https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #41 from Dexuan Cui ---
(In reply to mammootty from comment #40)
This sounds like a bug in the kernel function kern_kldload(): if it can't find
a mounted root file system, it should return some error rather than panic.
--
You
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #40 from mammootty ---
Thanks Wei Hu, We have made mlxen4 module as part of the kernel and commented
out the call to mlx4_request_modules() temporarily.
However we are looking for any change in namei() or vrefact() to handle
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #39 from Wei Hu ---
The cause of this problem is mlx4 driver was loaded before the root filesystem.
On larger VM the odds of hitting this increase as more things are in parallel.
Instead of adding 1 second delay in the
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
mammootty changed:
What|Removed |Added
CC||mammoot...@yahoo.com
--- Comment #38
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #37 from Wei Hu ---
I hit this bug on FreeBSD 12.1 image in Azure with Mellanox CX3 VF. Looks the
root file system is not available at the time mlx4 driver is trying to load
mlx4en kernel module. Adding one second sleep in
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #36 from Dexuan Cui ---
(In reply to pete from comment #35)
I guess nobody tried this yet and I doubt the issue can disappear automatically
with the new MLX driver, as the panic happened in namei() rather than in a MLX
driver
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #35 from p...@twisted.org.uk ---
(In reply to Dexuan Cui from comment #34)
So, I just saw a new version of the Mellanox driver was imported into 12.0
yesterday - has anyone tried this yet ? I am rebuilding on one of my Azure
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #34 from Dexuan Cui ---
(In reply to pete from comment #33)
Last Friday, I happened to reproduce the issue again with the latest stable/11.
UFS was used here.
I'm debugging it now.
--
You are receiving this mail because:
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #33 from p...@twisted.org.uk ---
(In reply to Dexuan Cui from comment #32)
You are trying to boot from ZFS, yes ? I get the panic when I have my root
filesystem as ZFS. It seems to work fine on UFS.
I will try and reproduce
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #32 from Dexuan Cui ---
(In reply to Andriy Gapon from comment #31)
Unluckily I still can't reproduce the issue recently. :-(
I created the Marketplace 11.1 VM on Azure, and can't reproduce the issue:
FreeBSD decui-bsd111
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #31 from Andriy Gapon ---
(In reply to Dexuan Cui from comment #30)
Perhaps. It depends on the documented contract of kern_kldload, if it exists.
In any case, kern_kldload is used very rarely and, except for mlx4 -- or rather
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #30 from Dexuan Cui ---
(In reply to Andriy Gapon from comment #28)
Even if the root filesystem is not mounted when the mlx4_core driver tries to
load mlx4en, I believe we should not get a panic like this. :-(
--
You are
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #29 from Dexuan Cui ---
(In reply to pete from comment #27)
@pete: You have done good tests! We know the issue exists in 12-CURRENT as
well. Thanks! I'm doing more tests too, trying to find out which exact line of
code causes
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #28 from Andriy Gapon ---
I do not know anything about this code and this problem, but I couldn't help
but notice that the driver attempts to load a module.
Does it ensure that the root filesystem is mounted?
--
You are
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #27 from p...@twisted.org.uk ---
Here is a 'bt' from the last ZFS boot attempt, in case it helps.
ZFS filesystem version: 5
ZFS storage pool version: features support (5000)
Timecounters tick every 10.000 msec
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #26 from p...@twisted.org.uk ---
More tests - I built a ZFS machine as D16s_v3 - that boots fine. Then I resized
both of them back to D4s_v3. The UFS machine now boots (which it didn't at that
size before) but the ZFS machine
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #25 from p...@twisted.org.uk ---
OK, this is interesting. With CURRENT booting from UFS:
Works fine if machine size is D16s_v3
Panics, as above, if machine size is D4s_v3
--
You are receiving this mail because:
You are the
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #24 from p...@twisted.org.uk ---
Also worth noting that the 3rd reboot was after a size change. Even on UFS on
11.2 I did find that when it booted sometimes the networking didn't work and I
had to redeploy a couple of times to
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #23 from p...@twisted.org.uk ---
(In reply to Dexuan Cui from comment #21)
So, I rebuilt with CURRENT, and it booted fine on UFS. Then I rebooted it a
couple of times, and the third time it did this:
mlxen0: Ethernet address:
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #22 from p...@twisted.org.uk ---
(In reply to Dexuan Cui from comment #21)
Its interesting that you got the panic on UFS previous - so did I, but I can't
reproduce it anymore either. However it is there on ZFS.
ZFS is superb -
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #21 from Dexuan Cui ---
(In reply to pete from comment #20)
I never used ZFS -- I even don't know how to format a ZFS partition :-), but
previously I did reproduce the panic when mlx VF driver loads: i.e. the panic
in namei(),
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #20 from p...@twisted.org.uk ---
(In reply to Dexuan Cui from comment #18)
It's ZFS.
I build a UFS machine the way you did, then binary upgrade to 11.2, then source
upgrade to STABLE. All fine. Then I copied the data to
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #19 from p...@twisted.org.uk ---
OK, I just did this and (to my surprise) I have a running machine, with
accelerated networking, on 11.2. Will try building from source on this version,
and then the version I got the panic on, to
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #18 from Dexuan Cui ---
(In reply to pete from comment #17)
I did both source code update and binary update, and can't repro the issue
either way.
For binary update, I referred
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #16 from Dexuan Cui ---
(In reply to Dexuan Cui from comment #15)
I also created a 11.1 VM from the MicrosoftOSTC:FreeBSD:11.1:latest image, and
updated it to 11.2 (11.2-RELEASE FreeBSD 11.2-RELEASE #0 r335510: Fri Jun 22
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #15 from Dexuan Cui ---
(In reply to pete from comment #12)
I installed a 11.2 VM on my local Hyper-V host with
http://ftp.freebsd.org/pub/FreeBSD/releases/ISO-IMAGES/11.2/FreeBSD-11.2-RELEASE-amd64-dvd1.iso,
and can't
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #14 from commit-h...@freebsd.org ---
A commit references this bug:
Author: dexuan
Date: Sat Jul 7 00:41:05 UTC 2018
New revision: 336054
URL: https://svnweb.freebsd.org/changeset/base/336054
Log:
hyperv: Fix boot-up after
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #13 from Dexuan Cui ---
(In reply to pete from comment #12)
Our test team can reproduce the issue. I'm going to dig into it.
--
You are receiving this mail because:
You are the assignee for the bug.
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #11 from Dexuan Cui ---
That first bad commit was reported by our test engineer. It turns out the
bisection in unreliable, but I do suspect something outside of the Hyper-V
drivers cause the panic, probably in the mlx driver,
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #10 from Dexuan Cui ---
(In reply to pete from comment #6)
> By the way, when I am using this, I should only have config in rc.conf
> for the mlxen0 device, and not simultaneously for hn0 if I understand
> this right ?
I
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #9 from p...@twisted.org.uk ---
Addendum - I created a complete new machine using the UFS image
"MicrosoftOSTC:FreeBSD:11.1:latest" and did a source upgrade to 11.2-RELEASE
using the version I compiled earlier.
This also panics
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #7 from p...@twisted.org.uk ---
__ _ _
| | | _ \ / | __ \
| |___ _ __ ___ ___ | |_) | (___ | | | |
| ___| '__/ _ \/ _ \| _ < \___ \| | | |
| | | | | __/ __/|
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #6 from p...@twisted.org.uk ---
I am building from 11.2 now. Looking at the commits you referenced, the big
change was for kernel malloc to return unelectable memory unless requested, yes
? If that was causing the panic though,
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #5 from Dexuan Cui ---
(In reply to pete from comment #4)
Building the latest 11.2 tree should be equivalent to the 11.2 iso.
11.2 will be released very soon. If there is a major issue with it, we must try
to fix it ASAP, and
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #4 from p...@twisted.org.uk ---
Am not sure how I would set about booting from an ISO in Azure, as have always
done source upgrades, but I can build the latest 11.2 tree, form which that ISO
was made, with no options and try
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=229167
--- Comment #3 from Dexuan Cui ---
(In reply to pete from comment #2)
Hi pete, can you please help to test the FreeBSD 11.2-RC3 iso
(https://download.freebsd.org/ftp/releases/amd64/amd64/ISO-IMAGES/11.2/FreeBSD-11.2-RC3-amd64-dvd1.iso
?
36 matches
Mail list logo
