The 6.7.3 stable kernel rebase contains new features, improved hardware support, and a number of important fixes across the tree.
Updates may require up to 24 hours to propagate to mirrors. If the following command doesn't work, please retry later:
sudo dnf upgrade --refresh --advisory=FEDORA-2024-2116a8468b
Please login to add feedback.
-1 | 13 | Test Case kernel regression |
This update has been submitted for testing by jforbes.
This update's test gating status has been changed to 'waiting'.
Host: KVM/QEMU Standard PC (Q35 + ICH9, 2009) (pc-q35-7.2) CPU: Intel(R) Core(TM) i7-4790K (8) @ 3.99 GHz Memory: 2.82 GiB
Regressions and Performance PASSED
Working on bare-metal.
This update's test gating status has been changed to 'passed'.
Working fine.
Acer Aspire Intel® Core™ i7-3632QM CPU @ 2.20GHz Graphics Processor: Mesa DRI Intel® HD Graphics 4000.
driver: i915 v: kernel arch: Gen-7 wi-fi: Qualcomm Atheros AR9485 Wireless Network.
Adapter vendor: Lite-On. driver: ath9k v: kernel.
Default fedora btrfs filesystem.
Did you see what Leight Scott said on the devel mailing list?
"Kernel-6.7.3 still has debugging enabled which will break the nvidia driver"
https://bugzilla.rpmfusion.org/show_bug.cgi?id=6859
https://forums.developer.nvidia.com/t/linux-6-7-beta-550-40-07-error-modpost-gpl-incompatible-module-nvidia-ko-uses-gpl-only-symbol-rcu-read-lock/280908
ASRock server with AMD 5950X and 128GB ECC RAM. Running many diverse services 24x7. Not regressions. Note that I previously could not use CPU boost on this server, but with the recent fixes to AMD pstate, It appears to work fine.
Works for me. The enabled Default and Performance tests pass OK.
Work Station, Asus prime mobo - Ryzen5 5600g/AMDGPU 400 Series Chipset, (UEFI) Secure boot, SSD's > RAID1 (ext4) running Plasma DE.
Works fine natively on a Lenovo T16 Gen1 AMD (AMD Ryzen 7 PRO 6850U, integrated Radeon graphics, Qualcomm QCNFA765, Realtek RTL8111/8168/8411).
KDE Spin F39 up to date as of now with stable repos plus the kernel 6.7.3-200.fc39.x86_64. Tainted = 0. No regression test done natively on the host.
Regression test was done in a KVM/QEMU VM with cpu-passthrough with updates-testing repos enabled & up to date:
./runtests.sh
PASS with 6.7.3-200.fc39.x86_64.AMD 3700X, Rog Strix B550-F, nvidia from rpmfusion. Not secure boot. UEFI boot.
Nvidia driver from rpmfusion will not build. It did build for 6.7.0, 6.7.1, and 6.7.2.
Regression tests pass, but as pointed out by johnh99 above, NVIDIA drivers won't build. For me on:
ASUS G750JS 1 x Intel(R) Core(TM) i7-4700HQ CPU @ 2.40GHz Intel Corporation 4th Gen Core Processor Integrated Graphics Controller (rev 06) NVIDIA Corporation GK104M [GeForce GTX 870M] (rev a1)
I need the NVIDIA drivers because the fall-back nouveau drivers run the card without the fans and the laptop cooks and crashes, so this kernel is a bust for me.
default & performance tests pass (KVM)
Unfortunately Nvidia is going to have to fix the Nvidia drivers. This isn't a bug in the upstream kernel, it is a valid bug fix that was backported to 6.6.15 and 6.7.3 (and I think 6.1.76). That fix brought in symbols which are EXPORT_SYMBOL_GPL. For many reasons, I am not going to change that export. From what I have heard, someone has already come up with a workaround for the Nvidia module which doesn't require any changes to the kernel, the question is whether that will be picked up by Nvidia (or rpmfusion) or they will have another solution. I will hold this update in testing for a bit to give Nvidia time to come up with a solution, but if push comes to shove, I can't hold back security updates and bug fixes for everyone else due to an out of tree driver not working.
Patched NVIDIA drivers from RPM Fusion (DLed of their koji site) are building against the kernel again.
This update has been pushed to testing.
This update can be pushed to stable now if the maintainer wishes
Works.
dmesg dmesg: read kernel buffer failed: Opération non permise
Works really well.
@micmor, it's not an error, dmesg is now requires sudo permission, but you can still use
journalctl -k
@agurenko , thank you for the information.
Thankyou to nixuser above. Nvidia akmod, kmod in rpmfusion koji were built late yesterday and work for 6.7.3. Link - https://koji.rpmfusion.org/koji/buildinfo?buildID=27690. Using it now.
I confirm that patch for nvidia drivers exist and works for me. Fedora 39,GTX 1060,kernel 6.7.3,drivers version 545.29.06
Framework 13 AMD 7840U GNOME Wayland, no apparent regressions. There are some issues reported with 6.7 and latest 780M graphics firmware, along with a revert to December firmware workaround, here: https://gitlab.freedesktop.org/drm/amd/-/issues/3097. I haven't reproduced this yet.
LGTM, Lenovo Yoga Slim 7 Pro 14ACH5, CPU AMD Ryzen 9 5900hx, GPU AMD Renoir, Wifi Mediatek mt7921e
jforbes edited this update.
Works fine on Lenovo ThinkPad P1 Gen 2
ok here
https://bugzilla.redhat.com/show_bug.cgi?id=2262577 This kernel broke suspend on AMD Ryzen 7 PRO 6850U with Radeon Graphics
https://bugzilla.redhat.com/show_bug.cgi?id=2262577 I can verify that this bug is legit. I have a laptop with the same CPU from HP (845 G9) and I ran into the same problem.
Works for me. (with nvidia)
In short how to : F39 & nv.drv. 550.40.07 & kernel 6.7.3-200.fc39.x86_64
Source files of 550.40.07 nv.drv are located at /var/lib/dkms/nvidia/550.40.07/sources
Here three files must be patched // as sudo
1) nv-linux.h located in common/inc Go to line 2076 before end of file.Press enter.Copy this piece of code & save changes:
'#if defined(CONFIG_HAVE_ARCH_PFN_VALID) || LINUX_VERSION_CODE < KERNEL_VERSION(6,1,76)
'# define nv_pfn_valid pfn_valid
'#else
'// pre-6.1.76 kernel pfn_valid version without GPL rcu_read_lock/unlock()
static inline int nv_pfn_valid(unsigned long pfn)
{
struct mem_section *ms;
}
'#endif
2) nv-mmap.c & os-mlock.c located in nvidia
2.1)Open nv-mmap.c & replace line 585 with : (nv_pfn_valid(PFN_DOWN(mmap_start))))
Save changes.
2.2) open os-mlock.c & replace - line 116 with : (!nv_pfn_valid(pfn)))
& line 190 with : if (nv_pfn_valid(pfn))
Save changes.
Now go to /var/lib/dkms/nvidia/550.40.07/sources & run make command
Copy the builed files nvidia*.ko to /var/lib/dkms/nvidia/550.40.07/6.7.3-200.fc39.x86_64/x86_64/modile
If directories 6.7.3-200.fc39.x86_64/x86_64/modile do`t exist create them.
Rebuild initramfs.
Reboot.
That's it.
Seems to work fine on bare-metal HP Z440, AMD RX570 with UEFI. Kernel regression tests both PASS, Bug 2252130 still present but OpenCL compute now working albeit with trace errors.
works fine with Thinkpad P1 gen3
This update has been submitted for stable by jforbes.
This update has been pushed to stable.
This kernel freezes during boot. I can enter a few characters onto the screen, but a moment later, it seems to freeze. If I press no keys, all I get is "Loading Linux 6.7.3" frozen onto the screen until rebooting and booting the older kernel in Grub. I am running Fedora 39 Workstation on a Framework Laptop 13 (Intel 12th gen i5).