From: Atom2 <ariel.atom2@web2web.at>
To: Doug Goldstein <cardoe@cardoe.com>, xen-devel@lists.xen.org
Subject: Re: HVM domains crash after upgrade from XEN 4.5.1 to 4.5.2
Date: Mon, 16 Nov 2015 21:14:22 +0100 [thread overview]
Message-ID: <564A391E.40308@web2web.at> (raw)
In-Reply-To: <564A32D5.6070704@cardoe.com>
Am 16.11.15 um 20:47 schrieb Doug Goldstein:
> I'm a little late to the thread but can you send me (you can do it
> off-list if you'd like) the USE flags you used for xen, xen-tools and
> seabios? Also emerge --info. You can kill two birds with one stone by
> using emerge --info xen.
>> Hi Doug,
>> here you go:
> Thanks. I'll use your configuration as a test point to update a few
> things with regard to the Gentoo ebuilds. I'm not the maintainer of Xen
> and SeaBIOS but I don't think the maintainers will have much issue with
> the changes.
Hi Doug,
that sounds great. If you require any more information, please do not
hesitate to contact me - either by mail or through this thread.
>> USE flags:
>> app-emulation/xen-4.5.2-r1::gentoo USE="-custom-cflags -debug -efi
>> -flask -xsm"
>> app-emulation/xen-tools-4.5.2::gentoo USE="hvm pam pygrub python qemu
>> screen system-seabios -api -custom-cflags -debug -doc -flask (-ocaml)
>> -ovmf -static-libs -system-qemu" PYTHON_TARGETS="python2_7"
>> sys-firmware/seabios-1.7.5::gentoo USE="binary"
> So looking at how SeaBIOS and friends are built I think we have an issue
> that needs to be addressed. That being said, you wouldn't have this
> issue if you did USE="-system-seabios -system-qemu". I believe you would
> also be ok if you had done USE="system-seabios system-qemu". But after a
> quick look at everything USE="system-seabios -system-qemu" will
> definitely do the wrong thing.
There was a reason why I did not use -system-seabios. The reason (and a
discussion about the issues I was facing back then) can be found in this
thread:
http://lists.xenproject.org/archives/html/xen-users/2015-01/msg00103.html.
This problem was then sorted by changing to the upstream SeaBIOS instead
of the bundled version (described in
http://lists.xenproject.org/archives/html/xen-users/2015-01/msg00131.html)
by changing the USE flag.
I don't know whether there were any relevant changes since xen-4.3.3,
but at that point in time it appeared to be my only chance to keep my
system up and running. Since then (January 2015) I did not find any
reason to revert the USE flag back to -system-seabios as everything was
working as expected - obviously up until now.
Thanks Atom2
>> emerge --info: Please see the attached file
>>> I'm not too familiar with the xen ebuilds but I was pretty sure that
>>> xen-tools is what builds hvmloader and it downloads a copy of SeaBIOS
>>> and builds it so that it remains consistent. But obviously your
>>> experience shows otherwise.
>> You are right, it's xen-tools that builds hvmloader. If I remember
>> correctly, the "system-seabios" USE flag (for xen-tools) specifies
>> whether sys-firmware/seabios is used and the latter downloads SeaBIOS in
>> it's binary form provided its "binary" USE flag is set. At least that's
>> my understanding.
>>> I'm looking at some ideas to improve SeaBIOS packaging on Gentoo and
>>> your info would be helpful.
>> Great. Whatever makes gentoo and xen stronger will be awesome. What
>> immediately springs to mind is to create a separate hvmloader package
>> and slot that (that's just an idea and probably not fully thought
>> through, but ss far as I understood Andrew, it would then be possible to
>> specify the specific firmware version [i.e. hvmloader] to use on xl's
>> command line by using firmware_override="full/path/to/firmware").
>>
>> I also found out that an upgrade to sys-firmware/seabios obviously does
>> not trigger an automatic re-emerge of xen-tools and thus hvmloader.
>> Shouldn't this also happen automatically as xen-tools depends on seabios?
>>
>> Thanks and best regards Atom2
>>
>>
>> P.S. If you prefer to take this off-list, just reply to my mail address.
next prev parent reply other threads:[~2015-11-16 20:14 UTC|newest]
Thread overview: 43+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-11-12 1:08 HVM domains crash after upgrade from XEN 4.5.1 to 4.5.2 Atom2
2015-11-12 12:52 ` Jan Beulich
2015-11-12 13:01 ` Andrew Cooper
2015-11-12 14:29 ` Atom2
2015-11-12 15:32 ` Jan Beulich
2015-11-12 16:43 ` Andrew Cooper
2015-11-12 23:00 ` Atom2
2015-11-13 7:25 ` Jan Beulich
2015-11-13 10:09 ` Andrew Cooper
2015-11-14 0:16 ` Atom2
2015-11-14 20:32 ` Andrew Cooper
2015-11-15 0:14 ` Atom2
2015-11-15 15:12 ` Andrew Cooper
2015-11-16 0:39 ` Atom2
2015-11-16 10:02 ` Andrew Cooper
2015-11-15 20:12 ` Doug Goldstein
2015-11-16 1:05 ` Atom2
2015-11-16 15:31 ` Konrad Rzeszutek Wilk
2015-11-16 19:16 ` Atom2
2015-11-16 19:25 ` Konrad Rzeszutek Wilk
2015-11-16 19:39 ` Doug Goldstein
2015-11-16 19:47 ` Konrad Rzeszutek Wilk
2015-11-16 19:45 ` Atom2
2015-11-16 23:01 ` Andrew Cooper
2015-11-16 23:10 ` Atom2
2015-11-18 22:51 ` Atom2
2015-11-18 23:17 ` Andrew Cooper
2015-11-19 0:31 ` Atom2
2015-11-19 1:06 ` Andrew Cooper
2015-11-19 20:02 ` Atom2
2015-11-19 23:53 ` Andrew Cooper
2015-11-24 11:53 ` Atom2
2015-11-19 10:24 ` Jan Beulich
2015-11-19 10:38 ` Andrew Cooper
2015-11-19 19:51 ` Atom2
2015-11-20 7:57 ` Jan Beulich
2015-11-24 10:32 ` Atom2
2015-11-24 10:43 ` Jan Beulich
2015-11-27 22:51 ` Atom2
2015-11-30 9:04 ` Jan Beulich
2015-11-16 19:47 ` Doug Goldstein
2015-11-16 20:14 ` Atom2 [this message]
2015-11-12 14:12 ` Atom2
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=564A391E.40308@web2web.at \
--to=ariel.atom2@web2web.at \
--cc=cardoe@cardoe.com \
--cc=xen-devel@lists.xen.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.