->nframes;
+ aoecmd_wreset(t);
t->nout = 0;
}
/* clean out the in-process request (if any) */
aoe_failip(d);
- d->htgt = NULL;
/* fast fail all pending I/O */
if (d->blkq) {
@@ -218,74 +236,159 @@ aoedev_downdev(struct aoedev *d)
set_capacity(d->gd, 0);
}
+/* return whether the user asked for this particular
+ * device to be flushed
+ */
+static int
+user_req(char *s, size_t slen, struct aoedev *d)
+{
+ const char *p;
+ size_t lim;
+
+ if (!d->gd)
+ return 0;
+ p = kbasename(d->gd->disk_name);
+ lim = sizeof(d->gd->disk_name);
+ lim -= p - d->gd->disk_name;
+ if (slen < lim)
+ lim = slen;
+
+ return !strncmp(s, p, lim);
+}
+
static void
-aoedev_freedev(struct aoedev *d)
+freedev(struct aoedev *d)
{
struct aoetgt **t, **e;
+ int freeing = 0;
+ unsigned long flags;
+
+ spin_lock_irqsave(&d->lock, flags);
+ if (d->flags & DEVFL_TKILL
+ && !(d->flags & DEVFL_FREEING)) {
+ d->flags |= DEVFL_FREEING;
+ freeing = 1;
+ }
+ spin_unlock_irqrestore(&d->lock, flags);
+ if (!freeing)
+ return;
- cancel_work_sync(&d->work);
+ del_timer_sync(&d->timer);
if (d->gd) {
+ aoedisk_rm_debugfs(d);
aoedisk_rm_sysfs(d);
del_gendisk(d->gd);
put_disk(d->gd);
blk_cleanup_queue(d->blkq);
}
t = d->targets;
- e = t + NTARGETS;
+ e = t + d->ntargets;
for (; t < e && *t; t++)
freetgt(d, *t);
if (d->bufpool)
mempool_destroy(d->bufpool);
skbpoolfree(d);
minor_free(d->sysminor);
- kfree(d);
+
+ spin_lock_irqsave(&d->lock, flags);
+ d->flags |= DEVFL_FREED;
+ spin_unlock_irqrestore(&d->lock, flags);
}
-int
-aoedev_flush(const char __user *str, size_t cnt)
+enum flush_parms {
+ NOT_EXITING = 0,
+ EXITING = 1,
+};
+
+static int
+flush(const char __user *str, size_t cnt, int exiting)
{
ulong flags;
struct aoedev *d, **dd;
- struct aoedev *rmd = NULL;
char buf[16];
int all = 0;
+ int specified = 0; /* flush a specific device */
+ unsigned int skipflags;
- if (cnt >= 3) {
+ skipflags = DEVFL_GDALLOC | DEVFL_NEWSIZE | DEVFL_TKILL;
+
+ if (!exiting && cnt >= 3) {
if (cnt > sizeof buf)
cnt = sizeof buf;
if (copy_from_user(buf, str, cnt))
return -EFAULT;
all = !strncmp(buf, "all", 3);
+ if (!all)
+ specified = 1;
}
+ flush_scheduled_work();
+ /* pass one: without sleeping, do aoedev_downdev */
spin_lock_irqsave(&devlist_lock, flags);
- dd = &devlist;
- while ((d = *dd)) {
+ for (d = devlist; d; d = d->next) {
spin_lock(&d->lock);
- if ((!all && (d->flags & DEVFL_UP))
- || (d->flags & (DEVFL_GDALLOC|DEVFL_NEWSIZE))
+ if (exiting) {
+ /* unconditionally take each device down */
+ } else if (specified) {
+ if (!user_req(buf, cnt, d))
+ goto cont;
+ } else if ((!all && (d->flags & DEVFL_UP))
+ || d->flags & skipflags
|| d->nopen
- || d->ref) {
- spin_unlock(&d->lock);
- dd = &d->next;
- continue;
- }
- *dd = d->next;
+ || d->ref)
+ goto cont;
+
aoedev_downdev(d);
d->flags |= DEVFL_TKILL;
+cont:
spin_unlock(&d->lock);
- d->next = rmd;
- rmd = d;
}
spin_unlock_irqrestore(&devlist_lock, flags);
- while ((d = rmd)) {
- rmd = d->next;
- del_timer_sync(&d->timer);
- aoedev_freedev(d); /* must be able to sleep */
+
+ /* pass two: call freedev, which might sleep,
+ * for aoedevs marked with DEVFL_TKILL
+ */
+restart:
+ spin_lock_irqsave(&devlist_lock, flags);
+ for (d = devlist; d; d = d->next) {
+ spin_lock(&d->lock);
+ if (d->flags & DEVFL_TKILL
+ && !(d->flags & DEVFL_FREEING)) {
+ spin_unlock(&d->lock);
+ spin_unlock_irqrestore(&devlist_lock, flags);
+ freedev(d);
+ goto restart;
+ }
+ spin_unlock(&d->lock);
}
+
+ /* pass three: remove aoedevs marked with DEVFL_FREED */
+ for (dd = &devlist, d = *dd; d; d = *dd) {
+ struct aoedev *doomed = NULL;
+
+ spin_lock(&d->lock);
+ if (d->flags & DEVFL_FREED) {
+ *dd = d->next;
+ doomed = d;
+ } else {
+ dd = &d->next;
+ }
+ spin_unlock(&d->lock);
+ if (doomed)
+ kfree(doomed->targets);
+ kfree(doomed);
+ }
+ spin_unlock_irqrestore(&devlist_lock, flags);
+
return 0;
}
+int
+aoedev_flush(const char __user *str, size_t cnt)
+{
+ return flush(str, cnt, NOT_EXITING);
+}
+
/* This has been confirmed to occur once with Tms=3*1000 due to the
* driver changing link and not processing its transmit ring. The
* problem is hard enough to solve by returning an error that I'm
@@ -332,13 +435,20 @@ aoedev_by_aoeaddr(ulong maj, int min, int do_alloc)
struct aoedev *d;
int i;
ulong flags;
- ulong sysminor;
+ ulong sysminor = 0;
spin_lock_irqsave(&devlist_lock, flags);
for (d=devlist; d; d=d->next)
if (d->aoemajor == maj && d->aoeminor == min) {
+ spin_lock(&d->lock);
+ if (d->flags & DEVFL_TKILL) {
+ spin_unlock(&d->lock);
+ d = NULL;
+ goto out;
+ }
d->ref++;
+ spin_unlock(&d->lock);
break;
}
if (d || !do_alloc || minor_get(&sysminor, maj, min) < 0)
@@ -346,6 +456,13 @@ aoedev_by_aoeaddr(ulong maj, int min, int do_alloc)
d = kcalloc(1, sizeof *d, GFP_ATOMIC);
if (!d)
goto out;
+ d->targets = kcalloc(NTARGETS, sizeof(*d->targets), GFP_ATOMIC);
+ if (!d->targets) {
+ kfree(d);
+ d = NULL;
+ goto out;
+ }
+ d->ntargets = NTARGETS;
INIT_WORK(&d->work, aoecmd_sleepwork);
spin_lock_init(&d->lock);
skb_queue_head_init(&d->skbpool);
@@ -359,10 +476,12 @@ aoedev_by_aoeaddr(ulong maj, int min, int do_alloc)
d->ref = 1;
for (i = 0; i < NFACTIVE; i++)
INIT_LIST_HEAD(&d->factive[i]);
+ INIT_LIST_HEAD(&d->rexmitq);
d->sysminor = sysminor;
d->aoemajor = maj;
d->aoeminor = min;
- d->mintimer = MINTIMER;
+ d->rttavg = RTTAVG_INIT;
+ d->rttdev = RTTDEV_INIT;
d->next = devlist;
devlist = d;
out:
@@ -396,21 +515,8 @@ freetgt(struct aoedev *d, struct aoetgt *t)
void
aoedev_exit(void)
{
- struct aoedev *d;
- ulong flags;
-
- aoe_flush_iocq();
- while ((d = devlist)) {
- devlist = d->next;
-
- spin_lock_irqsave(&d->lock, flags);
- aoedev_downdev(d);
- d->flags |= DEVFL_TKILL;
- spin_unlock_irqrestore(&d->lock, flags);
-
- del_timer_sync(&d->timer);
- aoedev_freedev(d);
- }
+ flush_scheduled_work();
+ flush(NULL, 0, EXITING);
}
int __init
diff --git a/drivers/block/aoe/aoemain.c b/drivers/block/aoe/aoemain.c
index 04793c2c701..4b987c2fefb 100644
--- a/drivers/block/aoe/aoemain.c
+++ b/drivers/block/aoe/aoemain.c
@@ -105,7 +105,7 @@ aoe_init(void)
aoechr_exit();
chr_fail:
aoedev_exit();
-
+
printk(KERN_INFO "aoe: initialisation failure.\n");
return ret;
}
diff --git a/drivers/block/aoe/aoenet.c b/drivers/block/aoe/aoenet.c
index 162c6471275..63773a90581 100644
--- a/drivers/block/aoe/aoenet.c
+++ b/drivers/block/aoe/aoenet.c
@@ -1,4 +1,4 @@
-/* Copyright (c) 2012 Coraid, Inc. See COPYING for GPL terms. */
+/* Copyright (c) 2013 Coraid, Inc. See COPYING for GPL terms. */
/*
* aoenet.c
* Ethernet portion of AoE driver
@@ -31,7 +31,7 @@ enum {
static char aoe_iflist[IFLISTSZ];
module_param_string(aoe_iflist, aoe_iflist, IFLISTSZ, 0600);
-MODULE_PARM_DESC(aoe_iflist, "aoe_iflist=\"dev1 [dev2 ...]\"");
+MODULE_PARM_DESC(aoe_iflist, "aoe_iflist=dev1[,dev2...]");
static wait_queue_head_t txwq;
static struct ktstate kts;
@@ -52,13 +52,18 @@ static struct sk_buff_head skbtxq;
/* enters with txlock held */
static int
-tx(void)
+tx(int id) __must_hold(&txlock)
{
struct sk_buff *skb;
+ struct net_device *ifp;
while ((skb = skb_dequeue(&skbtxq))) {
spin_unlock_irq(&txlock);
- dev_queue_xmit(skb);
+ ifp = skb->dev;
+ if (dev_queue_xmit(skb) == NET_XMIT_DROP && net_ratelimit())
+ pr_warn("aoe: packet could not be sent on %s. %s\n",
+ ifp ? ifp->name : "netif",
+ "consider increasing tx_queue_len");
spin_lock_irq(&txlock);
}
return 0;
@@ -119,8 +124,8 @@ aoenet_xmit(struct sk_buff_head *queue)
}
}
-/*
- * (1) len doesn't include the header by default. I want this.
+/*
+ * (1) len doesn't include the header by default. I want this.
*/
static int
aoenet_rcv(struct sk_buff *skb, struct net_device *ifp, struct packet_type *pt, struct net_device *orig_dev)
@@ -200,7 +205,8 @@ aoenet_init(void)
kts.lock = &txlock;
kts.fn = tx;
kts.waitq = &txwq;
- kts.name = "aoe_tx";
+ kts.id = 0;
+ snprintf(kts.name, sizeof(kts.name), "aoe_tx%d", kts.id);
if (aoe_ktstart(&kts))
return -EAGAIN;
dev_add_pack(&aoe_pt);
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2026-01-09 05:24:33 +0000