Re: [syzbot] kernel stack overflow in sock_close

On Fri, Dec 09, 2022 at 09:32:14PM +0800, Hillf Danton wrote:

On 9 Dec 2022 13:15:39 +0000 Mark Rutland [off-list ref]

quoted

On Thu, Dec 08, 2022 at 02:05:36AM -0800, syzbot wrote:

quoted

Hello,

syzbot found the following issue on:

HEAD commit:    e3cb714fb489 Merge branch 'for-next/core' into for-kernelci

This commit has a known-broken parent where some uaccess copies appeared to
result in stack corruption:

  https://lore.kernel.org/linux-arm-kernel/Y44gVm7IEMXqilef@FVFF77S0Q05N.cambridge.arm.com/ (local)

... which has now been dropped from the arm64 for-next/core branch, but
anything found on commit e3cb714fb489 will be suspect due to that.

This *might* a manifestation of the same issue; I'll have a go at reproducing
it locally.

Take a look at the reproducer [1] before kicking off your Harley.

[1] https://lore.kernel.org/lkml/00000000000073b14905ef2e7401@google.com/ (local)

Ah, yes; this is clearly not arm64-specific, and therefore has
nothing to do with the uaccess issue.

Sorry for the noise, and thanks for the pointer!

Mark.