AuxXxilium/linux_dsm_epyc7002
1
0
Fork 0
mirror of https://github.com/AuxXxilium/linux_dsm_epyc7002.git synced 2024-12-16 00:06:46 +07:00
Code Issues Actions Packages Projects Releases Wiki Activity

powerpc/64s/pseries: machine check convert to use common event code

Browse Source 
The common machine_check_event data structures and queues are mostly
platform independent, with powernv decoding SRR1/DSISR/etc., into
machine_check_event objects.

This patch converts pseries to use this infrastructure by decoding
fwnmi/rtas data into machine_check_event objects.

This allows queueing to be used by a subsequent change to delay the
virtual mode handling of machine checks that occur in kernel space
where it is unsafe to switch immediately to virtual mode, similarly
to powernv.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Fix implicit fallthrough warnings in mce_handle_error()]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20190802105709.27696-10-npiggin@gmail.com
This commit is contained in:
Nicholas Piggin 2019-08-02 20:56:34 +10:00 committed by Michael Ellerman
parent 7290f3b3d3
commit 9ca766f989
3 changed files with 242 additions and 276 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
arch/powerpc/include/asm/mce.h
View File

@ -30,6 +30,10 @@ enum MCE_Disposition {
enum MCE_Initiator { enum MCE_Initiator {
MCE_INITIATOR_UNKNOWN = 0, MCE_INITIATOR_UNKNOWN = 0,
MCE_INITIATOR_CPU = 1, MCE_INITIATOR_CPU = 1,
MCE_INITIATOR_PCI = 2,
MCE_INITIATOR_ISA = 3,
MCE_INITIATOR_MEMORY= 4,
MCE_INITIATOR_POWERMGM = 5,
}; };
enum MCE_ErrorType { enum MCE_ErrorType {
@ -41,6 +45,8 @@ enum MCE_ErrorType {
MCE_ERROR_TYPE_USER = 5, MCE_ERROR_TYPE_USER = 5,
MCE_ERROR_TYPE_RA = 6, MCE_ERROR_TYPE_RA = 6,
MCE_ERROR_TYPE_LINK = 7, MCE_ERROR_TYPE_LINK = 7,
MCE_ERROR_TYPE_DCACHE = 8,
MCE_ERROR_TYPE_ICACHE = 9,
}; };
enum MCE_ErrorClass { enum MCE_ErrorClass {

34
arch/powerpc/kernel/mce.c
View File

@ -325,7 +325,7 @@ static void machine_check_process_queued_event(struct irq_work *work)
void machine_check_print_event_info(struct machine_check_event *evt, void machine_check_print_event_info(struct machine_check_event *evt,
bool user_mode, bool in_guest) bool user_mode, bool in_guest)
{ {
const char *level, *sevstr, *subtype, *err_type; const char *level, *sevstr, *subtype, *err_type, *initiator;
uint64_t ea = 0, pa = 0; uint64_t ea = 0, pa = 0;
int n = 0; int n = 0;
char dar_str[50]; char dar_str[50];
@ -410,6 +410,28 @@ void machine_check_print_event_info(struct machine_check_event *evt,
break; break;
} }
switch(evt->initiator) {
case MCE_INITIATOR_CPU:
initiator = "CPU";
break;
case MCE_INITIATOR_PCI:
initiator = "PCI";
break;
case MCE_INITIATOR_ISA:
initiator = "ISA";
break;
case MCE_INITIATOR_MEMORY:
initiator = "Memory";
break;
case MCE_INITIATOR_POWERMGM:
initiator = "Power Management";
break;
case MCE_INITIATOR_UNKNOWN:
default:
initiator = "Unknown";
break;
}
switch (evt->error_type) { switch (evt->error_type) {
case MCE_ERROR_TYPE_UE: case MCE_ERROR_TYPE_UE:
err_type = "UE"; err_type = "UE";
@ -476,6 +498,14 @@ void machine_check_print_event_info(struct machine_check_event *evt,
if (evt->u.link_error.effective_address_provided) if (evt->u.link_error.effective_address_provided)
ea = evt->u.link_error.effective_address; ea = evt->u.link_error.effective_address;
break; break;
case MCE_ERROR_TYPE_DCACHE:
err_type = "D-Cache";
subtype = "Unknown";
break;
case MCE_ERROR_TYPE_ICACHE:
err_type = "I-Cache";
subtype = "Unknown";
break;
default: default:
case MCE_ERROR_TYPE_UNKNOWN: case MCE_ERROR_TYPE_UNKNOWN:
err_type = "Unknown"; err_type = "Unknown";
@ -508,6 +538,8 @@ void machine_check_print_event_info(struct machine_check_event *evt,
level, evt->cpu, evt->srr0, (void *)evt->srr0, pa_str); level, evt->cpu, evt->srr0, (void *)evt->srr0, pa_str);
} }
printk("%sMCE: CPU%d: Initiator %s\n", level, evt->cpu, initiator);
subtype = evt->error_class < ARRAY_SIZE(mc_error_class) ? subtype = evt->error_class < ARRAY_SIZE(mc_error_class) ?
mc_error_class[evt->error_class] : "Unknown"; mc_error_class[evt->error_class] : "Unknown";
printk("%sMCE: CPU%d: %s\n", level, evt->cpu, subtype); printk("%sMCE: CPU%d: %s\n", level, evt->cpu, subtype);

478
arch/powerpc/platforms/pseries/ras.c
View File

@ -76,6 +76,7 @@ struct pseries_mc_errorlog {
#define MC_ERROR_TYPE_UE 0x00 #define MC_ERROR_TYPE_UE 0x00
#define MC_ERROR_TYPE_SLB 0x01 #define MC_ERROR_TYPE_SLB 0x01
#define MC_ERROR_TYPE_ERAT 0x02 #define MC_ERROR_TYPE_ERAT 0x02
#define MC_ERROR_TYPE_UNKNOWN 0x03
#define MC_ERROR_TYPE_TLB 0x04 #define MC_ERROR_TYPE_TLB 0x04
#define MC_ERROR_TYPE_D_CACHE 0x05 #define MC_ERROR_TYPE_D_CACHE 0x05
#define MC_ERROR_TYPE_I_CACHE 0x07 #define MC_ERROR_TYPE_I_CACHE 0x07
@ -87,6 +88,9 @@ struct pseries_mc_errorlog {
#define MC_ERROR_UE_LOAD_STORE 3 #define MC_ERROR_UE_LOAD_STORE 3
#define MC_ERROR_UE_PAGE_TABLE_WALK_LOAD_STORE 4 #define MC_ERROR_UE_PAGE_TABLE_WALK_LOAD_STORE 4
#define UE_EFFECTIVE_ADDR_PROVIDED 0x40
#define UE_LOGICAL_ADDR_PROVIDED 0x20
#define MC_ERROR_SLB_PARITY 0 #define MC_ERROR_SLB_PARITY 0
#define MC_ERROR_SLB_MULTIHIT 1 #define MC_ERROR_SLB_MULTIHIT 1
#define MC_ERROR_SLB_INDETERMINATE 2 #define MC_ERROR_SLB_INDETERMINATE 2
@ -113,27 +117,6 @@ static inline u8 rtas_mc_error_sub_type(const struct pseries_mc_errorlog *mlog)
} }
} }
static
inline u64 rtas_mc_get_effective_addr(const struct pseries_mc_errorlog *mlog)
{
__be64 addr = 0;
switch (mlog->error_type) {
case MC_ERROR_TYPE_UE:
if (mlog->sub_err_type & 0x40)
addr = mlog->effective_address;
break;
case MC_ERROR_TYPE_SLB:
case MC_ERROR_TYPE_ERAT:
case MC_ERROR_TYPE_TLB:
if (mlog->sub_err_type & 0x80)
addr = mlog->effective_address;
default:
break;
}
return be64_to_cpu(addr);
}
/* /*
* Enable the hotplug interrupt late because processing them may touch other * Enable the hotplug interrupt late because processing them may touch other
* devices or systems (e.g. hugepages) that have not been initialized at the * devices or systems (e.g. hugepages) that have not been initialized at the
@ -511,149 +494,55 @@ int pSeries_system_reset_exception(struct pt_regs *regs)
return 0; /* need to perform reset */ return 0; /* need to perform reset */
} }
#define VAL_TO_STRING(ar, val) \
(((val) < ARRAY_SIZE(ar)) ? ar[(val)] : "Unknown")
static void pseries_print_mce_info(struct pt_regs *regs, static int mce_handle_error(struct pt_regs *regs, struct rtas_error_log *errp)
struct rtas_error_log *errp)
{ {
const char *level, *sevstr; struct mce_error_info mce_err = { 0 };
unsigned long eaddr = 0, paddr = 0;
struct pseries_errorlog *pseries_log; struct pseries_errorlog *pseries_log;
struct pseries_mc_errorlog *mce_log; struct pseries_mc_errorlog *mce_log;
int disposition = rtas_error_disposition(errp);
int initiator = rtas_error_initiator(errp);
int severity = rtas_error_severity(errp);
u8 error_type, err_sub_type; u8 error_type, err_sub_type;
u64 addr;
u8 initiator = rtas_error_initiator(errp);
int disposition = rtas_error_disposition(errp);
static const char * const initiators[] = { if (initiator == RTAS_INITIATOR_UNKNOWN)
[0] = "Unknown", mce_err.initiator = MCE_INITIATOR_UNKNOWN;
[1] = "CPU", else if (initiator == RTAS_INITIATOR_CPU)
[2] = "PCI", mce_err.initiator = MCE_INITIATOR_CPU;
[3] = "ISA", else if (initiator == RTAS_INITIATOR_PCI)
[4] = "Memory", mce_err.initiator = MCE_INITIATOR_PCI;
[5] = "Power Mgmt", else if (initiator == RTAS_INITIATOR_ISA)
}; mce_err.initiator = MCE_INITIATOR_ISA;
static const char * const mc_err_types[] = { else if (initiator == RTAS_INITIATOR_MEMORY)
[0] = "UE", mce_err.initiator = MCE_INITIATOR_MEMORY;
[1] = "SLB", else if (initiator == RTAS_INITIATOR_POWERMGM)
[2] = "ERAT", mce_err.initiator = MCE_INITIATOR_POWERMGM;
[3] = "Unknown", else
[4] = "TLB", mce_err.initiator = MCE_INITIATOR_UNKNOWN;
[5] = "D-Cache",
[6] = "Unknown",
[7] = "I-Cache",
};
static const char * const mc_ue_types[] = {
[0] = "Indeterminate",
[1] = "Instruction fetch",
[2] = "Page table walk ifetch",
[3] = "Load/Store",
[4] = "Page table walk Load/Store",
};
/* SLB sub errors valid values are 0x0, 0x1, 0x2 */ if (severity == RTAS_SEVERITY_NO_ERROR)
static const char * const mc_slb_types[] = { mce_err.severity = MCE_SEV_NO_ERROR;
[0] = "Parity", else if (severity == RTAS_SEVERITY_EVENT)
[1] = "Multihit", mce_err.severity = MCE_SEV_WARNING;
[2] = "Indeterminate", else if (severity == RTAS_SEVERITY_WARNING)
}; mce_err.severity = MCE_SEV_WARNING;
else if (severity == RTAS_SEVERITY_ERROR_SYNC)
mce_err.severity = MCE_SEV_SEVERE;
else if (severity == RTAS_SEVERITY_ERROR)
mce_err.severity = MCE_SEV_SEVERE;
else if (severity == RTAS_SEVERITY_FATAL)
mce_err.severity = MCE_SEV_FATAL;
else
mce_err.severity = MCE_SEV_FATAL;
/* TLB and ERAT sub errors valid values are 0x1, 0x2, 0x3 */ if (severity <= RTAS_SEVERITY_ERROR_SYNC)
static const char * const mc_soft_types[] = { mce_err.sync_error = true;
[0] = "Unknown", else
[1] = "Parity", mce_err.sync_error = false;
[2] = "Multihit",
[3] = "Indeterminate",
};
if (!rtas_error_extended(errp)) { mce_err.error_type = MCE_ERROR_TYPE_UNKNOWN;
pr_err("Machine check interrupt: Missing extended error log\n"); mce_err.error_class = MCE_ECLASS_UNKNOWN;
return;
}
pseries_log = get_pseries_errorlog(errp, PSERIES_ELOG_SECT_ID_MCE);
if (pseries_log == NULL)
return;
mce_log = (struct pseries_mc_errorlog *)pseries_log->data;
error_type = mce_log->error_type;
err_sub_type = rtas_mc_error_sub_type(mce_log);
switch (rtas_error_severity(errp)) {
case RTAS_SEVERITY_NO_ERROR:
level = KERN_INFO;
sevstr = "Harmless";
break;
case RTAS_SEVERITY_WARNING:
level = KERN_WARNING;
sevstr = "";
break;
case RTAS_SEVERITY_ERROR:
case RTAS_SEVERITY_ERROR_SYNC:
level = KERN_ERR;
sevstr = "Severe";
break;
case RTAS_SEVERITY_FATAL:
default:
level = KERN_ERR;
sevstr = "Fatal";
break;
}
#ifdef CONFIG_PPC_BOOK3S_64
/* Display faulty slb contents for SLB errors. */
if (error_type == MC_ERROR_TYPE_SLB)
slb_dump_contents(local_paca->mce_faulty_slbs);
#endif
printk("%s%s Machine check interrupt [%s]\n", level, sevstr,
disposition == RTAS_DISP_FULLY_RECOVERED ?
"Recovered" : "Not recovered");
if (user_mode(regs)) {
printk("%s NIP: [%016lx] PID: %d Comm: %s\n", level,
regs->nip, current->pid, current->comm);
} else {
printk("%s NIP [%016lx]: %pS\n", level, regs->nip,
(void *)regs->nip);
}
printk("%s Initiator: %s\n", level,
VAL_TO_STRING(initiators, initiator));
switch (error_type) {
case MC_ERROR_TYPE_UE:
printk("%s Error type: %s [%s]\n", level,
VAL_TO_STRING(mc_err_types, error_type),
VAL_TO_STRING(mc_ue_types, err_sub_type));
break;
case MC_ERROR_TYPE_SLB:
printk("%s Error type: %s [%s]\n", level,
VAL_TO_STRING(mc_err_types, error_type),
VAL_TO_STRING(mc_slb_types, err_sub_type));
break;
case MC_ERROR_TYPE_ERAT:
case MC_ERROR_TYPE_TLB:
printk("%s Error type: %s [%s]\n", level,
VAL_TO_STRING(mc_err_types, error_type),
VAL_TO_STRING(mc_soft_types, err_sub_type));
break;
default:
printk("%s Error type: %s\n", level,
VAL_TO_STRING(mc_err_types, error_type));
break;
}
addr = rtas_mc_get_effective_addr(mce_log);
if (addr)
printk("%s Effective address: %016llx\n", level, addr);
}
static int mce_handle_error(struct rtas_error_log *errp)
{
struct pseries_errorlog *pseries_log;
struct pseries_mc_errorlog *mce_log;
int disposition = rtas_error_disposition(errp);
u8 error_type;
if (!rtas_error_extended(errp)) if (!rtas_error_extended(errp))
goto out; goto out;
@ -664,6 +553,105 @@ static int mce_handle_error(struct rtas_error_log *errp)
mce_log = (struct pseries_mc_errorlog *)pseries_log->data; mce_log = (struct pseries_mc_errorlog *)pseries_log->data;
error_type = mce_log->error_type; error_type = mce_log->error_type;
err_sub_type = rtas_mc_error_sub_type(mce_log);
switch (mce_log->error_type) {
case MC_ERROR_TYPE_UE:
mce_err.error_type = MCE_ERROR_TYPE_UE;
switch (err_sub_type) {
case MC_ERROR_UE_IFETCH:
mce_err.u.ue_error_type = MCE_UE_ERROR_IFETCH;
break;
case MC_ERROR_UE_PAGE_TABLE_WALK_IFETCH:
mce_err.u.ue_error_type = MCE_UE_ERROR_PAGE_TABLE_WALK_IFETCH;
break;
case MC_ERROR_UE_LOAD_STORE:
mce_err.u.ue_error_type = MCE_UE_ERROR_LOAD_STORE;
break;
case MC_ERROR_UE_PAGE_TABLE_WALK_LOAD_STORE:
mce_err.u.ue_error_type = MCE_UE_ERROR_PAGE_TABLE_WALK_LOAD_STORE;
break;
case MC_ERROR_UE_INDETERMINATE:
default:
mce_err.u.ue_error_type = MCE_UE_ERROR_INDETERMINATE;
break;
}
if (mce_log->sub_err_type & UE_EFFECTIVE_ADDR_PROVIDED)
eaddr = be64_to_cpu(mce_log->effective_address);
if (mce_log->sub_err_type & UE_LOGICAL_ADDR_PROVIDED) {
paddr = be64_to_cpu(mce_log->logical_address);
} else if (mce_log->sub_err_type & UE_EFFECTIVE_ADDR_PROVIDED) {
unsigned long pfn;
pfn = addr_to_pfn(regs, eaddr);
if (pfn != ULONG_MAX)
paddr = pfn << PAGE_SHIFT;
}
break;
case MC_ERROR_TYPE_SLB:
mce_err.error_type = MCE_ERROR_TYPE_SLB;
switch (err_sub_type) {
case MC_ERROR_SLB_PARITY:
mce_err.u.slb_error_type = MCE_SLB_ERROR_PARITY;
break;
case MC_ERROR_SLB_MULTIHIT:
mce_err.u.slb_error_type = MCE_SLB_ERROR_MULTIHIT;
break;
case MC_ERROR_SLB_INDETERMINATE:
default:
mce_err.u.slb_error_type = MCE_SLB_ERROR_INDETERMINATE;
break;
}
if (mce_log->sub_err_type & 0x80)
eaddr = be64_to_cpu(mce_log->effective_address);
break;
case MC_ERROR_TYPE_ERAT:
mce_err.error_type = MCE_ERROR_TYPE_ERAT;
switch (err_sub_type) {
case MC_ERROR_ERAT_PARITY:
mce_err.u.erat_error_type = MCE_ERAT_ERROR_PARITY;
break;
case MC_ERROR_ERAT_MULTIHIT:
mce_err.u.erat_error_type = MCE_ERAT_ERROR_MULTIHIT;
break;
case MC_ERROR_ERAT_INDETERMINATE:
default:
mce_err.u.erat_error_type = MCE_ERAT_ERROR_INDETERMINATE;
break;
}
if (mce_log->sub_err_type & 0x80)
eaddr = be64_to_cpu(mce_log->effective_address);
break;
case MC_ERROR_TYPE_TLB:
mce_err.error_type = MCE_ERROR_TYPE_TLB;
switch (err_sub_type) {
case MC_ERROR_TLB_PARITY:
mce_err.u.tlb_error_type = MCE_TLB_ERROR_PARITY;
break;
case MC_ERROR_TLB_MULTIHIT:
mce_err.u.tlb_error_type = MCE_TLB_ERROR_MULTIHIT;
break;
case MC_ERROR_TLB_INDETERMINATE:
default:
mce_err.u.tlb_error_type = MCE_TLB_ERROR_INDETERMINATE;
break;
}
if (mce_log->sub_err_type & 0x80)
eaddr = be64_to_cpu(mce_log->effective_address);
break;
case MC_ERROR_TYPE_D_CACHE:
mce_err.error_type = MCE_ERROR_TYPE_DCACHE;
break;
case MC_ERROR_TYPE_I_CACHE:
mce_err.error_type = MCE_ERROR_TYPE_DCACHE;
break;
case MC_ERROR_TYPE_UNKNOWN:
default:
mce_err.error_type = MCE_ERROR_TYPE_UNKNOWN;
break;
}
#ifdef CONFIG_PPC_BOOK3S_64 #ifdef CONFIG_PPC_BOOK3S_64
if (disposition == RTAS_DISP_NOT_RECOVERED) { if (disposition == RTAS_DISP_NOT_RECOVERED) {
@ -682,99 +670,25 @@ static int mce_handle_error(struct rtas_error_log *errp)
slb_save_contents(local_paca->mce_faulty_slbs); slb_save_contents(local_paca->mce_faulty_slbs);
flush_and_reload_slb(); flush_and_reload_slb();
disposition = RTAS_DISP_FULLY_RECOVERED; disposition = RTAS_DISP_FULLY_RECOVERED;
rtas_set_disposition_recovered(errp);
break; break;
default: default:
break; break;
} }
} else if (disposition == RTAS_DISP_LIMITED_RECOVERY) {
/* Platform corrected itself but could be degraded */
printk(KERN_ERR "MCE: limited recovery, system may "
"be degraded\n");
disposition = RTAS_DISP_FULLY_RECOVERED;
} }
#endif #endif
out: out:
save_mce_event(regs, disposition == RTAS_DISP_FULLY_RECOVERED,
&mce_err, regs->nip, eaddr, paddr);
return disposition; return disposition;
} }
#ifdef CONFIG_MEMORY_FAILURE
static DEFINE_PER_CPU(int, rtas_ue_count);
static DEFINE_PER_CPU(unsigned long, rtas_ue_paddr[MAX_MC_EVT]);
#define UE_EFFECTIVE_ADDR_PROVIDED 0x40
#define UE_LOGICAL_ADDR_PROVIDED 0x20
static void pseries_hwpoison_work_fn(struct work_struct *work)
{
unsigned long paddr;
int index;
while (__this_cpu_read(rtas_ue_count) > 0) {
index = __this_cpu_read(rtas_ue_count) - 1;
paddr = __this_cpu_read(rtas_ue_paddr[index]);
memory_failure(paddr >> PAGE_SHIFT, 0);
__this_cpu_dec(rtas_ue_count);
}
}
static DECLARE_WORK(hwpoison_work, pseries_hwpoison_work_fn);
static void queue_ue_paddr(unsigned long paddr)
{
int index;
index = __this_cpu_inc_return(rtas_ue_count) - 1;
if (index >= MAX_MC_EVT) {
__this_cpu_dec(rtas_ue_count);
return;
}
this_cpu_write(rtas_ue_paddr[index], paddr);
schedule_work(&hwpoison_work);
}
static void pseries_do_memory_failure(struct pt_regs *regs,
struct pseries_mc_errorlog *mce_log)
{
unsigned long paddr;
if (mce_log->sub_err_type & UE_LOGICAL_ADDR_PROVIDED) {
paddr = be64_to_cpu(mce_log->logical_address);
} else if (mce_log->sub_err_type & UE_EFFECTIVE_ADDR_PROVIDED) {
unsigned long pfn;
pfn = addr_to_pfn(regs,
be64_to_cpu(mce_log->effective_address));
if (pfn == ULONG_MAX)
return;
paddr = pfn << PAGE_SHIFT;
} else {
return;
}
queue_ue_paddr(paddr);
}
static void pseries_process_ue(struct pt_regs *regs,
struct rtas_error_log *errp)
{
struct pseries_errorlog *pseries_log;
struct pseries_mc_errorlog *mce_log;
if (!rtas_error_extended(errp))
return;
pseries_log = get_pseries_errorlog(errp, PSERIES_ELOG_SECT_ID_MCE);
if (!pseries_log)
return;
mce_log = (struct pseries_mc_errorlog *)pseries_log->data;
if (mce_log->error_type == MC_ERROR_TYPE_UE)
pseries_do_memory_failure(regs, mce_log);
}
#else
static inline void pseries_process_ue(struct pt_regs *regs,
struct rtas_error_log *errp) { }
#endif /*CONFIG_MEMORY_FAILURE */
/* /*
* Process MCE rtas errlog event. * Process MCE rtas errlog event.
*/ */
@ -795,48 +709,50 @@ static void mce_process_errlog_event(struct irq_work *work)
* Return 1 if corrected (or delivered a signal). * Return 1 if corrected (or delivered a signal).
* Return 0 if there is nothing we can do. * Return 0 if there is nothing we can do.
*/ */
static int recover_mce(struct pt_regs *regs, struct rtas_error_log *err) static int recover_mce(struct pt_regs *regs, struct machine_check_event *evt)
{ {
int recovered = 0; int recovered = 0;
int disposition = rtas_error_disposition(err);
pseries_print_mce_info(regs, err);
if (!(regs->msr & MSR_RI)) { if (!(regs->msr & MSR_RI)) {
/* If MSR_RI isn't set, we cannot recover */ /* If MSR_RI isn't set, we cannot recover */
pr_err("Machine check interrupt unrecoverable: MSR(RI=0)\n"); pr_err("Machine check interrupt unrecoverable: MSR(RI=0)\n");
recovered = 0; recovered = 0;
} else if (evt->disposition == MCE_DISPOSITION_RECOVERED) {
} else if (disposition == RTAS_DISP_FULLY_RECOVERED) {
/* Platform corrected itself */ /* Platform corrected itself */
recovered = 1; recovered = 1;
} else if (evt->severity == MCE_SEV_FATAL) {
} else if (disposition == RTAS_DISP_LIMITED_RECOVERY) { /* Fatal machine check */
/* Platform corrected itself but could be degraded */ pr_err("Machine check interrupt is fatal\n");
printk(KERN_ERR "MCE: limited recovery, system may " recovered = 0;
"be degraded\n");
recovered = 1;
} else if (user_mode(regs) && !is_global_init(current) &&
rtas_error_severity(err) == RTAS_SEVERITY_ERROR_SYNC) {
/*
* If we received a synchronous error when in userspace
* kill the task. Firmware may report details of the fail
* asynchronously, so we can't rely on the target and type
* fields being valid here.
*/
printk(KERN_ERR "MCE: uncorrectable error, killing task "
"%s:%d\n", current->comm, current->pid);
_exception(SIGBUS, regs, BUS_MCEERR_AR, regs->nip);
recovered = 1;
} }
pseries_process_ue(regs, err); if (!recovered && evt->sync_error) {
/*
/* Queue irq work to log this rtas event later. */ * Try to kill processes if we get a synchronous machine check
irq_work_queue(&mce_errlog_process_work); * (e.g., one caused by execution of this instruction). This
* will devolve into a panic if we try to kill init or are in
* an interrupt etc.
*
* TODO: Queue up this address for hwpoisioning later.
* TODO: This is not quite right for d-side machine
* checks ->nip is not necessarily the important
* address.
*/
if ((user_mode(regs))) {
_exception(SIGBUS, regs, BUS_MCEERR_AR, regs->nip);
recovered = 1;
} else if (die_will_crash()) {
/*
* die() would kill the kernel, so better to go via
* the platform reboot code that will log the
* machine check.
*/
recovered = 0;
} else {
die("Machine check", regs, SIGBUS);
recovered = 1;
}
}
return recovered; return recovered;
} }
@ -853,14 +769,21 @@ static int recover_mce(struct pt_regs *regs, struct rtas_error_log *err)
*/ */
int pSeries_machine_check_exception(struct pt_regs *regs) int pSeries_machine_check_exception(struct pt_regs *regs)
{ {
struct rtas_error_log *errp; struct machine_check_event evt;
if (fwnmi_active) { if (!get_mce_event(&evt, MCE_EVENT_RELEASE))
fwnmi_release_errinfo(); return 0;
errp = fwnmi_get_errlog();
if (errp && recover_mce(regs, errp)) /* Print things out */
return 1; if (evt.version != MCE_V1) {
pr_err("Machine Check Exception, Unknown event version %d !\n",
evt.version);
return 0;
} }
machine_check_print_event_info(&evt, user_mode(regs), false);
if (recover_mce(regs, &evt))
return 1;
return 0; return 0;
} }
@ -877,7 +800,12 @@ long pseries_machine_check_realmode(struct pt_regs *regs)
* to panic. Hence we will call it as soon as we go into * to panic. Hence we will call it as soon as we go into
* virtual mode. * virtual mode.
*/ */
disposition = mce_handle_error(errp); disposition = mce_handle_error(regs, errp);
fwnmi_release_errinfo();
/* Queue irq work to log this rtas event later. */
irq_work_queue(&mce_errlog_process_work);
if (disposition == RTAS_DISP_FULLY_RECOVERED) if (disposition == RTAS_DISP_FULLY_RECOVERED)
return 1; return 1;
} }