On 03/22/2017 05:36 AM, kernel test robot wrote: > > FYI, we noticed the following commit: > > commit: b7ef62aa8a7aebb156ce093e3215fb821426fc1b ("mm, page_alloc: split > smallest stolen page in fallback") > https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git master > > in testcase: boot > > on test machine: qemu-system-x86_64 -enable-kvm -m 512M > > caused below changes (please refer to attached dmesg/kmsg for entire > log/backtrace): > > > +------------------------------------------------------------------+------------+------------+ > | | > f7b7744467 | b7ef62aa8a | > +------------------------------------------------------------------+------------+------------+ > | boot_successes | 0 > | 0 | > | boot_failures | 4 > | 4 | > | invoked_oom-killer:gfp_mask=0x | 4 > | | > | Mem-Info | 4 > | | > | Kernel_panic-not_syncing:Out_of_memory_and_no_killable_processes | 4 > | | > | BUG:kernel_hang_in_boot_stage | 0 > | 4 | > +------------------------------------------------------------------+------------+------------+ > > > > [ 2.398650] BTRFS: selftest: Running find delalloc tests > [ 2.413373] tsc: Refined TSC clocksource calibration: 2260.994 MHz > [ 2.414891] clocksource: tsc: mask: 0xffffffffffffffff max_cycles: > 0x209745a29ae, max_idle_ns: 440795217760 ns > > Elapsed time: 460 > BUG: kernel hang in boot stage
Thanks, I was able to reproduce and debug this. Andrew, please apply the following -fix. There will be conflicts on later patches, but with trivial resolution (pages -> free_pages). Thanks! ----8<---- commit 20a10571d2b020a3e0dee7ff236dadf4efe23026 Author: Vlastimil Babka <vba...@suse.cz> Date: Tue Mar 28 14:17:08 2017 +0200 mm, page_alloc: split smallest stolen page in fallback-fix The lkp-robot reported a test case stuck on boot due to the patch [1] which was due to endless loop in the modified __rmqueue(). It blindly expected that move_freepages_block() will succeed, but that can fail due to last pageblock pfn not belonging to the same zone as the fallback candidate page. This fix checks the result of move_freepages_block() and steals the single candidate page if it fails, which was also effectively done before [1]. [1] mmotm: mm-page_alloc-split-smallest-stolen-page-in-fallback.patch Signed-off-by: Vlastimil Babka <vba...@suse.cz> diff --git a/mm/page_alloc.c b/mm/page_alloc.c index fd7f865320c5..664cf85c51a3 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -1977,6 +1977,9 @@ static void steal_suitable_fallback(struct zone *zone, struct page *page, goto single_page; pages = move_freepages_block(zone, page, start_type); + /* moving whole block can fail due to zone boundary conditions */ + if (!pages) + goto single_page; /* Claim the whole block if over half of it is free */ if (pages >= (1 << (pageblock_order-1)) ||