On Thu, Jan 17, 2019 at 09:33:09AM +0100, Vlastimil Babka wrote: > > 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? > The error looks like compaction found a !PageBuddy on the free lists while the zone lock was held. That seems bad no matter what. I expect there will be a respin of the entire series relatively soon but none of the fixes so far would be for that level of damage. -- Mel Gorman SUSE Labs
