Re: [Xen-users] SATA drives will not start under xen

[Mike Lovell <mike@xxxxxxxxxxxx>]

Chris wrote:
> Quoting Mike Lovell <mike@xxxxxxxxxxxx>:
>
> > Chris wrote:
> > > Quoting Mike Lovell <mike@xxxxxxxxxxxx>:
> > >
> > > > On 7/18/2009 6:57 PM, Chris wrote:
> > > > > Hello.
> > > > >
> > > > > I am having a peculiar problem.  I am running a dual amd64 system 
> > > > >   on a Tyan Tomcat h1000s S3950 motherboard with 4 Gigs of ram 
> > > > > and   two SATA 3.0 g/s seagate drives (in RAID 1 configuration 
> > > > > using   LVM2).  I am using Gentoo Linux, current as of today.
> > > > > When I boot the system, everything looks good until it starts the 
> > > > >   drives, then the system chokes and spits out a bunch of errors  
> > > > > like  these:
> > > > > --------------------
> > > > >
> > > > > ACPI: PCI Interrupt 0000:01:0e.0[A] -> GSI 11 (level, low) -> IRQ 11
> > > > > ata1: SATA max UDMA/133 cmd 0xffffc2000002c000 ctl   
> > > > > 0xffffc2000002c020 bmdma 0xffffc2000002c031
> > > > > ata2: SATA max UDMA/133 cmd 0xffffc2000002c100 ctl   
> > > > > 0xffffc2000002c120 bmdma 0xffffc2000002c131
> > > > > ata3: SATA max UDMA/133 cmd 0xffffc2000002c200 ctl   
> > > > > 0xffffc2000002c220 bmdma 0xffffc2000002c231
> > > > > ata4: SATA max UDMA/133 cmd 0xffffc2000002c300 ctl   
> > > > > 0xffffc2000002c320 bmdma 0xffffc2000002c331
> > > > > scsi0: sata_svw
> > > > > ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
> > > > > ata1.00: ATA-7: ST3250410AS, 3.AAC, max UDMA/133
> > > > > ata1.00: 488397168 sectors, multi 16: LBA48 NCQ (depth 0/32)
> > > > > ata1.00: qc timeout (cmd0xef)
> > > > > ata1.00: failed to set xfermode (err_mask=0x4)
> > > > > ata1: failed to recover some devices retrying in 5 secs
> > > > > (starts over with SATA link up 1.5 gbps...)
> > > > > ------------------------------------------
> > > > > Eventually the system gives up (after struggling for like 10   
> > > > > minutes) and locks up hard.
> > > > > I found that if I disabled APIC in the BIOS, it will boot -- but  
> > > > > as  soon as I try to start the network, the logs fill with more  
> > > > > of the  same junk, and the system locks up hard.
> > > > > I installed a regular gentoo kernel, and used the exact same  
> > > > > config  file -- the system boots just fine.  No problems  
> > > > > what-so-ever.  So,  this is definitely a Xen thing...
> > > > > Anything I can do to fix this?
> > > > >
> > > > > btw: using kernel version 2.6.21-xen -- the most recent kernel in 
> > > > >   portage -- and xen 3.4.0
> > > > Do you get these same errors on an non-xenified kernel? Also, do the
> > > > errors always contains errors complaining about ata1 or scsi0? If this
> > > > is the case, then the first disk on the controller is bad and probably
> > > > should be replaced. These errors are usually the driver detecting a
> > > > problem disk and trying to handle the errors. Hope that answers your
> > > > question.
> > > >
> > > > mike
> > > As I stated, I have installed a regular gentoo kernel with the same 
> > >  exact .config file, and I do NOT have these problems.  Only with  
> > > Xen.  The controller is fine...  Gotta be a Xen issue.  It stinks  
> > > of an IRQ or some similar conflict to me...
> > > Any other ideas?
> > >
> > > Thanks.
> > My bad for not completely reading your email. Sry.
> >
> > Was the regular kernel you used also a 2.6.21 kernel? I am assuming so
> > since you said you used the exact same .config file and that is always
> > changing between kernel versions. I have never used a controller that
> > uses the sata_svw driver. One other option would be to use a newer
> > kernel. There may not be a newer version in portage, but there are
> > people that have made newer working tarballs.
> >
> > http://lists.xensource.com/archives/html/xen-users/2009-06/msg00200.html
> > http://code.google.com/p/gentoo-xen-kernel/
> >
> > Hunting around the mailing list archives should turn up some additional
> > resources. Hopefully these sites or the archives can help you out.
> >
> > mike
> Bad news -- I added the overlay for 2.6.30-r2 xen patched kernel, used 
> the same .config and recompiled...  it won't boot.  Goes right to a a 
> black screen, then reboots.  Tried remaking my initrd several times a 
> few different ways, just to see (a total shot in the dark) -- but no 
> dice.
> Also, I downloaded a fresh, vanilla 2.6.21 kernel for kernel.org, 
> copied over the same .config file I've been using -- and the system 
> boots without complaint.  The problem is definitely stemming from Xen. 
>  It either doesn't like the broadcom servworks sata_svw SATA chipset, 
> or I'm having an IRQ conflict (or both).  Since it will boot with APIC 
> turned off in the bios, I'm thinking the latter.  Starting up the 
> network (with APIC off) causes the conflict.  Booting with the APIC 
> turned on in the bios, the conflict just happens sooner... (well, as 
> good a theory as any other...)
> I have now put in about 30 hours on getting this thing to work.  I 
> have been a very big Xen fan... but I am sad to say, since I'm on a 
> deadline here, I may have to look to KVM or VMware... :(
> I appreciate the input, Mike.  If anyone has any great ideas here, I'd 
> love to hear them, and soon!
> Thanks.
>
> Chris
Which version of the xen hypervisor do you have installed? If you have 
an older one, the current xenified kernel release might not work.
mike


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