Definitely hang in graphic stack, it makes half paint and hangs, if you go to tty2 nad kill the hang window it continue to work, otherwise X is freeze. Can i test 5.4.8 in build repo or so ?
Kernel 5.4.7 issue
5.4.8 is up in [build], linux-next.
- Edited
It present in modesetting and in intel driver as well, bug in i915
[34862.967978] i915 0000:00:02.0: GPU recovery timed out, cancelling all in-flight rendering.
[34862.968624] i915 0000:00:02.0: Resetting chip for hang on rcs0
[34769.080002] Asynchronous wait on fence i915:kwin_x11[6512]:6a1d2 timed out (hint:intel_atomic_commit_ready+0x0/0x50 [i915])
[34722.936964] i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
[34722.937690] [drm:gen8_reset_engines [i915]] *ERROR* rcs0 reset request timed out: {request: 00000001, RESET_CTL: 00000001}
[34722.937736] i915 0000:00:02.0: Resetting chip for hang on rcs0
[34722.939472] [drm:gen8_reset_engines [i915]] *ERROR* rcs0 reset request timed out: {request: 00000001, RESET_CTL: 00000001}
[34722.940196] [drm:gen8_reset_engines [i915]] *ERROR* rcs0 reset request timed out: {request: 00000001, RESET_CTL: 00000001}
[34728.951963] i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
[34734.263959] Asynchronous wait on fence i915:kwin_x11[6512]:6a1d0 timed out (hint:intel_atomic_commit_ready+0x0/0x50 [i915])
[ 4105.944780] broken atomic modeset userspace detected, disabling atomic
https://gitlab.freedesktop.org/drm/intel/issues/673
It has a patch, it should be applied. (reverting changes it does not have proper fix for 5.4 branch)
- Edited
Can i get revert patch in kernel-next? I mean without need to self rebuild kernel.
The problem persists, in modern intel network card is down and up constantly, it does not detect full duplex link, This problem is caused by bug in the module. Only compiling the module version 3.6.0-NAPI of e1000e, and loading it with the parameter InterruptThrottleRate = 16000,16000, the network card, works perfect. This problem occurs after updating to kernel 5.4.7, probe 5.4.8 from build as linux-next and the same problem. With the kernel before 5.4.7 in KaOS, I had no problem.
5.4 branch is just broken.
It has been 12 days since I posted this problem with kernel 5.4.7, I currently have 5.4.13-next installed, which I have been updating since this problem arose, which still persists with the modern intel network interfaces, which they depend on module e1000e, which with each update of the next kernel, I have had to recompile from the sources of version 3.6.0-NAPI, which loading the module manually with the previous parameters of my other post, the network works again. I am still waiting for a definitive solution to this problem, and that the module works correctly and loads automatically.
[root@SeguRed-pc src]# modinfo e1000e
filename: /lib/modules/5.4.13-1-next/updates/drivers/net/ethernet/intel/e1000e/e1000e.ko.gz
version: 3.6.0-NAPI
license: GPL
description: Intel(R) PRO/1000 Network Driver
author: Intel Corporation, <linux.nics@intel.com>
srcversion: A304874A374FD7997DB2215
Linux-next is now at 5.5 (in the build repo), tested that one?
- Edited
demm I am testing 5.4.15 the problem persints. not view the 5.5-next kernel in repo build.
build is not synced. Do not update wait for you mirror to sync.
linuxcuba i am force sync, and install 5.5-next, Please wait to testing now.
At the moment everything is going well with kernel 5.5, the problem with the e1000e module of the intel network interfaces was solved. I keep trying it to see if everything works fine with virtualbox and the i915 intel video.
Thank you
Best Regards
No issues.