[SOLVED] Laptop no longer suspending

amr962 · 2024-03-01 22:44:17

I was not sure which was most recent, so I tried first and last one. These are the kernels I tested and their result:

r157:   no
r78:    yes
r39:    yes
r9:     yes

Everything was tested while plugged in. I didn't try r19, since I assumed it's older than r39 and does work. 6.7.6.arch1-2 didn't work.

seth · 2024-03-02 09:29:39

To decode the "yes"/"no" there: you mean that the bug shows up between "r78" (still worked) and "r157" (failed)?
So 5d08c617e59ab4dfbfa97a65dae2bad144fa7a6f is good and 00c48bfbd6b29b8ebf64edd059dbf9e95cedd5b1 is bad?
(You could have stopped after r78 then )

My blind guesswork (d855ceb6a5fde668c5431156bc60fae0cc52b764) would be right between those two and is pot. the cause.
You may try to just revert that first, otherwise you'd only have to bisect between the two points you've already tested.

amr962 · 2024-03-02 10:29:32

As said, I didn't know which of them was most recent, so I had to test both sides of the interval to properly binary search the last working version
Yes, r78 worked and r157 failed. I wouldn't be able to confirm which commit id is good and bad, as I couldn't even tell which kernel version was what.
If I understood correctly I should try and reverse this, how would I do that? Do I have to traditionally compile it or can I still use the PKGBUILD? Would that be simply doing

cd "$pkgname-$pkgver"
git apply -R <patch>

in the prepare() method on the commits patch?

seth · 2024-03-02 16:02:29

"git revert d855ceb6a5fde668c5431156bc60fae0cc52b764" should™ do.

https://wiki.archlinux.org/title/Kernel … ild_system for the general process.

amr962 · 2024-03-02 20:39:49

As I had suspected, trying to revert the commit didn't work, because the actual git tree is not there and the commit doesn't exist. So I yoinked the patch file and ran a

   patch -RNp1 < "../amd.patch"

in prepare() to revert it, which worked. I also removed some doc stuff as per the wiki.
The good news: suspending now works!
The maybe not so good news: I built the 6.7.7.arch1-1 version.
I will probably wait until 6.7.7 is available through pacman to confirm that it's also broken.

amr962 · 2024-03-03 10:31:47

6.7.7-arch1-1 is confirmed broken:

[amr@lenovo ~]$ uname -a
Linux lenovo 6.7.7-arch1-1 #1 SMP PREEMPT_DYNAMIC Fri, 01 Mar 2024 15:07:39 +0000 x86_64 GNU/Linux

My patched build which worked:

[amr@lenovo linux]$ git log
commit 3ee8900562726ca75feddde13bef02b5d046938d (HEAD -> main, tag: 6.7.7.arch1-1, origin/main, origin/HEAD)
Author: Jan Alexander Steffens (heftig) <heftig@archlinux.org>
Date:   Fri Mar 1 16:06:26 2024 +0100

    6.7.7.arch1-1

The changes I did:

[amr@lenovo linux]$ git diff
diff --git a/PKGBUILD b/PKGBUILD
index 64753de..16463ac 100644
--- a/PKGBUILD
+++ b/PKGBUILD
@@ -17,12 +17,6 @@ makedepends=(
   python
   tar
   xz
-
-  # htmldocs
-  graphviz
-  imagemagick
-  python-sphinx
-  texlive-latexextra
 )
 options=('!strip')
 _srcname=linux-${pkgver%.*}
@@ -69,6 +63,7 @@ prepare() {
     echo "Applying patch $src..."
     patch -Np1 < "../$src"
   done
+  patch -RNp1 < "../amd.patch"
 
   echo "Setting config..."
   cp ../config .config
@@ -83,7 +78,6 @@ build() {
   cd $_srcname
   make all
   make -C tools/bpf/bpftool vmlinux.h feature-clang-bpf-co-re=1
-  make htmldocs
 }
 
 _package() {
@@ -230,7 +224,6 @@ _package-docs() {
 pkgname=(
   "$pkgbase"
   "$pkgbase-headers"
-  "$pkgbase-docs"
 )
 for _p in "${pkgname[@]}"; do
   eval "package_$_p() {

The patch (copied from the actualy commit:

[amr@lenovo linux]$ cat amd.patch 
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 1 -
 drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c    | 9 ++++++++-
 2 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 5fe1df95dc389..19bc8d47317bb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -4496,7 +4496,6 @@ int amdgpu_device_suspend(struct drm_device *dev, bool fbcon)
 		drm_fb_helper_set_suspend_unlocked(adev_to_drm(adev)->fb_helper, true);
 
 	cancel_delayed_work_sync(&adev->delayed_init_work);
-	flush_delayed_work(&adev->gfx.gfx_off_delay_work);
 
 	amdgpu_ras_suspend(adev);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index b9674c57c4365..6ddc8e3360e22 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -723,8 +723,15 @@ void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable)
 
 		if (adev->gfx.gfx_off_req_count == 0 &&
 		   !adev->gfx.gfx_off_state) {
-			schedule_delayed_work(&adev->gfx.gfx_off_delay_work,
+			/* If going to s2idle, no need to wait */
+			if (adev->in_s0ix) {
+				if (!amdgpu_dpm_set_powergating_by_smu(adev,
+						AMD_IP_BLOCK_TYPE_GFX, true))
+					adev->gfx.gfx_off_state = true;
+			} else {
+				schedule_delayed_work(&adev->gfx.gfx_off_delay_work,
 					     delay);
+			}
 		}
 	} else {
 		if (adev->gfx.gfx_off_req_count == 0) {
-- 
cgit 1.2.3-korg

seth · 2024-03-03 12:30:54

\o/
=> https://gitlab.freedesktop.org/drm/amd/-/issues

amr962 · 2024-03-05 00:53:08

https://gitlab.freedesktop.org/drm/amd/-/issues/3242

amr962 · 2024-04-03 20:31:45

Running the upstream patch on both 6.8.1 and 6.8.2 has fixed the issue for me.

Arch Linux

#26 2024-03-01 22:44:17

Re: [SOLVED] Laptop no longer suspending

#27 2024-03-02 09:29:39

Re: [SOLVED] Laptop no longer suspending

#28 2024-03-02 10:29:32

Re: [SOLVED] Laptop no longer suspending

#29 2024-03-02 16:02:29

Re: [SOLVED] Laptop no longer suspending

#30 2024-03-02 20:39:49

Re: [SOLVED] Laptop no longer suspending

#31 2024-03-03 10:31:47

Re: [SOLVED] Laptop no longer suspending

#32 2024-03-03 12:30:54

Re: [SOLVED] Laptop no longer suspending

#33 2024-03-05 00:53:08

Re: [SOLVED] Laptop no longer suspending

#34 2024-04-03 20:31:45

Re: [SOLVED] Laptop no longer suspending

Board footer