On Tue, Jul 21, 2015 at 04:47:35PM -0700, David Rientjes wrote: > On Mon, 20 Jul 2015, Mel Gorman wrote: > > > From: Mel Gorman <mgorman@xxxxxxx> > > > > The zonelist cache (zlc) was introduced to skip over zones that were > > recently known to be full. At the time the paths it bypassed were the > > cpuset checks, the watermark calculations and zone_reclaim. The situation > > today is different and the complexity of zlc is harder to justify. > > > > 1) The cpuset checks are no-ops unless a cpuset is active and in general are > > a lot cheaper. > > > > 2) zone_reclaim is now disabled by default and I suspect that was a large > > source of the cost that zlc wanted to avoid. When it is enabled, it's > > known to be a major source of stalling when nodes fill up and it's > > unwise to hit every other user with the overhead. > > > > 3) Watermark checks are expensive to calculate for high-order > > allocation requests. Later patches in this series will reduce the cost of > > the watermark checking. > > > > 4) The most important issue is that in the current implementation it > > is possible for a failed THP allocation to mark a zone full for order-0 > > allocations and cause a fallback to remote nodes. > > > > The last issue could be addressed with additional complexity but it's > > not clear that we need zlc at all so this patch deletes it. If stalls > > due to repeated zone_reclaim are ever reported as an issue then we should > > introduce deferring logic based on a timeout inside zone_reclaim itself > > and leave the page allocator fast paths alone. > > > > Impact on page-allocator microbenchmarks is negligible as they don't hit > > the paths where the zlc comes into play. The impact was noticable in > > a workload called "stutter". One part uses a lot of anonymous memory, > > a second measures mmap latency and a third copies a large file. In an > > ideal world the latency application would not notice the mmap latency. > > On a 4-node machine the results of this patch are > > > > 4-node machine stutter > > 4.2.0-rc1 4.2.0-rc1 > > vanilla nozlc-v1r20 > > Min mmap 53.9902 ( 0.00%) 49.3629 ( 8.57%) > > 1st-qrtle mmap 54.6776 ( 0.00%) 54.1201 ( 1.02%) > > 2nd-qrtle mmap 54.9242 ( 0.00%) 54.5961 ( 0.60%) > > 3rd-qrtle mmap 55.1817 ( 0.00%) 54.9338 ( 0.45%) > > Max-90% mmap 55.3952 ( 0.00%) 55.3929 ( 0.00%) > > Max-93% mmap 55.4766 ( 0.00%) 57.5712 ( -3.78%) > > Max-95% mmap 55.5522 ( 0.00%) 57.8376 ( -4.11%) > > Max-99% mmap 55.7938 ( 0.00%) 63.6180 (-14.02%) > > Max mmap 6344.0292 ( 0.00%) 67.2477 ( 98.94%) > > Mean mmap 57.3732 ( 0.00%) 54.5680 ( 4.89%) > > > > Note the maximum stall latency which was 6 seconds and becomes 67ms with > > this patch applied. However, also note that it is not guaranteed this > > benchmark always hits pathelogical cases and the milage varies. There is > > a secondary impact with more direct reclaim because zones are now being > > considered instead of being skipped by zlc. > > > > 4.1.0 4.1.0 > > vanilla nozlc-v1r4 > > Swap Ins 838 502 > > Swap Outs 1149395 2622895 > > DMA32 allocs 17839113 15863747 > > Normal allocs 129045707 137847920 > > Direct pages scanned 4070089 29046893 > > Kswapd pages scanned 17147837 17140694 > > Kswapd pages reclaimed 17146691 17139601 > > Direct pages reclaimed 1888879 4886630 > > Kswapd efficiency 99% 99% > > Kswapd velocity 17523.721 17518.928 > > Direct efficiency 46% 16% > > Direct velocity 4159.306 29687.854 > > Percentage direct scans 19% 62% > > Page writes by reclaim 1149395.000 2622895.000 > > Page writes file 0 0 > > Page writes anon 1149395 2622895 > > > > The direct page scan and reclaim rates are noticable. It is possible > > this will not be a universal win on all workloads but cycling through > > zonelists waiting for zlc->last_full_zap to expire is not the right > > decision. > > > > Signed-off-by: Mel Gorman <mgorman@xxxxxxx> > > I don't use a config that uses cpusets to restrict memory allocation > anymore, but it'd be interesting to see the impact that the spinlock and > cpuset hierarchy scan has for non-hardwalled allocations. > > This removed the #define MAX_ZONELISTS 1 for UMA configs, which will cause > build errors, but once that's fixed: > The build error is now fixed. Thanks. -- Mel Gorman SUSE Labs -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>