Bug 242261

Summary: nvme0: aborting outstanding i/o
Product: Base System Reporter: Theron Tarigo <theron.tarigo>
Component: kernAssignee: freebsd-bugs (Nobody) <bugs>
Status: Closed FIXED    
Severity: Affects Only Me CC: imp, shuriku, theron.tarigo
Priority: ---    
Version: 12.1-RELEASE   
Hardware: Any   
OS: Any   

Description Theron Tarigo 2019-11-27 05:46:57 UTC
After resuming from suspend with 12.1-RELEASE, all processes accessing disk hang for 30 seconds, rendering the system frozen from user's perspective.  After this, dmesg has:

nvme0: Resetting controller due to a timeout.
nvme0: resetting controller
nvme0: aborting outstanding i/o
(... many repeats)
nvme0: aborting outstanding i/o
nvme0: async event occurred (type 0x0, info 0x00, page 0x01)
Comment 1 Oleksandr Kryvulia 2019-11-27 15:56:55 UTC
This was fixed on CURRENT and MFC-ed to 12-STABLE. See details on bug #211713

You can also try to decrease timeout:
sysctl dev.nvme.0.timeout_period=10
Comment 2 Warner Losh freebsd_committer freebsd_triage 2021-07-09 18:38:37 UTC
per comments, this issue has been fixed in the supported releases.
Comment 3 Theron Tarigo 2021-07-09 18:55:49 UTC
As of comment #1, it was not fixed; I have had to use the workaround instead (on 12.2-RELEASE and newer).  I haven't tried it since switching from Toshiba to Samsung nvme, but with same notebook and same buggy vendor ACPI.  Anyway, this is not something I care to spend further time investigating, as the notebook has mostly become a desktop machine, and it doesn't seem to affect anyone else anymore.
Comment 4 Warner Losh freebsd_committer freebsd_triage 2021-07-09 20:52:22 UTC
(In reply to Theron Tarigo from comment #3)

OK. If you upgrade to 13.0 or newer and still see this issue, please reopen this bug or at least drop a comment here. I believe everything should be in at least stable/13 to keep this from happening. stable/12 is older and may still suffer from this problem.