Regression booting winxp under Xen.

Regression booting winxp under Xen.

[Ian Campbell]

Xen's automated test system is reporting[1] that something between
bfb7b58b3068 and 136d4ec190af (for reference the previous pass is[2])
has broken booting winxp under Xen with Seabios.

Logs for one of the failures can be found at[3] (reached by following
the email's link to [4] and clicking the header of a failing column).
The VNC screenshot of the guest shows "Setup is inspecting your hardware
configuration..." (at least for those which have such a screenshot, not
sure why some don't).

I've not investigated more thoroughly yes, just posting in case
something obvious leaps out at someone. The automated test is currently
bisecting the issue, once it is done I'll let you know the result.

[1] http://lists.xen.org/archives/html/xen-devel/2014-10/msg03543.html
[2] http://lists.xen.org/archives/html/xen-devel/2014-10/msg01993.html
[3] http://www.chiark.greenend.org.uk/~xensrcts/logs/31223/test-amd64-i386-xl-qemuu-winxpsp3-vcpus1/info.html
[4] http://www.chiark.greenend.org.uk/~xensrcts/logs/31223/

Re: Regression booting winxp under Xen.

[Ian Campbell]

On Mon, 2014-11-03 at 09:59 +0000, Ian Campbell wrote:

> I've not investigated more thoroughly yes, just posting in case
> something obvious leaps out at someone. The automated test is currently
> bisecting the issue, once it is done I'll let you know the result.

If I'd read a bit further through my Monday morning INBOX I'd have
found:
        http://lists.xen.org/archives/html/xen-devel/2014-11/msg00001.html
        
which indicates that the bisector has fingered:

  commit 99cb8f3e9af516954b2f2fba97ce1856e3d7b93f
  Author: Kevin O'Connor <kevin@xxxxxxxxxxxx>
  Date:   Tue Oct 21 14:34:06 2014 -0400
  
      Do full BREGS backup/restore for pmm, pnp, and irqentry_extrastack
      
      Although these entry points only require backup and restore of the
      registers that the C code clobbers, there is no harm in backing up
      some additional registers.  This allows the BREGS macros to be used
      which makes the code a little more readable.
      
      Signed-off-by: Kevin O'Connor <kevin@xxxxxxxxxxxx>

Ian.

> [1] http://lists.xen.org/archives/html/xen-devel/2014-10/msg03543.html
> [2] http://lists.xen.org/archives/html/xen-devel/2014-10/msg01993.html
> [3] http://www.chiark.greenend.org.uk/~xensrcts/logs/31223/test-amd64-i386-xl-qemuu-winxpsp3-vcpus1/info.html
> [4] http://www.chiark.greenend.org.uk/~xensrcts/logs/31223/
> 
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel

Re: [SeaBIOS] Regression booting winxp under Xen.

[Kevin O'Connor]

On Mon, Nov 03, 2014 at 10:05:05AM +0000, Ian Campbell wrote:
> On Mon, 2014-11-03 at 09:59 +0000, Ian Campbell wrote:
> 
> > I've not investigated more thoroughly yes, just posting in case
> > something obvious leaps out at someone. The automated test is currently
> > bisecting the issue, once it is done I'll let you know the result.
> 
> If I'd read a bit further through my Monday morning INBOX I'd have
> found:
>         http://lists.xen.org/archives/html/xen-devel/2014-11/msg00001.html
>         
> which indicates that the bisector has fingered:
> 
>   commit 99cb8f3e9af516954b2f2fba97ce1856e3d7b93f

Sorry about that - I missed one of the stack offset conversions in the
PNP part of that change.  The fix is below and I just pushed it to
the main repo.

Thanks for catching it.
-Kevin


--- a/src/romlayout.S
+++ b/src/romlayout.S
@@ -286,7 +286,7 @@ entry_pnp_real:
         movw %cx, %ds
         leal BREGS_size-6+12(%esp), %eax  // %eax points to start of u16 args
         calll handle_pnp
-        movw %ax, 12(%esp)      // Modify %eax to return %ax
+        movw %ax, BREGS_eax(%esp)   // Modify %eax to return %ax
         POPBREGS
         popfl
         popl %esp

Re: [SeaBIOS] Regression booting winxp under Xen.

[Ian Campbell]

On Mon, 2014-11-03 at 09:59 -0500, Kevin O'Connor wrote:
> On Mon, Nov 03, 2014 at 10:05:05AM +0000, Ian Campbell wrote:
> > On Mon, 2014-11-03 at 09:59 +0000, Ian Campbell wrote:
> > 
> > > I've not investigated more thoroughly yes, just posting in case
> > > something obvious leaps out at someone. The automated test is currently
> > > bisecting the issue, once it is done I'll let you know the result.
> > 
> > If I'd read a bit further through my Monday morning INBOX I'd have
> > found:
> >         http://lists.xen.org/archives/html/xen-devel/2014-11/msg00001.html
> >         
> > which indicates that the bisector has fingered:
> > 
> >   commit 99cb8f3e9af516954b2f2fba97ce1856e3d7b93f
> 
> Sorry about that - I missed one of the stack offset conversions in the
> PNP part of that change.  The fix is below and I just pushed it to
> the main repo.
> 
> Thanks for catching it.

Thanks for the quick fix!

> -Kevin
> 
> 
> --- a/src/romlayout.S
> +++ b/src/romlayout.S
> @@ -286,7 +286,7 @@ entry_pnp_real:
>          movw %cx, %ds
>          leal BREGS_size-6+12(%esp), %eax  // %eax points to start of u16 args
>          calll handle_pnp
> -        movw %ax, 12(%esp)      // Modify %eax to return %ax
> +        movw %ax, BREGS_eax(%esp)   // Modify %eax to return %ax
>          POPBREGS
>          popfl
>          popl %esp

Re: [SeaBIOS] Regression booting winxp under Xen.

[Ian Campbell]

On Mon, 2014-11-03 at 15:42 +0000, Ian Campbell wrote:
> On Mon, 2014-11-03 at 09:59 -0500, Kevin O'Connor wrote:
> > On Mon, Nov 03, 2014 at 10:05:05AM +0000, Ian Campbell wrote:
> > > On Mon, 2014-11-03 at 09:59 +0000, Ian Campbell wrote:
> > > 
> > > > I've not investigated more thoroughly yes, just posting in case
> > > > something obvious leaps out at someone. The automated test is currently
> > > > bisecting the issue, once it is done I'll let you know the result.
> > > 
> > > If I'd read a bit further through my Monday morning INBOX I'd have
> > > found:
> > >         http://lists.xen.org/archives/html/xen-devel/2014-11/msg00001.html
> > >         
> > > which indicates that the bisector has fingered:
> > > 
> > >   commit 99cb8f3e9af516954b2f2fba97ce1856e3d7b93f
> > 
> > Sorry about that - I missed one of the stack offset conversions in the
> > PNP part of that change.  The fix is below and I just pushed it to
> > the main repo.
> > 
> > Thanks for catching it.
> 
> Thanks for the quick fix!

FYI the next Xen osstest run didn't suffer from this issue, thanks.

(in case you go looking at the results, it did fail for other Xen
related reasons, nothing to do with seabios, but this issue is
definitely gone.)

> 
> > -Kevin
> > 
> > 
> > --- a/src/romlayout.S
> > +++ b/src/romlayout.S
> > @@ -286,7 +286,7 @@ entry_pnp_real:
> >          movw %cx, %ds
> >          leal BREGS_size-6+12(%esp), %eax  // %eax points to start of u16 args
> >          calll handle_pnp
> > -        movw %ax, 12(%esp)      // Modify %eax to return %ax
> > +        movw %ax, BREGS_eax(%esp)   // Modify %eax to return %ax
> >          POPBREGS
> >          popfl
> >          popl %esp
> 
> 
> 
> _______________________________________________
> SeaBIOS mailing list
> SeaBIOS@seabios.org
> http://www.seabios.org/mailman/listinfo/seabios
>