Re: [regression] BUG: KASAN: use-after-free in lockdep_register_key+0x755/0x8f0

[Bjorn Helgaas <helgaas@kernel.org>]

On Sat, Jun 01, 2024 at 11:39:03AM +0300, Kalle Valo wrote:
> Kalle Valo <kvalo@kernel.org> writes:
> 
> > Dan Williams <dan.j.williams@intel.com> writes:
> >
> >> Kalle Valo wrote:
> >> [..]
> >>> >> The proposed fix for that is here:
> >>> >>
> >>> >> http://lore.kernel.org/r/66560aa9dbedb_195e294b0@dwillia2-mobl3.amr.corp.intel.com.notmuch
> >>> >
> >>> > I get "Not Found" from that link, is there a typo?
> >>> 
> >>> I found this fix from for-linus branch:
> >>> 
> >>> # PCI: Fix missing lockdep annotation for pci_cfg_access_trylock()for-linus
> >>> https://git.kernel.org/pub/scm/linux/kernel/git/pci/pci.git/commit/?h=for-linus&id=f941b9182c54a885a9d5d4cfd97af66873c98560
> >>> 
> >>> But at least that doesn't fix my crash.
> >>
> >> Sorry for the broken link I mistakenly used a message-id from an
> >> internal thread with the intel.com reporter. However, it is moot now
> >> because the new direction is to revert the lockdep infrastructure:
> >>
> >> https://lore.kernel.org/r/171711745834.1628941.5259278474013108507.stgit@dwillia2-xfh.jf.intel.com
> >>
> >> (that link works...)
> >
> > Thanks, that links works :) I did a quick test with the three patches
> > and I didn't see any crashes anymore. But to be confident I need to run
> > overnight tests, I'll provide my Tested-by after that.
> 
> Ok, I'm now quite confident that the issues I saw are solved so:
> 
> Tested-by: Kalle Valo <kvalo@kernel.org>

Thanks for reporting the issue and testing the fix!  Can you please
respond with your Tested-by to the actual patch(es) you tested?

Bjorn