Created attachment 123431 [details] dmesg With latest kernel, there are a lot of bcache-related error messages in dmesg. There wasn't something like that in 3.12.8. Steps to reproduce: 1. Boot 2. Wait, work a bit and stuff 3. Check dmesg
[ 1321.654865] INFO: task bcache_writebac:92 blocked for more than 120 seconds. [ 1321.654868] Tainted: G W O 3.13.0-1-ARCH #1 [ 1321.654869] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 1321.654871] bcache_writebac D 0000000000000000 0 92 2 0x00000000 [ 1321.654874] ffff8800ba5fbeb8 0000000000000046 ffff880037c09200 ffff8800ba5fbfd8 [ 1321.654877] 0000000000014440 0000000000014440 ffff880037c09200 0000000000000000 [ 1321.654878] 0000000000000000 ffff8800ba5fbe10 ffffffff81515dfb ffff880037c03d70 [ 1321.654881] Call Trace: [ 1321.654886] [<ffffffff81515dfb>] ? preempt_schedule+0x3b/0x60 [ 1321.654890] [<ffffffff8101d596>] ? ___preempt_schedule+0x56/0xb0 [ 1321.654894] [<ffffffff81083420>] ? kthread_create_on_node+0x60/0x180 [ 1321.654915] [<ffffffffa023b1b0>] ? read_dirty_endio+0x60/0x60 [bcache] [ 1321.654917] [<ffffffff81515ac9>] schedule+0x29/0x70 [ 1321.654919] [<ffffffff810835ed>] kthread+0xad/0xf0 [ 1321.654922] [<ffffffff81083540>] ? kthread_create_on_node+0x180/0x180 [ 1321.654924] [<ffffffff815212fc>] ret_from_fork+0x7c/0xb0 [ 1321.654927] [<ffffffff81083540>] ? kthread_create_on_node+0x180/0x180 That is just bcache getting blocked by something, rather than necessarily a bcache problem. Are all the wireless errors also in 3.12.8 ?
Hello, The wireless errors are another (already reported by me, testing patch now) bug. I'll check for this blocking again after recompiling/rebooting (on 3.12.8 now, no things like this there). Can I find by what it is blocked somehow?
Created attachment 123581 [details] another dmesg bcache blocks continue, even with cache_mode = none. Another dmesg attached, clean from iwlwifi errors.
I've bisected this bug down to commit f665c0f85, which seems to cause the bug. I've reverted it from 3.13, and the problem went away.
Created attachment 124061 [details] patch which reverts f665c0f85 For convenience of other users, here is the patch that reverts that change.
This bug is still present on 3.14.1
Created attachment 135181 [details] fix uninterruptible sleep in writeback thread This patch from mailing list fixes this issue.
Does this problem still show up in Linux v5.5 or v5.6-rc ? Thanks. Coly Li