https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222077
Conrad Meyer changed:
What|Removed |Added
Version|11.1-RELEASE|CURRENT
Assignee|freebsd-
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222126
--- Comment #8 from noah.bergba...@tum.de ---
The workaround I'm using now is:
echo "set timeout interval 2" | pfctl -mf - && echo "set timeout interval 1" |
pfctl -mf -
Every hour in crontab.
It causes the purge thread to wake up
(https:
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222126
--- Comment #7 from Kristof Provost ---
Hmm, that looks entirely normal. Your purge thread is sleeping in the
rw_sleep() call. It should be waking up every 10 seconds or so. It's either not
waking up, or it's not actually doing anything whe
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222126
--- Comment #6 from noah.bergba...@tum.de ---
# procstat -kk 16
16 100102 pf purge- mi_switch+0xe5
sleepq_timedwait+0x42 _sleep+0x235 pf_purge_thread+0x8e fork_exit+0x85
fork_trampoline+0xe
--
You are recei
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222126
--- Comment #5 from Kristof Provost ---
Okay, that's an interesting clue.
Let's see if we can work out where it's getting stuck. Can you do 'sudo
procstat -kk '?
--
You are receiving this mail because:
You are the assignee for the bug.
__
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222126
--- Comment #4 from noah.bergba...@tum.de ---
I just realized that because pf is compiled into the kernel, the dtrace probes
need to be fbt:kernel:* instead of fbt:pf:* . With that, it starts up:
dtrace: script './pf.dtrace' matched 8 probe
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222126
--- Comment #3 from Kristof Provost ---
(In reply to noah.bergbauer from comment #2)
The function may have wound up being inlined in your kernel config, so we can't
instrument it.
Just remove whatever functions dtrace can't find, we'll get
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222126
--- Comment #2 from noah.bergba...@tum.de ---
The purge thread is there:
root160,0 0,0 0 16 - DL 10:57 0:00,04 - [pf
purge]
The dtrace script fails:
dtrace: failed to compile script ./pf.dtrace: line 3: pr
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222126
Kristof Provost changed:
What|Removed |Added
CC||k...@freebsd.org
--- Comment #1
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222126
Bug ID: 222126
Summary: pf is not clearing expired states
Product: Base System
Version: 11.1-RELEASE
Hardware: amd64
OS: Any
Status: New
Severity: A
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222077
--- Comment #5 from Conrad Meyer ---
https://reviews.freebsd.org/D12269 is for the comment 3 issue. Does not
address the issue in the original description.
--
You are receiving this mail because:
You are the assignee for the bug.
___
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222077
--- Comment #4 from Conrad Meyer ---
(In reply to Maxim Khitrov from comment #3)
This issue is a userspace leak. It comes from g_metadata_store() in the geom
userspace code:
fd = g_open(name, 1);
...
sectorsize = g_sectors
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222077
--- Comment #3 from Maxim Khitrov ---
I think the same issue also affects geli metadata with 4k sector size. I was
able to reproduce this using the script below, but it took many iterations.
Once it happened, the same data was returned ever
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=219399
--- Comment #238 from Nils Beyer ---
(In reply to Don Lewis from comment #237)
okay, just for fun I increased the dead zone of the user page:
--
Index: sys/amd64/amd64
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=219399
--- Comment #237 from Don Lewis ---
(In reply to Nils Beyer from comment #236)
We now leave the boundary page unmapped, which is supposed to cure the problem
according to what AMD says. That's why the "program to cause Ryzen hang/reboot
on
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=219399
--- Comment #236 from Nils Beyer ---
(In reply to Don Lewis from comment #235)
> Yeah, I'd get in touch with AMD. I suspect that you'll have to go through
> some hardware troubleshooting again.
I haven't gotten any MCA messages yet with
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=219399
--- Comment #228 from Nils Beyer ---
(In reply to Nils Beyer from comment #225)
Froze again - black screen, fans still running...
--
You are receiving this mail because:
You are the assignee for the bug.
_
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=219399
--- Comment #235 from Don Lewis ---
You'll need the linux kmods for doing a full poudriere run I believe. The
lower bits really don't matter. The important thing is staying below
0x7000.
I think the only thing that AGESA 1.0.0.6b
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=219399
--- Comment #233 from Don Lewis ---
(In reply to Nils Beyer from comment #232)
Yeah that should be ok. I get this:
8
-1 12 0x7fffe000 8
As long as the third value ends with fexxx then you have the shared page
fix. There is some stra
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=219399
--- Comment #231 from Nils Beyer ---
(In reply to Don Lewis from comment #229)
can only guess:
---
root@asbach:/usr/local/poudriere/data/logs/bulk/11_1-default/201
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=219399
--- Comment #229 from Don Lewis ---
(In reply to Nils Beyer from comment #228)
That really sounds like the share page problem ...
Do you know what was building at the time?
--
You are receiving this mail because:
You are the assignee for
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=219399
--- Comment #234 from Nils Beyer ---
(In reply to Don Lewis from comment #233)
"linux" and "linux64" modules are indeed loaded:
root@asbach:/root/#kldstat
Id
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222039
Alan Somers changed:
What|Removed |Added
Flags||mfc-stable11?
--- Comment #2 from Al
23 matches
Mail list logo
