You are not logged in.
I've recently been trying to biscect two specific kernel versions, as an earlier version of the kernel 6.0.6-arch1-2 does not show this graphical issue I am getting, while 6.0.7-arch1-1 does.
I've followed the details at https://wiki.archlinux.org/title/Kernel … ild_System to retrieve the repo used to build the kernel, and I can build the latest version of the kernel with no problems. It's when it comes to biscecting it that I run into a problem.
I've tried combining https://wiki.archlinux.org/title/Bisect … s_with_Git with it. I went to
~/build/linux/archlinux-linux/and then running the following:
~/b/l/archlinux-linux (BARE:master) $ git bisect good v6.0.6-arch1
You need to start by "git bisect start"
Do you want me to do it for you [Y/n]? y
status: waiting for both good and bad commits
status: waiting for bad commit, 1 good commit known
~/b/l/archlinux-linux (BARE:master|BISECTING) $ git bisect bad v6.0.7-arch1
Bisecting: a merge base must be tested
[e90fbe65c6b31ed48a6f13c232b0ca26688218d5] Linux 6.0.6
~/b/l/archlinux-linux (BARE:master|BISECTING) $This looks OK to me? But I'm not sure as this is the first time I'm doing this.
I then went back a folder to where the PKGBUILD is again and then ran this:
~/b/linux $ makepkg -efsi
==> Making package: linux-glitch-bisect 6.0.12.arch1-1 (Mon 12 Dec 2022 00:18:14 GMT)
==> Checking runtime dependencies...
==> Checking buildtime dependencies...
==> WARNING: Using existing $srcdir/ tree
==> Removing existing $pkgdir/ directory...
==> Starting build()...It starts building, and seems to be going well, but then this happens:
DEPMOD /home/sonickyle27/build/linux/pkg/linux-glitch-bisect/usr/lib/modules/6.0.6-1-glitch-bisect
Warning: 'make modules_install' requires /doesnt/exist. Please install it.
This is probably in the kmod package.
rm: cannot remove '/home/sonickyle27/build/linux/pkg/linux-glitch-bisect/usr/lib/modules/6.0.12-arch1-1-glitch-bisect/source': No such file or directory
rm: cannot remove '/home/sonickyle27/build/linux/pkg/linux-glitch-bisect/usr/lib/modules/6.0.12-arch1-1-glitch-bisect/build': No such file or directory
==> ERROR: A failure occurred in package_linux-glitch-bisect().
Aborting...This is where I am currently stumped. I did check and that "kmod" package is installed, so I do not know how to proceed even after performing some hours of searching. I feel like I'm missing something obvious and that this process should be documented somewhere.
I've put a full log on https://0x0.st/ons5.txt
Last edited by Sonickyle27 (2023-12-03 15:49:47)
Offline
Offline
Thank you loqs, it looks like that worked!
Just to make sure, https://wiki.archlinux.org/title/Bisect … s_with_Git mentions that I may have to run
make clean, but then I remember seeing
makepkg -efsielsewhere. Do you do which would be correct in this case? I'd assume the makepkg one?
Last edited by Sonickyle27 (2022-12-12 08:21:22)
Offline
I used to use
makepkg -efsifor kernel bisections but I switched to having a git checkout to provide the commits and paste that into the PKGBUILD which I build in a clean chroot.
Edit:
https://drive.google.com/file/d/1OMnM_o … share_link linux-6.0.7-1.1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/1asZb_6 … share_link linux-headers-6.0.7-1.1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/1PF3Swu … share_link linux-6.0.6-1.1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/17HLelb … share_link linux-headers-6.0.6-1.1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/17M4_Dr … share_link linux-6.0.6.r120.g9352ad87a3a3-1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/1lw-1Fj … share_link linux-headers-6.0.6.r120.g9352ad87a3a3-1-x86_64.pkg.tar.zst
Last edited by loqs (2022-12-12 17:51:26)
Offline
Thank you. I have been trying to.use makepkg to bisect, but I was unsucessful in finding the commit that is causing the issue I reported in that gitlab link. Is there a way to confirm that what I am doing by following those bisect instructions is actually doing anything? I did notice each time I installed the kernel package after each build, the size varied, but somehow none of them showed the issue...
I did also try biscecting past 6.0.7, but I noticed that it was still using modules in a 6.0.6 folder even after running
git bisect reset?
Last edited by Sonickyle27 (2022-12-12 21:18:15)
Offline
In src/archlinux what is the output of
git status
git describe
make kernelreleaseAs an alternative I can keep building the bisection kernels for you. I just need to know which ones have the issue and which do not.
Offline
In src/archlinux what is the output of
git status git describe make kernelrelease
Sure, I did end up deleting my original directory and recreating it out of desperation, but this is what I see after I export the linux repo with asp. I'm doing this in the "~/build/linux/archlinux-linux" folder:
git status
fatal: this operation must be run in a work tree
git describe
fatal: No tags can describe '830b3c68c1fb1e9176028d02ef86f3cf76aa2476'.
Try --always, or create some tags.
make kernelrelease
make: *** No rule to make target 'kernelrelease'. Stop.After starting a bisect from v6.0.6-arch1 (good) to v6.0.7-arch1 (bad), all three output the same thing.
As an alternative I can keep building the bisection kernels for you. I just need to know which ones have the issue and which do not.
If that is OK, I would really appreciate it as you definitely know what you are doing compared to me haha
Offline
loqs wrote:In src/archlinux what is the output of
git status git describe make kernelreleaseSure, I did end up deleting my original directory and recreating it out of desperation, but this is what I see after I export the linux repo with asp. I'm doing this in the "~/build/linux/archlinux-linux" folder:
That directory is the mirror of the linux git repo. The checkout is in "~/build/linux/src/archlinux-linux".
loqs wrote:As an alternative I can keep building the bisection kernels for you. I just need to know which ones have the issue and which do not.
If that is OK, I would really appreciate it as you definitely know what you are doing compared to me haha
No problem just check which of the three kernels I have uploaded to Google Drive has the issue.
Offline
Sonickyle27 wrote:loqs wrote:In src/archlinux what is the output of
git status git describe make kernelreleaseSure, I did end up deleting my original directory and recreating it out of desperation, but this is what I see after I export the linux repo with asp. I'm doing this in the "~/build/linux/archlinux-linux" folder:
That directory is the mirror of the linux git repo. The checkout is in "~/build/linux/src/archlinux-linux".
Ah ok, please see below from after when I started the bisect:
~/b/l/s/archlinux-linux (makepkg|✔) $ git status
On branch makepkg
Untracked files:
(use "git add <file>..." to include in what will be committed)
localversion.10-pkgrel
localversion.20-pkgname
version
nothing added to commit but untracked files present (use "git add" to track)
~/b/l/s/archlinux-linux (makepkg|✔) $ git describe
v6.1-arch1
~/b/l/s/archlinux-linux (makepkg|✔) $ make kernelrelease
6.1.0-arch1-1-glitch-bisectI will check those versions out when I can.
Offline
No problem just check which of the three kernels I have uploaded to Google Drive has the issue.
Thank you, just tried them all.
linux-6.0.6-1.1: Good
linux-6.0.6.r120.g9352ad87a3a3-1: Bad
linux-headers-6.0.7-1.1: Bad
Offline
$ git bisect start
status: waiting for both good and bad commits
$ git bisect good v6.0.6
status: waiting for bad commit, 1 good commit known
$ git bisect bad v6.0.7
Bisecting: 120 revisions left to test after this (roughly 7 steps)
[9352ad87a3a3ba9b58e49ce1a0e291619ab0d26e] media: sun6i-csi: Add a Kconfig dependency on RESET_CONTROLLER
$ git bisect bad
Bisecting: 59 revisions left to test after this (roughly 6 steps)
[731e6ea3ec1391b706163c24089ea5dfbcbed79a] drm/i915: Extend Wa_1607297627 to Alderlake-Phttps://drive.google.com/file/d/1ADpOml … share_link linux-6.0.6.r60.g731e6ea3ec13-1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/1TstF-M … share_link linux-headers-6.0.6.r60.g731e6ea3ec13-1-x86_64.pkg.tar.zst
Offline
$ git bisect start status: waiting for both good and bad commits $ git bisect good v6.0.6 status: waiting for bad commit, 1 good commit known $ git bisect bad v6.0.7 Bisecting: 120 revisions left to test after this (roughly 7 steps) [9352ad87a3a3ba9b58e49ce1a0e291619ab0d26e] media: sun6i-csi: Add a Kconfig dependency on RESET_CONTROLLER $ git bisect bad Bisecting: 59 revisions left to test after this (roughly 6 steps) [731e6ea3ec1391b706163c24089ea5dfbcbed79a] drm/i915: Extend Wa_1607297627 to Alderlake-Plinux-6.0.6.r60.g731e6ea3ec13-1-x86_64.pkg.tar.zst
linux-headers-6.0.6.r60.g731e6ea3ec13-1-x86_64.pkg.tar.zst
Thanks, that one is good.
Offline
Now that I think about it, should I have been running the git bisect commands in the "~/build/linux/src/archlinux-linux" folder instead of "~/build/linux/archlinux-linux/"?
Offline
Now that I think about it, should I have been running the git bisect commands in the "~/build/linux/src/archlinux-linux" folder instead of "~/build/linux/archlinux-linux/"?
Yes
git bisect good
Bisecting: 29 revisions left to test after this (roughly 5 steps)
[37fd7341de6c4d93b8be940efe428424de22b3a8] mm/kmemleak: prevent soft lockup in kmemleak_scan()'s object iteration loopshttps://drive.google.com/file/d/1L9pZwM … share_link linux-6.0.6.r90.g37fd7341de6c-1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/19m20vC … share_link linux-headers-6.0.6.r90.g37fd7341de6c-1-x86_64.pkg.tar.zst
Edit:
Assuming the next one will be bad, please confirm this:
git bisect bad
Bisecting: 14 revisions left to test after this (roughly 4 steps)
[ca2893296ae7c8997d67f1bda9080a7f96d716e8] crypto: x86/polyval - Fix crashes when keys are not 16-byte alignedhttps://drive.google.com/file/d/1zE9bSs … share_link linux-6.0.6.r75.gca2893296ae7-1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/13NTCrh … share_link linux-headers-6.0.6.r75.gca2893296ae7-1-x86_64.pkg.tar.zst
Edit2:
Assuming another bad one, please confirm this:
git bisect bad
Bisecting: 7 revisions left to test after this (roughly 3 steps)
[0a30a47741b6df1f9555a0fac6aebb7e8c363bad] drm/msm: fix use-after-free on probe deferralhttps://drive.google.com/file/d/1ypY2Ny … share_link linux-6.0.6.r67.g0a30a47741b6-1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/1G47HhN … share_link linux-headers-6.0.6.r67.g0a30a47741b6-1-x86_64.pkg.tar.zst
Edit3:
Assuming another bad one, again please confirm this:
git bisect bad
Bisecting: 3 revisions left to test after this (roughly 2 steps)
[6607901fce79959130db485e7549dacaaeef034e] drm/amdgpu: fix pstate setting issuehttps://drive.google.com/file/d/1MzPo_- … share_link linux-6.0.6.r63.g6607901fce79-1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/15ppCNO … share_link linux-headers-6.0.6.r63.g6607901fce79-1-x86_64.pkg.tar.zst
I can not predict this one, amdgpu related commits on both sides of the bisection point.
Last edited by loqs (2022-12-13 22:05:08)
Offline
Sonickyle27 wrote:Now that I think about it, should I have been running the git bisect commands in the "~/build/linux/src/archlinux-linux" folder instead of "~/build/linux/archlinux-linux/"?
Yes
[...]
I can not predict this one, amdgpu related commits on both sides of the bisection point.
This is what I got:
R90: bad
R75: bad
R67: bad
R63: good
Offline
git bisect good
Bisecting: 1 revision left to test after this (roughly 1 step)
[e60a87e401e5c06cc492bd91b54d579c78df46b3] drm/amdkfd: update gfx1037 Lx cache settinghttps://drive.google.com/file/d/1e3YHwc … share_link linux-6.0.6.r65.ge60a87e401e5-1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/1hUJdIi … share_link linux-headers-6.0.6.r65.ge60a87e401e5-1-x86_64.pkg.tar.zst
Offline
git bisect good Bisecting: 1 revision left to test after this (roughly 1 step) [e60a87e401e5c06cc492bd91b54d579c78df46b3] drm/amdkfd: update gfx1037 Lx cache settinglinux-6.0.6.r65.ge60a87e401e5-1-x86_64.pkg.tar.zst
linux-headers-6.0.6.r65.ge60a87e401e5-1-x86_64.pkg.tar.zst
This one is bad.
Thank you so much for taking the time to build these for me.
Offline
git bisect bad
Bisecting: 0 revisions left to test after this (roughly 0 steps)
[c46dda486e4f9b86eeac40aaea65a6f93dc44dad] drm/amd/display: Revert logic for plane modifiershttps://drive.google.com/file/d/1Vc4019 … share_link linux-6.0.6.r64.gc46dda486e4f-1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/1IPGVOW … share_link linux-headers-6.0.6.r64.gc46dda486e4f-1-x86_64.pkg.tar.zst
Edit:
If this one is bad
$ git bisect bad
c46dda486e4f9b86eeac40aaea65a6f93dc44dad is the first bad commit
commit c46dda486e4f9b86eeac40aaea65a6f93dc44dad
Author: Joaquín Ignacio Aramendía <samsagax@gmail.com>
Date: Mon Oct 24 08:33:59 2022 -0300
drm/amd/display: Revert logic for plane modifiers
commit 809734c110548dca410fb0cca52e6b1540319f5e upstream.
This file was split in commit 5d945cbcd4b16a29d6470a80dfb19738f9a4319f
("drm/amd/display: Create a file dedicated to planes") and the logic in
dm_plane_format_mod_supported() function got changed by a switch logic.
That change broke drm_plane modifiers setting on series 5000 APUs
(tested on OXP mini AMD 5800U and HP Dev One 5850U PRO)
leading to Gamescope not working as reported on GitHub[1]
To reproduce the issue, enter a TTY and run:
$ gamescope -- vkcube
With said commit applied it will abort. This one restores the old logic,
fixing the issue that affects Gamescope.
[1](https://github.com/Plagman/gamescope/issues/624)
Cc: <stable@vger.kernel.org> # 6.0.x
Signed-off-by: Joaquín Ignacio Aramendía <samsagax@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
.../drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c | 50 +++-------------------
1 file changed, 7 insertions(+), 43 deletions(-)
$ git bisect log
git bisect start
# status: waiting for both good and bad commits
# good: [e90fbe65c6b31ed48a6f13c232b0ca26688218d5] Linux 6.0.6
git bisect good e90fbe65c6b31ed48a6f13c232b0ca26688218d5
# status: waiting for bad commit, 1 good commit known
# bad: [3a2fa3c01fc7c2183eb3278bd912e5bcec20eb2a] Linux 6.0.7
git bisect bad 3a2fa3c01fc7c2183eb3278bd912e5bcec20eb2a
# bad: [9352ad87a3a3ba9b58e49ce1a0e291619ab0d26e] media: sun6i-csi: Add a Kconfig dependency on RESET_CONTROLLER
git bisect bad 9352ad87a3a3ba9b58e49ce1a0e291619ab0d26e
# good: [731e6ea3ec1391b706163c24089ea5dfbcbed79a] drm/i915: Extend Wa_1607297627 to Alderlake-P
git bisect good 731e6ea3ec1391b706163c24089ea5dfbcbed79a
# bad: [37fd7341de6c4d93b8be940efe428424de22b3a8] mm/kmemleak: prevent soft lockup in kmemleak_scan()'s object iteration loops
git bisect bad 37fd7341de6c4d93b8be940efe428424de22b3a8
# bad: [ca2893296ae7c8997d67f1bda9080a7f96d716e8] crypto: x86/polyval - Fix crashes when keys are not 16-byte aligned
git bisect bad ca2893296ae7c8997d67f1bda9080a7f96d716e8
# bad: [0a30a47741b6df1f9555a0fac6aebb7e8c363bad] drm/msm: fix use-after-free on probe deferral
git bisect bad 0a30a47741b6df1f9555a0fac6aebb7e8c363bad
# good: [6607901fce79959130db485e7549dacaaeef034e] drm/amdgpu: fix pstate setting issue
git bisect good 6607901fce79959130db485e7549dacaaeef034e
# bad: [e60a87e401e5c06cc492bd91b54d579c78df46b3] drm/amdkfd: update gfx1037 Lx cache setting
git bisect bad e60a87e401e5c06cc492bd91b54d579c78df46b3
# bad: [c46dda486e4f9b86eeac40aaea65a6f93dc44dad] drm/amd/display: Revert logic for plane modifiers
git bisect bad c46dda486e4f9b86eeac40aaea65a6f93dc44dad
# first bad commit: [c46dda486e4f9b86eeac40aaea65a6f93dc44dad] drm/amd/display: Revert logic for plane modifiersIf this one is good
git bisect good
e60a87e401e5c06cc492bd91b54d579c78df46b3 is the first bad commit
commit e60a87e401e5c06cc492bd91b54d579c78df46b3
Author: Prike Liang <Prike.Liang@amd.com>
Date: Thu Oct 20 14:44:26 2022 +0800
drm/amdkfd: update gfx1037 Lx cache setting
commit 9656db1b933caf6ffaaef10322093fe018359090 upstream.
Update the gfx1037 L1/L2 cache setting.
Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/gpu/drm/amd/amdkfd/kfd_crat.c | 53 ++++++++++++++++++++++++++++++++++-
1 file changed, 52 insertions(+), 1 deletion(-)
$ git bisect log
git bisect start
# status: waiting for both good and bad commits
# good: [e90fbe65c6b31ed48a6f13c232b0ca26688218d5] Linux 6.0.6
git bisect good e90fbe65c6b31ed48a6f13c232b0ca26688218d5
# bad: [3a2fa3c01fc7c2183eb3278bd912e5bcec20eb2a] Linux 6.0.7
git bisect bad 3a2fa3c01fc7c2183eb3278bd912e5bcec20eb2a
# bad: [9352ad87a3a3ba9b58e49ce1a0e291619ab0d26e] media: sun6i-csi: Add a Kconfig dependency on RESET_CONTROLLER
git bisect bad 9352ad87a3a3ba9b58e49ce1a0e291619ab0d26e
# good: [731e6ea3ec1391b706163c24089ea5dfbcbed79a] drm/i915: Extend Wa_1607297627 to Alderlake-P
git bisect good 731e6ea3ec1391b706163c24089ea5dfbcbed79a
# bad: [37fd7341de6c4d93b8be940efe428424de22b3a8] mm/kmemleak: prevent soft lockup in kmemleak_scan()'s object iteration loops
git bisect bad 37fd7341de6c4d93b8be940efe428424de22b3a8
# bad: [ca2893296ae7c8997d67f1bda9080a7f96d716e8] crypto: x86/polyval - Fix crashes when keys are not 16-byte aligned
git bisect bad ca2893296ae7c8997d67f1bda9080a7f96d716e8
# bad: [0a30a47741b6df1f9555a0fac6aebb7e8c363bad] drm/msm: fix use-after-free on probe deferral
git bisect bad 0a30a47741b6df1f9555a0fac6aebb7e8c363bad
# good: [6607901fce79959130db485e7549dacaaeef034e] drm/amdgpu: fix pstate setting issue
git bisect good 6607901fce79959130db485e7549dacaaeef034e
# bad: [e60a87e401e5c06cc492bd91b54d579c78df46b3] drm/amdkfd: update gfx1037 Lx cache setting
git bisect bad e60a87e401e5c06cc492bd91b54d579c78df46b3
# good: [c46dda486e4f9b86eeac40aaea65a6f93dc44dad] drm/amd/display: Revert logic for plane modifiers
git bisect good c46dda486e4f9b86eeac40aaea65a6f93dc44dad
# first bad commit: [e60a87e401e5c06cc492bd91b54d579c78df46b3] drm/amdkfd: update gfx1037 Lx cache settingEdit2:
The kernel issue should then be reported upstream at https://gitlab.freedesktop.org/drm/amd/-/issues (you might want to test linux 6.1 first to see if the issue has been fixed)
Last edited by loqs (2022-12-14 00:32:37)
Offline
Thank you so much for helping with this. R64 is bad too, so that'd be c46dda486e4f9b86eeac40aaea65a6f93dc44dad that is the bad commit.
I'll mention this in the issue I created in the mesa issue tracker at https://gitlab.freedesktop.org/mesa/mes … te_1680917 then look at reporting it in https://gitlab.freedesktop.org/drm/amd/-/issues
I believe I have checked 6.1 before but I will double-check.
Again thank you for building all these versions and telling me the correct folder to run the bisect commands in.
Offline