Sorry for the delay, I was ill... On Aug 24, 09 14:18:23 -0500, David C. Rankin wrote:
Sorry for the delay in getting this worked. I am stuck. I did the bisect at "start d19a10b753d3 1.2.5" and the radeonhd driver still fails for the RS690M (x1200) card.
git bisect start d19a10b753d3 1.2.5 git bisect run make git bisect run sudo make install
Err - no, you cannot automate the test in this case (because you have to check the results manually) - which is what git bisect run will do. So you have to do: - "git checkout d19a10b753d3" - compile, install, run, confirm that it's working right. - "git checkout 1.2.5" - compile, install, run, confirm that it's broken - "git bisect start d19a10b753d3 1.2.5" - compile, install, run, check whether it's broken or good when it's broken: "git bisect bad" when it's good: "git bisect good" - git bisect bad/good will provide you a new version, so repeat the last step until it tells you that we have found the culprit.
There were also some concerning messages during the compile. Specifically there is a "fatal: ref HEAD is not a symbolic ref" error. I don't know if that is one of the causes of the driver failure, but "fatal" is never good. I have posted the full make output here:
This is because you were no longer in a git bisect.
The 'git bisect run XYZ' does the bisecting automatically until the very
end, after that the bisect process is done. This is only possible if the
output (broken or good) can be determined automatically.
CU
Matthias
--
Matthias Hopf