Re: INFO: task hung in btrfs_alloc_tree_block

From: Hao Sun
Date: Tue Sep 14 2021 - 03:12:55 EST


Qu Wenruo <quwenruo.btrfs@xxxxxxx> 于2021年9月14日周二 下午12:45写道:
>
>
>
> On 2021/9/14 上午11:22, Hao Sun wrote:
> > Qu Wenruo <quwenruo.btrfs@xxxxxxx> 于2021年9月14日周二 上午11:13写道:
> >>
> >>
> >>
> >> On 2021/9/14 上午10:44, Hao Sun wrote:
> >>> Hello,
> >>>
> >>> When using Healer to fuzz the latest Linux kernel, the following crash
> >>> was triggered.
> >>>
> >>> HEAD commit: 6880fa6c5660 Linux 5.15-rc1
> >>> git tree: upstream
> >>> console output:
> >>> https://drive.google.com/file/d/1U3ei_jCODG9N5UHOspSRmykrEDSey3Qn/view?usp=sharing
> >>> kernel config: https://drive.google.com/file/d/1rUzyMbe5vcs6khA3tL9EHTLJvsUdWcgB/view?usp=sharing
> >>
> >> Any recorded info for the injected errors during the test?
> >>
> >> It's hanging on a tree lock, without knowing the error injected, it's
> >> really hard to find out what's the cause.
> >>
> >
> > The `task hang` happened without any fault injection.
> > Based on the recorded logs
> > (https://drive.google.com/file/d/1x7u4JfyeL8WhetacBsPDVXm48SvVJUo7/view?usp=sharing
> > and https://drive.google.com/file/d/1U3ei_jCODG9N5UHOspSRmykrEDSey3Qn/view?usp=sharing),
> > no fault-injection log was printed before the task hang.
>
> OK, then it seems like a big problem.
>
> Any workload log from the fuzzer so we can try to reproduce?
>
> Or just using the tool?
>

Execution history:
https://drive.google.com/file/d/1yq_hKNHBbOh8NU_ZRNFYjP9NmnRv0RPL/view?usp=sharing

The above is the execution history with the latest 1024 progs saved
before the task hang happened.
However, it is always hard to get useful information from that and
also hard for fuzzer to reproduce the `task hang`.
I'll keep track of this bug and send you the reproducer program once
Healer found it.

Regards
Hao