>-----Original Message----- >From: Fengguang Wu [mailto:fengguang.wu@xxxxxxxxx] >Sent: 2013年10月9日 22:12 >To: Lisa Du >Cc: KOSAKI Motohiro; linux-mm@xxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx >Subject: 6e543d5780e fixed a boot hang > >Greetings, > >FYI, this commit seem to fix a boot hang problem here. > >commit 6e543d5780e36ff5ee56c44d7e2e30db3457a7ed >Author: Lisa Du <cldu@xxxxxxxxxxx> >Date: Wed Sep 11 14:22:36 2013 -0700 > > mm: vmscan: fix do_try_to_free_pages() livelock > > > [ 1.394871] pci 0000:00:02.0: Boot video device > [ 1.395883] PCI: CLS 0 bytes, default 64 > >In parent commit, it will hang right here. > >With this commit, it will continue to emit the below OOM messages (which is not a surprise to me because the boot test runs in a small >memory KVM and the kconfig builds in lots of drivers). I think you may meet the same issue as mine. Direct reclaim loop forever with zone->all_unreclaimable = 0(as kswapd sleeps forever). And at the boot stage, no one detect and terminate it, so you see the boot hang. After apply this patch, you see there's oom-killer invoked as direct reclaim would break when zone was unreclaimable. > > [ 1.631892] swapper/0 invoked oom-killer: gfp_mask=0x2000d0, order=1, oom_score_adj=0 > [ 1.633549] swapper/0 cpuset=/ mems_allowed=0 > [ 1.634443] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 3.12.0-rc4-00019-g8b5ede6 #126 > [ 1.635982] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 > [ 1.637088] 0000000000000002 ffff88001dd41b28 ffffffff82c8d78f ffff88001ef7c040 > [ 1.638955] ffff88001dd41ba8 ffffffff82c8395f ffffffff83c54680 ffff88001dd41b60 > [ 1.640830] ffffffff810f3f06 0000000000001eb4 0000000000000246 ffff88001dd41b98 > [ 1.642687] Call Trace: > [ 1.643313] [<ffffffff82c8d78f>] dump_stack+0x54/0x74 > [ 1.644331] [<ffffffff82c8395f>] dump_header.isra.10+0x7a/0x1ba > [ 1.645443] [<ffffffff810f3f06>] ? lock_release_holdtime.part.27+0x4c/0x50 > [ 1.646685] [<ffffffff810f795a>] ? lock_release+0x189/0x1d1 > [ 1.647744] [<ffffffff811530a8>] out_of_memory+0x39e/0x3ee > [ 1.648882] [<ffffffff811579f5>] __alloc_pages_nodemask+0x668/0x7de > [ 1.650385] [<ffffffff8118eb53>] kmem_getpages+0x75/0x16c > [ 1.651429] [<ffffffff81190d20>] fallback_alloc+0x12c/0x1ea > [ 1.652528] [<ffffffff810f38e8>] ? trace_hardirqs_off+0xd/0xf > [ 1.653627] [<ffffffff81190be5>] ____cache_alloc_node+0x14a/0x159 > [ 1.654783] [<ffffffff817059fb>] ? dma_debug_init+0x1ef/0x29a > [ 1.655928] [<ffffffff8119162c>] kmem_cache_alloc_trace+0x83/0x11a > [ 1.657108] [<ffffffff817059fb>] dma_debug_init+0x1ef/0x29a > [ 1.658182] [<ffffffff841ac38b>] pci_iommu_init+0x16/0x52 > [ 1.659263] [<ffffffff841ac375>] ? iommu_setup+0x27d/0x27d > [ 1.660342] [<ffffffff810020d2>] do_one_initcall+0x93/0x137 > [ 1.661415] [<ffffffff810bd300>] ? param_set_charp+0x92/0xd8 > [ 1.662503] [<ffffffff810bd52e>] ? parse_args+0x189/0x247 > [ 1.663555] [<ffffffff8419fed1>] kernel_init_freeable+0x15e/0x1df > [ 1.664724] [<ffffffff8419f729>] ? do_early_param+0x88/0x88 > [ 1.665814] [<ffffffff82c77867>] ? rest_init+0xdb/0xdb > [ 1.666824] [<ffffffff82c77875>] kernel_init+0xe/0xdb > [ 1.667824] [<ffffffff82cbc57c>] ret_from_fork+0x7c/0xb0 > [ 1.668911] [<ffffffff82c77867>] ? rest_init+0xdb/0xdb > [ 1.669925] Mem-Info: > [ 1.670508] Node 0 DMA per-cpu: > >Thanks, >Fengguang ?韬{.n???檩jg???a?旃???)钋???骅w+h?璀?ｙ/i?⒏??⒎???Щ ??m?? ?)钋???痂?^??觥??ザ?v???O璁?f??i?⒏?