On Mon 27-02-17 16:36:43, Robert Kudyba wrote: > Feb 27 04:36:54 curie kernel: rsync: page allocation stalls for 10699ms, > order:0, mode:0x2420848(GFP_NOFS|__GFP_NOFAIL|__GFP_HARDWALL|__GFP_MOVABLE) > Feb 27 04:36:54 curie kernel: CPU: 2 PID: 32649 Comm: rsync Tainted: G > L 4.9.10-200.fc25.i686+PAE #1 [...]
This is a lowmem request (aka only zones DMA and Normal can be used). > Feb 27 04:36:55 curie kernel: DMA free:5584kB min:2580kB low:3224kB > high:3868kB active_anon:0kB inactive_anon:0kB active_file:148kB > inactive_file:0kB unevictable:0kB writepending:0kB present:15996kB > managed:15916kB mlocked:0kB slab_reclaimable:10032kB slab_unreclaimable:120kB > kernel_stack:0kB pagetables:0kB bounce:0kB free_pcp:0kB local_pcp:0kB > free_cma:0kB > Feb 27 04:36:55 curie kernel: lowmem_reserve[]: 0 751 8055 8055 this one is protected by the lowmem_reserve > Feb 27 04:36:55 curie kernel: Normal free:128408kB min:128488kB low:160608kB > high:192728kB active_anon:0kB inactive_anon:0kB active_file:44340kB > inactive_file:4720kB unevictable:0kB writepending:0kB present:892920kB > managed:791608kB mlocked:0kB slab_reclaimable:569704kB > slab_unreclaimable:29448kB kernel_stack:1168kB pagetables:0kB bounce:0kB > free_pcp:1568kB local_pcp:236kB free_cma:0kB and this one is hitting the min watermark while there is not really much to reclaim. Only the page cache which might be pinned and not reclaimable from this context because this is GFP_NOFS request. It is not all that surprising the reclaim context fights to get some memory. There is a huge amount of the reclaimable slab which probably just makes a slow progress. That is not something completely surprsing on 32b system I am afraid. Btw. is the stall repeating with the increased time or it gets resolved eventually? -- Michal Hocko SUSE Labs
