Re: kernel BUG at mm/page_alloc.c:LINE!

[Vlastimil Babka]

On 1/17/19 3:33 AM, syzbot wrote:
> Hello> syzbot found the following crash on:
> 
> HEAD commit:    b808822a75a3 Add linux-next specific files for 20190111
> git tree:       linux-next
> console output: https://syzkaller.appspot.com/x/log.txt?x=16a471d8c00000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=c052ead0aed5001b
> dashboard link: https://syzkaller.appspot.com/bug?extid=80dd4798c16c634daf15
> compiler:       gcc (GCC) 9.0.0 20181231 (experimental)
> 
> Unfortunately, I don't have any reproducer for this crash yet.
> 
> IMPORTANT: if you fix the bug, please add the following tag to the commit:
> Reported-by: syzbot+80dd4798c16c634daf15@xxxxxxxxxxxxxxxxxxxxxxxxx
> 
> ------------[ cut here ]------------
> kernel BUG at mm/page_alloc.c:3112!

Why does the mail subject say LINE, anyway?

> invalid opcode: 0000 [#1] PREEMPT SMP KASAN
> CPU: 0 PID: 1043 Comm: kcompactd0 Not tainted 5.0.0-rc1-next-20190111 #10
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS  
> Google 01/01/2011
> RIP: 0010:__isolate_free_page+0x4a8/0x680 mm/page_alloc.c:3112

That's BUG_ON(!PageBuddy(page)); in __isolate_free_page().

> Code: 4c 39 e3 77 c0 0f b6 8d 74 ff ff ff b8 01 00 00 00 48 d3 e0 e9 11 fd  
> ff ff 48 c7 c6 a0 63 52 88 4c 89 e7 e8 6a 14 10 00 0f 0b <0f> 0b 48 c7 c6  
> c0 64 52 88 4c 89 e7 e8 57 14 10 00 0f 0b 48 89 cf
> RSP: 0000:ffff8880a78e6f58 EFLAGS: 00010007
> RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffff88812fffc7e0
> RDX: 1ffff11025fff8fc RSI: 0000000000000007 RDI: ffff88812fffc7b0
> RBP: ffff8880a78e7018 R08: ffff8880a78ce000 R09: ffffed1014f1cdf2
> R10: ffffed1014f1cdf1 R11: 0000000000000003 R12: ffff88812fffc7b0
> R13: 1ffff11014f1cdf2 R14: ffff88812fffc7b0 R15: ffff8880a78e6ff0
> FS:  0000000000000000(0000) GS:ffff8880ae600000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 0000000000438ca0 CR3: 0000000009871000 CR4: 00000000001426f0
> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000600
> Call Trace:
>   fast_isolate_freepages mm/compaction.c:1356 [inline]

Mel's new code... but might be just a victim of e.g. bad struct page
initialization?

>   isolate_freepages mm/compaction.c:1429 [inline]
>   compaction_alloc+0xd05/0x2970 mm/compaction.c:1541
>   unmap_and_move mm/migrate.c:1177 [inline]
>   migrate_pages+0x48e/0x2cc0 mm/migrate.c:1417
>   compact_zone+0x2207/0x3e90 mm/compaction.c:2173
>   kcompactd_do_work+0x6de/0x1200 mm/compaction.c:2564
>   kcompactd+0x251/0x970 mm/compaction.c:2657
>   kthread+0x357/0x430 kernel/kthread.c:247
>   ret_from_fork+0x3a/0x50 arch/x86/entry/entry_64.S:352
> Modules linked in:
> 
> ======================================================
> WARNING: possible circular locking dependency detected
> 5.0.0-rc1-next-20190111 #10 Not tainted
> ------------------------------------------------------

Dunno about that, but doesn't seem to be the root cause anyway.

> -> #0 (console_owner){-.-.}:
>         lock_acquire+0x1db/0x570 kernel/locking/lockdep.c:3860
>         console_lock_spinning_enable kernel/printk/printk.c:1647 [inline]
>         console_unlock+0x516/0x1040 kernel/printk/printk.c:2452
>         vprintk_emit+0x370/0x960 kernel/printk/printk.c:1978
>         vprintk_default+0x28/0x30 kernel/printk/printk.c:2005
>         vprintk_func+0x7e/0x189 kernel/printk/printk_safe.c:398
>         printk+0xba/0xed kernel/printk/printk.c:2038
>         report_bug.cold+0x11/0x5e lib/bug.c:191
>         fixup_bug arch/x86/kernel/traps.c:178 [inline]
>         fixup_bug arch/x86/kernel/traps.c:173 [inline]
>         do_error_trap+0x11b/0x200 arch/x86/kernel/traps.c:271
>         do_invalid_op+0x37/0x50 arch/x86/kernel/traps.c:290
>         invalid_op+0x14/0x20 arch/x86/entry/entry_64.S:973
>         __ClearPageBuddy include/linux/page-flags.h:706 [inline]

So that's VM_BUG_ON_PAGE(!Page##uname(page), page); in
__ClearPage##uname, so another problem with !PageBuddy.

>         rmv_page_order mm/page_alloc.c:744 [inline]
>         rmv_page_order mm/page_alloc.c:742 [inline]
>         __isolate_free_page+0x4a8/0x680 mm/page_alloc.c:3134

But this is later in the function than the first BUG_ON, so something
has raced with us?

Also two kcompactd crashes with slightly different stacktraces, that
would have to be a NUMA system with multiple kcompactd's?

>         fast_isolate_freepages mm/compaction.c:1356 [inline]
>         isolate_freepages mm/compaction.c:1429 [inline]
>         compaction_alloc+0xd05/0x2970 mm/compaction.c:1541
>         unmap_and_move mm/migrate.c:1177 [inline]
>         migrate_pages+0x48e/0x2cc0 mm/migrate.c:1417
>         compact_zone+0x2207/0x3e90 mm/compaction.c:2173
>         kcompactd_do_work+0x6de/0x1200 mm/compaction.c:2564
>         kcompactd+0x251/0x970 mm/compaction.c:2657
>         kthread+0x357/0x430 kernel/kthread.c:247
>         ret_from_fork+0x3a/0x50 arch/x86/entry/entry_64.S:352