Re: [Xen-devel] [RFC][PATCH 1/1] Add IOREQ_TYPE_VMWARE_PORT

[Don Slutz <dslutz@xxxxxxxxxxx>]

On 09/29/14 07:13, Paolo Bonzini wrote:
> Il 26/09/2014 20:51, Don Slutz ha scritto:
> > +    memcpy(&regs->rax, &vio->io_data, vio->io_size);
> Shouldn't this match the code you use below for case
> HVMIO_handle_vmport_awaiting_completion?
Nope.  This path is either:

vio->io_state == HVMIO_completed at entry or the the error path
for "If there is no backing DM".  In both cases what to put in rax
is in vio->io_data with vio->io_size set also.

This code comes from hvmemul_do_io().

   -Don Slutz

> Paolo
>
> > +    return X86EMUL_OKAY;
> > +}
> > +
> >   static int hvmemul_do_io(
> >       int is_mmio, paddr_t addr, unsigned long *reps, int size,
> >       paddr_t ram_gpa, int dir, int df, void *p_data)
> > diff --git a/xen/arch/x86/hvm/io.c b/xen/arch/x86/hvm/io.c
> > index 9f565d6..96ef3ff 100644
> > --- a/xen/arch/x86/hvm/io.c
> > +++ b/xen/arch/x86/hvm/io.c
> > @@ -206,6 +206,25 @@ void hvm_io_assist(ioreq_t *p)
> >           else
> >               memcpy(&guest_cpu_user_regs()->rax, &p->data, vio->io_size);
> >           break;
> > +    case HVMIO_handle_vmport_awaiting_completion:
> > +    {
> > +        struct cpu_user_regs *regs = guest_cpu_user_regs();
> > +
> > +        /* Always zero extension for eax */
> > +        regs->rax = (uint32_t)(p->addr >> 32);
> > +        /* Only zero extension if 32bit register changed */
> > +        if ( (uint32_t)regs->rbx != (uint32_t)p->addr )
> > +            regs->rbx = (uint32_t)p->addr;
> > +        if ( (uint32_t)regs->rcx != (uint32_t)p->count )
> > +            regs->rcx = (uint32_t)p->count;
> > +        if ( (uint32_t)regs->rdx != (uint32_t)p->size )
> > +            regs->rdx = (uint32_t)p->size;
> > +        if ( (uint32_t)regs->rsi != (uint32_t)(p->data >> 32) )
> > +            regs->rsi = (uint32_t)(p->data >> 32);
> > +        if ( (uint32_t)regs->rdi != (uint32_t)p->data )
> > +            regs->rdi = (uint32_t)p->data;
> > +    }

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel