Kernel bug in Intel driver

Christopher James Halse Rogers raof at
Sat Apr 17 12:05:14 UTC 2010

On Fri, 2010-04-16 at 09:54 +0100, Mark Ellse wrote:
> Also, what about this one, which freezes graphics on a large number of
> Intel boards? (Is Intel a small, insignificant manufacturer whose
> products are not a priority?)
I'm not entirely sure what more you expect from that bug - it's
importance high, has had an upstream developer participating, and an
awesome member of the x-swat team providing PPA testing packages.  The
fact that it remains unfixed is not due to any lack of work on it!

The X team have been monitoring this, and other similar problems - the
i8xx chips have not had a good time of Lucid.  As you see on that bug
there has been lots of work upstream to isolate and fix the cause.  

Tracking the various upstream bugs you'll see that there have been a
number of false starts, but it looks like there *might* be a fix soon.
That fix won't be going into Lucid's release, as it's a big patch which
touches code for all the Intel cards, and hence might introduce new
problems in the i9xx chips.  If all goes well, this might make it into
lucid-updates and 10.04.1.

Because of the serious stability issues i845 and i855 users have been
reporting we're testing disabling both 3D and KMS on these chips.  If
feedback shows that this still isn't stable, we've got a fallback plan
to switch to vesa on those cards, which has been reported stable.
That's a serious functionality regression, but might be necessary to
provide a sufficiently stable desktop.  In this case, users will be able
to manually enable the -intel driver, KMS, and 3D if they wish - these
bugs are incredibly timing-specific, with some setups crashing almost
immediately after starting X, and others crashing less than once a week.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: This is a digitally signed message part
URL: <>

More information about the Ubuntu-devel-discuss mailing list