Heads-up: PCI regression introduced in or around 5.15.129 (commit 40613da52b13)
by jwoithe from LinuxQuestions.org on (#6H5QM)
After upgrading one of my Slackware64 15.0 machines from 5.15.72 to 5.15.139, the console would freeze while booting. The boot process continued behind the scenes, with the system ultimately being reachable on the network (via ssh for example). It also responded to a keyboard-initiated reboot. The kernel logs indicated:
Code:radeon 0000:4b:00.0: Fatal error during GPU init
radeon: probe of 0000:4b:00.0 failed with error -12It turns out that there's a regression in the kernel's PCI subsystem which is causing various issues. While I experienced a console freeze on bare metal due to a crash of the GPU, others have seen problems associated with the hot-adding of SCSI discs. These all encountered problems allocating memory, as was the problem on my machine with the Radeon failure. I suspect that other hardware combinations could give rise to similar failures.
I have reported the problem to the PCI maintainers[1] who are working on a fix. In the meantime, the offending commits[2] have been reverted[3]. The revert will obviously land in the 5.15.x stable series at some point, with an alternative solution to the problem that 40613da52b13 was trying to address following in due course.
If one upgrades to 5.15.139 and experiences boot problems along the lines of those mentioned above, the solution is to return to a kernel before 1.15.129 until the issue is resolved.
[1] https://lore.kernel.org/linux-pci/20....redhat.com/T/
[2] 40613da52b13, cc22522fd55e
[3] https://lore.kernel.org/linux-pci/20....redhat.com/T/
Code:radeon 0000:4b:00.0: Fatal error during GPU init
radeon: probe of 0000:4b:00.0 failed with error -12It turns out that there's a regression in the kernel's PCI subsystem which is causing various issues. While I experienced a console freeze on bare metal due to a crash of the GPU, others have seen problems associated with the hot-adding of SCSI discs. These all encountered problems allocating memory, as was the problem on my machine with the Radeon failure. I suspect that other hardware combinations could give rise to similar failures.
I have reported the problem to the PCI maintainers[1] who are working on a fix. In the meantime, the offending commits[2] have been reverted[3]. The revert will obviously land in the 5.15.x stable series at some point, with an alternative solution to the problem that 40613da52b13 was trying to address following in due course.
If one upgrades to 5.15.139 and experiences boot problems along the lines of those mentioned above, the solution is to return to a kernel before 1.15.129 until the issue is resolved.
[1] https://lore.kernel.org/linux-pci/20....redhat.com/T/
[2] 40613da52b13, cc22522fd55e
[3] https://lore.kernel.org/linux-pci/20....redhat.com/T/