← Back to team overview

kernel-packages team mailing list archive

[Bug 1370425] [NEW] kernel bug seen while try to use madvise system call with MADV_HWPOISON mode

 

You have been subscribed to a public bug:

Problem Description
====================
kernel bug seen while try to use madvise system call with MADV_HWPOISON mode
 
---uname output---
Linux u10thp 3.16.0-9-generic #14-Ubuntu SMP Fri Aug 15 15:03:36 UTC 2014 ppc64le ppc64le ppc64le GNU/Linux
 
Machine Type = Power 8 
 
Steps to Reproduce
====================
1.  Install Ubuntu 14.10 guest on PowerKVM.
2.  Setup hugepage backing guest VM.
3.  Try madv_poison.c code to test madvise sys. call with HWPOISON mode(test code is attached).
	gcc -o madv_poison madv_poison.c
	./madv_poison -C -i 1 		(1 - shm_test)

Ubuntu 14.10 LE throws kernel bug :
root@u10thp:~# ./madv_poison -C -i 1
vm.memory_failure_early_kill = 0
[pid 2301] start page-poisoning test
[pid 2301] there are 1 shm_child
[pid 2301] have spawned 1 processes
[pid 2301] wait for Pid 2304
[pid 2304] shm dirty poisoning page 0x3fffa7ce0000
[ 7905.009001] Injecting memory failure for page 0xe6a7 at 0x3fffa7ce0000
[ 7905.009359] MCE 0xe6a7: dirty LRU page recovery: Recovered
[pid 2304] writing 2
[ 7905.009901] ------------[ cut here ]------------
[ 7905.010164] kernel BUG at /build/buildd/linux-3.16.0/arch/powerpc/mm/fault.c:180!
[ 7905.010396] Oops: Exception in kernel mode, sig: 5 [#234]
[ 7905.010438] SMP NR_CPUS=2048 NUMA pSeries
[ 7905.010480] Modules linked in: pseries_rng rtc_generic ohci_pci
[ 7905.010614] CPU: 0 PID: 2304 Comm: madv_poison Tainted: G      D       3.16.0-9-generic #14-Ubuntu
[ 7905.010686] task: c0000000e0a92a60 ti: c0000000e09e8000 task.ti: c0000000e09e8000
[ 7905.010746] NIP: c0000000009e3314 LR: c0000000009e2e54 CTR: 0000000000000000
[ 7905.010864] REGS: c0000000e09eb990 TRAP: 0700   Tainted: G      D        (3.16.0-9-generic)
[ 7905.010924] MSR: 8000000000029033 <SF,EE,ME,IR,DR,RI,LE>  CR: 28002882  XER: 00000000
[ 7905.011125] CFAR: c0000000009e3170 SOFTE: 1 
GPR00: c0000000009e2e54 c0000000e09ebc10 c0000000013742e0 0000000000000010 
GPR04: c0000000e0b37ff8 00003fffa7ce0000 00000000000000a9 0000000000000000 
GPR08: 0000000000000000 0000000000000010 c0000000e0a92a60 0000000000000020 
GPR12: 0000000048002884 c00000000fe40000 0000000000000000 0000000000000000 
GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000000000000 
GPR20: 00000000000000a9 0000000000000000 c0000000e0597a40 c0000000e022b060 
GPR24: 0000000000000010 c0000000e022b000 c000000000009568 00003fffa7ce0000 
GPR28: 0000000000000000 0000000000000000 0000000002000000 c0000000e09ebea0 
[ 7905.012189] NIP [c0000000009e3314] do_page_fault+0x984/0x990
[ 7905.012241] LR [c0000000009e2e54] do_page_fault+0x4c4/0x990
[ 7905.012281] Call Trace:
[ 7905.012361] [c0000000e09ebc10] [c0000000009e2e54] do_page_fault+0x4c4/0x990 (unreliable)
[ 7905.012434] [c0000000e09ebe30] [c000000000009568] handle_page_fault+0x10/0x30
[ 7905.012494] Instruction dump:
[ 7905.012580] e92d0290 e8690460 38630060 4b7274d9 60000000 e93f0108 3bc00000 792a97e3 
[ 7905.012683] 4082f77c 3bc00009 60000000 4bfff774 <0fe00000> 00000000 00000000 3c4c0099 
[ 7905.012845] ---[ end trace a48a199a061eed79 ]---
[ 7905.019084] 
[pid 2301] Ins 0: Pid 2304: failed - shared memory test
[pid 2301] 	!!! Page Poisoning Test is FAILED (1 failures found). !!!

[pid 2301] page-poisoning test done!
root@u10thp:~# 

== Comment: #1 - Kalpana Shetty <kalshett@xxxxxxxxxx> -  ==
The test code works fine with x86/Ubuntu VM so if it is not supported on power then it should have thrown an error not supported as it does with PowerKVM / RHEL 7 VM.

Intel/Ubuntu 14.04 VM:			=================================> Working fine.
root@u04vm14:~# ./madv_poison -C -i 1               (shm_test case)
vm.memory_failure_early_kill = 0
[pid 7325] start page-poisoning test
[pid 7325] there are 1 shm_child
[pid 7325] have spawned 1 processes
[pid 7325] wait for Pid 7328
[pid 7328] shm dirty poisoning page 0x7f60ca8ea000
[pid 7328] writing 2
[pid 7328] signal 7 code 4 addr 0x7f60ca8ea000
[pid 7328] pass: recovered
[pid 7325] Ins 0: Pid 7328: pass - shared memory test
[pid 7325] 	!!! Page Poisoning Test got PASS. !!!

[pid 7325] page-poisoning test done!

PowerKVM / RHEL 7 VM:
[root@rhel7-web-VM1 ~]# ./madv_poison -C -i 1
sysctl: cannot stat /proc/sys/vm/memory_failure_early_kill: No such file or directory
[pid 11512] start page-poisoning test
[pid 11512] there are 1 shm_child
[pid 11512] have spawned 1 processes
[pid 11514] shm dirty poisoning page 0x3fff84d60000
[pid 11512] wait for Pid 11514
[pid 11514] failed: Kernel doesn't support poison injection ============================> unsupported error.
[pid 11512] Ins 0: Pid 11514: failed - shared memory test
[pid 11512] 	!!! Page Poisoning Test is FAILED (1 failures found). !!!

[pid 11512] page-poisoning test done!

** Affects: linux (Ubuntu)
     Importance: Undecided
         Status: New


** Tags: architecture-ppc64le bot-comment bugnameltc-115345 severity-high targetmilestone-inin---
-- 
kernel bug seen while try to use madvise system call with MADV_HWPOISON mode
https://bugs.launchpad.net/bugs/1370425
You received this bug notification because you are a member of Kernel Packages, which is subscribed to linux in Ubuntu.