Bug#783437: linux-image-3.16.0-4-amd64: buggy debian patch to kernel upstream.
Package: linux-image-3.16.0-4-amd64
Version: 3.16.7-ckt9-3~deb8u1
Severity: normal
Not sure what is being changed in the kernel going for jessie, but from
what I can tell for two upgrades I had to resort to a newer kernel from
debian's experimental.
I have got two systems here that get the same problem, and this relates
to the video/Xorg driver issue. Though it may sound remotely unrelated,
I think something that debian is doing with upstream is the reason why
there is the same video driver issue.
kernel linux-image-3.16.0-4-amd64 is problematic (as well as the others
of amd64 3.14+ from backports or updates,.. no different result), for
nvidia and intel gfx. The driver loads up shown in X log, and there is
nothing extra more in detail to know really -- same symptom going all
the way back from kernel 3.2 and up. So it's been happening before
jessie but the user can manage to find a workaround with it.(though not
easily, it's difficult unless one is familiar with it)
1- Video driver loads
2- Xorg attempts to start, no "edid" information is ever picked up.
Graphic server fails to find any "screens" configured because for some
reason it knows there is a "video card"(using the desired driver) but
doesn't apply to using it at all and X immediately quits. (I don't
bother to using fallback drivers as there's no point to using them when
accelerated graphics is instead needed)
If the same thing is broken (two machines, one nvidia, the other stocked
intel) and I know it has something to do with debian's adjustments to
the upstream because when I recompile outside debian's standard
repository (backports or stable), the problem is then resolved.
but maybe someone out there knows it better than I can-- something in
3.17 rc5 from experimental is enabling more support for video drivers to
be working in general, at least for the two systems here being tested.
What's common between these two machines. One uses the nvidia driver
with the 3.17rc5 kernel, the other uses just this same kernel and uses
the stocked intel video driver. Both systems then get their video
detection resolved as E-did can be picked up by the X server..
I know there's a lot of blame on nvidia drivers, but this also happens
with a kernel's stocked intel driver which should be working regardless
the problems with nouveau and nvidia. Since there is a problem with
even the stocked intel driver in a debian repackaging for
Jessie(3.16.0-4), and using 3.17rc5 actually overcomes it, then perhaps
this might offer a clue to something wrong with one of the patches going
into 3.16.
so please have a look
thanks
Reply to: