Perhaps someone has ideas on this topic, a recent attempt to play with Xen was a
reather unlucky event, all what I will be able to demonstrate on that system is
a PV ONNV domU which
ikely will not be very attractive to the audience :(
HW: U40M2, 2 x 2 core AMD revF procs, 8GB MEM, Phonix BIOS 1.5 (latest)
1 x 200GB internal SATA drive
SW: dom0 ONNV build 82, latest VirtManager from Opensolaris.org
attempts:
1) PV (non-graphics) domU - ONNV build 82 (source = iso image ondisk)
- installation into a ZPOOL/ZVOL - success
- installation into a raw partition /dev/dsk - failed with panic in SATA driver
- installation into a file (mkfile) - failed
suddenly I/O stopped during Solaris Installation,
domU still listed in VirtManager but domU frozen no more I/O on the system
2) HVW FreeBSD 6.3 (source = iso image ondisk)
- installation into a ZPOOL/ZVOL, raw device /dev/dsk - failed,
FreeBSD kernel boots then suddenly I/O stopped, domU
still listet in VirtManager but domU frozen, no more I/O on the system
3) HVM Windows 98 (source = iso image on disk)
- installation into a ZPOOL/ZVOL, raw partition /dev/dsk
and file (mkfile) and ''qemu-img'' image _all_ failed,
Windows 86 boots up, Installation starts, as soon as I/O happens
suddenly I/O stopped, domU still listed in Virtmanager but domU
frozen no more I/O on the system
4) HVM Windows XP 64 (source = iso image on disk)
- installation into a ZPOOL/ZVOL, get the blue ''windows setup''
screen
and the I/O suddenly stopped, domU still visible in VirtManager but
no more I/O happens, domU appears frozen
5) HVM OpenSuse 10.3 Linux (source = iso image on disk, on cd)
- installation into a ZPOOL/ZVOL, raw device /dev/dsk - failed,
Linux boots all installation clicks can been made, installation starts
and then suddenly I/O stopped, domU
still listet in VirtManager but domU frozen, no more I/O on the system
in the middle we also faced a ''black-screen of death'' ie.
system
spontaniously rebooted without notice or dump I believe during some steps in 5)
So to sump-up, it seems that everytime an installation attempted
to start I/O to the pseudo-domU disk of whatever nature, I/O
just stopped and the domU''s gone out to luch while the dom0 is up
(apart from the panic and reboot cases)
We even tried to put the target onto a 2nd internal SATA disk, without
any avail, same unlucky results.
I''d really appreciate any hints or explanations....
thanks
frankB
This message posted from opensolaris.org
> Perhaps someone has ideas on this topic, a recent > attempt to play with Xen was a reather unlucky event, > all what I will be able to demonstrate on that system > is a PV ONNV domU which > ikely will not be very attractive to the audience :( > > W: U40M2, 2 x 2 core AMD revF procs, 8GB MEM, Phonix > BIOS 1.5 (latest) > 1 x 200GB internal SATA drive > : dom0 ONNV build 82, latest VirtManager from > Opensolaris.org > > ttempts: > > ) PV (non-graphics) domU - ONNV build 82 (source > iso image ondisk) > > installation into a ZPOOL/ZVOL - successHad you to modify /etc/system at DomU side to get proper networking between SNV82 DomU and SNV82 Dom0 ?> installation into a raw partition /dev/dsk - failed > with panic in SATA driver > > - installation into a file (mkfile) - failed > suddenly I/O stopped during Solaris Installation, > domU still listed in VirtManager but domU frozen no > more I/O on the system > > ) HVW FreeBSD 6.3 (source = iso image ondisk) > > installation into a ZPOOL/ZVOL, raw device /dev/dsk > - failed, > FreeBSD kernel boots then suddenly I/O stopped, > domU > still listet in VirtManager but domU frozen, no > more I/O on the system > > 3) HVM Windows 98 (source = iso image on disk)HVM Windows 2003,XP (32 bit) would work fine.> > installation into a ZPOOL/ZVOL, raw partition > /dev/dsk > and file (mkfile) and ''qemu-img'' image _all_ > failed, > Windows 86 boots up, Installation starts, as soon > as I/O happens > suddenly I/O stopped, domU still listed in > Virtmanager but domU > frozen no more I/O on the system > > 4) HVM Windows XP 64 (source = iso image on disk) > > installation into a ZPOOL/ZVOL, get the blue > ''windows setup'' screen > and the I/O suddenly stopped, domU still visible in > VirtManager but > no more I/O happens, domU appears frozen > > 5) HVM OpenSuse 10.3 Linux (source = iso image on > disk, on cd) > > installation into a ZPOOL/ZVOL, raw device /dev/dsk > - failed,View:- http://lists.xensource.com/archives/html/xen-users/2008-01/msg00699.html It''s known bug.> Linux boots all installation clicks can been made, > installation starts > and then suddenly I/O stopped, domU > still listet in VirtManager but domU frozen, no more > I/O on the system > > in the middle we also faced a ''black-screen of death'' > ie. system > spontaniously rebooted without notice or dump I > believe during some steps in 5) > > o to sump-up, it seems that everytime an installation > attempted > to start I/O to the pseudo-domU disk of whatever > nature, I/O > just stopped and the domU''s gone out to luch while > the dom0 is up > (apart from the panic and reboot cases) > > e even tried to put the target onto a 2nd internal > SATA disk, without > any avail, same unlucky results. > > ''d really appreciate any hints or explanations.... > > thanks > frankBThis message posted from opensolaris.org
On Wed, Feb 20, 2008 at 07:39:45AM -0800, Frank Batschulat wrote:> 1) PV (non-graphics) domU - ONNV build 82 (source = iso image ondisk) > > - installation into a ZPOOL/ZVOL - success > - installation into a raw partition /dev/dsk - failed with panic in SATA driverCan you file a bug with the dump please?> 2) HVW FreeBSD 6.3 (source = iso image ondisk)I don''t believe we''ve tested FreeBSD HVM at all.> 5) HVM OpenSuse 10.3 Linux (source = iso image on disk, on cd)This is broken everywhere in Xen.> 3) HVM Windows 98 (source = iso image on disk) > 4) HVM Windows XP 64 (source = iso image on disk)I think these should work fine in the forthcoming 3.1 For 3.0.4, Windows XP works pretty well. regards john
On Wed, Feb 20, 2008 at 11:38:38AM -0800, Boris Derzhavets wrote:> > ) PV (non-graphics) domU - ONNV build 82 (source > > iso image ondisk) > > > > installation into a ZPOOL/ZVOL - success > Had you to modify /etc/system at DomU side to get proper networking between > SNV82 DomU and SNV82 Dom0 ?There should be no need for any such changes. regards john
John Levon wrote:>> 5) HVM OpenSuse 10.3 Linux (source = iso image on disk, on cd) > > This is broken everywhere in Xen.Only on Intel VT - should work fine on AMD. You can apparently get it to work on Intel with a lot of hard work to get around the graphical installer. It should presumably work Ok with the real mode patches. Evan
On Wed, 20 Feb 2008 20:49:22 +0100, John Levon <john.levon@sun.com> wrote:> On Wed, Feb 20, 2008 at 07:39:45AM -0800, Frank Batschulat wrote: > >> 1) PV (non-graphics) domU - ONNV build 82 (source = iso image ondisk) >> >> - installation into a ZPOOL/ZVOL - success >> - installation into a raw partition /dev/dsk - failed with panic in SATA driver > > Can you file a bug with the dump please?will do once I have the 2 dumps in SWAN and finished a little bit of digging to make judgement to which cat/subcat they belong to.>> 2) HVW FreeBSD 6.3 (source = iso image ondisk) > > I don''t believe we''ve tested FreeBSD HVM at all.hmmm I arrived at that idea based on: http://opensolaris.org/os/community/xen/docs/specs/ Supported configurations include the following: Solaris dom0, Solaris domU, Linux domU, FreeBSD domU>> 5) HVM OpenSuse 10.3 Linux (source = iso image on disk, on cd) > > This is broken everywhere in Xen.does that mean using fedora 8 or CentOS 5.1 or Ubunthu 7.10 make me happy ? I have them all handy here ;-)>> 3) HVM Windows 98 (source = iso image on disk) >> 4) HVM Windows XP 64 (source = iso image on disk) > > I think these should work fine in the forthcoming 3.1 > > For 3.0.4, Windows XP works pretty well.is windows 98 supposed to work at all ? is there a difference between 32 and 64 bit XP ? is service pack 2 for XP required ? thanks! frankB
Frank Batschulat (Home) wrote:> On Wed, 20 Feb 2008 20:49:22 +0100, John Levon <john.levon@sun.com> wrote: > >> On Wed, Feb 20, 2008 at 07:39:45AM -0800, Frank Batschulat wrote: >> >>> 1) PV (non-graphics) domU - ONNV build 82 (source = iso image ondisk) >>> >>> - installation into a ZPOOL/ZVOL - success >>> - installation into a raw partition /dev/dsk - failed with panic in SATA driver >> Can you file a bug with the dump please? > > will do once I have the 2 dumps in SWAN and finished a little bit of digging > to make judgement to which cat/subcat they belong to. > >>> 2) HVW FreeBSD 6.3 (source = iso image ondisk) >> I don''t believe we''ve tested FreeBSD HVM at all. > > hmmm I arrived at that idea based on: > > http://opensolaris.org/os/community/xen/docs/specs/ > > Supported configurations include the following: > > Solaris dom0, Solaris domU, Linux domU, FreeBSD domUThere is a FreeBSD thread on a list somewhere. I thought that 7.x worked and there was a workaround for 6.3. Something to do with the hard disk driver mayybe?>>> 5) HVM OpenSuse 10.3 Linux (source = iso image on disk, on cd) >> This is broken everywhere in Xen. > > does that mean using fedora 8 or CentOS 5.1 or Ubunthu 7.10 > make me happy ? I have them all handy here ;-)Here''s my list; all of which work on my Intel system. : alpha[1]#; xm list Name ID Mem VCPUs State Time(s) Domain-0 0 3886 2 r----- 96.7 asteriskNOW 512 1 0.0 centos3 512 1 0.0 centos4 512 1 0.0 centos5 512 1 0.0 centos5-1 512 1 0.0 debian4 512 1 0.0 embedded 512 1 0.0 fedora6 512 1 0.0 fedora7 512 1 0.0 fedora8 512 1 0.0 gentoo 512 1 0.0 path-lamp 512 1 0.0 pkg 512 1 0.0 rhel3 512 1 0.0 rhel4 512 1 0.0 rhel5 512 1 0.0 s10u3 512 1 0.0 s10u4 512 1 0.0 s10u5 512 1 0.0 Nevada 512 1 0.0 ubuntu 512 1 0.0 vista 1024 1 0.0 winxp 512 1 0.0 : alpha[1]#;>>> 3) HVM Windows 98 (source = iso image on disk) >>> 4) HVM Windows XP 64 (source = iso image on disk) >> I think these should work fine in the forthcoming 3.1 >> >> For 3.0.4, Windows XP works pretty well. > > is windows 98 supposed to work at all ?Not supported. Didn''t work the last time I tried. Maybe in the future.> is there a difference between 32 and 64 bit XP ?Yes.> is service pack 2 for XP required ?No sure. It''s hard to keep track of all the different versions of everything. I am using SP2. MRJ
> On Wed, Feb 20, 2008 at 11:38:38AM -0800, Boris > Derzhavets wrote: > > > > ) PV (non-graphics) domU - ONNV build 82 (source > > > > iso image ondisk) > > > > > > installation into a ZPOOL/ZVOL - success > > Had you to modify /etc/system at DomU side to get > proper networking between > > SNV82 DomU and SNV82 Dom0 ? > > There should be no need for any such changes.John, This issue does exist in SNV81. Looks like, it also mentioned on the list of Known Bugs and issues. 1.Support Solaris domU, Solaris dom0 on Xen. Why looking at bugs fixed in SNV82, i cannot find the one above ? Thanks. -b> > regards > john > _______________________________________________ > xen-discuss mailing list > xen-discuss@opensolaris.orgThis message posted from opensolaris.org
On Thu, Feb 21, 2008 at 03:20:26AM -0800, Boris Derzhavets wrote:> > > Had you to modify /etc/system at DomU side to get > > proper networking between > > > SNV82 DomU and SNV82 Dom0 ? > > > > There should be no need for any such changes. > John, > This issue does exist in SNV81.What issue specifically?> Looks like, it also mentioned on the list of Known Bugs and issues.That applies to various older releases, not necessarily just the current one. regards john
> > This issue does exist in SNV81. > What issue specifically?Installing SNV81 DomU at SNV81 Dom0, i had to add the line :- set xnf:xnf_cksum_offload = 0 to /etc/system on SNV81 DomU side and reboot DomU one more time. When it had been done network between SNV81 DomU and SNV81 Dom0 worked fine. Otherwise, just ssh connection from SNV81 DomU to SNV81 Dom0 caused DomU malfunctioning right away.> > Looks like, it also mentioned on the list of Known > Bugs and issues. > > That applies to various older releases, not > necessarily just the current > one. > > regards > john > _______________________________________________ > xen-discuss mailing list > xen-discuss@opensolaris.orgThis message posted from opensolaris.org
On Wed, 20 Feb 2008 20:49:22 +0100, John Levon <john.levon@sun.com> wrote:> On Wed, Feb 20, 2008 at 07:39:45AM -0800, Frank Batschulat wrote: > >> 1) PV (non-graphics) domU - ONNV build 82 (source = iso image ondisk) >> >> - installation into a ZPOOL/ZVOL - success >> - installation into a raw partition /dev/dsk - failed with panic in SATA driver > > Can you file a bug with the dump please?done, filed bug against the sata driver: 6669666 sata:sata_scsi_destroy_pkt panics during Xen PV install into raw device typical stacktrace looked like: panic string: BAD TRAP: type=e (#pf Page fault) rp=ffffff000f160560 addr=38 occurred in module "unix" due to a NULL pointer dereference ==== panic interrupt thread: 0xffffff000f160c80 PID: 0 on CPU: 1 (last_swtch: -0.04s) PIL: 5 ===cmd: sched t_procp: 0xfffffffffbc3dd70(proc_sched) p_as: 0xfffffffffbc3f9e0(kas) zone: address translation failed for zone_name addr: 8 bytes @ 0x0 t_stk: 0xffffff000f160c70 sp: 0xffffff000f1602c0 t_stkbase: 0xffffff000f15c000 t_pri: 104(SYS) pctcpu: 0.000000 t_lwp: 0x0 psrset: 0 last CPU: 1 idle: 1940 ticks (19.40 seconds) start: Sun Feb 17 18:14:54 2008 age: 884 seconds (14 minutes 44 seconds) interrupted (pinned) thread: 0xffffff000f12ac80 tstate: TS_ONPROC - thread is being run on a processor tflg: T_INTR_THREAD - thread is an interrupt thread T_TALLOCSTK - thread structure allocated from stk T_PANIC - thread initiated a system panic tpflg: none set tsched: TS_LOAD - thread is in memory TS_DONT_SWAP - thread/LWP should not be swapped pflag: SSYS - system resident process pc: 0xfffffffffb84f37d unix:vpanic_common+0x13b startpc: 0xfffffffffb86edb0 unix:thread_create_intr+0x0 unix:vpanic_common+0x13b() unix:panic+0x9c() unix:die+0xc8() unix:trap+0x13cf() unix:cmntrap_pushed+0x7d() -- panic trap data type: 0xe (Page fault) addr 0x38 rp 0xffffff000f160560 savfp 0xffffff000f160690 savpc 0xfffffffffb8344cb %rbp 0xffffff000f160690 %rsp 0xffffff000f160650 %rip 0xfffffffffb8344cb (unix:add_page_to_pool+0x6b) 0%rdi 0xfffffffffbc132c8 1%rsi 0 2%rdx 0xffffff000f160c80 3%rcx 0x7 4%r8 0 5%r9 0xffffff02c22c3000 %rax 0 %rbx 0 %r10 0x274f1 %r11 0x7ffffffffde1 %r12 0xfffffffffbc132c8 %r13 0xffffff0007342e60 %r14 0xfffffffffbc3fac0 %r15 0xfffffffffbc132e8 %cs 0xe030 (GDT unknown,KPL) %ds 0x4b (UCS_SEL) %es 0x4b (UCS_SEL) %fs 0 (KFS_SEL) %gs 0 (KFS_SEL) %ss 0xe02b (GDT unknown,UPL) trapno 0xe (Page fault) err 0 (page not present,read,supervisor) %rfl 0x10246 (parity|zero|interrupt enable|resume) fsbase 0xffff000f160690ff gsbase 0xffff000f160690ff <trap>unix:add_page_to_pool+0x6b() unix:page_destroy_io+0x6a() unix:segkmem_free_vn+0xdc() unix:segkmem_xfree+0x2a() unix:segkmem_free_io+0x20() genunix:vmem_xfree+0x10c() genunix:vmem_free+0x25() unix:kfreea+0x5a() unix:i_ddi_mem_free+0x5d() rootnex:rootnex_teardown_copybuf+0x20() rootnex:rootnex_dma_unbindhdl+0xbe() genunix:ddi_dma_unbind_handle+0x22() sata:sata_scsi_destroy_pkt+0x83() scsi:scsi_destroy_pkt+0x1d() sd:sd_destroypkt_for_buf+0x1c() sd:sd_return_command+0x13a() sd:sdintr+0x50a() sata:sata_txlt_rw_completion+0x17f() nv_sata:nv_complete_io+0x7d() nv_sata:mcp55_packet_complete_intr+0x175() nv_sata:mcp55_intr_port+0x69() nv_sata:mcp55_intr+0x20() unix:av_dispatch_autovect+0x78() unix:dispatch_hardint+0x33() unix:switch_sp_and_call+0x13() unix:do_interrupt+0x9b() unix:xen_callback_handler+0x370() unix:xen_callback+0xcd() unix:HYPERVISOR_sched_op+0x29() unix:HYPERVISOR_block+0x11() unix:mach_cpu_idle+0x12() unix:cpu_idle+0xcc() unix:idle+0x10e() unix:thread_start+0x8() --- frankB
On Fri, 29 Feb 2008 15:56:01 +0100, John Levon <john.levon@sun.com> wrote:> On Fri, Feb 29, 2008 at 02:51:27PM +0100, Frank Batschulat (Home) wrote: > >> 6669666 sata:sata_scsi_destroy_pkt panics during Xen PV install into raw device > > Thanks. > >> %rfl 0x10246 (parity|zero|interrupt enable|resume) >> fsbase 0xffff000f160690ff gsbase 0xffff000f160690ff >> <trap>unix:add_page_to_pool+0x6b() >> unix:page_destroy_io+0x6a() >> unix:segkmem_free_vn+0xdc() > > I /think/ this is a known bug that Stu has a fix for.yepp, correct, I closed it as dup of: 6665404 Xen svn b83a Dom0 kernel panics on scp seconds ago ;-) thanks! frankB
On Fri, Feb 29, 2008 at 02:51:27PM +0100, Frank Batschulat (Home) wrote:> 6669666 sata:sata_scsi_destroy_pkt panics during Xen PV install into raw deviceThanks.> %rfl 0x10246 (parity|zero|interrupt enable|resume) > fsbase 0xffff000f160690ff gsbase 0xffff000f160690ff > <trap>unix:add_page_to_pool+0x6b() > unix:page_destroy_io+0x6a() > unix:segkmem_free_vn+0xdc()I /think/ this is a known bug that Stu has a fix for. regards john