- Kernel - mailweb.openeuler.org

[PATCH openEuler-1.0-LTS] iomap: add support to track dirty state of sub pages
by Yu Kuai 04 Jan '24

04 Jan '24

hulk inclusion category: performance bugzilla: https://gitee.com/openeuler/kernel/issues/I8TKTW CVE: NA ------------------------------------------ commit 9dc55f1389f9 ("iomap: add support for sub-pagesize buffered I/O without buffer heads") replace the per-block structure buffer_head with the per-page structure iomap_page. However, iomap_page can't track the dirty state of sub pages, which will cause performance issue since sub pages will be writeback even if they are not dirty. For example, if block size is 4k and page size is 64k: dd if=/dev/zero of=testfile bs=4k count=16 oflag=sync With buffer_head implementation, the above dd cmd will writeback 4k in each round. However, with iomap_page implementation, the range of writeback in each round is from the start of the page to the end offset we just wrote. Thus add support to track dirty state in iomap_page. test environment: platform: arm64 pagesize: 64k blocksize: 4k test case: dd if=/dev/zero of=/mnt/testfile bs=1M count=128 fio --ioengine=sync --rw=randwrite --iodepth=64 --name=test \ --filename=/mnt/testfile --bs=4k --fsync=1 The test result is: xfs with patch: WRITE: bw=4609KiB/s (4720kB/s) xfs without patch WRITE: bw=2714KiB/s (2780kB/s) ext4: WRITE: bw=3840KiB/s (3932kB/s) Signed-off-by: Yu Kuai <yukuai3(a)huawei.com> --- fs/iomap.c | 70 ++++++++++++++++++++++++++++++++++++------- fs/xfs/xfs_aops.c | 3 +- include/linux/iomap.h | 15 ++++++++-- 3 files changed, 74 insertions(+), 14 deletions(-) diff --git a/fs/iomap.c b/fs/iomap.c index 7078add7bbf9..d7234c7ac95e 100644 --- a/fs/iomap.c +++ b/fs/iomap.c @@ -121,8 +121,8 @@ iomap_page_create(struct inode *inode, struct page *page) iop = kmalloc(sizeof(*iop), GFP_NOFS | __GFP_NOFAIL); atomic_set(&iop->read_count, 0); atomic_set(&iop->write_count, 0); - spin_lock_init(&iop->uptodate_lock); - bitmap_zero(iop->uptodate, PAGE_SIZE / SECTOR_SIZE); + spin_lock_init(&iop->state_lock); + bitmap_zero(iop->state, IOMAP_STATE_ARRAY_SIZE); /* * migrate_page_move_mapping() assumes that pages with private data have @@ -175,7 +175,7 @@ iomap_adjust_read_range(struct inode *inode, struct iomap_page *iop, /* move forward for each leading block marked uptodate */ for (i = first; i <= last; i++) { - if (!test_bit(i, iop->uptodate)) + if (!test_bit(i, iop->state)) break; *pos += block_size; poff += block_size; @@ -185,7 +185,7 @@ iomap_adjust_read_range(struct inode *inode, struct iomap_page *iop, /* truncate len if we find any trailing uptodate block(s) */ for ( ; i <= last; i++) { - if (test_bit(i, iop->uptodate)) { + if (test_bit(i, iop->state)) { plen -= (last - i + 1) * block_size; last = i - 1; break; @@ -209,6 +209,54 @@ iomap_adjust_read_range(struct inode *inode, struct iomap_page *iop, *lenp = plen; } +static void +iomap_set_range_dirty(struct page *page, unsigned int off, + unsigned int len) +{ + struct inode *inode = page->mapping->host; + unsigned int first = DIRTY_BITS(off >> inode->i_blkbits); + unsigned int last = DIRTY_BITS((off + len - 1) >> inode->i_blkbits); + unsigned long flags; + struct iomap_page *iop; + + if (PageError(page)) + return; + + if (len) + iomap_set_page_dirty(page); + + if (!page_has_private(page)) + return; + + iop = to_iomap_page(page); + spin_lock_irqsave(&iop->state_lock, flags); + bitmap_set(iop->state, first, last - first + 1); + spin_unlock_irqrestore(&iop->state_lock, flags); +} + +void +iomap_clear_range_dirty(struct page *page, unsigned int off, + unsigned int len) +{ + struct inode *inode = page->mapping->host; + unsigned int first = DIRTY_BITS(off >> inode->i_blkbits); + unsigned int last = DIRTY_BITS((off + len - 1) >> inode->i_blkbits); + unsigned long flags; + struct iomap_page *iop; + + if (PageError(page)) + return; + + if (!page_has_private(page)) + return; + + iop = to_iomap_page(page); + spin_lock_irqsave(&iop->state_lock, flags); + bitmap_clear(iop->state, first, last - first + 1); + spin_unlock_irqrestore(&iop->state_lock, flags); +} +EXPORT_SYMBOL_GPL(iomap_clear_range_dirty); + static void iomap_iop_set_range_uptodate(struct page *page, unsigned off, unsigned len) { @@ -220,17 +268,17 @@ iomap_iop_set_range_uptodate(struct page *page, unsigned off, unsigned len) unsigned long flags; unsigned int i; - spin_lock_irqsave(&iop->uptodate_lock, flags); + spin_lock_irqsave(&iop->state_lock, flags); for (i = 0; i < PAGE_SIZE / i_blocksize(inode); i++) { if (i >= first && i <= last) - set_bit(i, iop->uptodate); - else if (!test_bit(i, iop->uptodate)) + set_bit(i, iop->state); + else if (!test_bit(i, iop->state)) uptodate = false; } if (uptodate) SetPageUptodate(page); - spin_unlock_irqrestore(&iop->uptodate_lock, flags); + spin_unlock_irqrestore(&iop->state_lock, flags); } static void @@ -544,7 +592,7 @@ iomap_is_partially_uptodate(struct page *page, unsigned long from, if (iop) { for (i = first; i <= last; i++) - if (!test_bit(i, iop->uptodate)) + if (!test_bit(i, iop->state)) return 0; return 1; } @@ -760,7 +808,7 @@ __iomap_write_end(struct inode *inode, loff_t pos, unsigned len, if (unlikely(copied < len && !PageUptodate(page))) return 0; iomap_set_range_uptodate(page, offset_in_page(pos), len); - iomap_set_page_dirty(page); + iomap_set_range_dirty(page, offset_in_page(pos), len); return copied; } @@ -1096,7 +1144,7 @@ iomap_page_mkwrite_actor(struct inode *inode, loff_t pos, loff_t length, } else { WARN_ON_ONCE(!PageUptodate(page)); iomap_page_create(inode, page); - set_page_dirty(page); + iomap_set_range_dirty(page, offset_in_page(pos), length); } return length; diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c index 7a6396bd8114..554c74726552 100644 --- a/fs/xfs/xfs_aops.c +++ b/fs/xfs/xfs_aops.c @@ -751,7 +751,7 @@ xfs_writepage_map( for (i = 0, file_offset = page_offset(page); i < (PAGE_SIZE >> inode->i_blkbits) && file_offset < end_offset; i++, file_offset += len) { - if (iop && !test_bit(i, iop->uptodate)) + if (iop && !test_bit(DIRTY_BITS(i), iop->state)) continue; error = xfs_map_blocks(wpc, inode, file_offset); @@ -800,6 +800,7 @@ xfs_writepage_map( */ set_page_writeback_keepwrite(page); } else { + iomap_clear_range_dirty(page, 0, PAGE_SIZE); clear_page_dirty_for_io(page); set_page_writeback(page); } diff --git a/include/linux/iomap.h b/include/linux/iomap.h index eed29ae891b1..4ce1290ac0c3 100644 --- a/include/linux/iomap.h +++ b/include/linux/iomap.h @@ -7,6 +7,7 @@ #include <linux/bitmap.h> #include <linux/mm.h> #include <linux/types.h> +#include <linux/blkdev.h> struct address_space; struct fiemap_extent_info; @@ -53,6 +54,9 @@ struct vm_fault; */ #define IOMAP_NULL_ADDR -1ULL /* addr is not valid */ +#define DIRTY_BITS(x) ((x) + PAGE_SIZE / SECTOR_SIZE) +#define IOMAP_STATE_ARRAY_SIZE (PAGE_SIZE * 2 / SECTOR_SIZE) + struct iomap { u64 addr; /* disk offset of mapping, bytes */ loff_t offset; /* file offset of mapping, bytes */ @@ -114,8 +118,12 @@ struct iomap_ops { struct iomap_page { atomic_t read_count; atomic_t write_count; - spinlock_t uptodate_lock; - DECLARE_BITMAP(uptodate, PAGE_SIZE / 512); + spinlock_t state_lock; + /* + * The first half bits are used to track sub-page uptodate status, + * the second half bits are for dirty status. + */ + DECLARE_BITMAP(state, IOMAP_STATE_ARRAY_SIZE); }; static inline struct iomap_page *to_iomap_page(struct page *page) @@ -136,6 +144,9 @@ int iomap_is_partially_uptodate(struct page *page, unsigned long from, int iomap_releasepage(struct page *page, gfp_t gfp_mask); void iomap_invalidatepage(struct page *page, unsigned int offset, unsigned int len); +void iomap_clear_range_dirty(struct page *page, unsigned int off, + unsigned int len); + #ifdef CONFIG_MIGRATION int iomap_migrate_page(struct address_space *mapping, struct page *newpage, struct page *page, enum migrate_mode mode); -- 2.39.2

2 1

[PATCH OLK-6.6 v3 0/3] block: show info about opening a mounted device for write
by Li Lingfeng 04 Jan '24

04 Jan '24

Both writing a mounted device and writing part0 while mounting a partition will provide a prompt message. v1->v2: 1) Replace BLK_DEV_DUMPINFO with BLK_DEV_WRITE_MOUNTED_QUIET, which is logically opposite; 2) Make BLK_DEV_WRITE_MOUNTED_QUIET dependent on BLK_DEV_WRITE_MOUNTED, and decide whether to set bd_writers by BLK_DEV_WRITE_MOUNTED_QUIET; 3) Move bdev_dump_info() into bdev_may_open(); 4) Add the cmdline interface to control whether to show info; 5) Pass the state of "writes blocked" to part0 when mounting a partition. v2->v3: Add restriction for modifying bd_writers. Li Lingfeng (3): block: Add config to show info about opening a mounted device for write block: detect confilt of write and mount between partitions and part0 block: show info of confilt of write and mount between partitions and part0 block/Kconfig | 9 ++++ block/bdev.c | 120 +++++++++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 122 insertions(+), 7 deletions(-) -- 2.31.1

2 4

[PATCH OLK-6.6 00/57] backport mainline md patch
by Li Nan 04 Jan '24

04 Jan '24

Alex Lyakas (1): md: Whenassemble the array, consult the superblock of the freshest device David Jeffery (1): md/raid6: use valid sector values to determine if an I/O should wait on the reshape Denis Plotnikov (1): md-cluster: check for timeout while a new disk adding Gou Hao (1): md/raid1: remove unnecessary null checking Junxiao Bi (2): md: bypass block throttle for superblock update Revert "md/raid5: Wait for MD_SB_CHANGE_PENDING in raid5d" Justin Stitt (1): md: replace deprecated strncpy with memcpy Li Nan (2): md: factor out a helper exceed_read_errors() to check read_errors md/raid1: support read error check Mariusz Tkaczyk (1): md: do not require mddev_lock() for all options in array_state_store() Yu Kuai (47): md: use separate work_struct for md_start_sync() md: factor out a helper to choose sync action from md_check_recovery() md: delay choosing sync action to md_start_sync() md: factor out a helper rdev_removeable() from remove_and_add_spares() md: factor out a helper rdev_is_spare() from remove_and_add_spares() md: factor out a helper rdev_addable() from remove_and_add_spares() md: delay remove_and_add_spares() for read only array to md_start_sync() md: initialize 'active_io' while allocating mddev md: initialize 'writes_pending' while allocating mddev md-bitmap: remove the checking of 'pers->quiesce' from location_store() md-bitmap: suspend array earlier in location_store() md: don't check 'mddev->pers' from suspend_hi_store() md: don't check 'mddev->pers' and 'pers->quiesce' from suspend_lo_store() md: factor out a helper from mddev_put() md: simplify md_seq_ops md/raid1: don't split discard io for write behind md: use READ_ONCE/WRITE_ONCE for 'suspend_lo' and 'suspend_hi' md/raid5-cache: use READ_ONCE/WRITE_ONCE for 'conf->log' md: replace is_md_suspended() with 'mddev->suspended' in md_check_recovery() md: add new helpers to suspend/resume array md: add new helpers to suspend/resume and lock/unlock array md/dm-raid: use new apis to suspend array md/md-bitmap: use new apis to suspend array for location_store() md/raid5-cache: use new apis to suspend array md/raid5: use new apis to suspend array md: use new apis to suspend array for sysfs apis md: use new apis to suspend array for adding/removing rdev from state_store() md: use new apis to suspend array for ioctls involed array reconfiguration md: use new apis to suspend array before mddev_create/destroy_serial_pool md: cleanup mddev_create/destroy_serial_pool() md/md-linear: cleanup linear_add() md/raid5: replace suspend with quiesce() callback md: suspend array in md_start_sync() if array need reconfiguration md: remove old apis to suspend the array md: rename __mddev_suspend/resume() back to mddev_suspend/resume() md: cleanup pers->prepare_suspend() md: remove flag RemoveSynchronized md/raid10: remove rcu protection to access rdev from conf md/raid1: remove rcu protection to access rdev from conf md/raid5: remove rcu protection to access rdev from conf md/md-multipath: remove rcu protection to access rdev from conf md: synchronize flush io with array reconfiguration md: fix missing flush of sync_work md: don't leave 'MD_RECOVERY_FROZEN' in error path of md_set_readonly() md: fix stopping sync thread md: split MD_RECOVERY_NEEDED out of mddev_resume dm-raid: delay flushing event_work() after reconfig_mutex is released drivers/md/md.h | 75 +-- drivers/md/raid5.h | 4 +- drivers/md/dm-raid.c | 20 +- drivers/md/md-autodetect.c | 4 +- drivers/md/md-bitmap.c | 61 +- drivers/md/md-cluster.c | 15 +- drivers/md/md-linear.c | 2 - drivers/md/md-multipath.c | 32 +- drivers/md/md.c | 1089 ++++++++++++++++++++---------------- drivers/md/raid1-10.c | 54 ++ drivers/md/raid1.c | 97 ++-- drivers/md/raid10.c | 274 ++------- drivers/md/raid5-cache.c | 75 ++- drivers/md/raid5-ppl.c | 16 +- drivers/md/raid5.c | 310 +++------- 15 files changed, 995 insertions(+), 1133 deletions(-) -- 2.39.2

2 58

[PATCH OLK-6.6 0/2] Add huge page allocation limit
by Liu Shixin 04 Jan '24

04 Jan '24

Ensure the system has sufficient available memory. Kefeng Wang (2): hugetlb: Add huge page allocation limit openeuler_defconfig: enable HUGETLB_ALLOC_LIMIT .../admin-guide/kernel-parameters.txt | 10 +++ arch/arm64/configs/openeuler_defconfig | 1 + arch/x86/configs/openeuler_defconfig | 1 + fs/Kconfig | 14 ++++ mm/hugetlb.c | 78 +++++++++++++++++++ 5 files changed, 104 insertions(+) -- 2.25.1

2 3

[PATCH OLK-6.6 V1 00/12] infrastructure for scheduler bpf
by Guan Jing 04 Jan '24

04 Jan '24

infrastructure for scheduler bpf Guan Jing (12): sched: programmable: Introduce bpf sched sched: programmable: Add a tag for the task sched: programmable: Add a tag for the task group sched: programmable: Add user interface of task group tag sched: programmable: Add user interface of task tag sched: basic infrastructure for scheduler bpf sched: introduce bpf_sched_enable() libbpf: add support for scheduler bpf programs bpftool: recognize scheduler programs sched: Add helper functions to get cpu statistics sched: programmable: Add hook in select_task_rq_fair() sched: programmable: Add hook in can_migrate_task() arch/arm64/configs/openeuler_defconfig | 1 + arch/x86/configs/openeuler_defconfig | 1 + fs/proc/base.c | 64 ++++++++++++++ include/linux/bpf_sched.h | 52 +++++++++++ include/linux/bpf_types.h | 4 + include/linux/sched.h | 38 ++++++++ include/linux/sched_hook_defs.h | 3 + include/uapi/linux/bpf.h | 9 ++ init/init_task.c | 3 + kernel/bpf/Kconfig | 10 +++ kernel/bpf/btf.c | 1 + kernel/bpf/syscall.c | 27 ++++++ kernel/bpf/trampoline.c | 1 + kernel/bpf/verifier.c | 11 ++- kernel/sched/bpf_sched.c | 103 +++++++++++++++++++++ kernel/sched/build_utility.c | 4 + kernel/sched/core.c | 118 +++++++++++++++++++++++++ kernel/sched/fair.c | 46 ++++++++++ kernel/sched/sched.h | 10 +++ scripts/bpf_doc.py | 4 + tools/include/uapi/linux/bpf.h | 9 ++ tools/lib/bpf/bpf.c | 1 + tools/lib/bpf/libbpf.c | 23 ++++- tools/lib/bpf/libbpf.h | 2 + tools/lib/bpf/libbpf.map | 1 + 25 files changed, 544 insertions(+), 2 deletions(-) create mode 100644 include/linux/bpf_sched.h create mode 100644 include/linux/sched_hook_defs.h create mode 100644 kernel/sched/bpf_sched.c -- 2.34.1

2 13

[PATCH OLK-6.6 00/18] md mainline patch backport (1)
by Li Nan 04 Jan '24

04 Jan '24

Justin Stitt (1): md: replace deprecated strncpy with memcpy Mariusz Tkaczyk (1): md: do not require mddev_lock() for all options in array_state_store() Yu Kuai (16): md: use separate work_struct for md_start_sync() md: factor out a helper to choose sync action from md_check_recovery() md: delay choosing sync action to md_start_sync() md: factor out a helper rdev_removeable() from remove_and_add_spares() md: factor out a helper rdev_is_spare() from remove_and_add_spares() md: factor out a helper rdev_addable() from remove_and_add_spares() md: delay remove_and_add_spares() for read only array to md_start_sync() md: initialize 'active_io' while allocating mddev md: initialize 'writes_pending' while allocating mddev md-bitmap: remove the checking of 'pers->quiesce' from location_store() md-bitmap: suspend array earlier in location_store() md: don't check 'mddev->pers' from suspend_hi_store() md: don't check 'mddev->pers' and 'pers->quiesce' from suspend_lo_store() md: factor out a helper from mddev_put() md: simplify md_seq_ops md/raid1: don't split discard io for write behind drivers/md/md.h | 9 +- drivers/md/dm-raid.c | 7 +- drivers/md/md-bitmap.c | 47 ++-- drivers/md/md.c | 578 +++++++++++++++++++++++------------------ drivers/md/raid1.c | 6 +- drivers/md/raid10.c | 3 - drivers/md/raid5.c | 3 - 7 files changed, 359 insertions(+), 294 deletions(-) -- 2.39.2

2 19

[PATCH openEuler-1.0-LTS 0/3] Fix bugs from LTS patches
by Zhang Zekun 04 Jan '24

04 Jan '24

Bugfix from linux stable patches Mukesh Ojha (2): devcoredump : Serialize devcd_del work devcoredump: Send uevent once devcd is ready Saravana Kannan (1): driver core: Release all resources during unbind before updating device links drivers/base/dd.c | 4 +- drivers/base/devcoredump.c | 86 +++++++++++++++++++++++++++++++++++++- 2 files changed, 86 insertions(+), 4 deletions(-) -- 2.17.1

2 4

[PATCH OLK-5.10 00/11] md: mainline backport
by Li Nan 04 Jan '24

04 Jan '24

Christoph Hellwig (6): md: convert to blk_alloc_disk/blk_cleanup_disk md: add the bitmap group to the default groups for the md kobject md: extend disks_mutex coverage md: properly unwind when failing to add the kobject in md_alloc md: fix mddev->kobj lifetime md: fix error handling in md_alloc Eric Dumazet (1): md: use msleep() in md_notify_reboot() Luis Chamberlain (1): md: add error handling support for add_disk() Rikard Falkeborn (1): md: Constify attribute_group structs Yu Kuai (1): md/raid10: fix improper BUG_ON() in raise_barrier() zhangyue (1): md: fix double free of mddev->private in autorun_array() drivers/md/md.h | 4 +- drivers/md/md-bitmap.c | 2 +- drivers/md/md.c | 108 ++++++++++++++++++++++++----------------- drivers/md/raid10.c | 2 +- drivers/md/raid5.c | 2 +- 5 files changed, 68 insertions(+), 50 deletions(-) -- 2.39.2

2 12

[PATCH OpenEuler-1.0-LTS 0/3] Fix bugs from LTS patches
by Zhang Zekun 04 Jan '24

04 Jan '24

Bugfix from linux stable patches Mukesh Ojha (2): devcoredump : Serialize devcd_del work devcoredump: Send uevent once devcd is ready Saravana Kannan (1): driver core: Release all resources during unbind before updating device links drivers/base/dd.c | 4 +- drivers/base/devcoredump.c | 86 +++++++++++++++++++++++++++++++++++++- 2 files changed, 86 insertions(+), 4 deletions(-) -- 2.17.1

2 4

[PATCH OLK-6.6] livepatch/core: Fix miss disable ro for MOD_RO_AFTER_INIT memory
by Zheng Yejian 04 Jan '24

04 Jan '24

hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I8HNHM -------------------------------- It was missed to set MOD_RO_AFTER_INIT type memory to be writable when porting c8783f92771c ("livepatch/core: Revert module_enable_ro and module_disable_ro"). Thanks the report from laokz in https://gitee.com/laokz/klpmake/issues/I8HFMR. Fixes: c8783f92771c ("livepatch/core: Revert module_enable_ro and module_disable_ro") Signed-off-by: Zheng Yejian <zhengyejian1(a)huawei.com> --- kernel/module/strict_rwx.c | 1 + 1 file changed, 1 insertion(+) diff --git a/kernel/module/strict_rwx.c b/kernel/module/strict_rwx.c index 397e18f50517..575bf99c723a 100644 --- a/kernel/module/strict_rwx.c +++ b/kernel/module/strict_rwx.c @@ -46,6 +46,7 @@ void module_disable_ro(const struct module *mod) module_set_memory(mod, MOD_INIT_TEXT, set_memory_rw); module_set_memory(mod, MOD_RODATA, set_memory_rw); module_set_memory(mod, MOD_INIT_RODATA, set_memory_rw); + module_set_memory(mod, MOD_RO_AFTER_INIT, set_memory_rw); } #endif /* CONFIG_LIVEPATCH_WO_FTRACE */ -- 2.25.1

2 1