Xen devel - Apr 2010 - Xen-4.0.0, qemu-xen-4.0, pvops 2.6.32.10 combo on debian-lenny

Hi,
Just done installing .deb packages of Xen-4.0, xen-qemu-dm-4.0, with pvops
kernel-2.6.32.10 with backported nouveau.

Enable XEN_GNTDEV in kernel config - or sdl-enabled domUs will be mere
blank windows.
This is changed in qemu-dm-4.0 from qemu-dm-3.4

tap:aio (or tap:tapdisk:aio) both work (code in BlktapController.py
forces subtyp to tapdisk
if not given) with pvops domU. Previously (with xen-3.4), tap:aio:
worked only for disk-file
images and not partition-file images.

Two blktap0 nodes are created in /dev/xen andin /dev/xen/blktap-2. One
creation is by udev
and the other by mknod in the xen-tools.But the blktap0 and tapdev0 in
/dev/xen/blktap-2 are
not removed on exit of domU.

BTW, in a recent post on page-sharing, it was said that MEMSHRLIBS is
not enabled in
blktap2 - but tapdisk2 IS linked with it. Maybe that leads to the foll. problem.

HVM WinXP (with GPL PV drivers), which worked before with tap:aio now
crash the whole
system horribly. Apparently, assertion of list_empty(&vreq->next),
line 1822 in tapdisk-vbd.c
failed - causing tapdisk2 to segfault -  leading to kernel-bug. GOOD
NEWS is that the same HVM
works fine with tap:aio changed to file: .

kernel: ------------[ cut here ]------------
kernel: Kernel BUG at ffffffff81158cd3 [verbose debug info unavailable]
kernel: invalid opcode: 0000 [#1] PREEMPT SMP
kernel: last sysfs file: /sys/devices/platform/coretemp.0/temp1_input
kernel: CPU 1
kernel: Modules linked in: nf_conntrack_ipv4 nf_defrag_ipv4 xt_state
nf_conntrack xt_physdev iptable_filter ip_tables x_tables tun fbcon
font bitblit softcursor nouveau ttm drm_kms_helper drm cfbcopyarea
cfbimgblt cfbfillrect bridge stp llc ipv6 nfsd lockd nfs_acl
auth_rpcgss sunrpc exportfs fuse skge sky2
kernel: Pid: 6807, comm: qemu-dm Not tainted
2.6.32.10-xen0-git20100410+asusp5wd #1 System Product Name
kernel: RIP: e030:[<ffffffff81158cd3>]  [<ffffffff81158cd3>]
blk_finish_request+0x2c/0x235
kernel: RSP: e02b:ffff88002cf41b88  EFLAGS: 00010087
kernel: RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffffffffffffffff
kernel: RDX: 00000000ffffff01 RSI: 0000000000000000 RDI: ffff88003e0a19d0
kernel: RBP: ffff88003e0a19d0 R08: 0000000000000000 R09: 00000000003bcad7
kernel: R10: 0000000000000082 R11: 0000000000000019 R12: ffff880037dd4b70
kernel: R13: ffff880037b4c9d0 R14: 0000000000000082 R15: 0000000000000000
kernel: FS:  00007fbf532c96f0(0063) GS:ffff880001d05000(0000)
knlGS:0000000000000000
kernel: CS:  e033 DS: 0000 ES: 0000 CR0: 000000008005003b
kernel: CR2: 0000000000000000 CR3: 000000002fb79000 CR4: 0000000000002660
kernel: DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
kernel: DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
kernel: Process qemu-dm (pid: 6807, threadinfo ffff88002cf40000, task
ffff88003f9952c0)
kernel: Stack:
kernel:  ffff880037b4c9d0 0000000000000000 ffff88003e0a19d0 ffff880037dd4b70
kernel: <0> ffff880037b4c9d0 0000000000000082 ffffffff8107c999
ffffffff811590bd
kernel: <0> ffff880037dd4b70 ffff88003e0a19d0 ffff880037b4c800
ffffffff811ca5b9
kernel: Call Trace:
kernel:  [<ffffffff8107c999>] ? sync_page_killable+0x0/0x2f
kernel:  [<ffffffff811590bd>] ? __blk_end_request+0x22/0x2a
kernel:  [<ffffffff811ca5b9>] ? blktap_device_do_request+0x275/0x294
kernel:  [<ffffffff8135e7dd>] ? _spin_unlock_irqrestore+0x19/0x3d
kernel:  [<ffffffff81046dbb>] ? del_timer+0x69/0x72
kernel:  [<ffffffff8100dda9>] ? xen_force_evtchn_callback+0x9/0xa
kernel:  [<ffffffff8100e5d2>] ? check_events+0x12/0x20
kernel:  [<ffffffff8107c999>] ? sync_page_killable+0x0/0x2f
kernel:  [<ffffffff81159960>] ? generic_unplug_device+0x21/0x30
kernel:  [<ffffffff8107c98f>] ? sync_page+0x3c/0x46
kernel:  [<ffffffff8135e7dd>] ? _spin_unlock_irqrestore+0x19/0x3d
kernel:  [<ffffffff8107c9a2>] ? sync_page_killable+0x9/0x2f
kernel:  [<ffffffff8135d256>] ? __wait_on_bit_lock+0x3f/0x84
kernel:  [<ffffffff8107c8dc>] ? __lock_page_killable+0x5d/0x63
kernel:  [<ffffffff8105044b>] ? wake_bit_function+0x0/0x23
kernel:  [<ffffffff8107e09c>] ? generic_file_aio_read+0x3bc/0x548
kernel:  [<ffffffff810aa019>] ? do_sync_read+0xce/0x113
kernel:  [<ffffffff811ba35c>] ? notify_remote_via_irq+0x2d/0x75
kernel:  [<ffffffff81048ad1>] ? signal_wake_up+0x24/0x33
kernel:  [<ffffffff8105041d>] ? autoremove_wake_function+0x0/0x2e
kernel:  [<ffffffff81028424>] ? pvclock_clocksource_read+0x3a/0x70
kernel:  [<ffffffff8100e32d>] ? xen_clocksource_read+0x35/0x53
kernel:  [<ffffffff810aaa40>] ? vfs_read+0xaa/0x146
kernel:  [<ffffffff810aab33>] ? sys_pread64+0x57/0x77
kernel:  [<ffffffff81010e42>] ? system_call_fastpath+0x16/0x1b
kernel: Code: 57 41 89 f7 41 56 41 55 41 54 55 48 89 fd 53 48 83 ec 08
f6 47 49 10 74 0c 48 8b 7f 40 48 89 ee e8 14 1a 00 00 48 39 6d 00 74
04 <0f> 0b eb fe 83 3d 3a 15 4d 00 00 74 0b 83 7d 4c 01 75 05 e8 62
kernel: RIP  [<ffffffff81158cd3>] blk_finish_request+0x2c/0x235
kernel:  RSP <ffff88002cf41b88>
kernel: ---[ end trace cc62d0967b8547ab ]---
ONLY power-switch control active at this point.

On xen-4.0.0 packaging:
Followed the debian way of modifying prefixes, rpaths to move
libraries used by xen-utils
into PRIVATE_LIBDIR and executables into PRIVATE_BINDIR. If linking to
these libs
are required, the ''dev'' package installs static libraries in
the
system LIBDIR. Executable
are linked through a wrapper-script.
TODO: libxenlight, libxlutil and xl patches - which cosily tie up together.
If anyone is interested, I can mail the debian-directory.tar.gz which
contains all the patches
that gets applied before a debian build.

I am posting this info ''cos I went on some goose-chase trying to
figure out dependencies, ...
and trying to figure out what was missing in getting it up and running
- what with the qemu-dm
git being hosed (now it is OK). Now I know that this combo is sufficient.

Next task - get a new virsh up:-)  Never tried it before. Or track down fbdev:-(

Enlightenment on dom0 is COOL.
Arvind.

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On Sun, Apr 11, 2010 at 07:52:41PM +0530, Arvind R
wrote:
> Hi,
> Just done installing .deb packages of Xen-4.0, xen-qemu-dm-4.0, with pvops
> kernel-2.6.32.10 with backported nouveau.
> 
> Enable XEN_GNTDEV in kernel config - or sdl-enabled domUs will be mere
> blank windows.
> This is changed in qemu-dm-4.0 from qemu-dm-3.4
Ahh, I remember seeing those errors on QEMU and never tracked them down
further than that..
.. snip ..
> Next task - get a new virsh up:-)  Never tried it before. Or track down
fbdev:-(
By fbdev you mean QEMU SDL dependency on the GNTDEV? Or is this issues
with the Dom0 fbdev itself so the barebone SDL apps on top of fb don''t
work properly?

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On Mon, Apr 12, 2010 at 8:24 PM, Konrad Rzeszutek Wilk
<konrad.wilk@oracle.com> wrote:
> On Sun, Apr 11, 2010 at 07:52:41PM +0530, Arvind R wrote:
>> Hi,
>> Just done installing .deb packages of Xen-4.0, xen-qemu-dm-4.0, with
pvops
>> kernel-2.6.32.10 with backported nouveau.
>>
>> Enable XEN_GNTDEV in kernel config - or sdl-enabled domUs will be mere
>> blank windows.
>> This is changed in qemu-dm-4.0 from qemu-dm-3.4
>
> Ahh, I remember seeing those errors on QEMU and never tracked them down
> further than that..
> .. snip ..
>> Next task - get a new virsh up:-)  Never tried it before. Or track down
fbdev:-(
>
> By fbdev you mean QEMU SDL dependency on the GNTDEV? Or is this issues
> with the Dom0 fbdev itself so the barebone SDL apps on top of fb
don''t
> work properly?
>
The latter. There are issues with KMS Xorg server (1.7.5+) that
prevents fbdev (DDX)
from working on Xen.

The dependency on GNTDEV is ok if documented, and is because qemu-dm
has some new xc_gnttab_map_gnt_ref() code in xen_console.c and xen_backend.c
- haven''t pursued it.

Meanwhile, have packaged all but remus. xl create fails on waiting for
tap creation
but xl destroy works. Lots of failure modes overall - very unstable I guess.

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel