diff options
Diffstat (limited to 'arch/ia64/kernel/mca_drv.c')
| -rw-r--r-- | arch/ia64/kernel/mca_drv.c | 77 |
1 files changed, 50 insertions, 27 deletions
diff --git a/arch/ia64/kernel/mca_drv.c b/arch/ia64/kernel/mca_drv.c index afc1403799c..94f8bf777af 100644 --- a/arch/ia64/kernel/mca_drv.c +++ b/arch/ia64/kernel/mca_drv.c @@ -3,7 +3,7 @@ * Purpose: Generic MCA handling layer * * Copyright (C) 2004 FUJITSU LIMITED - * Copyright (C) Hidetoshi Seto (seto.hidetoshi@jp.fujitsu.com) + * Copyright (C) 2004 Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com> * Copyright (C) 2005 Silicon Graphics, Inc * Copyright (C) 2005 Keith Owens <kaos@sgi.com> * Copyright (C) 2006 Russ Anderson <rja@sgi.com> @@ -14,7 +14,6 @@ #include <linux/interrupt.h> #include <linux/irq.h> #include <linux/kallsyms.h> -#include <linux/smp_lock.h> #include <linux/bootmem.h> #include <linux/acpi.h> #include <linux/timer.h> @@ -23,12 +22,12 @@ #include <linux/smp.h> #include <linux/workqueue.h> #include <linux/mm.h> +#include <linux/slab.h> #include <asm/delay.h> #include <asm/machvec.h> #include <asm/page.h> #include <asm/ptrace.h> -#include <asm/system.h> #include <asm/sal.h> #include <asm/mca.h> @@ -159,7 +158,8 @@ mca_handler_bh(unsigned long paddr, void *iip, unsigned long ipsr) ia64_mlogbuf_dump(); printk(KERN_ERR "OS_MCA: process [cpu %d, pid: %d, uid: %d, " "iip: %p, psr: 0x%lx,paddr: 0x%lx](%s) encounters MCA.\n", - raw_smp_processor_id(), current->pid, current->uid, + raw_smp_processor_id(), current->pid, + from_kuid(&init_user_ns, current_uid()), iip, ipsr, paddr, current->comm); spin_lock(&mca_bh_lock); @@ -349,7 +349,7 @@ init_record_index_pools(void) /* - 3 - */ slidx_pool.max_idx = (rec_max_size/sect_min_size) * 2 + 1; - slidx_pool.buffer = (slidx_list_t *) + slidx_pool.buffer = kmalloc(slidx_pool.max_idx * sizeof(slidx_list_t), GFP_KERNEL); return slidx_pool.buffer ? 0 : -ENOMEM; @@ -439,7 +439,7 @@ is_mca_global(peidx_table_t *peidx, pal_bus_check_info_t *pbci, * @peidx: pointer of index of processor error section * * Return value: - * target address on Success / 0 on Failue + * target address on Success / 0 on Failure */ static u64 get_target_identifier(peidx_table_t *peidx) @@ -602,11 +602,40 @@ recover_from_platform_error(slidx_table_t *slidx, peidx_table_t *peidx, default: break; } + } else if (psp->cc && !psp->bc) { /* Cache error */ + status = recover_from_read_error(slidx, peidx, pbci, sos); } return status; } +/* + * recover_from_tlb_check + * @peidx: pointer of index of processor error section + * + * Return value: + * 1 on Success / 0 on Failure + */ +static int +recover_from_tlb_check(peidx_table_t *peidx) +{ + sal_log_mod_error_info_t *smei; + pal_tlb_check_info_t *ptci; + + smei = (sal_log_mod_error_info_t *)peidx_tlb_check(peidx, 0); + ptci = (pal_tlb_check_info_t *)&(smei->check_info); + + /* + * Look for signature of a duplicate TLB DTC entry, which is + * a SW bug and always fatal. + */ + if (ptci->op == PAL_TLB_CHECK_OP_PURGE + && !(ptci->itr || ptci->dtc || ptci->itc)) + return fatal_mca("Duplicate TLB entry"); + + return mca_recovered("TLB check recovered"); +} + /** * recover_from_processor_error * @platform: whether there are some platform error section or not @@ -618,13 +647,6 @@ recover_from_platform_error(slidx_table_t *slidx, peidx_table_t *peidx, * Return value: * 1 on Success / 0 on Failure */ -/* - * Later we try to recover when below all conditions are satisfied. - * 1. Only one processor error section is exist. - * 2. BUS_CHECK is exist and the others are not exist.(Except TLB_CHECK) - * 3. The entry of BUS_CHECK_INFO is 1. - * 4. "External bus error" flag is set and the others are not set. - */ static int recover_from_processor_error(int platform, slidx_table_t *slidx, @@ -652,38 +674,39 @@ recover_from_processor_error(int platform, slidx_table_t *slidx, return fatal_mca("error not contained"); /* + * Look for recoverable TLB check + */ + if (psp->tc && !(psp->cc || psp->bc || psp->rc || psp->uc)) + return recover_from_tlb_check(peidx); + + /* * The cache check and bus check bits have four possible states * cc bc - * 0 0 Weird record, not recovered - * 1 0 Cache error, not recovered - * 0 1 I/O error, attempt recovery * 1 1 Memory error, attempt recovery + * 1 0 Cache error, attempt recovery + * 0 1 I/O error, attempt recovery + * 0 0 Other error type, not recovered */ - if (psp->bc == 0 || pbci == NULL) - return fatal_mca("No bus check"); + if (psp->cc == 0 && (psp->bc == 0 || pbci == NULL)) + return fatal_mca("No cache or bus check"); /* - * Sorry, we cannot handle so many. + * Cannot handle more than one bus check. */ if (peidx_bus_check_num(peidx) > 1) return fatal_mca("Too many bus checks"); - /* - * Well, here is only one bus error. - */ + if (pbci->ib) return fatal_mca("Internal Bus error"); - if (pbci->cc) - return fatal_mca("Cache-cache error"); if (pbci->eb && pbci->bsi > 0) return fatal_mca("External bus check fatal status"); /* - * This is a local MCA and estimated as recoverble external bus error. - * (e.g. a load from poisoned memory) - * This means "there are some platform errors". + * This is a local MCA and estimated as a recoverable error. */ if (platform) return recover_from_platform_error(slidx, peidx, pbci, sos); + /* * On account of strange SAL error record, we cannot recover. */ |