Created attachment 119159 [details] dmesg of a broken boot As requested from https://bugzilla.kernel.org/show_bug.cgi?id=106301 im posting here. I tried to bisect this. but whenever i reach the i915 drm changes computer locks during boot which defeats the purpose (skipped many commits w/o success). Im attaching a dmesg of a booting kernel and a picture of the screen. a letterbox of garbled screen from about [0,10] to [1024,768] (aprox) is garbled. the rest of the screen works as it should. Trying to take a screenshot with the gnome-screenshot app produces a correct image including the area that is garbled. 00:02.1 Display controller [0380]: Intel Corporation Mobile 945GM/GMS/GME, 943/940GML Express Integrated Graphics Controller [8086:27a6] (rev 03) git bisect start # good: [64291f7db5bd8150a74ad2036f1037e6a0428df2] Linux 4.2 git bisect good 64291f7db5bd8150a74ad2036f1037e6a0428df2 # bad: [6ff33f3902c3b1c5d0db6b1e2c70b6d76fba357f] Linux 4.3-rc1 git bisect bad 6ff33f3902c3b1c5d0db6b1e2c70b6d76fba357f # good: [dd5cdb48edfd34401799056a9acf61078d773f90] Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next git bisect good dd5cdb48edfd34401799056a9acf61078d773f90 # bad: [f377ea88b862bf7151be96d276f4cb740f8e1c41] Merge branch 'drm-next' of git://people.freedesktop.org/~airlied/linux git bisect bad f377ea88b862bf7151be96d276f4cb740f8e1c41 # good: [abebcdfb64f1b39eeeb14282d9cd4aad1ed86f8d] Merge tag 'sound-4.3-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound git bisect good abebcdfb64f1b39eeeb14282d9cd4aad1ed86f8d # bad: [bef2c7bd578e91c9c10983e0c15c4501127b77ca] Merge tag 'drm/tegra/for-4.3-rc1' of git://anongit.freedesktop.org/tegra/linux into drm-next git bisect bad bef2c7bd578e91c9c10983e0c15c4501127b77ca # bad: [97d3308ab245c51ae237b3444afa7ae87aa9bcd4] drm/i915: Add HAS_CORE_RING_FREQ macro git bisect bad 97d3308ab245c51ae237b3444afa7ae87aa9bcd4 # skip: [75289874e4484cd4702b3341b654b45b4a09b9d3] drm/i915: Update add_request() to take a request structure git bisect skip 75289874e4484cd4702b3341b654b45b4a09b9d3 # skip: [e62925567c7926e78bc8ca976cde5c28ea265a49] drm/i915/bxt: BUNs related to port PLL git bisect skip e62925567c7926e78bc8ca976cde5c28ea265a49 # skip: [76c3916887f1225db4b4960c59820db1a995f3cb] drm/i915: Update deferred context creation to do explicit request management git bisect skip 76c3916887f1225db4b4960c59820db1a995f3cb # skip: [2e906beac6be0116c557f96d1c07cb7a955f8059] drm/i915/gtt: Reorder page alloc/free/init functions git bisect skip 2e906beac6be0116c557f96d1c07cb7a955f8059
Created attachment 119160 [details] picture of garbled screen
Tested 4.3-rc7 and problem exists there.
tested 4.3 and bug is still present.
Created attachment 119541 [details] [review] Disable primary plane harder in noatomic. Uh oh, the hardware doesn't look too impressed at how we disabled the primary plane.. Does the attached patch work?
*** Bug 92097 has been marked as a duplicate of this bug. ***
(In reply to Maarten Lankhorst from comment #4) > Created attachment 119541 [details] [review] [review] > Disable primary plane harder in noatomic. > > Uh oh, the hardware doesn't look too impressed at how we disabled the > primary plane.. > > Does the attached patch work? Hello. Yes it does work. Tested on top of 4.3 final.
Can you give me a tested-by line for reference in the patch?
(In reply to Maarten Lankhorst from comment #7) > Can you give me a tested-by line for reference in the patch? Sorry im no developer. how am i supposed to do this?
(In reply to Tomas M. from comment #8) > (In reply to Maarten Lankhorst from comment #7) > > Can you give me a tested-by line for reference in the patch? > > Sorry im no developer. how am i supposed to do this? Basically your comment #6 was just that. Maarten, I don't usually ask for a separate confirmation, I just pick it up from the bugzilla name/email, and add the Tested-by: line myself.
Yeah but I don't think M is his full last name. :)
(In reply to Maarten Lankhorst from comment #10) > Yeah but I don't think M is his full last name. :) No its not. you can get my full name from my e-mail address. Its an easy way to detect spam (how they address me!) Tested by: Tomas Mezzadra Regards and thanks for all your help!
(In reply to Maarten Lankhorst from comment #10) > Yeah but I don't think M is his full last name. :) Fair enough, although I don't really care all that much about real/full names for Tested-bys. Reviewed/Signed-off/Acked-by are another story. In any case, thanks for testing this, Tomas!
Waiting for merge. The patch works. Device: DELL XPS M1210/945GM Kernel: Arch Linux/4.3-ck (patched) Tested by: Tatsuyuki Ishi
Maarten, please reference the commit message that fixed this. Thanks.
(In reply to Jani Nikula from comment #14) > Maarten, please reference the commit message that fixed this. Thanks. commit *id* that is
Oops was a bit overzealous. Patch on ML to fix it is "[PATCH] drm/i915: Do a better job at disabling primary plane in the noatomic case."
seems to be fixed by 634b3a4a476e96816d5d6cd5bb9f8900a53f56ba, right?
(In reply to Felix Schwarz from comment #17) > seems to be fixed by 634b3a4a476e96816d5d6cd5bb9f8900a53f56ba, right? Yes. That commit was the fix
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.