Xen devel - May 2007 - [RFC][PATCH 2/6] HVM PCI Passthrough (non-IOMMU)

misc.patch
    
    - Some fixes in order to solve some compilation errors
    - Imperative for the 1:1 layout

Signed-off-by: Guy Zana <guy@neocleus.com>


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

On Thu, May 31, 2007 at 07:05:01PM -0400, Guy Zana wrote:
> --- a/xen/arch/x86/hvm/platform.c	Sat May 05 13:48:05 2007 +0100
> +++ b/xen/arch/x86/hvm/platform.c	Thu May 31 21:04:53 2007 +0300
> @@ -844,6 +844,9 @@ void send_pio_req(unsigned long port, un
>      struct vcpu *v = current;
>      vcpu_iodata_t *vio;
>      ioreq_t *p;
> +#ifdef CONFIG_NEO
> +    struct cpu_user_regs *regs =
&v->arch.hvm_vcpu.io_op.io_context;
> +#endif
>  
>      if ( size == 0 || count == 0 ) {
>          printk("null pio request? port %lx, count %lx, "
> @@ -870,6 +873,11 @@ void send_pio_req(unsigned long port, un
>      p->addr = port;
>      p->count = count;
>      p->df = df;
> +
> +#ifdef CONFIG_NEO
> +    p->eip = regs->eip;
> +    p->cs = regs->cs;
> +#endif
For merging it would be good if the CONFIG_XXX bits could go away
unless absolutely necessary.
>      p->io_count++;
>  
> --- a/xen/include/asm-x86/e820.h	Sat May 05 13:48:05 2007 +0100
> +++ b/xen/include/asm-x86/e820.h	Thu May 31 21:04:53 2007 +0300
> @@ -3,13 +3,6 @@
>  
>  #include <public/hvm/e820.h>
>  
> -#define E820MAX	128
> -
> -struct e820map {
> -    int nr_map;
> -    struct e820entry map[E820MAX];
> -};
> -
Why does this need to move to hvm/e820.h?

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

> -----Original Message-----
> From: Muli Ben-Yehuda [mailto:muli@il.ibm.com] 
> Sent: Monday, June 04, 2007 9:28 PM
> To: Guy Zana
> Cc: xen-devel@lists.xensource.com
> Subject: Re: [Xen-devel] [RFC][PATCH 2/6] HVM PCI Passthrough 
> (non-IOMMU)
> 
> > +
> > +#ifdef CONFIG_NEO
> > +    p->eip = regs->eip;
> > +    p->cs = regs->cs;
> > +#endif
> 
> For merging it would be good if the CONFIG_XXX bits could go 
> away unless absolutely necessary.
Ok.
> 
> >      p->io_count++;
> >  
> > --- a/xen/include/asm-x86/e820.h	Sat May 05 13:48:05 2007 +0100
> > +++ b/xen/include/asm-x86/e820.h	Thu May 31 21:04:53 2007 +0300
> > @@ -3,13 +3,6 @@
> >  
> >  #include <public/hvm/e820.h>
> >  
> > -#define E820MAX	128
> > -
> > -struct e820map {
> > -    int nr_map;
> > -    struct e820entry map[E820MAX];
> > -};
> > -
> 
> Why does this need to move to hvm/e820.h?
It''s more logical this way, all e820 structures on the same file.
I don''t have the code in front of me but I think I needed struct
e820map in libxc.

Thanks,
Guy.

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

On Mon, Jun 04, 2007 at 05:48:35PM -0400, Guy Zana
wrote:
> > > --- a/xen/include/asm-x86/e820.h	Sat May 05 13:48:05 2007 +0100
> > > +++ b/xen/include/asm-x86/e820.h	Thu May 31 21:04:53 2007 +0300
> > > @@ -3,13 +3,6 @@
> > >  
> > >  #include <public/hvm/e820.h>
> > >  
> > > -#define E820MAX	128
> > > -
> > > -struct e820map {
> > > -    int nr_map;
> > > -    struct e820entry map[E820MAX];
> > > -};
> > > -
> > 
> > Why does this need to move to hvm/e820.h?
> 
> It''s more logical this way, all e820 structures on the same file. 
I
> don''t have the code in front of me but I think I needed struct
> e820map in libxc.
Sounds like a useful separate patch.

Cheers,
Muli

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