Re: [Fedora-xen] expected disk performance?
by Ben
Well, I changed the io scheduling on the dom0 to use deadline and got
a 30% improvement. I then noticed that the domU was using the noop
scheduler, so for kicks I changed that to be deadline as well. Now
I'm going about 3 times faster than when I had been this morning. At
6MB/s scap speed over 100Mb ethernet, that's not exactly where I want
it, but it's an awful lot better.
Thanks for this suggestion!
On May 24, 2006, at 12:58 PM, Brian Stein wrote:
> Ben wrote:
>> It's an adaptec card, and I've seen it push over 25MB/s before I
>> went to xen, so it's not the card.
>
> Try changing the I/O elevator to deadline for dom0 and report back
> on the thruput. As there is planned tuning work this could change,
> but for now it may dramatically help.
>
17 years, 11 months
CPU0 FATAL TRAP 6 -> Not possible to run FC5 guest on kernel-xen0-2.6.16-1.2122_FC5smp
by Thomas von Steiger
Hello,
I try to run FC5 guest on dom0 kernel-xen0-2.6.16-1.2122_FC5smp.
FC5 Guest boot's up without any problem only with single core kernel.
If I do something like "yum update" or bonnie++ on this guest I get always
the same "CPU0 FATAL TRAP 6".
On version kernel-xen0-2.6.16-1.2096_FC5 I can not start a FC5 guest ->
guest stops with boot up.
With kernel-xen0-2.6.16-1.2096_FC5 I can run rhel4, rh72 guests stable (not
smp).
Message on remote logserver:
(XEN) HVM_PIT: guest freq in cycles=2999809
(XEN) Assertion '(sp == 0) || (pending_eoi[cpu][sp-1].vector < vector)'
failed, line 189, file irq.c
(XEN) BUG at irq.c:189
(XEN) (file=extable.c, line=77) Pre-exception: ff13173f -> 00000000
(XEN) ----[ Xen-3.0.2-2 Not tainted ]----
(XEN) CPU: 0
(XEN) EIP: e008:[<ff13173f>] __do_IRQ_guest+0x192/0x2af
(XEN) EFLAGS: 00010086 CONTEXT: hypervisor
(XEN) eax: ff1a7138 ebx: ff117d4b ecx: 00004001 edx: 00005535
(XEN) esi: c040b520 edi: 0000000a ebp: ff1b9eac esp: ff1b9e74
(XEN) cr0: 8005003b cr3: 58493000
(XEN) ds: e010 es: e010 fs: e010 gs: e010 ss: e010 cs: e008
(XEN) Xen stack trace from esp=ff1b9e74:
(XEN) ff197519 ff197527 000000bd ff197527 ffbfee80 00000002 00000013
ff1d2d00
(XEN) ffbb5280 ff1e57a0 ff1b9ecc 00000001 00000000 ff117d4b ff1b9edc
ff131250
(XEN) 000000b0 00000000 ff1b9edc ff11cfb0 ff1c3684 ff1c1110 000000b0
ff1d2d00
(XEN) ff1c3680 ff117d4b 00e46103 ff12cdd5 ff1b9ee8 ff117d4b 00000003
ffbbb204
(XEN) c040b520 0000000a ff1b9f7c 00000000 00b00000 ff11b6b1 0000e008
00000246
(XEN) ff1c1100 ffbbb204 5db0a035 00000143 ffb1c080 00000020 00000007
ff11af3f
(XEN) ffbbb194 ff1bfe84 00000001 00000001 5dcb73f9 00000143 ffbbb180
0007a120
(XEN) 00000000 ffb1c080 ffbbb180 00000000 5db0a035 00000143 ffb31788
0007a120
(XEN) 00000000 ffb1c080 ff1b9fac ff11bb79 00000001 ff1c2500 00e46037
ff1b9fb4
(XEN) ff1b9fac ff17e171 00000002 00000001 00000000 00000002 00000000
ff17e376
(XEN) 00000002 00000000 c03b9000 c040b520 0000000a 00000000 c03e6000
00880000
(XEN) 000000f9 00fb0000 00000000 0000e008 00000202 0000007b 0000007b
00000000
(XEN) 00000000 00000000 ffb1c080
(XEN) Xen call trace:
(XEN) [<ff13173f>] __do_IRQ_guest+0x192/0x2af
(XEN) [<ff131250>] do_IRQ+0x58/0x18e
(XEN) [<ff12cdd5>] common_interrupt+0x45/0x50
(XEN) [<ff11b6b1>] __enter_scheduler+0x3d8/0x48c
(XEN) [<ff11bb79>] do_softirq+0xa1/0xb8
(XEN)
(XEN) ************************************
(XEN) CPU0 FATAL TRAP 6 (invalid opcode), ERROR_CODE 0000, IN INTERRUPT
CONTEXT.
(XEN) System shutting down -- need manual reset.
(XEN) ************************************
What can i do ?
Has anybody see the same problem ?
regards,
Thomas
17 years, 11 months
FW: [Fedora-ia64-list] Xen and Xen0 boot on FC6 IPF/Xen
by Zhang, Xiantao
>From Fenghua, Good news!
We also found the initrd issue, the rhel4-u2 ‘s mkinitrd version is too old. We are upgrading it to try, in addition, we modified xen0’s configuration to CONFIG_EXT3_FS=y and CONFIG_JDB=y instead of module there.
Hopefully, this minor change can pass panic : )
Thanks
-Xiantao
________________________________
From: fedora-ia64-list-bounces(a)redhat.com [mailto:fedora-ia64-list-bounces@redhat.com] On Behalf Of Yu, Fenghua
Sent: 2006年5月25日 11:10
To: fedora-ia64-list(a)redhat.com
Subject: [Fedora-ia64-list] Xen and Xen0 boot on FC6 IPF/Xen
Hypervisor and Xen0 from latest IPF/Xen rpms can boot on FC6/Montecito now. Minor changes needed in configuration and initrd to pass kernel panic. And xend will cause system hang. I disable it for the time being. So we can focus more on XenU.
Thanks.
-Fenghua
--
Fedora-ia64-list mailing list
Fedora-ia64-list(a)redhat.com
https://www.redhat.com/mailman/listinfo/fedora-ia64-list
17 years, 11 months
RE: [Fedora-xen] [IA64] [Xen-ia64-devel] Add ptc.l emulation for 2.6.17 xenlinux to fix federon-xen-ia64 boot
by Zhang, Xiantao
Great!
Thanks
-Xiantao
> -----Original Message-----
> From: Aron Griffis [mailto:aron@hp.com]
> Sent: 2006年5月25日 8:33
> To: Zhang, Xiantao
> Cc: fedora-xen(a)redhat.com
> Subject: Re: [Fedora-xen] [IA64] [Xen-ia64-devel] Add ptc.l emulation for
> 2.6.17 xenlinux to fix federon-xen-ia64 boot
>
> Zhang, Xiantao wrote: [Wed May 24 2006, 09:47:54AM EDT]
> > Hi Aron,
>
> Hi Xiantao,
>
> > With this patch, federon-xen-ia64 with rhel4-u2 can boot to "switch to
> > new root". Seems this error is related to FC6 OS, maybe initrd has
> > something wrong in rhel4-u2. Due to lack of FC6 at hand, we couldn't
> > try to boot it on FC6. Anyway, this patch is a must for boot 2.6.17
> > kernel.
> > In addition, I attached the boot log, anyone can help to try on FC6 with
> > this patch?
>
> Thanks you for this patch! Alex applied it to xen-ia64-unstable
> earlier today, and following that I updated my rpm with the new
> changeset.
>
> I am seeing the exact same issue that you reported (/dev/root mounting
> problems) on fedora rawhide. I will let you know if I can debug it.
>
> Regards,
> Aron
17 years, 11 months
[IA64] [Xen-ia64-devel] Add ptc.l emulation for 2.6.17 xenlinux to fix federon-xen-ia64 boot
by Zhang, Xiantao
This patch adds support for ptc.l emulation. In 2.6.16 kernel
flush_tlb_range will call global_tlb_purge directly, so ptc.l shouldn't
be used when CONFIG_SMP enable.
But in order to enhance performance (maybe), 2.6.17 kernel in smp
environment will do mm check first. If mm is current->active_mm and the
mm
(corresponding process) just runs on the local processor, kernel only
needs to do ptc.l at local processor instead of global purge. So ptc.l
emulation is necessary for 2.6.17 kernel.
Hi Aron,
With this patch, federon-xen-ia64 with rhel4-u2 can boot to "switch to
new root". Seems this error is related to FC6 OS, maybe initrd has
something wrong in rhel4-u2. Due to lack of FC6 at hand, we couldn't
try to boot it on FC6. Anyway, this patch is a must for boot 2.6.17
kernel.
In addition, I attached the boot log, anyone can help to try on FC6 with
this patch?
Thanks
-Xiantao
17 years, 11 months
Document the lomount utility from Xen
by Aleksander Adamowski
Hi!
When trying to repair a messed up guest domain's disk image in FC5, I've
discovered a very usefule but undocumented utility from the Xen RPM:
# lomount
You must specify -diskimage and -partition
ex. lomount -t fs-type -diskimage hda.img -partition 1 /mnt
Since the disk images created with xenguest-install.py are images of
whole disk (including the partition table, MBR etc.), they can't simply
be mounted like in the Xen official documentation.
This tool parses the partition table and allows to mount a chosen
partition from such a disk image. It automatically computes the needed
"offset" mount option.
Using lomount I have successfully repaired the guest domain.
I think lomount should be documented in the Wiki
(http://fedoraproject.org/wiki/FedoraXenQuickstartFC5).
If you want, I can write that hint over there myself - just give me edit
permissions in the Wiki (user: AleksanderAdamowski).
--
Best Regards,
Aleksander Adamowski
GG#: 274614
ICQ UIN: 19780575
http://olo.ab.altkom.pl
17 years, 11 months
sparse-merge tutorial
by Aron Griffis
Here is the process I use for porting xenlinux to the kernel version
used by fedora rawhide. I suspect this is very similar to what Juan
is doing, but I haven't talked with him yet about it.
Aron Griffis wrote: [Sat May 20 2006, 03:34:35PM EDT]
> 1. Upstream xen is presently based on 2.6.16.13. Fedora kernel is (or
> was yesterday) based on 2.6.17-rc4-git5. To port xen forward, the
> most maintainable method seems to be to do the port in the context
> of a xen-ia64-unstable mercurial clone (xen-ia64-unstable-2.6.17
> above). Using this method makes it relatively easy to:
>
> (a) port forward to a new kernel at any time using the
> sparse-merge script
>
> (b) pull new changes from upstream xen and avoid most manual
> merging
>
> (c) extract a patch at any time that represents the forward-port
> of xen to a new kernel
>
> (d) generate a patch at any time that adds xen support to the
> fedora kernel (linux-2.6-xen.patch generated with "make
> mkpatches")
First, determine what version the Fedora kernel is based on. Get the
src.rpm and look inside it to see what version it is based on:
$ wget http://download.fedora.redhat.com/pub/fedora/linux/core/development/SRPMS...
$ mkdir tmp
$ cd tmp
$ rpm2cpio ../kernel-2.6.16-1.2208_FC6.src.rpm | cpio -imd
$ ls *.bz2
linux-2.6.16.tar.bz2
patch-2.6.17-rc4-git9.bz2
patch-2.6.17-rc4.bz2
xen-20060503.tar.bz2
So we see that 2208 is based on 2.6.17-rc4-git9. We will need
a tarball for xen to unpack later, so make it now:
$ tar xjf linux-2.6.16.tar.bz2
$ mv linux-2.6.16 linux-2.6.17-rc4-git9
$ cd linux-2.6.17-rc4-git9
$ bzip2 -dc ../patch-2.6.17-rc4-git9.bz2 | patch -p1
$ cd ..
$ tar cjf ../linux-2.6.17-rc4-git9.tar.bz2 linux-2.6.17-rc4-git9
Save off the patch for later use, and remove the temporary directory:
$ mv patch-2.6.17-rc4-git9.bz2 ..
$ cd ..
$ rm -r tmp
Next clone a mercurial kernel tree. This will be for the sparse-merge
script to reference:
$ hg clone http://kernel.org/hg/linux-2.6
And clone the xen repo you want to use (xen-ia64-unstable or
xen-unstable):
$ hg clone http://xenbits.xensource.com/ext/xen-ia64-unstable.hg
Next use the sparse-merge script to port the tree forward. Don't
forget the "v" in NEWTAG. OLDTAG is fetched automatically from
buildconfigs/mk.linux-2.6-xen. ARCH defaults to ia64, so override it
to "." to match all files in the sparse tree.
$ cd xen-ia64-unstable.hg
$ LINUXDIR=$HOME/linux-2.6 NEWTAG=v2.6.17-rc4-git9 ARCH=. \
xen/arch/ia64/tools/sparse-merge
At this point sparse-merge will do its magic. For the most part, the
port will go forward smoothly, but it will also announce rejects as it
goes. When it finishes, find the rejects and fix each one
appropriately:
$ find linux-2.6-xen-sparse -name \*.rej
When you're done, you can remove the rejects and diffs from the sparse
tree. You might save them first in case you'd like to reference them
later:
$ find linux-2.6-xen-sparse -name \*.rej -o -name \*.diff \
| xargs -r tar cjf ../diffs-rejs-2.6.17-rc4-git9.tar.bz2
$ find linux-2.6-xen-sparse -name \*.rej -o -name \*.diff \
| xargs -r rm -f
Next update the buildconfig to reflect the new version that xen is
based on:
$ sed -i '/^LINUX_VER/s/=.*/= 2.6.17-rc4-git9/' \
buildconfigs/mk.linux-2.6-xen
Next do a test build:
$ make world
If it fails, you'll have to investigate and then fix things in the
sparse tree, then do:
$ make kdelete world
You can do this until you get a clean build, then it's time to commit
your changes and make the patch:
$ hg add linux-2.6-xen-sparse
$ hg remove --after linux-2.6-xen-sparse
$ hg st
$ hg ci -m "sparse-merge to 2.6.17-rc4-git9"
$ make mkpatches
At this point you have a patch which should apply to the Fedora
kernel. Here is what I do personally:
$ cd $HOME
$ hg clone http://free.linux.hp.com/~agriffis/fedora-kernel-ia64
$ cp xen-ia64-unstable.hg/linux-2.6-xen.patch \
fedora-kernel-ia64/SOURCES/linux-2.6-xen-ia64-10138.patch
$ rpmbuild -ba fedora-kernel-ia64/SPECS/kernel-2.6.spec
This assumes the patch was made based on (upstream) xen-ia64-unstable
cset 10138, and doesn't count any spec changes you might need to make.
I hope this helps. At this point it's easy to either update the patch
within the existing tree for a new kernel version using sparse-merge,
or update to a newer xen-ia64-unstable cset with hg pull, hg merge.
Hope that helps,
Aron
17 years, 11 months
FW: [Xen-ia64-devel] Add ptc.l emulation for 2.6.17 xenlinux to fixfederon-xen-ia64 boot
by Yang, Fred
Re-forward to the correct mail list
-----Original Message-----
From: Zhang, Xiantao
Sent: Wednesday, May 24, 2006 6:59 AM
Subject: RE: [Xen-ia64-devel] Add ptc.l emulation for 2.6.17 xenlinux to
fixfederon-xen-ia64 boot
After many times' try, I think the main issue is the initrd
which created for booting FC6 couldn't match with rhel4-u2. Maybe their
module format is different. I think we should set up FC6 environment to
try next step. If so, I think debugging work maybe easier :)
> >> -----Original Message-----
> >> From: xen-ia64-devel-bounces(a)lists.xensource.com
> >> [mailto:xen-ia64-devel-bounces@lists.xensource.com] On
> >Behalf Of Zhang,
> >> Xiantao
> >> Sent: Wednesday, May 24, 2006 4:26 AM
> >> To: xen-ia64-devel(a)lists.xensource.com; fedora-ia64-list(a)redhat.com
> >> Subject: [Xen-ia64-devel] Add ptc.l emulation for 2.6.17 xenlinux
to
> >> fixfederon-xen-ia64 boot
> >>
> >> This patch adds support for ptc.l emulation. In 2.6.16 kernel
> >> flush_tlb_range will call global_tlb_purge directly, so
> >ptc.l shouldn't
> >> be used when CONFIG_SMP enable.
> >> But in order to enhance performance (maybe), 2.6.17 kernel in smp
> >> environment
> >> will do mm check first. If mm is current->active_mm and the mm
> >> (corresponding process) just runs on the local processor, kernel
only
> >> needs to do ptc.l at local processor instead of global
> >purge. So ptc.l
> >> emulation is necessary for 2.6.17 kernel.
> >>
> >> Hi Aron,
> >> With this patch, federon-xen-ia64 with rhel4-u2 can boot to
> >"switch to
> >> new root". Seems this error is related to FC6 OS, due to
> >lack of FC6 at
> >> hand, we couldn't try to boot it on FC6. Anyway, this patch is a
must
> >> for boot 2.6.17 kernel.
> >> In addition, I attached the boot log, anyone can help to try
> >on FC6 with
> >> this patch?
> >>
> >> Thanks
> >> -Xiantao
> >
> >
17 years, 11 months
Can't boot rawhide guest on FC5 host
by Orion Poplawski
Not sure if this is expected or not, but I've been unable to boot a
development guest kernel on a FC5 x86_64 host.
[root@hammer ~]# xm create -c fcdev-x86_64
Using config file "/etc/xen/fcdev-x86_64".
Going to boot Fedora Core (2.6.16-1.2208_FC6xenU)
kernel: /vmlinuz-2.6.16-1.2208_FC6xenU
initrd: /initrd-2.6.16-1.2208_FC6xenU.img
Started domain fcdev-x86_64
[root@hammer ~]#
--
Orion Poplawski
System Administrator 303-415-9701 x222
Colorado Research Associates/NWRA FAX: 303-415-9702
3380 Mitchell Lane, Boulder CO 80301 http://www.co-ra.com
17 years, 11 months
