Re: [PATCH 6.8 000/340] 6.8.10-rc2 review

[Greg Kroah-Hartman <gregkh@linuxfoundation.org>]

On Wed, May 15, 2024 at 05:37:15PM +0100, Mark Brown wrote:
> On Wed, May 15, 2024 at 10:27:21AM +0200, Greg Kroah-Hartman wrote:
> > This is the start of the stable review cycle for the 6.8.10 release.
> > There are 340 patches in this series, all will be posted as a response
> > to this one.  If anyone has any issues with these being applied, please
> > let me know.
> 
> I'm seeing issues with the ftrace "Test file and directory owership
> changes for eventfs" test on several platforms, including both 32 Arm
> and whatever random x86_64 box Linaro have in their lab.  The logs
> aren't terribly helpful since they just log a "not ok", example here:
> 
>   https://lava.sirena.org.uk/scheduler/job/265221#L3252
> 
> Bisects land on "eventfs: Do not differentiate the toplevel events
> directory" as having introduced the issue.  Other stables don't seem to
> be affected.
> 
> Bisect log (this one from a Raspberry Pi 3 in 32 bit mode):
> 
> # bad: [cfe824b75b3d9d13a891ad1c4a2d6fe0eceed1e9] Linux 6.8.10-rc2
> # good: [f3d61438b613b87afb63118bea6fb18c50ba7a6b] Linux 6.8.9
> # good: [428b806127e00d1c39ed72cbae36dbb4598e58dd] usb: dwc3: core: Prevent phy suspend during init
> # good: [a336529a6498c3e7208415b1c2710872aebf04aa] drm/vmwgfx: Fix invalid reads in fence signaled events
> # good: [dcca5ac4f5de7cca371138049a4a5877a6a3af97] hv_netvsc: Don't free decrypted memory
> git bisect start 'cfe824b75b3d9d13a891ad1c4a2d6fe0eceed1e9' 'f3d61438b613b87afb63118bea6fb18c50ba7a6b' '428b806127e00d1c39ed72cbae36dbb4598e58dd' 'a336529a6498c3e7208415b1c2710872aebf04aa' 'dcca5ac4f5de7cca371138049a4a5877a6a3af97'
> # bad: [cfe824b75b3d9d13a891ad1c4a2d6fe0eceed1e9] Linux 6.8.10-rc2
> git bisect bad cfe824b75b3d9d13a891ad1c4a2d6fe0eceed1e9
> # good: [00dfda4fc19df6f2235723e9529efa94cbc122a2] nvme-pci: Add quirk for broken MSIs
> git bisect good 00dfda4fc19df6f2235723e9529efa94cbc122a2
> # bad: [1239a1c5dc96166a0010de49e4769e08bc6d75b3] Bluetooth: qca: fix wcn3991 device address check
> git bisect bad 1239a1c5dc96166a0010de49e4769e08bc6d75b3
> # good: [a2ede3c7da39a8ab359cd23ebba04603e119ac59] ksmbd: do not grant v2 lease if parent lease key and epoch are not set
> git bisect good a2ede3c7da39a8ab359cd23ebba04603e119ac59
> # bad: [21b410a9ae24348d143dbfe3062eae67d52d5a76] eventfs: Do not differentiate the toplevel events directory
> git bisect bad 21b410a9ae24348d143dbfe3062eae67d52d5a76
> # good: [801cdc1467e661f2b151eeb8a25042593a487c78] tracefs: Still use mount point as default permissions for instances
> git bisect good 801cdc1467e661f2b151eeb8a25042593a487c78
> # first bad commit: [21b410a9ae24348d143dbfe3062eae67d52d5a76] eventfs: Do not differentiate the toplevel events directory

Thanks for the bisection, I'll go drop this from 6.8 and 6.6 queues.

Hopefully 6.9 doesn't also have this issue.

greg k-h