| Message ID | 20240326185032.72159-1-ryncsn@gmail.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | mm/swap: always use swap cache for synchronization | expand |
Hi, Kairui, Kairui Song <ryncsn@gmail.com> writes: > From: Kairui Song <kasong@tencent.com> > > A month ago a bug was fixed for SWP_SYNCHRONOUS_IO swapin (swap cache > bypass swapin): > https://lore.kernel.org/linux-mm/20240219082040.7495-1-ryncsn@gmail.com/ > > Because we have to spin on the swap map on race, and swap map is too small > to contain more usable info, an ugly schedule_timeout_uninterruptible(1) > is added. It's not the first time a hackish workaround was added for cache > bypass swapin and not the last time. I did many experiments locally to > see if the swap cache bypass path can be dropped while keeping the > performance still comparable. And it seems doable. > In general, I think that it's a good idea to unify cache bypass swapin and normal swapin. But I haven't dive into the implementation yet. > This series does the following things: > 1. Remove swap cache bypass completely. > 2. Apply multiple optimizations after that, these optimizations are > either undoable or very difficult to do without dropping the cache > bypass swapin path. > 3. Use swap cache as a synchronization layer, also unify some code > with page cache (filemap). > > As a result, we have: > 1. A comparable performance, some tests are even faster. > 2. Multi-index support for swap cache. > 3. Removed many hackish workarounds including above long tailing > issue is gone. > > Sending this as RFC to collect some discussion, suggestion, or rejection > early, this seems need to be split into multiple series, but the > performance is not good until the last patch so I think start by > seperating them may make this approach not very convincing. And there > are still some (maybe further) TODO items and optimization space > if we are OK with this approach. > > This is based on my another series, for reusing filemap code for swapcache: > [PATCH v2 0/4] mm/filemap: optimize folio adding and splitting > https://lore.kernel.org/linux-mm/20240325171405.99971-1-ryncsn@gmail.com/ > > Patch 1/10, introduce a helper from filemap side to be used later. > Patch 2/10, 3/10 are clean up and prepare for removing the swap cache > bypass swapin path. > Patch 4/10, removed the swap cache bypass swapin path, and the > performance drop heavily (-28%). > Patch 5/10, apply the first optimization after the removal, since all > folios goes through swap cache now, there is no need to explicit shadow > clearing any more. > Patch 6/10, apply another optimization after clean up shadow clearing > routines. Now swapcache is very alike page cache, so just reuse page > cache code and we will have multi-index support. Shadow memory usage > dropped a lot. > Patch 7/10, just rename __read_swap_cache_async, it will be refactored > and a key part of this series, and the naming is very confusing to me. > Patch 8/10, make swap cache as a synchronization layer, introduce two > helpers for adding folios to swap cache, caller will either succeed or > get a folio to wait on. > Patch 9/10, apply another optimization. With above two helpers, looking > up of swapcache can be optimized and avoid false looking up, which > helped improve the performance. > Patch 10/10, apply a major optimization for SWP_SYNCHRONOUS_IO devices, > after this commit, performance for simple swapin/swapout is basically > same as before. > > Test 1, sequential swapin/out of 30G zero page on ZRAM: > > Before (us) After (us) > Swapout: 33619409 33886008 > Swapin: 32393771 32465441 (- 0.2%) > Swapout (THP): 7817909 6899938 (+11.8%) > Swapin (THP) : 32452387 33193479 (- 2.2%) If my understanding were correct, we don't have swapin (THP) support, yet. Right? > And after swapping out 30G with THP, the radix node usage dropped by a > lot: > > Before: radix_tree_node 73728K > After: radix_tree_node 7056K (-94%) Good! > Test 2: > Mysql (16g buffer pool, 32G ZRAM SWAP, 4G memcg, Zswap disabled, THP never) > sysbench /usr/share/sysbench/oltp_read_only.lua --mysql-user=root \ > --mysql-password=1234 --mysql-db=sb --tables=36 --table-size=2000000 \ > --threads=48 --time=300 --report-interval=10 run > > Before: transactions: 4849.25 per sec > After: transactions: 4849.40 per sec > > Test 3: > Mysql (16g buffer pool, NVME SWAP, 4G memcg, Zswap enabled, THP never) > echo never > /sys/kernel/mm/transparent_hugepage/enabled > echo 100 > /sys/module/zswap/parameters/max_pool_percent > echo 1 > /sys/module/zswap/parameters/enabled > echo y > /sys/module/zswap/parameters/shrinker_enabled > > sysbench /usr/share/sysbench/oltp_read_only.lua --mysql-user=root \ > --mysql-password=1234 --mysql-db=sb --tables=36 --table-size=2000000 \ > --threads=48 --time=600 --report-interval=10 run > > Before: transactions: 1662.90 per sec > After: transactions: 1726.52 per sec 3.8% improvement. Good! > Test 4: > Mysql (16g buffer pool, NVME SWAP, 4G memcg, Zswap enabled, THP always) > echo always > /sys/kernel/mm/transparent_hugepage/enabled > echo 100 > /sys/module/zswap/parameters/max_pool_percent > echo 1 > /sys/module/zswap/parameters/enabled > echo y > /sys/module/zswap/parameters/shrinker_enabled > > sysbench /usr/share/sysbench/oltp_read_only.lua --mysql-user=root \ > --mysql-password=1234 --mysql-db=sb --tables=36 --table-size=2000000 \ > --threads=48 --time=600 --report-interval=10 run > > Before: transactions: 2860.90 per sec. > After: transactions: 2802.55 per sec. > > Test 5: > Memtier / memcached (16G brd SWAP, 8G memcg, THP never): > > memcached -u nobody -m 16384 -s /tmp/memcached.socket -a 0766 -t 16 -B binary & > > memtier_benchmark -S /tmp/memcached.socket \ > -P memcache_binary -n allkeys --key-minimum=1 \ > --key-maximum=24000000 --key-pattern=P:P -c 1 -t 16 \ > --ratio 1:0 --pipeline 8 -d 1000 > > Before: 106730.31 Ops/sec > After: 106360.11 Ops/sec > > Test 5: > Memtier / memcached (16G brd SWAP, 8G memcg, THP always): > > memcached -u nobody -m 16384 -s /tmp/memcached.socket -a 0766 -t 16 -B binary & > > memtier_benchmark -S /tmp/memcached.socket \ > -P memcache_binary -n allkeys --key-minimum=1 \ > --key-maximum=24000000 --key-pattern=P:P -c 1 -t 16 \ > --ratio 1:0 --pipeline 8 -d 1000 > > Before: 83193.11 Ops/sec > After: 82504.89 Ops/sec > > These tests are tested under heavy memory stress, and the performance > seems basically same as before,very slightly better/worse for certain > cases, the benefits of multi-index are basically erased by > fragmentation and workingset nodes usage is slightly lower. > > Some (maybe further) TODO items if we are OK with this approach: > > - I see a slight performance regression for THP tests, > could identify a clear hotspot with perf, my guess is the > content on the xa_lock is an issue (we have a xa_lock for > every 64M swap cache space), THP handling needs to take the lock > longer than usual. splitting the xa_lock to be more > fine-grained seems a good solution. We have > SWAP_ADDRESS_SPACE_SHIFT = 14 which is not an optimal value. > Considering XA_CHUNK_SHIFT is 6, we will have three layer of Xarray > just for 2 extra bits. 12 should be better to always make use of > the whole XA chunk and having two layers at most. But duplicated > address_space struct also wastes more memory and cacheline. > I see an observable performance drop (~3%) after change > SWAP_ADDRESS_SPACE_SHIFT to 12. Might be a good idea to > decouple swap cache xarray from address_space (there are > too many user for swapcache, shouldn't come too dirty). > > - Actually after patch Patch 4/10, the performance is much better for > tests limited with memory cgroup, until 10/10 applied the direct swap > cache freeing logic for SWP_SYNCHRONOUS_IO swapin. Because if the swap > device is not near full, swapin doesn't clear up the swapcache, so > repeated swapout doesn't need to re-alloc a swap entry, make things > faster. This may indicate that lazy freeing of swap cache could benifit > certain workloads and may worth looking into later. > > - Now SWP_SYNCHRONOUS_IO swapin will bypass readahead and force drop > swap cache after swapin is done, which can be cleaned up and optimized > further after this patch. Device type will only determine the > readahead logic, and swap cache drop check can be based purely on swap > count. > > - Recent mTHP swapin/swapout series should have no fundamental > conflict with this. > > Kairui Song (10): > mm/filemap: split filemap storing logic into a standalone helper > mm/swap: move no readahead swapin code to a stand-alone helper > mm/swap: convert swapin_readahead to return a folio > mm/swap: remove cache bypass swapin > mm/swap: clean shadow only in unmap path > mm/swap: switch to use multi index entries > mm/swap: rename __read_swap_cache_async to swap_cache_alloc_or_get > mm/swap: use swap cache as a synchronization layer > mm/swap: delay the swap cache look up for swapin > mm/swap: optimize synchronous swapin > > include/linux/swapops.h | 5 +- > mm/filemap.c | 161 +++++++++----- > mm/huge_memory.c | 78 +++---- > mm/internal.h | 2 + > mm/memory.c | 133 ++++------- > mm/shmem.c | 44 ++-- > mm/swap.h | 71 ++++-- > mm/swap_state.c | 478 +++++++++++++++++++++------------------- > mm/swapfile.c | 64 +++--- > mm/vmscan.c | 8 +- > mm/workingset.c | 2 +- > mm/zswap.c | 4 +- > 12 files changed, 540 insertions(+), 510 deletions(-) -- Best Regards, Huang, Ying
On Wed, Mar 27, 2024 at 10:54 AM Huang, Ying <ying.huang@intel.com> wrote: > > Hi, Kairui, > > Kairui Song <ryncsn@gmail.com> writes: > > > From: Kairui Song <kasong@tencent.com> > > > > A month ago a bug was fixed for SWP_SYNCHRONOUS_IO swapin (swap cache > > bypass swapin): > > https://lore.kernel.org/linux-mm/20240219082040.7495-1-ryncsn@gmail.com/ > > > > Because we have to spin on the swap map on race, and swap map is too small > > to contain more usable info, an ugly schedule_timeout_uninterruptible(1) > > is added. It's not the first time a hackish workaround was added for cache > > bypass swapin and not the last time. I did many experiments locally to > > see if the swap cache bypass path can be dropped while keeping the > > performance still comparable. And it seems doable. > > > > In general, I think that it's a good idea to unify cache bypass swapin > and normal swapin. But I haven't dive into the implementation yet. Thanks! This series might be a bit too large, I can try to split it for easier reviewing later, just if we are OK with this idea. > > > This series does the following things: > > 1. Remove swap cache bypass completely. > > 2. Apply multiple optimizations after that, these optimizations are > > either undoable or very difficult to do without dropping the cache > > bypass swapin path. > > 3. Use swap cache as a synchronization layer, also unify some code > > with page cache (filemap). > > > > As a result, we have: > > 1. A comparable performance, some tests are even faster. > > 2. Multi-index support for swap cache. > > 3. Removed many hackish workarounds including above long tailing > > issue is gone. > > > > Sending this as RFC to collect some discussion, suggestion, or rejection > > early, this seems need to be split into multiple series, but the > > performance is not good until the last patch so I think start by > > seperating them may make this approach not very convincing. And there > > are still some (maybe further) TODO items and optimization space > > if we are OK with this approach. > > > > This is based on my another series, for reusing filemap code for swapcache: > > [PATCH v2 0/4] mm/filemap: optimize folio adding and splitting > > https://lore.kernel.org/linux-mm/20240325171405.99971-1-ryncsn@gmail.com/ > > > > Patch 1/10, introduce a helper from filemap side to be used later. > > Patch 2/10, 3/10 are clean up and prepare for removing the swap cache > > bypass swapin path. > > Patch 4/10, removed the swap cache bypass swapin path, and the > > performance drop heavily (-28%). > > Patch 5/10, apply the first optimization after the removal, since all > > folios goes through swap cache now, there is no need to explicit shadow > > clearing any more. > > Patch 6/10, apply another optimization after clean up shadow clearing > > routines. Now swapcache is very alike page cache, so just reuse page > > cache code and we will have multi-index support. Shadow memory usage > > dropped a lot. > > Patch 7/10, just rename __read_swap_cache_async, it will be refactored > > and a key part of this series, and the naming is very confusing to me. > > Patch 8/10, make swap cache as a synchronization layer, introduce two > > helpers for adding folios to swap cache, caller will either succeed or > > get a folio to wait on. > > Patch 9/10, apply another optimization. With above two helpers, looking > > up of swapcache can be optimized and avoid false looking up, which > > helped improve the performance. > > Patch 10/10, apply a major optimization for SWP_SYNCHRONOUS_IO devices, > > after this commit, performance for simple swapin/swapout is basically > > same as before. > > > > Test 1, sequential swapin/out of 30G zero page on ZRAM: > > > > Before (us) After (us) > > Swapout: 33619409 33886008 > > Swapin: 32393771 32465441 (- 0.2%) > > Swapout (THP): 7817909 6899938 (+11.8%) > > Swapin (THP) : 32452387 33193479 (- 2.2%) > > If my understanding were correct, we don't have swapin (THP) support, > yet. Right? Yes, this series doesn't change how swapin/swapout works with THP in general, but now THP swapout will leave shadows with large order, so it needs to be splitted upon swapin, that will slow down later swapin by a little bit but I think that's worth it. If we can do THP swapin in the future, this split on swapin can be saved to make the performance even better. > > > And after swapping out 30G with THP, the radix node usage dropped by a > > lot: > > > > Before: radix_tree_node 73728K > > After: radix_tree_node 7056K (-94%) > > Good! > > > Test 2: > > Mysql (16g buffer pool, 32G ZRAM SWAP, 4G memcg, Zswap disabled, THP never) > > sysbench /usr/share/sysbench/oltp_read_only.lua --mysql-user=root \ > > --mysql-password=1234 --mysql-db=sb --tables=36 --table-size=2000000 \ > > --threads=48 --time=300 --report-interval=10 run > > > > Before: transactions: 4849.25 per sec > > After: transactions: 4849.40 per sec > > > > Test 3: > > Mysql (16g buffer pool, NVME SWAP, 4G memcg, Zswap enabled, THP never) > > echo never > /sys/kernel/mm/transparent_hugepage/enabled > > echo 100 > /sys/module/zswap/parameters/max_pool_percent > > echo 1 > /sys/module/zswap/parameters/enabled > > echo y > /sys/module/zswap/parameters/shrinker_enabled > > > > sysbench /usr/share/sysbench/oltp_read_only.lua --mysql-user=root \ > > --mysql-password=1234 --mysql-db=sb --tables=36 --table-size=2000000 \ > > --threads=48 --time=600 --report-interval=10 run > > > > Before: transactions: 1662.90 per sec > > After: transactions: 1726.52 per sec > > 3.8% improvement. Good! > > > Test 4: > > Mysql (16g buffer pool, NVME SWAP, 4G memcg, Zswap enabled, THP always) > > echo always > /sys/kernel/mm/transparent_hugepage/enabled > > echo 100 > /sys/module/zswap/parameters/max_pool_percent > > echo 1 > /sys/module/zswap/parameters/enabled > > echo y > /sys/module/zswap/parameters/shrinker_enabled > > > > sysbench /usr/share/sysbench/oltp_read_only.lua --mysql-user=root \ > > --mysql-password=1234 --mysql-db=sb --tables=36 --table-size=2000000 \ > > --threads=48 --time=600 --report-interval=10 run > > > > Before: transactions: 2860.90 per sec. > > After: transactions: 2802.55 per sec. > > > > Test 5: > > Memtier / memcached (16G brd SWAP, 8G memcg, THP never): > > > > memcached -u nobody -m 16384 -s /tmp/memcached.socket -a 0766 -t 16 -B binary & > > > > memtier_benchmark -S /tmp/memcached.socket \ > > -P memcache_binary -n allkeys --key-minimum=1 \ > > --key-maximum=24000000 --key-pattern=P:P -c 1 -t 16 \ > > --ratio 1:0 --pipeline 8 -d 1000 > > > > Before: 106730.31 Ops/sec > > After: 106360.11 Ops/sec > > > > Test 5: > > Memtier / memcached (16G brd SWAP, 8G memcg, THP always): > > > > memcached -u nobody -m 16384 -s /tmp/memcached.socket -a 0766 -t 16 -B binary & > > > > memtier_benchmark -S /tmp/memcached.socket \ > > -P memcache_binary -n allkeys --key-minimum=1 \ > > --key-maximum=24000000 --key-pattern=P:P -c 1 -t 16 \ > > --ratio 1:0 --pipeline 8 -d 1000 > > > > Before: 83193.11 Ops/sec > > After: 82504.89 Ops/sec > > > > These tests are tested under heavy memory stress, and the performance > > seems basically same as before,very slightly better/worse for certain > > cases, the benefits of multi-index are basically erased by > > fragmentation and workingset nodes usage is slightly lower. > > > > Some (maybe further) TODO items if we are OK with this approach: > > > > - I see a slight performance regression for THP tests, > > could identify a clear hotspot with perf, my guess is the > > content on the xa_lock is an issue (we have a xa_lock for > > every 64M swap cache space), THP handling needs to take the lock > > longer than usual. splitting the xa_lock to be more > > fine-grained seems a good solution. We have > > SWAP_ADDRESS_SPACE_SHIFT = 14 which is not an optimal value. > > Considering XA_CHUNK_SHIFT is 6, we will have three layer of Xarray > > just for 2 extra bits. 12 should be better to always make use of > > the whole XA chunk and having two layers at most. But duplicated > > address_space struct also wastes more memory and cacheline. > > I see an observable performance drop (~3%) after change > > SWAP_ADDRESS_SPACE_SHIFT to 12. Might be a good idea to > > decouple swap cache xarray from address_space (there are > > too many user for swapcache, shouldn't come too dirty). > > > > - Actually after patch Patch 4/10, the performance is much better for > > tests limited with memory cgroup, until 10/10 applied the direct swap > > cache freeing logic for SWP_SYNCHRONOUS_IO swapin. Because if the swap > > device is not near full, swapin doesn't clear up the swapcache, so > > repeated swapout doesn't need to re-alloc a swap entry, make things > > faster. This may indicate that lazy freeing of swap cache could benifit > > certain workloads and may worth looking into later. > > > > - Now SWP_SYNCHRONOUS_IO swapin will bypass readahead and force drop > > swap cache after swapin is done, which can be cleaned up and optimized > > further after this patch. Device type will only determine the > > readahead logic, and swap cache drop check can be based purely on swap > > count. > > > > - Recent mTHP swapin/swapout series should have no fundamental > > conflict with this. > > > > Kairui Song (10): > > mm/filemap: split filemap storing logic into a standalone helper > > mm/swap: move no readahead swapin code to a stand-alone helper > > mm/swap: convert swapin_readahead to return a folio > > mm/swap: remove cache bypass swapin > > mm/swap: clean shadow only in unmap path > > mm/swap: switch to use multi index entries > > mm/swap: rename __read_swap_cache_async to swap_cache_alloc_or_get > > mm/swap: use swap cache as a synchronization layer > > mm/swap: delay the swap cache look up for swapin > > mm/swap: optimize synchronous swapin > > > > include/linux/swapops.h | 5 +- > > mm/filemap.c | 161 +++++++++----- > > mm/huge_memory.c | 78 +++---- > > mm/internal.h | 2 + > > mm/memory.c | 133 ++++------- > > mm/shmem.c | 44 ++-- > > mm/swap.h | 71 ++++-- > > mm/swap_state.c | 478 +++++++++++++++++++++------------------- > > mm/swapfile.c | 64 +++--- > > mm/vmscan.c | 8 +- > > mm/workingset.c | 2 +- > > mm/zswap.c | 4 +- > > 12 files changed, 540 insertions(+), 510 deletions(-) > > -- > Best Regards, > Huang, Ying
[...] >>> Test 1, sequential swapin/out of 30G zero page on ZRAM: >>> >>> Before (us) After (us) >>> Swapout: 33619409 33886008 >>> Swapin: 32393771 32465441 (- 0.2%) >>> Swapout (THP): 7817909 6899938 (+11.8%) >>> Swapin (THP) : 32452387 33193479 (- 2.2%) >> >> If my understanding were correct, we don't have swapin (THP) support, >> yet. Right? > > Yes, this series doesn't change how swapin/swapout works with THP in > general, but now THP swapout will leave shadows with large order, so > it needs to be splitted upon swapin, that will slow down later swapin > by a little bit but I think that's worth it. > > If we can do THP swapin in the future, this split on swapin can be > saved to make the performance even better. I'm confused by this (clearly my understanding of how this works is incorrect). Perhaps you can help me understand: When you talk about "shadows" I assume you are referring to the swap cache? It was my understanding that swapping out a THP would always leave the large folio in the swap cache, so this is nothing new? And on swap-in, if the target page is in the swap cache, even if part of a large folio, why does it need to be split? I assumed the single page would just be mapped? (and if all the other pages subsequently fault, then you end up with a fully mapped large folio back in the process)? Perhaps I'm misunderstanding what "shadows" are?
Ryan Roberts <ryan.roberts@arm.com> writes: > [...] > >>>> Test 1, sequential swapin/out of 30G zero page on ZRAM: >>>> >>>> Before (us) After (us) >>>> Swapout: 33619409 33886008 >>>> Swapin: 32393771 32465441 (- 0.2%) >>>> Swapout (THP): 7817909 6899938 (+11.8%) >>>> Swapin (THP) : 32452387 33193479 (- 2.2%) >>> >>> If my understanding were correct, we don't have swapin (THP) support, >>> yet. Right? >> >> Yes, this series doesn't change how swapin/swapout works with THP in >> general, but now THP swapout will leave shadows with large order, so >> it needs to be splitted upon swapin, that will slow down later swapin >> by a little bit but I think that's worth it. >> >> If we can do THP swapin in the future, this split on swapin can be >> saved to make the performance even better. > > I'm confused by this (clearly my understanding of how this works is incorrect). > Perhaps you can help me understand: > > When you talk about "shadows" I assume you are referring to the swap cache? It > was my understanding that swapping out a THP would always leave the large folio > in the swap cache, so this is nothing new? > > And on swap-in, if the target page is in the swap cache, even if part of a large > folio, why does it need to be split? I assumed the single page would just be > mapped? (and if all the other pages subsequently fault, then you end up with a > fully mapped large folio back in the process)? > > Perhaps I'm misunderstanding what "shadows" are? Perhaps, shadow is used to support workingset protection/detection on the anonymous LRU list as in the following patchset (merged). https://lore.kernel.org/all/1595490560-15117-5-git-send-email-iamjoonsoo.kim@lge.com/T/#m962395eb5968c74b0c4c8e41d4b0dcdd3f28b2e6 -- Best Regards, Huang, Ying
On 27/03/2024 08:32, Huang, Ying wrote: > Ryan Roberts <ryan.roberts@arm.com> writes: > >> [...] >> >>>>> Test 1, sequential swapin/out of 30G zero page on ZRAM: >>>>> >>>>> Before (us) After (us) >>>>> Swapout: 33619409 33886008 >>>>> Swapin: 32393771 32465441 (- 0.2%) >>>>> Swapout (THP): 7817909 6899938 (+11.8%) >>>>> Swapin (THP) : 32452387 33193479 (- 2.2%) >>>> >>>> If my understanding were correct, we don't have swapin (THP) support, >>>> yet. Right? >>> >>> Yes, this series doesn't change how swapin/swapout works with THP in >>> general, but now THP swapout will leave shadows with large order, so >>> it needs to be splitted upon swapin, that will slow down later swapin >>> by a little bit but I think that's worth it. >>> >>> If we can do THP swapin in the future, this split on swapin can be >>> saved to make the performance even better. >> >> I'm confused by this (clearly my understanding of how this works is incorrect). >> Perhaps you can help me understand: >> >> When you talk about "shadows" I assume you are referring to the swap cache? It >> was my understanding that swapping out a THP would always leave the large folio >> in the swap cache, so this is nothing new? >> >> And on swap-in, if the target page is in the swap cache, even if part of a large >> folio, why does it need to be split? I assumed the single page would just be >> mapped? (and if all the other pages subsequently fault, then you end up with a >> fully mapped large folio back in the process)? >> >> Perhaps I'm misunderstanding what "shadows" are? > > Perhaps, shadow is used to support workingset protection/detection on > the anonymous LRU list as in the following patchset (merged). > > https://lore.kernel.org/all/1595490560-15117-5-git-send-email-iamjoonsoo.kim@lge.com/T/#m962395eb5968c74b0c4c8e41d4b0dcdd3f28b2e6 Thanks! Although after reading the cover letter I still don't really understand the need for splitting. The LRU applies to whole folios. > > -- > Best Regards, > Huang, Ying
On Wed, Mar 27, 2024 at 4:27 PM Ryan Roberts <ryan.roberts@arm.com> wrote: > > [...] > > >>> Test 1, sequential swapin/out of 30G zero page on ZRAM: > >>> > >>> Before (us) After (us) > >>> Swapout: 33619409 33886008 > >>> Swapin: 32393771 32465441 (- 0.2%) > >>> Swapout (THP): 7817909 6899938 (+11.8%) > >>> Swapin (THP) : 32452387 33193479 (- 2.2%) > >> > >> If my understanding were correct, we don't have swapin (THP) support, > >> yet. Right? > > > > Yes, this series doesn't change how swapin/swapout works with THP in > > general, but now THP swapout will leave shadows with large order, so > > it needs to be splitted upon swapin, that will slow down later swapin > > by a little bit but I think that's worth it. > > > > If we can do THP swapin in the future, this split on swapin can be > > saved to make the performance even better. > > I'm confused by this (clearly my understanding of how this works is incorrect). > Perhaps you can help me understand: > > When you talk about "shadows" I assume you are referring to the swap cache? It > was my understanding that swapping out a THP would always leave the large folio > in the swap cache, so this is nothing new? > > And on swap-in, if the target page is in the swap cache, even if part of a large > folio, why does it need to be split? I assumed the single page would just be > mapped? (and if all the other pages subsequently fault, then you end up with a > fully mapped large folio back in the process)? > > Perhaps I'm misunderstanding what "shadows" are? Hi Ryan My bad I haven't made this clear. Ying have posted the link to the commit that added "shadow" support for anon pages, it has become a very important part for LRU activation / workingset tracking. Basically when folios are removed from the cache xarray (eg. after swap writeback is done), instead of releasing the xarray slot, an unsigned long / void * is stored to it, recording some info that will be used when refault happens, to decide how to handle the folio from LRU / workingset side. And about large folio in swapcahce: if you look at the current version of add_to_swap_cache in mainline (it adds a folio of any order into swap cache), it calls xas_create_range(&xas) which fill all xarray slots in entire range covered by the folio. But xarray supports multi-index storing, making use of the nature of the radix tree to save a lot of slots. eg. for a 2M THP page, previously 8 + 512 slots (8 extra xa nodes) is needed to store it, after this series it only needs 8 slots by using a multi-index store. (not sure if I did the math right). Same for shadow, when folio is being deleted, __delete_from_swap_cache will currently walk the xarray with xas_next update all 8 + 512 slots one by one, after this series only 8 stores are needed (ignoring fragmentation). And upon swapin, I was talking about swapin 1 sub page of a THP folio, and the folio is gone, leaving a few multi-index shadow slots. The multi-index slots need to be splitted (multi-index slot have to be updated as a whole or split first, __filemap_add_folio handles such split), I optimize and reused routine in __filemap_add_folio in this series so without too much work it works perfectly for swapcache.
From: Kairui Song <kasong@tencent.com> A month ago a bug was fixed for SWP_SYNCHRONOUS_IO swapin (swap cache bypass swapin): https://lore.kernel.org/linux-mm/20240219082040.7495-1-ryncsn@gmail.com/ Because we have to spin on the swap map on race, and swap map is too small to contain more usable info, an ugly schedule_timeout_uninterruptible(1) is added. It's not the first time a hackish workaround was added for cache bypass swapin and not the last time. I did many experiments locally to see if the swap cache bypass path can be dropped while keeping the performance still comparable. And it seems doable. This series does the following things: 1. Remove swap cache bypass completely. 2. Apply multiple optimizations after that, these optimizations are either undoable or very difficult to do without dropping the cache bypass swapin path. 3. Use swap cache as a synchronization layer, also unify some code with page cache (filemap). As a result, we have: 1. A comparable performance, some tests are even faster. 2. Multi-index support for swap cache. 3. Removed many hackish workarounds including above long tailing issue is gone. Sending this as RFC to collect some discussion, suggestion, or rejection early, this seems need to be split into multiple series, but the performance is not good until the last patch so I think start by seperating them may make this approach not very convincing. And there are still some (maybe further) TODO items and optimization space if we are OK with this approach. This is based on my another series, for reusing filemap code for swapcache: [PATCH v2 0/4] mm/filemap: optimize folio adding and splitting https://lore.kernel.org/linux-mm/20240325171405.99971-1-ryncsn@gmail.com/ Patch 1/10, introduce a helper from filemap side to be used later. Patch 2/10, 3/10 are clean up and prepare for removing the swap cache bypass swapin path. Patch 4/10, removed the swap cache bypass swapin path, and the performance drop heavily (-28%). Patch 5/10, apply the first optimization after the removal, since all folios goes through swap cache now, there is no need to explicit shadow clearing any more. Patch 6/10, apply another optimization after clean up shadow clearing routines. Now swapcache is very alike page cache, so just reuse page cache code and we will have multi-index support. Shadow memory usage dropped a lot. Patch 7/10, just rename __read_swap_cache_async, it will be refactored and a key part of this series, and the naming is very confusing to me. Patch 8/10, make swap cache as a synchronization layer, introduce two helpers for adding folios to swap cache, caller will either succeed or get a folio to wait on. Patch 9/10, apply another optimization. With above two helpers, looking up of swapcache can be optimized and avoid false looking up, which helped improve the performance. Patch 10/10, apply a major optimization for SWP_SYNCHRONOUS_IO devices, after this commit, performance for simple swapin/swapout is basically same as before. Test 1, sequential swapin/out of 30G zero page on ZRAM: Before (us) After (us) Swapout: 33619409 33886008 Swapin: 32393771 32465441 (- 0.2%) Swapout (THP): 7817909 6899938 (+11.8%) Swapin (THP) : 32452387 33193479 (- 2.2%) And after swapping out 30G with THP, the radix node usage dropped by a lot: Before: radix_tree_node 73728K After: radix_tree_node 7056K (-94%) Test 2: Mysql (16g buffer pool, 32G ZRAM SWAP, 4G memcg, Zswap disabled, THP never) sysbench /usr/share/sysbench/oltp_read_only.lua --mysql-user=root \ --mysql-password=1234 --mysql-db=sb --tables=36 --table-size=2000000 \ --threads=48 --time=300 --report-interval=10 run Before: transactions: 4849.25 per sec After: transactions: 4849.40 per sec Test 3: Mysql (16g buffer pool, NVME SWAP, 4G memcg, Zswap enabled, THP never) echo never > /sys/kernel/mm/transparent_hugepage/enabled echo 100 > /sys/module/zswap/parameters/max_pool_percent echo 1 > /sys/module/zswap/parameters/enabled echo y > /sys/module/zswap/parameters/shrinker_enabled sysbench /usr/share/sysbench/oltp_read_only.lua --mysql-user=root \ --mysql-password=1234 --mysql-db=sb --tables=36 --table-size=2000000 \ --threads=48 --time=600 --report-interval=10 run Before: transactions: 1662.90 per sec After: transactions: 1726.52 per sec Test 4: Mysql (16g buffer pool, NVME SWAP, 4G memcg, Zswap enabled, THP always) echo always > /sys/kernel/mm/transparent_hugepage/enabled echo 100 > /sys/module/zswap/parameters/max_pool_percent echo 1 > /sys/module/zswap/parameters/enabled echo y > /sys/module/zswap/parameters/shrinker_enabled sysbench /usr/share/sysbench/oltp_read_only.lua --mysql-user=root \ --mysql-password=1234 --mysql-db=sb --tables=36 --table-size=2000000 \ --threads=48 --time=600 --report-interval=10 run Before: transactions: 2860.90 per sec. After: transactions: 2802.55 per sec. Test 5: Memtier / memcached (16G brd SWAP, 8G memcg, THP never): memcached -u nobody -m 16384 -s /tmp/memcached.socket -a 0766 -t 16 -B binary & memtier_benchmark -S /tmp/memcached.socket \ -P memcache_binary -n allkeys --key-minimum=1 \ --key-maximum=24000000 --key-pattern=P:P -c 1 -t 16 \ --ratio 1:0 --pipeline 8 -d 1000 Before: 106730.31 Ops/sec After: 106360.11 Ops/sec Test 5: Memtier / memcached (16G brd SWAP, 8G memcg, THP always): memcached -u nobody -m 16384 -s /tmp/memcached.socket -a 0766 -t 16 -B binary & memtier_benchmark -S /tmp/memcached.socket \ -P memcache_binary -n allkeys --key-minimum=1 \ --key-maximum=24000000 --key-pattern=P:P -c 1 -t 16 \ --ratio 1:0 --pipeline 8 -d 1000 Before: 83193.11 Ops/sec After: 82504.89 Ops/sec These tests are tested under heavy memory stress, and the performance seems basically same as before,very slightly better/worse for certain cases, the benefits of multi-index are basically erased by fragmentation and workingset nodes usage is slightly lower. Some (maybe further) TODO items if we are OK with this approach: - I see a slight performance regression for THP tests, could identify a clear hotspot with perf, my guess is the content on the xa_lock is an issue (we have a xa_lock for every 64M swap cache space), THP handling needs to take the lock longer than usual. splitting the xa_lock to be more fine-grained seems a good solution. We have SWAP_ADDRESS_SPACE_SHIFT = 14 which is not an optimal value. Considering XA_CHUNK_SHIFT is 6, we will have three layer of Xarray just for 2 extra bits. 12 should be better to always make use of the whole XA chunk and having two layers at most. But duplicated address_space struct also wastes more memory and cacheline. I see an observable performance drop (~3%) after change SWAP_ADDRESS_SPACE_SHIFT to 12. Might be a good idea to decouple swap cache xarray from address_space (there are too many user for swapcache, shouldn't come too dirty). - Actually after patch Patch 4/10, the performance is much better for tests limited with memory cgroup, until 10/10 applied the direct swap cache freeing logic for SWP_SYNCHRONOUS_IO swapin. Because if the swap device is not near full, swapin doesn't clear up the swapcache, so repeated swapout doesn't need to re-alloc a swap entry, make things faster. This may indicate that lazy freeing of swap cache could benifit certain workloads and may worth looking into later. - Now SWP_SYNCHRONOUS_IO swapin will bypass readahead and force drop swap cache after swapin is done, which can be cleaned up and optimized further after this patch. Device type will only determine the readahead logic, and swap cache drop check can be based purely on swap count. - Recent mTHP swapin/swapout series should have no fundamental conflict with this. Kairui Song (10): mm/filemap: split filemap storing logic into a standalone helper mm/swap: move no readahead swapin code to a stand-alone helper mm/swap: convert swapin_readahead to return a folio mm/swap: remove cache bypass swapin mm/swap: clean shadow only in unmap path mm/swap: switch to use multi index entries mm/swap: rename __read_swap_cache_async to swap_cache_alloc_or_get mm/swap: use swap cache as a synchronization layer mm/swap: delay the swap cache look up for swapin mm/swap: optimize synchronous swapin include/linux/swapops.h | 5 +- mm/filemap.c | 161 +++++++++----- mm/huge_memory.c | 78 +++---- mm/internal.h | 2 + mm/memory.c | 133 ++++------- mm/shmem.c | 44 ++-- mm/swap.h | 71 ++++-- mm/swap_state.c | 478 +++++++++++++++++++++------------------- mm/swapfile.c | 64 +++--- mm/vmscan.c | 8 +- mm/workingset.c | 2 +- mm/zswap.c | 4 +- 12 files changed, 540 insertions(+), 510 deletions(-)