On 2012年05月08日 11:46, Benjamin Herrenschmidt wrote:
> Hi Wang !
>
> Does this patch fixes it for you ?
>
Sorry, this patch doesn't work. And my system crashed again with the patch.
======================================================
# kernel BUG at /usr/src/kernels/linux/arch/powerpc/kernel/irq.c:188!
cpu 0x0: Vector: 700 (Program Check) at [c00000026ffebbb0]
pc: c00000000000ea9c: .__check_irq_replay+0x7c/0x90
lr: c000000000010578: .arch_local_irq_restore+0x38/0x90
sp: c00000026ffebe30
msr: 8000000000029032
current = 0xc000000000e27be0
paca = 0xc000000003580000 softe: 0 irq_happened: 0x01
pid = 0, comm = swapper/0
kernel BUG at /usr/src/kernels/linux/arch/powerpc/kernel/irq.c:188!
enter ? for help
[link register ] c000000000010578 .arch_local_irq_restore+0x38/0x90
[c00000026ffebe30] c000000000f42100 softirq_vec+0x0/0x80 (unreliable)
[c00000026ffebea0] c000000000085854 .__do_softirq+0xa4/0x2a0
[c00000026ffebf90] c0000000000229b8 .call_do_softirq+0x14/0x24
[c000000000edf870] c0000000000106c8 .do_softirq+0xf8/0x130
[c000000000edf910] c000000000085544 .irq_exit+0xc4/0xf0
[c000000000edf990] c0000000000100a4 .do_IRQ+0xe4/0x310
[c000000000edfa50] c0000000000038c0 hardware_interrupt_common+0x140/0x180
--- Exception: 501 (Hardware Interrupt) at c0000000000105b4
.arch_local_irq_restore+0x74/0x90
[c000000000edfd40] c000000000058480 .pSeries_idle+0x10/0x40 (unreliable)
[c000000000edfdb0] c000000000017d70 .cpu_idle+0x190/0x290
[c000000000edfe70] c00000000000b308 .rest_init+0x88/0xa0
[c000000000edfef0] c0000000008c0d1c .start_kernel+0x554/0x574
[c000000000edff90] c000000000009658 .start_here_common+0x20/0x48
0:mon> e
cpu 0x0: Vector: 700 (Program Check) at [c00000026ffebbb0]
pc: c00000000000ea9c: .__check_irq_replay+0x7c/0x90
lr: c000000000010578: .arch_local_irq_restore+0x38/0x90
sp: c00000026ffebe30
msr: 8000000000029032
current = 0xc000000000e27be0
paca = 0xc000000003580000 softe: 0 irq_happened: 0x01
pid = 0, comm = swapper/0
kernel BUG at /usr/src/kernels/linux/arch/powerpc/kernel/irq.c:188!
0:mon> r
R00 = 0000000000000001 R16 = 0000000003680000
R01 = c00000026ffebe30 R17 = 000000000021ed0f
R02 = c000000000edd228 R18 = 000000000021efbb
R03 = 0000000000000500 R19 = 000000000021ee84
R04 = 0000000000000000 R20 = c000000000f42100
R05 = 00000000000007ea R21 = 0000000000000000
R06 = 00000000273f6d30 R22 = c000000000955b80
R07 = 00363d0e68097e11 R23 = c000000000955b80
R08 = 00000000008c0000 R24 = 000000000000000a
R09 = c000000003580000 R25 = 0000000000000000
R10 = 0000000000000001 R26 = c000000000edc100
R11 = 0000000000000000 R27 = c00000026ffe8000
R12 = 0000000000000002 R28 = 0000000000000000
R13 = c000000003580000 R29 = c000000000f42100
R14 = 0000000002e1fa78 R30 = c000000000e60890
R15 = 0000000001173000 R31 = 0000000000000040
pc = c00000000000ea9c .__check_irq_replay+0x7c/0x90
cfar= c00000000000ea3c .__check_irq_replay+0x1c/0x90
lr = c000000000010578 .arch_local_irq_restore+0x38/0x90
msr = 8000000000029032 cr = 28000048
ctr = c000000000063f70 xer = 0000000000000001 trap = 700
0:mon> t
[link register ] c000000000010578 .arch_local_irq_restore+0x38/0x90
[c00000026ffebe30] c000000000f42100 softirq_vec+0x0/0x80 (unreliable)
[c00000026ffebea0] c000000000085854 .__do_softirq+0xa4/0x2a0
[c00000026ffebf90] c0000000000229b8 .call_do_softirq+0x14/0x24
[c000000000edf870] c0000000000106c8 .do_softirq+0xf8/0x130
[c000000000edf910] c000000000085544 .irq_exit+0xc4/0xf0
[c000000000edf990] c0000000000100a4 .do_IRQ+0xe4/0x310
[c000000000edfa50] c0000000000038c0 hardware_interrupt_common+0x140/0x180
--- Exception: 501 (Hardware Interrupt) at c0000000000105b4
.arch_local_irq_restore+0x74/0x90
[c000000000edfd40] c000000000058480 .pSeries_idle+0x10/0x40 (unreliable)
[c000000000edfdb0] c000000000017d70 .cpu_idle+0x190/0x290
[c000000000edfe70] c00000000000b308 .rest_init+0x88/0xa0
[c000000000edfef0] c0000000008c0d1c .start_kernel+0x554/0x574
[c000000000edff90] c000000000009658 .start_here_common+0x20/0x48
0:mon> di
> From 249f8649bf95a4c3e6637284754a165c1d83c394 Mon Sep 17 00:00:00 2001
> From: Benjamin Herrenschmidt <b...@kernel.crashing.org>
> Date: Tue, 8 May 2012 13:31:59 +1000
> Subject: [PATCH 2/3] powerpc/irq: Fix bug with new lazy IRQ handling code
>
> We had a case where we could turn on hard interrupts while
> leaving the PACA_IRQ_HARD_DIS bit set in the PACA. This can
> in turn cause a BUG_ON() to hit in __check_irq_replay() due
> to interrupt state getting out of sync.
>
> The assembly code was also way too convoluted. Instead, we
> now leave it to the C code to do the right thing which ends
> up being smaller and more readable.
>
> Signed-off-by: Benjamin Herrenschmidt <b...@kernel.crashing.org>
> ---
> arch/powerpc/kernel/entry_64.S | 18 ------------------
> arch/powerpc/kernel/irq.c | 8 +++++++-
> 2 files changed, 7 insertions(+), 19 deletions(-)
>
> diff --git a/arch/powerpc/kernel/entry_64.S b/arch/powerpc/kernel/entry_64.S
> index fd46046..29f1357 100644
> --- a/arch/powerpc/kernel/entry_64.S
> +++ b/arch/powerpc/kernel/entry_64.S
> @@ -763,16 +763,6 @@ do_work:
> SOFT_DISABLE_INTS(r3,r4)
> 1: bl .preempt_schedule_irq
>
> - /* Hard-disable interrupts again (and update PACA) */
> -#ifdef CONFIG_PPC_BOOK3E
> - wrteei 0
> -#else
> - ld r10,PACAKMSR(r13) /* Get kernel MSR without EE */
> - mtmsrd r10,1
> -#endif /* CONFIG_PPC_BOOK3E */
> - li r0,PACA_IRQ_HARD_DIS
> - stb r0,PACAIRQHAPPENED(r13)
> -
> /* Re-test flags and eventually loop */
> clrrdi r9,r1,THREAD_SHIFT
> ld r4,TI_FLAGS(r9)
> @@ -783,14 +773,6 @@ do_work:
> user_work:
> #endif /* CONFIG_PREEMPT */
>
> - /* Enable interrupts */
> -#ifdef CONFIG_PPC_BOOK3E
> - wrteei 1
> -#else
> - ori r10,r10,MSR_EE
> - mtmsrd r10,1
> -#endif /* CONFIG_PPC_BOOK3E */
> -
> andi. r0,r4,_TIF_NEED_RESCHED
> beq 1f
> bl .restore_interrupts
> diff --git a/arch/powerpc/kernel/irq.c b/arch/powerpc/kernel/irq.c
> index 5ec1b23..3717fb5 100644
> --- a/arch/powerpc/kernel/irq.c
> +++ b/arch/powerpc/kernel/irq.c
> @@ -260,11 +260,17 @@ EXPORT_SYMBOL(arch_local_irq_restore);
> * if they are currently disabled. This is typically called before
> * schedule() or do_signal() when returning to userspace. We do it
> * in C to avoid the burden of dealing with lockdep etc...
> + *
> + * NOTE: This is called with interrupts hard disabled but not marked
> + * as such in paca->irq_happened, so we need to resync this.
> */
> void restore_interrupts(void)
> {
> - if (irqs_disabled())
> + if (irqs_disabled()) {
> + local_paca->irq_happened |= PACA_IRQ_HARD_DIS;
> local_irq_enable();
> + } else
> + __hard_irq_enable();
> }
>
> #endif /* CONFIG_PPC64 */
_______________________________________________
Linuxppc-dev mailing list
Linuxppc-dev@lists.ozlabs.org
https://lists.ozlabs.org/listinfo/linuxppc-dev
