diff for duplicates of <1462309239.21143.6.camel@linux.intel.com> diff --git a/a/1.txt b/N1/1.txt index 23867d9..effc484 100644 --- a/a/1.txt +++ b/N1/1.txt @@ -4,24 +4,24 @@ by page basis, e.g.: 1. The acquisition of the mapping tree lock in swap cache when adding a page to swap cache, and then again when deleting a page from swap cache after -it has been swapped out.A +it has been swapped out. 2. The acquisition of the lock on swap device to allocate a swap slot for -a page to be swapped out.A +a page to be swapped out. -With the advent of high speed block devices that's several ordersA A +With the advent of high speed block devices that's several orders of magnitude faster than the old spinning disks, these bottlenecks become fairly significant, especially on server class machines -with many theads running.A A To reduce these locking costs, this patch +with many theads running. To reduce these locking costs, this patch series attempt to batch the pages on the following oprations needed on for swap: 1. Allocate swap slots in large batches, so locks on the swap device -don't need to be acquired as often.A -2. Add anonymous pages to the swap cache for the same swap device inA A A A A A A A A A A A A +don't need to be acquired as often. +2. Add anonymous pages to the swap cache for the same swap device in batches, so the mapping tree lock can be acquired less. 3. Delete pages from swap cache also in batches. We experimented the effect of this patches. We set up N threads to access -memory in excess of memory capcity, causing swap.A A In experiments using +memory in excess of memory capcity, causing swap. In experiments using a single pmem based fast block device on a 2 socket machine, we saw that for 1 thread, there is a ~25% increase in swap throughput and for 16 threads, the swap throughput increase by ~85%, when compared with the @@ -36,27 +36,21 @@ Tim Tim Chen (7): -A mm: Cleanup - Reorganize the shrink_page_list code into smaller -A A A A functions -A mm: Group the processing of anonymous pages to be swapped in -A A A A shrink_page_list -A mm: Add new functions to allocate swap slots in batches -A mm: Shrink page list batch allocates swap slots for page swapping -A mm: Batch addtion of pages to swap cache -A mm: Cleanup - Reorganize code to group handling of page -A mm: Batch unmapping of pages that are in swap cache - -A include/linux/swap.h |A A 29 ++- -A mm/swap_state.cA A A A A A | 253 +++++++++++++----- -A mm/swapfile.cA A A A A A A A | 215 +++++++++++++-- -A mm/vmscan.cA A A A A A A A A A | 725 ++++++++++++++++++++++++++++++++++++++------------- -A 4 files changed, 945 insertions(+), 277 deletions(-) - ---A + mm: Cleanup - Reorganize the shrink_page_list code into smaller + functions + mm: Group the processing of anonymous pages to be swapped in + shrink_page_list + mm: Add new functions to allocate swap slots in batches + mm: Shrink page list batch allocates swap slots for page swapping + mm: Batch addtion of pages to swap cache + mm: Cleanup - Reorganize code to group handling of page + mm: Batch unmapping of pages that are in swap cache + + include/linux/swap.h | 29 ++- + mm/swap_state.c | 253 +++++++++++++----- + mm/swapfile.c | 215 +++++++++++++-- + mm/vmscan.c | 725 ++++++++++++++++++++++++++++++++++++++------------- + 4 files changed, 945 insertions(+), 277 deletions(-) + +-- 2.5.5 - --- -To unsubscribe, send a message with 'unsubscribe linux-mm' in -the body to majordomo@kvack.org. For more info on Linux MM, -see: http://www.linux-mm.org/ . -Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> diff --git a/a/content_digest b/N1/content_digest index 08a4e01..403024f 100644 --- a/a/content_digest +++ b/N1/content_digest @@ -22,24 +22,24 @@ "\n" "1. The acquisition of the mapping tree lock in swap cache when adding\n" "a page to swap cache, and then again when deleting a page from swap cache after\n" - "it has been swapped out.A \n" + "it has been swapped out.\302\240\n" "2. The acquisition of the lock on swap device to allocate a swap slot for\n" - "a page to be swapped out.A \n" + "a page to be swapped out.\302\240\n" "\n" - "With the advent of high speed block devices that's several ordersA A \n" + "With the advent of high speed block devices that's several orders\302\240\302\240\n" "of magnitude faster than the old spinning disks, these bottlenecks\n" "become fairly significant, especially on server class machines\n" - "with many theads running.A A To reduce these locking costs, this patch\n" + "with many theads running.\302\240\302\240To reduce these locking costs, this patch\n" "series attempt to batch the pages on the following oprations needed\n" "on for swap:\n" "1. Allocate swap slots in large batches, so locks on the swap device\n" - "don't need to be acquired as often.A \n" - "2. Add anonymous pages to the swap cache for the same swap device inA A A A A A A A A A A A A \n" + "don't need to be acquired as often.\302\240\n" + "2. Add anonymous pages to the swap cache for the same swap device in\302\240\302\240\302\240\302\240\302\240\302\240\302\240\302\240\302\240\302\240\302\240\302\240\302\240\n" "batches, so the mapping tree lock can be acquired less.\n" "3. Delete pages from swap cache also in batches.\n" "\n" "We experimented the effect of this patches. We set up N threads to access\n" - "memory in excess of memory capcity, causing swap.A A In experiments using\n" + "memory in excess of memory capcity, causing swap.\302\240\302\240In experiments using\n" "a single pmem based fast block device on a 2 socket machine, we saw\n" "that for 1 thread, there is a ~25% increase in swap throughput and for\n" "16 threads, the swap throughput increase by ~85%, when compared with the\n" @@ -54,29 +54,23 @@ "\n" "\n" "Tim Chen (7):\n" - "A mm: Cleanup - Reorganize the shrink_page_list code into smaller\n" - "A A A A functions\n" - "A mm: Group the processing of anonymous pages to be swapped in\n" - "A A A A shrink_page_list\n" - "A mm: Add new functions to allocate swap slots in batches\n" - "A mm: Shrink page list batch allocates swap slots for page swapping\n" - "A mm: Batch addtion of pages to swap cache\n" - "A mm: Cleanup - Reorganize code to group handling of page\n" - "A mm: Batch unmapping of pages that are in swap cache\n" + "\302\240 mm: Cleanup - Reorganize the shrink_page_list code into smaller\n" + "\302\240\302\240\302\240\302\240functions\n" + "\302\240 mm: Group the processing of anonymous pages to be swapped in\n" + "\302\240\302\240\302\240\302\240shrink_page_list\n" + "\302\240 mm: Add new functions to allocate swap slots in batches\n" + "\302\240 mm: Shrink page list batch allocates swap slots for page swapping\n" + "\302\240 mm: Batch addtion of pages to swap cache\n" + "\302\240 mm: Cleanup - Reorganize code to group handling of page\n" + "\302\240 mm: Batch unmapping of pages that are in swap cache\n" "\n" - "A include/linux/swap.h |A A 29 ++-\n" - "A mm/swap_state.cA A A A A A | 253 +++++++++++++-----\n" - "A mm/swapfile.cA A A A A A A A | 215 +++++++++++++--\n" - "A mm/vmscan.cA A A A A A A A A A | 725 ++++++++++++++++++++++++++++++++++++++-------------\n" - "A 4 files changed, 945 insertions(+), 277 deletions(-)\n" + "\302\240include/linux/swap.h |\302\240\302\24029 ++-\n" + "\302\240mm/swap_state.c\302\240\302\240\302\240\302\240\302\240\302\240| 253 +++++++++++++-----\n" + "\302\240mm/swapfile.c\302\240\302\240\302\240\302\240\302\240\302\240\302\240\302\240| 215 +++++++++++++--\n" + "\302\240mm/vmscan.c\302\240\302\240\302\240\302\240\302\240\302\240\302\240\302\240\302\240\302\240| 725 ++++++++++++++++++++++++++++++++++++++-------------\n" + "\302\2404 files changed, 945 insertions(+), 277 deletions(-)\n" "\n" - "--A \n" - "2.5.5\n" - "\n" - "--\n" - "To unsubscribe, send a message with 'unsubscribe linux-mm' in\n" - "the body to majordomo@kvack.org. For more info on Linux MM,\n" - "see: http://www.linux-mm.org/ .\n" - "Don't email: <a href=mailto:\"dont@kvack.org\"> email@kvack.org </a>" + "--\302\240\n" + 2.5.5 -7ffb907d3b3479d70b31703095aaecab6e1840f8a5926b1159f9d39a0abb3f4f +0d0694a53c9faa2037554cdc185e7ec31edac3367213ff99961b4c61fa211a6c
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.