"important" because of segfaults and crashes; not RC for the usual
reason of this not applying to all users.

I have a ThinkPad X1 Carbon, second generation, Haswell, with 2560x1440
internal display on eDP, and a mini-DP port.  I can attach to an
external 1080p HDMI display (television) via a mini-DP to HDMI cable.

With this version of xserver-xorg-video-intel (in jessie and sid),
displaying on the external monitor, I can reliably cause errors like the
following by running something graphically intensive and fullscreen like
neverball:

Failed to open BO for returned DRI2 buffer (1920x1080, dri2 back buffer, named 7).
This is likely a bug in the X Server that will lead to a crash soon.
Segmentation fault

(I think that's a message from Mesa; grep turns up "Failed to open BO
for returned DRI2 buffer" in /usr/lib/x86_64-linux-gnu/dri/i965_dri.so
(and other DRI drivers as well).

That error corresponds to the following in dmesg:

[   86.499802] neverball[1166]: segfault at 208 ip 00007f0847c69ccd sp 00007fff5c62f370 error 4 in i965_dri.so[7f0847916000+51e000]
[   86.554523] gnome-shell[618]: segfault at 3c ip 00007fd4d784122c sp 00007fff3fdb7230 error 4 in i965_dri.so[7fd4d74ad000+51e000]

(When neverball crashes, gnome-shell crashes and restarts.)

I also got a similar error (and gnome-shell crash) from a fullscreen
glxgears, though it occurred after already having one such error via
neverball, so I don't know if I can get it to crash the first time via
just glxgears.  But I can *reliably* hit this by running neverball for a
few seconds.

If I install xserver-xorg-video-intel 2:2.99.916+git20141119-1~exp1 from
experimental, restart gdm, and change *nothing* else, I can no longer
reproduce this problem.  That doesn't necessarily mean the bug lies in
xserver-xorg-video-intel, though; it's also possible that changing that
one thing affects some bug elsewhere.  There is at the very least *also*
a bug in Mesa for segfaulting because of a DRI2 failure, and for that
matter that another application (gnome-shell) segfaulted due to the same
error.

Nothing shows up in Xorg.0.log from this crash.