On Tue, 26 Aug 2025 at 16:45, Greg Kroah-Hartman gregkh@linuxfoundation.org wrote:
On Tue, Aug 26, 2025 at 02:06:04PM +0200, Bartosz Golaszewski wrote:
On Tue, 19 Aug 2025 at 13:52, Greg Kroah-Hartman gregkh@linuxfoundation.org wrote:
On Tue, Aug 19, 2025 at 01:30:46PM +0200, Bartosz Golaszewski wrote:
On Tue, 19 Aug 2025 at 12:02, Naresh Kamboju naresh.kamboju@linaro.org wrote:
On Tue, 19 Aug 2025 at 00:18, Naresh Kamboju naresh.kamboju@linaro.org wrote:
Boot regression: stable-rc 6.15.11-rc1 arm64 Qualcomm Dragonboard 410c Unable to handle kernel NULL pointer dereference qcom_scm_shm_bridge_enable
I have reverted the following patch and the regression got fixed.
firmware: qcom: scm: initialize tzmem before marking SCM as available [ Upstream commit 87be3e7a2d0030cda6314d2ec96b37991f636ccd ]
Hi! I'm on vacation, I will look into this next week. I expect there to be a fix on top of this commit.
Ok, I'll go and drop this one from the queues now, thanks.
greg k-h
Hi!
The issue was caused by only picking up commit 7ab36b51c6bee ("firmware: qcom: scm: request the waitqueue irq *after* initializing SCM") into stable, while the following four must be applied instead:
23972da96e1ee ("firmware: qcom: scm: remove unused arguments from SHM bridge routines") dc3f4e75c54c1 ("firmware: qcom: scm: take struct device as argument in SHM bridge enable") 87be3e7a2d003 ("firmware: qcom: scm: initialize tzmem before marking SCM as available") 7ab36b51c6bee ("firmware: qcom: scm: request the waitqueue irq *after* initializing SCM")
6.15.y is long end-of-life, so is anything still to be done here?
The same applies to all other branches for which Naresh reported an issue: 6.12, 6.15 and 6.16. I think only 6.16 is still relevant.
Bartosz