|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [Xen-devel] [PATCH v3] x86: show remote CPU state upon fatal NMI or unknown MCE
Quite frequently the watchdog would hit an innocent CPU, e.g. one
trying to acquire a spin lock a remote CPU holds for extended periods
of time, or a random CPU in TSC calbration rendezvous. In such cases
the register and stack dump for that CPU doesn't really help in the
analysis of the problem.
To keep things reasonable on large systems, only log CS:RIP by default.
This can be overridden via a new command line option such that full
register/stack state would get logged.
Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
Reviewed-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
---
v3: Add timeout. (Andrew, I hope the R-b can be kept despite the logic
not being a 1:1 clone of what the crash dump path does.)
v2: Pass flag to fatal_trap().
--- a/docs/misc/xen-command-line.markdown
+++ b/docs/misc/xen-command-line.markdown
@@ -198,6 +198,14 @@ Permit Xen to use Address Space Identifi
tags the TLB entries with an ID per vcpu. This allows for guest TLB flushes
to be performed without the overhead of a complete TLB flush.
+### async-show-all
+> `= <boolean>`
+
+> Default: `false`
+
+Forces all CPUs' full state to be logged upon certain fatal asynchronous
+exceptions (watchdog NMIs and unexpected MCEs).
+
### ats
> `= <boolean>`
--- a/xen/arch/x86/cpu/mcheck/mce.c
+++ b/xen/arch/x86/cpu/mcheck/mce.c
@@ -77,7 +77,7 @@ static void unexpected_machine_check(con
{
console_force_unlock();
printk("Unexpected Machine Check Exception\n");
- fatal_trap(regs);
+ fatal_trap(regs, 1);
}
--- a/xen/arch/x86/nmi.c
+++ b/xen/arch/x86/nmi.c
@@ -488,7 +488,7 @@ bool_t nmi_watchdog_tick(const struct cp
console_force_unlock();
printk("Watchdog timer detects that CPU%d is stuck!\n",
smp_processor_id());
- fatal_trap(regs);
+ fatal_trap(regs, 1);
}
}
else
--- a/xen/arch/x86/traps.c
+++ b/xen/arch/x86/traps.c
@@ -525,6 +525,25 @@ void vcpu_show_execution_state(struct vc
vcpu_unpause(v);
}
+static cpumask_t show_state_mask;
+static bool_t opt_show_all;
+boolean_param("async-show-all", opt_show_all);
+
+static int nmi_show_execution_state(const struct cpu_user_regs *regs, int cpu)
+{
+ if ( !cpumask_test_cpu(cpu, &show_state_mask) )
+ return 0;
+
+ if ( opt_show_all )
+ show_execution_state(regs);
+ else
+ printk(XENLOG_ERR "CPU%d @ %04x:%08lx (%pS)\n", cpu, regs->cs,
regs->rip,
+ guest_mode(regs) ? _p(regs->rip) : NULL);
+ cpumask_clear_cpu(cpu, &show_state_mask);
+
+ return 1;
+}
+
static const char *trapstr(unsigned int trapnr)
{
static const char * const strings[] = {
@@ -544,7 +563,7 @@ static const char *trapstr(unsigned int
* are disabled). In such situations we can't do much that is safe. We try to
* print out some tracing and then we just spin.
*/
-void fatal_trap(const struct cpu_user_regs *regs)
+void fatal_trap(const struct cpu_user_regs *regs, bool_t show_remote)
{
static DEFINE_PER_CPU(char, depth);
unsigned int trapnr = regs->entry_vector;
@@ -570,6 +589,34 @@ void fatal_trap(const struct cpu_user_re
printk("Faulting linear address: %p\n", _p(cr2));
show_page_walk(cr2);
}
+
+ if ( show_remote )
+ {
+ unsigned int msecs, pending;
+
+ cpumask_andnot(&show_state_mask, &cpu_online_map,
+ cpumask_of(smp_processor_id()));
+ set_nmi_callback(nmi_show_execution_state);
+ /* Ensure new callback is set before sending out the NMI. */
+ smp_wmb();
+ smp_send_nmi_allbutself();
+
+ /* Wait at most 10ms for some other CPU to respond. */
+ msecs = 10;
+ pending = cpumask_weight(&show_state_mask);
+ while ( pending && msecs-- )
+ {
+ unsigned int left;
+
+ mdelay(1);
+ left = cpumask_weight(&show_state_mask);
+ if ( left < pending )
+ {
+ pending = left;
+ msecs = 10;
+ }
+ }
+ }
}
panic("FATAL TRAP: vector = %d (%s)\n"
@@ -1711,7 +1758,7 @@ void do_page_fault(struct cpu_user_regs
{
console_start_sync();
printk("Xen SM%cP violation\n", (pf_type == smep_fault) ? 'E' :
'A');
- fatal_trap(regs);
+ fatal_trap(regs, 0);
}
if ( pf_type != real_fault )
@@ -1782,7 +1829,7 @@ void __init do_early_page_fault(struct c
console_start_sync();
printk("Early fatal page fault at %04x:%p (cr2=%p, ec=%04x)\n",
regs->cs, _p(regs->eip), _p(cr2), regs->error_code);
- fatal_trap(regs);
+ fatal_trap(regs, 0);
}
}
@@ -3598,7 +3645,7 @@ static void pci_serr_error(const struct
default: /* 'fatal' */
console_force_unlock();
printk("\n\nNMI - PCI system error (SERR)\n");
- fatal_trap(regs);
+ fatal_trap(regs, 0);
}
}
@@ -3613,7 +3660,7 @@ static void io_check_error(const struct
default: /* 'fatal' */
console_force_unlock();
printk("\n\nNMI - I/O ERROR\n");
- fatal_trap(regs);
+ fatal_trap(regs, 0);
}
outb((inb(0x61) & 0x0f) | 0x08, 0x61); /* clear-and-disable IOCK */
@@ -3633,7 +3680,7 @@ static void unknown_nmi_error(const stru
console_force_unlock();
printk("Uhhuh. NMI received for unknown reason %02x.\n", reason);
printk("Do you have a strange power saving mode enabled?\n");
- fatal_trap(regs);
+ fatal_trap(regs, 0);
}
}
--- a/xen/arch/x86/x86_64/entry.S
+++ b/xen/arch/x86/x86_64/entry.S
@@ -558,6 +558,7 @@ exception_with_ints_disabled:
/* No special register assumptions. */
FATAL_exception_with_ints_disabled:
+ xorl %esi,%esi
movq %rsp,%rdi
call fatal_trap
BUG /* fatal_trap() shouldn't return. */
--- a/xen/include/asm-x86/processor.h
+++ b/xen/include/asm-x86/processor.h
@@ -541,7 +541,7 @@ void show_registers(const struct cpu_use
void show_execution_state(const struct cpu_user_regs *regs);
#define dump_execution_state() run_in_exception_handler(show_execution_state)
void show_page_walk(unsigned long addr);
-void noreturn fatal_trap(const struct cpu_user_regs *regs);
+void noreturn fatal_trap(const struct cpu_user_regs *regs, bool_t show_remote);
void compat_show_guest_stack(struct vcpu *v,
const struct cpu_user_regs *regs, int lines);
Attachment:
x86-NMI-show-global-state.patch _______________________________________________ Xen-devel mailing list Xen-devel@xxxxxxxxxxxxx http://lists.xen.org/xen-devel
|
 |
Lists.xenproject.org is hosted with RackSpace, monitoring our |