Bug 206847 - drivers/char/random.c commit breaks syslog-ng and metalog kernel.log logging
Summary: drivers/char/random.c commit breaks syslog-ng and metalog kernel.log logging
Status: RESOLVED CODE_FIX
Alias: None
Product: Drivers
Classification: Unclassified
Component: Console/Framebuffers (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: James Simmons
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2020-03-13 23:24 UTC by 0strodamus
Modified: 2020-04-14 22:11 UTC (History)
2 users (show)

See Also:
Kernel Version: 5.5.6+ and 5.4.22+
Subsystem:
Regression: No
Bisected commit-id:


Attachments
output from lscpu (2.32 KB, text/plain)
2020-04-12 08:16 UTC, Lars W.
Details
output from lscpu (2.20 KB, text/plain)
2020-04-12 20:42 UTC, 0strodamus
Details

Description 0strodamus 2020-03-13 23:24:24 UTC
commit 1b710b1b10eff9d46666064ea25f079f70bc67a8 (https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=1b710b1b10eff9d46666064ea25f079f70bc67a8) is breaking both syslog-ng and metalog from detecting real-time kernel log updates which results in no logs captured to /var/log/kernel.log. dmesg and /dev/kmsg do show log updates.


If this commit is reversed, normal logging occurs. I have tested reversing this patch with kernel versions 5.5.6, 5.5.9, 5.4.24, and 5.4.25.


I am running kernels compiled locally on a modified Arch Linux AMD x64 setup, using sysvinit and OpenRC.


Thanks for providing the kernel and any help fixing this bug.
Comment 1 Lars W. 2020-04-12 08:16:00 UTC
Created attachment 288369 [details]
output from lscpu

I have the exact same issue with "dmesg --follow" (haven't checked my /var/log/kern.log being affected).
Reverting the mentioned commit also fixes this for me.

I am having this issue on a machine with AMD Opteron CPUs running Gentoo Linux with sysvinit/openrc. I do _not_ have this issue on my Notebook with Intel Core i7 CPU running the same Gentoo Linux sysvinit/openrc combination.

I had a talk in IRC with another user being affected by this and we happen to have the same AMD CPU family and Model on our affected systems. Attached is my CPU information.
Comment 2 0strodamus 2020-04-12 20:42:23 UTC
Created attachment 288399 [details]
output from lscpu

Adding output from lscpu from my system. Since my initial report, I have additionally successfully tested reversing this patch with kernel versions up to 5.6.3.
Comment 3 Sergey Senozhatsky 2020-04-14 09:29:07 UTC
Commit ab6f762f0f53162 5.7-rc1 (Linus' HEAD) should fix the problem.

-ss
Comment 4 0strodamus 2020-04-14 22:11:19 UTC
Thanks! I applied commit ab6f762f0f53162 to 5.6.4 and verified that this fixes the problem.

Note You need to log in before you can comment on or make changes to this bug.