show_mem panic
I just found that I can reproduce the show_mem panic with the discontig version of show_mem. Here's the backtrace:
Free swap: 0kB
Node ID: 0
Unable to handle kernel paging request at virtual address
a0007ff69f650000
swapper[1]: Oops 11012296146944 [1]
Pid: 1, CPU 0, comm: swapper
psr : 0000121008022018 ifs : 800000000000068f ip : [<a000000100054761>] Not tainted
ip is at show_mem+0x161/0x320
unat: 0000000000000000 pfs : 000000000000068f rsc : 0000000000000003
rnat: 0000000000000000 bsps: 0000000000000000 pr : 00000000000069a5
ldrs: 0000000000000000 ccv : 0000000000000000 fpsr: 0009804c8a70033f csd : 0000000000000000 ssd : 0000000000000000
b0 : a000000100054700 b6 : e0001030042713c0 b7 : e000000000fffc00 f6 : 10023f8d5875e00000000 f7 : 10008fa00000000000000
f8 : 1003e0000000007f672ec f9 : 10019fece5d8f5c0f735c
f10 : 0fffbccccccccc8c00000 f11 : 1003e0000000000000002
r1 : a00000010098cdc0 r2 : ffffffffffff01c8 r3 : 0000000000000000 r8 : 000000000000000b r9 : 0000000000000001 r10 : 0000000000000000 r11 : a00000010079eb98 r12 : e000103014997b40 r13 : e000103014990000 r14 : a0000001006c0d28 r15 : a0007ff69f650000 r16 : a0007ff69f650020 r17 : 0000000000000000 r18 : a0007ff69f650000 r19 : 000000000001f000 r20 : a0000001006c4698 r21 : e000103006855720 r22 : a00000010078d018 r23 : 0000000000000000 r24 : 000000000000000b r25 : a00000010079cf70 r26 : e000103014997a68 r27 : e000103014997a60 r28 : a00000010083f0c4 r29 : 000000000000fe2c r30 : 0000000000000000 r31 : a00000010079eb50
Call Trace:
[<a000000100014ea0>] show_stack+0x80/0xa0
spà00103014997710 bspà001030149913c8
[<a000000100039730>] die+0x170/0x200
spà001030149978e0 bspà00103014991390
[<a0000001000532e0>] ia64_do_page_fault+0x340/0x940
spà001030149978e0 bspà00103014991328
[<a00000010000de20>] ia64_leave_kernel+0x0/0x260
spà00103014997970 bspà00103014991328
[<a000000100054760>] show_mem+0x160/0x320
spà00103014997b40 bspà001030149912b0
[<a0000001003a79c0>] sysrq_handle_showmem+0x20/0x40
spà00103014997b40 bspà00103014991298
[<a0000001003a7ef0>] __handle_sysrq_nolock+0x110/0x260
spà00103014997b40 bspà00103014991250
[<a0000001003a7db0>] handle_sysrq+0x70/0xa0
spà00103014997b40 bspà00103014991220
[<a0000001003a9030>] sn_receive_chars+0x330/0x3c0
spà00103014997b40 bspà00103014991198
[<a0000001003a9ae0>] sn_sal_interrupt+0x220/0x2a0
spà00103014997b50 bspà00103014991150
[<a000000100011600>] handle_IRQ_event+0xa0/0x120
spà00103014997c00 bspà00103014991108
[<a000000100012040>] do_IRQ+0x2a0/0x3a0
spà00103014997c00 bspà001030149910b8
[<a000000100013e90>] ia64_handle_irq+0xb0/0x1a0
spà00103014997c00 bspà00103014991080
[<a00000010000de20>] ia64_leave_kernel+0x0/0x260
spà00103014997c00 bspà00103014991080
[<a0000001000c0700>] panic+0x2e0/0x320
spà00103014997dd0 bspà00103014990ff0
[<a00000010061a050>] mount_block_root+0x150/0x220
spà00103014997e10 bspà00103014990f80
[<a00000010061a270>] prepare_namespace+0x70/0x160
spà00103014997e30 bspà00103014990f60
[<a000000100009320>] init+0xe0/0x2a0
spà00103014997e30 bspà00103014990f48
[<a000000100016b50>] kernel_thread_helper+0xd0/0x100
spà00103014997e30 bspà00103014990f20
[<a000000100009040>] ia64_invoke_kernel_thread_helper+0x20/0x40
spà00103014997e30 bspà00103014990f20
[<e0001030042713c0>] 0xe0001030042713c0
spà00103014997e30 bspà00103014990f20
[<a0000001000c05a0>] panic+0x180/0x320
spà00103014997e30 bspà00103014990f20
<0>Kernel panic: Aiee, killing interrupt handler!In interrupt handler - not syncing
It's in the inner loop looking at the pages in the first node's node_mem_map. I'm almost positive that this code worked in the past, Andreas, when you reproduced this on a tiger machine was it with CONFIG_VIRTUAL_MEM_MAP enabled?
Thanks,
Jesse
Quoted 4 lines · Jesse Barnes
It's in the inner loop looking at the pages in the first node's
node_mem_map. I'm almost positive that this code worked in the past, Andreas, when you reproduced this on a tiger machine was it withCONFIG_VIRTUAL_MEM_MAPenabled?
Bjorn just informed me that John Marvin was taking a look at this. It looks like we need to make sure the page we're about to look at is valid for the virtual mem map case. And btw, it looks like the contig version of show_mem should be fixed: total shouldn't include non-existant pages (if we add pfn_valid checks to both the discontig and contig versions that should be an easy fix).
Jesse
Quoted 4 lines · [email protected] (Jesse Barnes) writes:
It's in the inner loop looking at the pages in the first node's
node_mem_map. I'm almost positive that this code worked in the past, Andreas, when you reproduced this on a tiger machine was it withCONFIG_VIRTUAL_MEM_MAPenabled?
Yes, that's our standard configuration.
Andreas.
Signature · 4 lines
Andreas Schwab, SuSE Labs, [email protected] SuSE Linux AG, Maxfeldstraße 5, 90409 Nürnberg, Germany Key fingerprint = 58CA 54C7 6D53 942B 1756 01D3 44D5 214B 8276 4ED5 "And now for something completely different."