Re: [Qemu-devel] [PATCH 0/1] qemu:pci-assign: try to pci-assign.c

[Laszlo Ersek]

On 08/27/14 11:13, Tiejun Chen wrote:
> As you know I'm working on supporting IGD passthrough.
> 
> Here we need load VGABIOS to work out IGD case. Obviously something may
> be duplicated to kvm codes, we should unify some codes but looks its not
> easy to finish that in short time. So as Michael suggestion, at least
> we'd better split assigned_dev_load_option_rom to reuse on both kvm
> and xen.
> 
> I don't finish all IGD stuff patches but here I'd like to post
> some related codes to show how to use assigned_dev_load_option_rom()
> lately.
> 
> +static int get_vgabios(XenPCIPassthroughState *s, void *ptr,
> +                       XenHostPCIDevice *dev)
> +{
> +    int size = 0;
> +
> +    size = dev_load_option_rom(&s->dev, OBJECT(dev), ptr, dev->domain,
> +                               dev->bus, dev->dev, dev->func);
> +
> +    return size;
> +}
> +
> +int xen_pt_setup_vga(XenPCIPassthroughState *s, XenHostPCIDevice *dev)
> +{
> +    void *bios = NULL;
> +    int bios_size = 0;
> +    int rc = 0;
> +
> +    if (!is_vga_passthrough(dev)) {
> +        return rc;
> +    }
> +
> +    bios_size = get_vgabios(s, bios, dev);
> +    if (!bios || !bios_size) {
> +        XEN_PT_ERR(NULL, "VGA: getting VBIOS!\n");
> +        rc = -1;
> +        goto out;
> +    }
> ...
> 
> ----------------------------------------------------------------
> Tiejun Chen (1):
>       hw/pci-assign: split pci-assign.c
> 
>  hw/i386/kvm/pci-assign.c    | 170 
> +++--------------------------------------------------------------------------------------------------------------------------------------
>  include/hw/pci/pci_assign.h | 204 
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 207 insertions(+), 167 deletions(-)
>  create mode 100644 include/hw/pci/pci_assign.h
> 
> Thanks
> Tiejun
> 

Just to make it clear: I'm assuming that you CC'd me because
get-maintainer.pl listed my name, as a committer. I had one series in
this area that improved error message propagation and error reporting.

I have minimal knowledge about device assignment proper, so I will not
review your patch. Please don't wait for my review, and if you have
further patches in the area, there's no need to CC me -- I can't help.

Thanks,
Laszlo