The patch titled memcg: fix false positive VM_BUG on non-SMP has been added to the -mm tree. Its filename is memcg-fix-false-positive-vm_bug-on-non-smp.patch Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/SubmitChecklist when testing your code *** See http://userweb.kernel.org/~akpm/stuff/added-to-mm.txt to find out what to do about this The current -mm tree may be found at http://userweb.kernel.org/~akpm/mmotm/ ------------------------------------------------------ Subject: memcg: fix false positive VM_BUG on non-SMP From: Kirill A. Shutemov <kirill@xxxxxxxxxxxxx> ------------[ cut here ]------------ kernel BUG at mm/memcontrol.c:2155! invalid opcode: 0000 [#1] last sysfs file: Pid: 18, comm: sh Not tainted 2.6.37-rc3 #3 /Bochs EIP: 0060:[<c10731b2>] EFLAGS: 00000246 CPU: 0 EIP is at mem_cgroup_move_account+0xe2/0xf0 EAX: 00000004 EBX: c6f931d4 ECX: c681c300 EDX: c681c000 ESI: c681c300 EDI: ffffffea EBP: c681c000 ESP: c46f3e30 DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 0068 Process sh (pid: 18, ti=c46f2000 task=c6826e60 task.ti=c46f2000) Stack: 00000155 c681c000 0805f000 c46ee180 c46f3e5c c7058820 c1074d37 00000000 08060000 c46db9a0 c46ec080 c7058820 0805f000 08060000 c46f3e98 c1074c50 c106c75e c46f3e98 c46ec080 08060000 0805ffff c46db9a0 c46f3e98 c46e0340 Call Trace: [<c1074d37>] ? mem_cgroup_move_charge_pte_range+0xe7/0x130 [<c1074c50>] ? mem_cgroup_move_charge_pte_range+0x0/0x130 [<c106c75e>] ? walk_page_range+0xee/0x1d0 [<c10725d6>] ? mem_cgroup_move_task+0x66/0x90 [<c1074c50>] ? mem_cgroup_move_charge_pte_range+0x0/0x130 [<c1072570>] ? mem_cgroup_move_task+0x0/0x90 [<c1042616>] ? cgroup_attach_task+0x136/0x200 [<c1042878>] ? cgroup_tasks_write+0x48/0xc0 [<c1041e9e>] ? cgroup_file_write+0xde/0x220 [<c101398d>] ? do_page_fault+0x17d/0x3f0 [<c108a79d>] ? alloc_fd+0x2d/0xd0 [<c1041dc0>] ? cgroup_file_write+0x0/0x220 [<c1077ba2>] ? vfs_write+0x92/0xc0 [<c1077c81>] ? sys_write+0x41/0x70 [<c1140e3d>] ? syscall_call+0x7/0xb Code: 03 00 74 09 8b 44 24 04 e8 1c f1 ff ff 89 73 04 8d 86 b0 00 00 00 b9 01 00 00 00 89 da 31 ff e8 65 f5 ff ff e9 4d ff ff ff 0f 0b <0f> 0b 0f 0b 0f 0b 90 8d b4 26 00 00 00 00 83 ec 10 8b 0d f4 e3 EIP: [<c10731b2>] mem_cgroup_move_account+0xe2/0xf0 SS:ESP 0068:c46f3e30 ---[ end trace 7daa1582159b6532 ]--- lock_page_cgroup and unlock_page_cgroup are implemented using bit_spinlock. bit_spinlock doesn't touch the bit if we are on non-SMP machine, so we can't use the bit to check whether the lock was taken. Let's introduce is_page_cgroup_locked based on bit_spin_is_locked instead of PageCgroupLocked to fix it. Signed-off-by: Kirill A. Shutemov <kirill@xxxxxxxxxxxxx> Cc: Daisuke Nishimura <nishimura@xxxxxxxxxxxxxxxxx> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx> Cc: Balbir Singh <balbir@xxxxxxxxxxxxxxxxxx> Cc: <stable@xxxxxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- include/linux/page_cgroup.h | 7 +++++-- mm/memcontrol.c | 2 +- 2 files changed, 6 insertions(+), 3 deletions(-) diff -puN include/linux/page_cgroup.h~memcg-fix-false-positive-vm_bug-on-non-smp include/linux/page_cgroup.h --- a/include/linux/page_cgroup.h~memcg-fix-false-positive-vm_bug-on-non-smp +++ a/include/linux/page_cgroup.h @@ -59,8 +59,6 @@ static inline void ClearPageCgroup##unam static inline int TestClearPageCgroup##uname(struct page_cgroup *pc) \ { return test_and_clear_bit(PCG_##lname, &pc->flags); } -TESTPCGFLAG(Locked, LOCK) - /* Cache flag is set only once (at allocation) */ TESTPCGFLAG(Cache, CACHE) CLEARPCGFLAG(Cache, CACHE) @@ -104,6 +102,11 @@ static inline void unlock_page_cgroup(st bit_spin_unlock(PCG_LOCK, &pc->flags); } +static inline int is_page_cgroup_locked(struct page_cgroup *pc) +{ + return bit_spin_is_locked(PCG_LOCK, &pc->flags); +} + #else /* CONFIG_CGROUP_MEM_RES_CTLR */ struct page_cgroup; diff -puN mm/memcontrol.c~memcg-fix-false-positive-vm_bug-on-non-smp mm/memcontrol.c --- a/mm/memcontrol.c~memcg-fix-false-positive-vm_bug-on-non-smp +++ a/mm/memcontrol.c @@ -2152,7 +2152,7 @@ static void __mem_cgroup_move_account(st { VM_BUG_ON(from == to); VM_BUG_ON(PageLRU(pc->page)); - VM_BUG_ON(!PageCgroupLocked(pc)); + VM_BUG_ON(!is_page_cgroup_locked(pc)); VM_BUG_ON(!PageCgroupUsed(pc)); VM_BUG_ON(pc->mem_cgroup != from); _ Patches currently in -mm which might be from kirill@xxxxxxxxxxxxx are memcg-fix-false-positive-vm_bug-on-non-smp.patch linux-next.patch -- To unsubscribe from this list: send the line "unsubscribe mm-commits" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html