You are not logged in.
gothmog123 wrote:sorry maybe i'm slow. what exactly is the solution? i've been having freezes, kernel panics, everything with ryzen 3 for a few days.
If your issues matches the oops in post #5 then see the upstream report in post #14 and there is a built kernel with the fix linked in the edit to post #17.
i see thanks. do you think the issue is gonna be fixed in the next offical kernel?
Offline
i see thanks. do you think the issue is gonna be fixed in the next offical kernel?
No. https://gitlab.freedesktop.org/drm/amd/ … te_1979495
Very rough guess a few weeks.
Offline
One of my AMD PC running Arch had a close related problem since upgrade to 6.3.9.arch1-1. I thought the kernel was the problem but the system was running fine on TTY3 (console cli). Every time I switched to graphic mode it was hanging after a few seconds. I checked in /var/log/pacman.log to see what was upgraded alongside the kernel and the culprit seems to be mesa.
I downgrade mesa from mesa 23.1.3-1 to mesa 23.1.2-1 and the system seems back to normal.
Maybe I'm off topic here, I hope not, I'm just giving a hint...
Sous Linux depuis 1998 / Using Linux since 1998
Offline
One of my AMD PC running Arch had a close related problem since upgrade to 6.3.9.arch1-1. I thought the kernel was the problem but the system was running fine on TTY3 (console cli). Every time I switched to graphic mode it was hanging after a few seconds. I checked in /var/log/pacman.log to see what was upgraded alongside the kernel and the culprit seems to be mesa.
I downgrade mesa from mesa 23.1.3-1 to mesa 23.1.2-1 and the system seems back to normal.
Maybe I'm off topic here, I hope not, I'm just giving a hint...
Thanks for this suggestion! It's likely not related to OP's issue so sorry for the offtopic also, but I came to the forums looking for help as I have struggled with a hanging system since yesterday. Saw on htop how the shared memory usage kept creeping up until 100% memory utilization which led to problems as I don't use swap.
Figured it had something to do with graphics as e.g. trying video playback on FF led to this issue almost instantly. Downgrading to mesa 23.1.2-1 seems to have fixed it instantly.
Offline
Fix will be added in 6.5-rc1. not fixed in 6.4.0.
Offline
6.4.1 with the following commits applied (everything currently in agd5f/drm-fixes-6.4 which did not make it into 6.4.1)
git cherry-pick -n e2ad8e2df432498b1cee2af04df605723f4d75e6 # drm/amdgpu: make sure BOs are locked in amdgpu_vm_get_memory
git cherry-pick -n ca0b954a4315ca2228001c439ae1062561c81989 # drm/amdgpu: make sure that BOs have a backing store
git cherry-pick -n ef3c36a6e025e9b16ca3321479ba016841fa17a0 # drm/amdgpu: Skip mark offset for high priority rings
git cherry-pick -n ea2c3c08554601b051d91403a241266e1cf490a5 # drm/amdgpu: fix clearing mappings for BOs that are always valid in VM
git cherry-pick -n fd21987274463a439c074b8f3c93d3b132e4c031 # drm/amd/pm: revise the ASPM settings for thunderbolt attached scenario
git cherry-pick -n e5df16d9428f5c6d2d0b1eff244d6c330ba9ef3a # drm/amdgpu/sdma4: set align mask to 255
git cherry-pick -n 0f48a4b83610cb0e4e0bc487800ab69f51b4aca6 # drm/amd/display: add a NULL pointer check
git cherry-pick -n 26518b39181876064850209ecdab48c0ee5924b1 # drm/amd/display: disable seamless boot if force_odm_combine is enabled
git cherry-pick -n ed83fe2abcace898fdec5c2ba0455703178ac9a3 # drm/amd/display: Fix 128b132b link loss handling
git cherry-pick -n 1af3d0a8e8b8db855ee3c98d210f8ee01b2bb80f # drm/amd/pm: update the LC_L1_INACTIVITY setting to address possible noise issue
git cherry-pick -n 072030b1783056b5de8b0fac5303a5e9dbc6cfde # drm/amd: Disable PSR-SU on Parade 0803 TCON
git cherry-pick -n 2da0036ea99bccb27f7fe3cf2aa2900860e9be46 # drm/amd/pm: add abnormal fan detection for smu 13.0.0
git cherry-pick -n 4ff96bcc0d40b66bf3ddd6010830e9a4f9b85d53 # drm/amdgpu: check RAS irq existence for VCN/JPEG
git cherry-pick -n af22d6a869cc26b519bfdcd54293c53f2e491870 # drm/amd/display: perform a bounds check before filling dirty rectangles
git cherry-pick -n 570b295248b00c3cf4cf59e397de5cb2361e10c2 # drm/amdgpu: fix number of fence calculations
git cherry-pick -n 5c6d52ff4b61e5267b25be714eb5a9ba2a338199 # drm/amd: Don't try to enable secure display TA multiple times
https://drive.google.com/file/d/1Hp7Lp_ … sp=sharing linux-6.4.3.arch1-1.1-x86_64.pkg.tar.zst
https://drive.google.com/file/d/1LLsscj … sp=sharing linux-headers-6.4.3.arch1-1.1-x86_64.pkg.tar.zst
Edit:
Updated for 6.4.3 which does not contain the fix.
Last edited by loqs (2023-07-13 13:52:52)
Offline
Also having this issue, Thank you for providing the kernel with patches applied for fixing it
Offline
Even older versions of Ryzen CPUs face the same issue, 3xxx series with Vega graphics. Thanks for patched kernel, but I will simply ignore linux package till mainline gets patched.
Offline
sorry but why don't the arch devs patch the kernel?
Offline
Looks like the patch did not find its way to 6.4.3: https://git.kernel.org/pub/scm/linux/ke … inux-6.4.y
Offline
@loqs, thanks for the kernel. i have read your post too late and already built the 6.4.3-arch1 kernel for my own. But how did you change the kernel version in PKGBUILD and how can i get rid of this dirty tag? And how did you change the version? And can i rename my kernel without rebuilding?
I think the solution is to make an branch and commit everything?
Last edited by arthurw (2023-07-13 12:44:19)
Offline
But how did you change the kernel version in PKGBUILD
I have been changing the pkgver https://gitlab.archlinux.org/archlinux/ … KGBUILD#L4 and pkgrel https://gitlab.archlinux.org/archlinux/ … KGBUILD#L5
and how can i get rid of this dirty tag?
Put all patches after the block ending on https://gitlab.archlinux.org/archlinux/ … GBUILD#L60 which fixes the version
And can i rename my kernel without rebuilding?
No.
This is a diff of the changes I made for linux-6.4.3.arch1-1.1-x86_64.pkg.tar.zst which also include not building the docs linux-docs package.
diff --git a/PKGBUILD b/PKGBUILD
index 66231b0..a9ea965 100644
--- a/PKGBUILD
+++ b/PKGBUILD
@@ -2,7 +2,7 @@
pkgbase=linux
pkgver=6.4.3.arch1
-pkgrel=1
+pkgrel=1.1
pkgdesc='Linux'
_srctag=v${pkgver%.*}-${pkgver##*.}
url="https://github.com/archlinux/linux/commits/$_srctag"
@@ -16,14 +16,9 @@ makedepends=(
libelf
pahole
perl
+ python
tar
xz
-
- # htmldocs
- graphviz
- imagemagick
- python-sphinx
- texlive-latexextra
)
options=('!strip')
_srcname=archlinux-linux
@@ -59,6 +54,23 @@ prepare() {
make -s kernelrelease > version
make mrproper
+ git cherry-pick -n e2ad8e2df432498b1cee2af04df605723f4d75e6 # drm/amdgpu: make sure BOs are locked in amdgpu_vm_get_memory
+ git cherry-pick -n ca0b954a4315ca2228001c439ae1062561c81989 # drm/amdgpu: make sure that BOs have a backing store
+ git cherry-pick -n ef3c36a6e025e9b16ca3321479ba016841fa17a0 # drm/amdgpu: Skip mark offset for high priority rings
+ git cherry-pick -n ea2c3c08554601b051d91403a241266e1cf490a5 # drm/amdgpu: fix clearing mappings for BOs that are always valid in VM
+ git cherry-pick -n fd21987274463a439c074b8f3c93d3b132e4c031 # drm/amd/pm: revise the ASPM settings for thunderbolt attached scenario
+ git cherry-pick -n e5df16d9428f5c6d2d0b1eff244d6c330ba9ef3a # drm/amdgpu/sdma4: set align mask to 255
+ git cherry-pick -n 0f48a4b83610cb0e4e0bc487800ab69f51b4aca6 # drm/amd/display: add a NULL pointer check
+ git cherry-pick -n 26518b39181876064850209ecdab48c0ee5924b1 # drm/amd/display: disable seamless boot if force_odm_combine is enabled
+ git cherry-pick -n ed83fe2abcace898fdec5c2ba0455703178ac9a3 # drm/amd/display: Fix 128b132b link loss handling
+ git cherry-pick -n 1af3d0a8e8b8db855ee3c98d210f8ee01b2bb80f # drm/amd/pm: update the LC_L1_INACTIVITY setting to address possible noise issue
+ git cherry-pick -n 072030b1783056b5de8b0fac5303a5e9dbc6cfde # drm/amd: Disable PSR-SU on Parade 0803 TCON
+ git cherry-pick -n 2da0036ea99bccb27f7fe3cf2aa2900860e9be46 # drm/amd/pm: add abnormal fan detection for smu 13.0.0
+ git cherry-pick -n 4ff96bcc0d40b66bf3ddd6010830e9a4f9b85d53 # drm/amdgpu: check RAS irq existence for VCN/JPEG
+ git cherry-pick -n af22d6a869cc26b519bfdcd54293c53f2e491870 # drm/amd/display: perform a bounds check before filling dirty rectangles
+ git cherry-pick -n 570b295248b00c3cf4cf59e397de5cb2361e10c2 # drm/amdgpu: fix number of fence calculations
+ git cherry-pick -n 5c6d52ff4b61e5267b25be714eb5a9ba2a338199 # drm/amd: Don't try to enable secure display TA multiple times
+
local src
for src in "${source[@]}"; do
src="${src%%::*}"
@@ -79,7 +91,6 @@ prepare() {
build() {
cd $_srcname
_make all
- _make htmldocs
}
_package() {
@@ -204,29 +215,9 @@ _package-headers() {
ln -sr "$builddir" "$pkgdir/usr/src/$pkgbase"
}
-_package-docs() {
- pkgdesc="Documentation for the $pkgdesc kernel"
-
- cd $_srcname
- local builddir="$pkgdir/usr/lib/modules/$(<version)/build"
-
- echo "Installing documentation..."
- local src dst
- while read -rd '' src; do
- dst="${src#Documentation/}"
- dst="$builddir/Documentation/${dst#output/}"
- install -Dm644 "$src" "$dst"
- done < <(find Documentation -name '.*' -prune -o ! -type d -print0)
-
- echo "Adding symlink..."
- mkdir -p "$pkgdir/usr/share/doc"
- ln -sr "$builddir/Documentation" "$pkgdir/usr/share/doc/$pkgbase"
-}
-
pkgname=(
"$pkgbase"
"$pkgbase-headers"
- "$pkgbase-docs"
)
for _p in "${pkgname[@]}"; do
eval "package_$_p() {
Edit:
https://lore.kernel.org/stable/20230712 … 9b@gregkh/
Last edited by loqs (2023-07-13 13:53:10)
Offline
@loqs, thank you! +1 for
sorry but why don't the arch devs patch the kernel?
Offline
sorry but why don't the arch devs patch the kernel?
Few devs read the forums. Is there a bug report on https://bugs.archlinux.org for this issue?
Offline
No i don't think so. Nothing found.
Ohhh did anybody tried 6.4.3 with the patch and an amd cpu? regarding: https://bugs.archlinux.org/task/79090
Last edited by arthurw (2023-07-13 13:48:07)
Offline
Even older versions of Ryzen CPUs face the same issue, 3xxx series with Vega graphics. Thanks for patched kernel, but I will simply ignore linux package till mainline gets patched.
I'm thinking of doing that but is it safe? Not gonna cause problems with other updated packages depending on the newer kernels?
Offline
lpr1 wrote:I will simply ignore linux package till mainline gets patched.
I'm thinking of doing that but is it safe? Not gonna cause problems with other updated packages depending on the newer kernels?
Userspace packages are in general independent of the kernel. The one update that might cause an issue is gcc has just been updated, so DKMS package updates built using the new gcc would fail for the old kernel as the kernel requires matching gcc versions. So I would suggest ignoring dkms packages along with the kernel.
Offline
gothmog123 wrote:lpr1 wrote:I will simply ignore linux package till mainline gets patched.
I'm thinking of doing that but is it safe? Not gonna cause problems with other updated packages depending on the newer kernels?
Userspace packages are in general independent of the kernel. The one update that might cause an issue is gcc has just been updated, so DKMS package updates built using the new gcc would fail for the old kernel as the kernel requires matching gcc versions. So I would suggest ignoring dkms packages along with the kernel.
i switched to lts kernel in the meantime. (alongside to mainline in my case, and mainline not updated) Anybody experiences with 6.4.3 and the patch with amd?
Last edited by arthurw (2023-07-16 17:24:20)
Offline
Fix is queued for 6.4.4 https://git.kernel.org/pub/scm/linux/ke … ings.patch
Offline
gothmog123 wrote:lpr1 wrote:I will simply ignore linux package till mainline gets patched.
I'm thinking of doing that but is it safe? Not gonna cause problems with other updated packages depending on the newer kernels?
Userspace packages are in general independent of the kernel. The one update that might cause an issue is gcc has just been updated, so DKMS package updates built using the new gcc would fail for the old kernel as the kernel requires matching gcc versions. So I would suggest ignoring dkms packages along with the kernel.
Good to know, did not face any issue so far, but I don't have dkms installed at all, now about packages, I guess I have none of them either.
Offline
Is the issue resolved with 6.4.4.arch1-1?
Offline
I had the problems with the 6.3.9 kernel (AMD Athlon 3000G CPU), I just installed the 6.4.4. First 3 reboots went ok with 6.4.4, give me some time to check whether it remains stable.
Offline
Is the issue resolved with 6.4.4.arch1-1?
I had the issue as well, so far with 6.4.4 all good.
Offline
@loqs, thanks very much for your help. Works. Marked as resolved.
But i am still on lts because of this bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2701
Last edited by arthurw (2023-07-22 09:12:31)
Offline