lkml.org 
[lkml]   [2020]   [Feb]   [21]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: [PATCH 5.5 020/399] f2fs: call f2fs_balance_fs outside of locked page
Hello,

On Fri, Feb 21, 2020 at 08:35:45AM +0100, Greg Kroah-Hartman wrote:
> From: Jaegeuk Kim <jaegeuk@kernel.org>
>
> [ Upstream commit bdf03299248916640a835a05d32841bb3d31912d ]

I have somes issues with this patch.

It causes panics due to hung tasks on 5.6. I guess it fixes one deadlock, but
causes other one? Not sure backporting it to stable branches is a good idea.

regards,
o.


INFO: task kworker/u16:2:341 blocked for more than 122 seconds.
Not tainted 5.6.0-rc2-00254-g9a029a493dc16 #4
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
kworker/u16:2 D 0 341 2 0x00000000
Workqueue: writeback wb_workfn (flush-179:0)
Backtrace:
[<c0912bd0>] (__schedule) from [<c0913274>] (schedule+0x78/0xf4)
r10:ede1a000 r9:00000000 r8:ede1ba60 r7:ec417290 r6:00000002 r5:ede1a000
r4:ee8e8000
[<c09131fc>] (schedule) from [<c017ec74>] (rwsem_down_write_slowpath+0x24c/0x4c0)
r5:00000001 r4:ec417280
[<c017ea28>] (rwsem_down_write_slowpath) from [<c0915f6c>] (down_write+0x6c/0x70)
r10:ec417280 r9:ede1bd80 r8:ee128000 r7:00000001 r6:00000000 r5:eff0afc4
r4:ec417280
[<c0915f00>] (down_write) from [<c0435b68>] (f2fs_write_single_data_page+0x608/0x7ac)
r5:eff0afc4 r4:ec4170e0
[<c0435560>] (f2fs_write_single_data_page) from [<c0435fc0>] (f2fs_write_cache_pages+0x2b4/0x7c4)
r10:ede1bc28 r9:ec4171e0 r8:ec4170e0 r7:00000001 r6:ede1bd80 r5:00000001
r4:eff0afc4
[<c0435d0c>] (f2fs_write_cache_pages) from [<c0436814>] (f2fs_write_data_pages+0x344/0x35c)
r10:0000012c r9:ee12802c r8:ee128000 r7:00000004 r6:ec4171e0 r5:ec4170e0
r4:ede1bd80
[<c04364d0>] (f2fs_write_data_pages) from [<c0267fa0>] (do_writepages+0x3c/0xd4)
r10:0000012c r9:c0e03d00 r8:00001400 r7:c0264e94 r6:ede1bd80 r5:ec4171e0
r4:ec4170e0
[<c0267f64>] (do_writepages) from [<c0310d24>] (__writeback_single_inode+0x44/0x454)
r7:ec4171e0 r6:ede1beac r5:ede1bd80 r4:ec4170e0
[<c0310ce0>] (__writeback_single_inode) from [<c0311338>] (writeback_sb_inodes+0x204/0x4b0)
r10:0000012c r9:c0e03d00 r8:ec417148 r7:ec4170e0 r6:ede1beac r5:ec417188
r4:eebed848
[<c0311134>] (writeback_sb_inodes) from [<c0311634>] (__writeback_inodes_wb+0x50/0xe4)
r10:ee7128e8 r9:c0e03d00 r8:eebed85c r7:ede1beac r6:00000000 r5:eebed848
r4:ee120000
[<c03115e4>] (__writeback_inodes_wb) from [<c031195c>] (wb_writeback+0x294/0x338)
r10:00020800 r9:ede1a000 r8:c0e04e64 r7:eebed848 r6:000192d0 r5:ede1beac
r4:eebed848
[<c03116c8>] (wb_writeback) from [<c0312e98>] (wb_workfn+0x3e0/0x54c)
r10:ee894005 r9:eebed84c r8:eebed948 r7:eebed848 r6:00000000 r5:eebed954
r4:00002b6e
[<c0312ab8>] (wb_workfn) from [<c014f2b8>] (process_one_work+0x214/0x544)
r10:ee894005 r9:00000200 r8:00000000 r7:ee894000 r6:ef044400 r5:edb1c700
r4:eebed954
[<c014f0a4>] (process_one_work) from [<c014f634>] (worker_thread+0x4c/0x574)
r10:ef044400 r9:c0e03d00 r8:ef044418 r7:00000088 r6:ef044400 r5:edb1c714
r4:edb1c700
[<c014f5e8>] (worker_thread) from [<c01564fc>] (kthread+0x144/0x170)
r10:ef125e90 r9:ec0f235c r8:edb1c700 r7:ede1a000 r6:00000000 r5:ec0f2300
r4:ec0f2340
[<c01563b8>] (kthread) from [<c01010e8>] (ret_from_fork+0x14/0x2c)
Exception stack(0xede1bfb0 to 0xede1bff8)
bfa0: 00000000 00000000 00000000 00000000
bfc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
bfe0: 00000000 00000000 00000000 00000000 00000013 00000000
r10:00000000 r9:00000000 r8:00000000 r7:00000000 r6:00000000 r5:c01563b8
r4:ec0f2300
NMI backtrace for cpu 2
CPU: 2 PID: 52 Comm: khungtaskd Not tainted 5.6.0-rc2-00254-g9a029a493dc16 #4
Hardware name: Allwinner A83t board
Backtrace:
[<c010db5c>] (dump_backtrace) from [<c010dee0>] (show_stack+0x20/0x24)
r7:00000000 r6:60060013 r5:00000000 r4:c0e9ab10

> Otherwise, we can hit deadlock by waiting for the locked page in
> move_data_block in GC.
>
> Thread A Thread B
> - do_page_mkwrite
> - f2fs_vm_page_mkwrite
> - lock_page
> - f2fs_balance_fs
> - mutex_lock(gc_mutex)
> - f2fs_gc
> - do_garbage_collect
> - ra_data_block
> - grab_cache_page
> - f2fs_balance_fs
> - mutex_lock(gc_mutex)
>
> Fixes: 39a8695824510 ("f2fs: refactor ->page_mkwrite() flow")
> Reviewed-by: Chao Yu <yuchao0@huawei.com>
> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
> Signed-off-by: Sasha Levin <sashal@kernel.org>
> ---
> fs/f2fs/file.c | 7 ++++---
> 1 file changed, 4 insertions(+), 3 deletions(-)
>
> diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c
> index 33c412d178f0f..6c4436a5ce797 100644
> --- a/fs/f2fs/file.c
> +++ b/fs/f2fs/file.c
> @@ -50,7 +50,7 @@ static vm_fault_t f2fs_vm_page_mkwrite(struct vm_fault *vmf)
> struct page *page = vmf->page;
> struct inode *inode = file_inode(vmf->vma->vm_file);
> struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
> - struct dnode_of_data dn = { .node_changed = false };
> + struct dnode_of_data dn;
> int err;
>
> if (unlikely(f2fs_cp_error(sbi))) {
> @@ -63,6 +63,9 @@ static vm_fault_t f2fs_vm_page_mkwrite(struct vm_fault *vmf)
> goto err;
> }
>
> + /* should do out of any locked page */
> + f2fs_balance_fs(sbi, true);
> +
> sb_start_pagefault(inode->i_sb);
>
> f2fs_bug_on(sbi, f2fs_has_inline_data(inode));
> @@ -120,8 +123,6 @@ static vm_fault_t f2fs_vm_page_mkwrite(struct vm_fault *vmf)
> out_sem:
> up_read(&F2FS_I(inode)->i_mmap_sem);
>
> - f2fs_balance_fs(sbi, dn.node_changed);
> -
> sb_end_pagefault(inode->i_sb);
> err:
> return block_page_mkwrite_return(err);
> --
> 2.20.1
>
>
>

\
 
 \ /
  Last update: 2020-02-22 05:42    [W:0.883 / U:3.332 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site