[Bug 215927] New: kernel deadlock when mounting the image

[Bug 215927] New: kernel deadlock when mounting the image

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=215927

            Bug ID: 215927
           Summary: kernel deadlock when mounting the image
           Product: File System
           Version: 2.5
    Kernel Version: 5.17
          Hardware: All
                OS: Linux
              Tree: Mainline
            Status: NEW
          Severity: normal
          Priority: P1
         Component: XFS
          Assignee: filesystem_xfs@kernel-bugs.kernel.org
          Reporter: yanming@tju.edu.cn
        Regression: No

The kernel (v5.17) deadlocked when I was going to mount a XFS image. I'd tried
to kill the mount process, but failed. Nothing I can do to recover the kernel
but reboot. I am wondering is there a bug in the kernel?

I have uploaded the XFS image to google net disk
(https://drive.google.com/file/d/16pFbMkWkx692DWQX-3cE7yRO5JRsCzd0/view?usp=sharing).
You can reproduce this issue by running:
losetup /dev/loop0 case.img
mount /dev/loop0 /mnt/test/

Hoping someone can help me with this issue.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 215927] kernel deadlock when mounting the image

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=215927

--- Comment #1 from Artem S. Tashkinov (aros@gmx.com) ---
Created attachment 300862
  --> https://bugzilla.kernel.org/attachment.cgi?id=300862&action=edit
Filesystem image

Confirming.

[   74.606226] watchdog: BUG: soft lockup - CPU#1 stuck for 26s! [mount:677]
[  101.997591] watchdog: BUG: soft lockup - CPU#1 stuck for 51s! [mount:677]
[  129.388962] watchdog: BUG: soft lockup - CPU#1 stuck for 77s! [mount:677]

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 215927] kernel deadlock when mounting the image

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=215927

Artem S. Tashkinov (aros@gmx.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |high

--- Comment #2 from Artem S. Tashkinov (aros@gmx.com) ---
When trying to mount the image the whole system deadlocks and nothing works.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 215927] kernel deadlock when mounting the image

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=215927

--- Comment #3 from Artem S. Tashkinov (aros@gmx.com) ---
XFS maintainers? This looks like a serious issue.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

Re: [Bug 215927] kernel deadlock when mounting the image

[Darrick J. Wong]

On Thu, May 05, 2022 at 05:46:45AM +0000, bugzilla-daemon@kernel.org wrote:
> https://bugzilla.kernel.org/show_bug.cgi?id=215927
> 
> --- Comment #3 from Artem S. Tashkinov (aros@gmx.com) ---
> XFS maintainers? This looks like a serious issue.

Well then, you'd better help us triage and fix this.

"If you are going to run some scripted tool to randomly
corrupt the filesystem to find failures, then you have an
ethical and moral responsibility to do some of the work to
narrow down and identify the cause of the failure, not just
throw them at someone to do all the work."

--D

> 
> -- 
> You may reply to this email to add a comment.
> 
> You are receiving this mail because:
> You are watching the assignee of the bug.

Re: [Bug 215927] kernel deadlock when mounting the image

[Dave Chinner]

On Thu, May 05, 2022 at 05:46:45AM +0000, bugzilla-daemon@kernel.org wrote:
> https://bugzilla.kernel.org/show_bug.cgi?id=215927
> 
> --- Comment #3 from Artem S. Tashkinov (aros@gmx.com) ---
> XFS maintainers? This looks like a serious issue.

A fix has already been written and pushed into the upstream tree.

It's not a real world problem - exposing this issue requires
significant malicious tampering with multiple filesystem structures.
We can't (and have never tried to) defend against such a threat
model - our verification mechanisms are intended to defend against
known storage corruption vectors and software bugs, not malicious
actors.

If anyone wants credit for discovering fuzzer induced bugs like
this, then they need to be responsible in how they report them and
perform some initial triage work to determine the scope of the issue
they have discovered before they report it.  Making potentially
malicious reproducer scripts public without any warning does not win
any friends or gain influence.

We're tired of having to immediately jump to investigate issues
found by format verification attack tools that have been dumped in
public with zero analysis, zero warning and, apparently, no clue
of how serious the problem discovered might be.

The right process for reporting issues found by format verification
attacks is called "responsible disclosure". This is the process that
reporting any issue that has potential system security impacts
should use.

-Dave.
-- 
Dave Chinner
david@fromorbit.com

[Bug 215927] kernel deadlock when mounting the image

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=215927

--- Comment #4 from Darrick J. Wong (djwong@kernel.org) ---
On Thu, May 05, 2022 at 05:46:45AM +0000, bugzilla-daemon@kernel.org wrote:
> https://bugzilla.kernel.org/show_bug.cgi?id=215927
> 
> --- Comment #3 from Artem S. Tashkinov (aros@gmx.com) ---
> XFS maintainers? This looks like a serious issue.

Well then, you'd better help us triage and fix this.

"If you are going to run some scripted tool to randomly
corrupt the filesystem to find failures, then you have an
ethical and moral responsibility to do some of the work to
narrow down and identify the cause of the failure, not just
throw them at someone to do all the work."

--D

> 
> -- 
> You may reply to this email to add a comment.
> 
> You are receiving this mail because:
> You are watching the assignee of the bug.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 215927] kernel deadlock when mounting the image

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=215927

--- Comment #5 from Dave Chinner (david@fromorbit.com) ---
On Thu, May 05, 2022 at 05:46:45AM +0000, bugzilla-daemon@kernel.org wrote:
> https://bugzilla.kernel.org/show_bug.cgi?id=215927
> 
> --- Comment #3 from Artem S. Tashkinov (aros@gmx.com) ---
> XFS maintainers? This looks like a serious issue.

A fix has already been written and pushed into the upstream tree.

It's not a real world problem - exposing this issue requires
significant malicious tampering with multiple filesystem structures.
We can't (and have never tried to) defend against such a threat
model - our verification mechanisms are intended to defend against
known storage corruption vectors and software bugs, not malicious
actors.

If anyone wants credit for discovering fuzzer induced bugs like
this, then they need to be responsible in how they report them and
perform some initial triage work to determine the scope of the issue
they have discovered before they report it.  Making potentially
malicious reproducer scripts public without any warning does not win
any friends or gain influence.

We're tired of having to immediately jump to investigate issues
found by format verification attack tools that have been dumped in
public with zero analysis, zero warning and, apparently, no clue
of how serious the problem discovered might be.

The right process for reporting issues found by format verification
attacks is called "responsible disclosure". This is the process that
reporting any issue that has potential system security impacts
should use.

-Dave.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.