* [Adeos-main] [pull request extension] x86-64: Fix potential irq stack corruption
@ 2009-12-15 9:43 Jan Kiszka
0 siblings, 0 replies; only message in thread
From: Jan Kiszka @ 2009-12-15 9:43 UTC (permalink / raw)
To: Philippe Gerum; +Cc: adeos-main
While this still does not heal the irq stack corruption I see here with
CONFIG_PREEMPT, it remains a bug that may have triggered severe issues,
though only very infrequently. OK, three I-pipe bugs fixed, but the big
one is still open. Sigh...
The following changes since commit b62b3fcb9ec752c89a5185e7863e15f004c32d42:
Jan Kiszka (1):
x86: Drop redundant ipipe_suspend_domain from cpu_idle
are available in the git repository at:
git://git.kiszka.org/ipipe-2.6 queues/2.6.31-x86
Jan Kiszka (1):
x86: Make stack switch in call_softirq atomic
arch/x86/include/asm/irqflags.h | 2 ++
arch/x86/kernel/entry_64.S | 4 ++++
2 files changed, 6 insertions(+), 0 deletions(-)
------
x86: Make stack switch in call_softirq atomic
The interrupt stack must always be switched with hard IRQs disabled as
I-pipe uses it as well. If not, we risk subtle stack corruptions when
preempting call_softirq at the wrong instruction.
Signed-off-by: Jan Kiszka <jan.kiszka@domain.hid>
---
arch/x86/include/asm/irqflags.h | 2 ++
arch/x86/kernel/entry_64.S | 4 ++++
2 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/arch/x86/include/asm/irqflags.h b/arch/x86/include/asm/irqflags.h
index 1baceba..da1e655 100644
--- a/arch/x86/include/asm/irqflags.h
+++ b/arch/x86/include/asm/irqflags.h
@@ -223,12 +223,14 @@ static inline unsigned long __raw_local_irq_save(void)
#define ENABLE_INTERRUPTS(clobbers) sti
#endif /* CONFIG_X86_64 */
#define ENABLE_INTERRUPTS_HW_COND sti
+#define DISABLE_INTERRUPTS_HW_COND cli
#define DISABLE_INTERRUPTS_HW(clobbers) cli
#define ENABLE_INTERRUPTS_HW(clobbers) sti
#else /* !CONFIG_IPIPE */
#define ENABLE_INTERRUPTS(x) sti
#define DISABLE_INTERRUPTS(x) cli
#define ENABLE_INTERRUPTS_HW_COND
+#define DISABLE_INTERRUPTS_HW_COND
#define DISABLE_INTERRUPTS_HW(clobbers) DISABLE_INTERRUPTS(clobbers)
#define ENABLE_INTERRUPTS_HW(clobbers) ENABLE_INTERRUPTS(clobbers)
#endif /* !CONFIG_IPIPE */
diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S
index b876c16..5237029 100644
--- a/arch/x86/kernel/entry_64.S
+++ b/arch/x86/kernel/entry_64.S
@@ -1413,14 +1413,18 @@ ENTRY(call_softirq)
CFI_REL_OFFSET rbp,0
mov %rsp,%rbp
CFI_DEF_CFA_REGISTER rbp
+ DISABLE_INTERRUPTS_HW_COND
incl PER_CPU_VAR(irq_count)
cmove PER_CPU_VAR(irq_stack_ptr),%rsp
+ ENABLE_INTERRUPTS_HW_COND
push %rbp # backlink for old unwinder
call __do_softirq
+ DISABLE_INTERRUPTS_HW_COND
leaveq
CFI_DEF_CFA_REGISTER rsp
CFI_ADJUST_CFA_OFFSET -8
decl PER_CPU_VAR(irq_count)
+ ENABLE_INTERRUPTS_HW_COND
ret
CFI_ENDPROC
END(call_softirq)
--
1.6.0.2
^ permalink raw reply related [flat|nested] only message in thread
only message in thread, other threads:[~2009-12-15 9:43 UTC | newest]
Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-12-15 9:43 [Adeos-main] [pull request extension] x86-64: Fix potential irq stack corruption Jan Kiszka
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.