[Bug 95911] Recursive error in radeon device driver module after resume from hibernation

[bugzilla-daemon at bugzilla.kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=95911

--- Comment #17 from gitne at excite.co.jp ---
(In reply to Alex Deucher from comment #16)
> Created attachment 178141 [details]
> possible fix
> 
> Does the attached patch help?

Thank you for proposing a possible solution. Unfortunately, I do not know if
this patch is going to help because I am not a kernel developer. I do not build
nor intend to build any kernels. However, I can test fully functional Fedora
RPM packages.

Nevertheless, I have tested kernel version 3.19.7. As it turns out, hibernation
and wake up seem to work properly if no Firefox is running. Latest versions of
Firefox always use some sort of hardware accelerated graphics. This makes me
assume that the radeon device driver (or the device itself) is put into some
inconsistent or invalid state from user-space, meaning, input values to
kernel-space are not checked properly by the device driver. This invalid state
can probably be saved but cannot be restored on wake up because it is - well -
invalid by definition. So, although I am not a kernel developer I pretty much
doubt that the proposed patch is going to fix this problem because it seems to
put the device just into a deeper sleep mode. This change has no impact on any
previously saved invalid state, regardless while saving or while restoring.
But, as always, I may be wrong.

Can you build a test package?

-- 
You are receiving this mail because:
You are watching the assignee of the bug.