[Bug 99737] [PATCH] Fix_stuck_on_bsd_ring for Clarkdale/Ironlake and kernel 4.4.x

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=99737

--- Comment #6 from Rainer Fiebig <jrf at mailbox.org> ---
(In reply to Jani Nikula from comment #4)
> This bugzilla is about bugs that are unresolved in our current upstream
> kernel tree. I'll take your word for how long it took to fix the bug, it
> certainly seems it took too long, but as I understand it, it's fixed now.
> 

Lapsus memoriae on my side: "stuck on bsd ring" started with 4.3 not 4.2 (I
kept a log). So, only 5 instead of 6 kernels until fix. But given the amount of
time and quality of life I lost due to "stuck...", I won't apologize. ;)

> The stable kernels (with the three level version numbers x.y.z) are based on
> released upstream kernels (x.y) with backports of fixes from later kernels.
> How big the .z is merely the number of times a batch of fixes has been
> backported to the stable kernel. It's not a measure of quality or how much
> development effort has been put to it.
> 
> The fixes to be backported to stable kernels must adhere to a pretty strict
> set of rules [1]. For example, the fix must exist in a later upstream
> kernel, and it cannot be bigger than 100 lines, with context. Your proposed
> fix is 500+ lines, and as such won't be considered. (And we don't maintain
> the stable kernels or backports.)
> 

Alright, then it won't get fixed. The a/m 2 committs cannot be backported
easily, the differences between 4.7/4.8 and 4.4 are too big. The patch is 500
lines for a reason. That doesn't mean that it cannot be compressed down to 100
lines. You i915-devs surely can but I can't.

> The kernel is a fast moving target, and from upstream perspective v4.4 is
> old. Just in drm/i915 the diffstat since then is 211 files changed, 103334
> insertions(+), 48260 deletions(-). It's not trivial to assess the validity
> of a backport across such a diff.
> 

Not trivial but not impossible either. If *I* as average user could analyze the
problem, find the 2 committs, solve the puzzle between vastly apart kernels and
come up with a patch despite huge differences in code, any of you experienced
i915-devs should be able to check and improve that patch with significantly
less effort and time than it took me to produce it.

> I can only recommend to consider upgrading to a more recent kernel release
> (and that does not mean a more recent stable .z release of an old kernel).
> 

The "stuck on bsd ring"-story recommends exactly the contrary.

For me it's no longer a problem: I have a patch - and I can easily switch to
post-4.7 or pre-4.3. 
But not all users of affected distros can.

I wouldn't mind if you closed this well-intentioned but obviously ill-placed
report.

> [1] https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html

-- 
You are receiving this mail because:
You are on the CC list for the bug.
You are the assignee for the bug.
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20170403/0c619e2b/attachment.html>