Comment 26 for bug 1386695

Revision history for this message
N1ck 7h0m4d4k15 (nicktux) wrote :

@jsalisbury

thanks for offering assistance and willing to help.

I've finished the process of reverse bisecting and I have spotted some good(bad) commits.

With the **priceless** help of @apw in IRC I have learned the correct procedure of reverse bisecting (it was a real learning curve!)

Here is the new log

git bisect start
# good: [fe82dcec644244676d55a1384c958d5f67979adb] Linux 3.17-rc7
git bisect good fe82dcec644244676d55a1384c958d5f67979adb
# bad: [bfe01a5ba2490f299e1d2d5508cbbbadd897bbe9] Linux 3.17
git bisect bad bfe01a5ba2490f299e1d2d5508cbbbadd897bbe9
# good: [abc40bd2eeb77eb7c2effcaf63154aad929a1d5f] mm: numa: Do not mark PTEs pte_numa when splitting huge pages
git bisect good abc40bd2eeb77eb7c2effcaf63154aad929a1d5f
# good: [58586869599f6bb38aeca71a847cd77bfea74808] Merge tag 'pm+acpi-3.17-final' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
git bisect good 58586869599f6bb38aeca71a847cd77bfea74808
# bad: [7d1419f30cc5106196e54a282d7e115e698c95f6] Merge branch 'for-linus' of git://git.samba.org/sfrench/cifs-2.6
git bisect bad 7d1419f30cc5106196e54a282d7e115e698c95f6
# bad: [eee0815dabbdd7d584bea8275f5758d25c97cb9b] Merge tag 'drm-intel-fixes-2014-10-02' of git://anongit.freedesktop.org/drm-intel into drm-fixes
git bisect bad eee0815dabbdd7d584bea8275f5758d25c97cb9b
# bad: [634ffcccfbe59d77652804e1beb415d3329b1bc6] drm/nouveau: punt fbcon resume out to a workqueue
git bisect bad 634ffcccfbe59d77652804e1beb415d3329b1bc6
# bad: [f2f9a2cbaf019481feefe231f996d3602612fa99] drm/nouveau: fix regression on original nv50 board
git bisect bad f2f9a2cbaf019481feefe231f996d3602612fa99
# bad: [5838ae610ff36777b8fce6f353c2417980c1a1fa] drm/nv50/disp: fix dpms regression on certain boards
git bisect bad 5838ae610ff36777b8fce6f353c2417980c1a1fa
# first bad commit: [5838ae610ff36777b8fce6f353c2417980c1a1fa] drm/nv50/disp: fix dpms regression on certain boards

Now I have to test the commits beginning by the first bad (good) and building-installing-testing the kernel(s).

But is it possible that in a case, 2 or 3 commits combined would be needed, in order to fix a problem?

The only commits relevant with my problem, as I can understand, are 3 (drm/nouveau/fbcon /nv50 /nv50/disp).

But maybe I need to combine them all together in order to fix the resume problem, I'm not sure.