RE: [Xen-devel] IGD passthrough

["Kay, Allen M" <allen.m.kay@xxxxxxxxx>]

I agree with you Sergey.  I will submit a patch to fix this after I do some 
testing with it.

-----Original Message-----
From: Sergey Tovpeko [mailto:tsv.devel@xxxxxxxxx] 
Sent: Friday, March 04, 2011 4:22 AM
To: Kay, Allen M
Cc: xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: Re: [Xen-devel] IGD passthrough

Allen,

I'm concerning about the init value
u32 val = -1;

in pt_pci_host_read

u32 pt_pci_host_read(int bus, int dev, int fn, u32 addr, int len)
{

struct pci_dev *pci_dev;
u32 val = -1;

pci_access_init();
pci_dev = pci_get_dev(dpci_infos.pci_access, 0, bus, dev, fn);
if ( !pci_dev )
return 0;

pci_read_block(pci_dev, addr, (u8 *) &val, len);
return val;
}


Initing with (-1) will fail when reading less then 4 bytes in 
pci_read_block. I run into this in register_vga_regions function when it 
returned vendor_id 0xffff8086 for igd, which left me without opregion at 
all.

It seems to me zero-value in pt_pci_host_read would be more correct.

Sergey.
>
> That's good to hear. I assume both systems are using the same Win7-64 
> IGD driver. I know the IGD drivers for XP and Win7 are different. I 
> have not tried XP environment so there might be problems.
>
> *From:* Sergey Tovpeko [mailto:tsv.devel@xxxxxxxxx]
> *Sent:* Wednesday, March 02, 2011 11:58 AM
> *To:* Kay, Allen M
> *Cc:* xen-devel@xxxxxxxxxxxxxxxxxxx
> *Subject:* Re: [Xen-devel] IGD passthrough
>
> I checked IGD passthrough with native driver on other notebook i5-540 
> cpu. Win7 64 driver works and which make me happy and confuse. I don't 
> know the name of second notebook, it was provided us by our 
> partners.At least both notebooks have the same CPUs.
>
> Thank you for your advice.
>
> Sergey.
>
> On Wed, Mar 2, 2011 at 7:30 AM, Sergey Tovpeko <tsv.devel@xxxxxxxxx 
> <mailto:tsv.devel@xxxxxxxxx>> wrote:
>
> Thank you, Allen for your replay.
>
> I'll try to check this on other hardware.
>
> As for Win7-64. It looks differently how IGD works. I mean when driver 
> starts, I could recognize the windows' greeting and login. But the 
> point is that the display is not normal, not usual. It looks like the 
> brightness is so low, that I could hardly recognize what is output on 
> the screen.
>
> Attached qemu's log for win7.
>
> Sergey.
>
>
>
> On Tue, Mar 1, 2011 at 8:46 PM, Kay, Allen M <allen.m.kay@xxxxxxxxx 
> <mailto:allen.m.kay@xxxxxxxxx>> wrote:
>
> Sergy,
>
> I noticed you are using WinXP guest. Have you tried Win7-64 guest? Do 
> you have another Calpella laptop (i.e. Thinkpad T41) to try on?
>
> It would be interesting to know if this problem is WinXP specific or 
> HP 8440p specific. I know there are there are quite a bit of 
> difference between XP and Win7 IGD driver.
>
> I have ready access to ThinkPad and Dell Calepella laptop.
>
> Allen
>
>
> -----Original Message-----
> From: Sergey Tovpeko [mailto:tsv.devel@xxxxxxxxx 
> <mailto:tsv.devel@xxxxxxxxx>]
> Sent: Tuesday, March 01, 2011 4:08 AM
> To: xen-devel@xxxxxxxxxxxxxxxxxxx <mailto:xen-devel@xxxxxxxxxxxxxxxxxxx>
>
> Cc: Kay, Allen M; weidong.han@xxxxxxxxx <mailto:weidong.han@xxxxxxxxx>
> Subject: [Xen-devel] IGD passthrough
>
> Hello, everybody!
>
> I've got an issue concerning IGD passthrough on calpella/sandybridge 
> platform.
> The IGD have passthrough-ed well on Win7 and WinXp until I setup 
> native driver for GPU.
>
> After installing drivers and rebooting domain the display got black 
> when driver starts.
>
> I found the similar issue
> http://lists.xensource.com/archives/html/xen-devel/2010-07/msg01145.html
>
> Could you provide any advice to make graphic works?
>
>
> My system:
> Hardware : EliteBook 8440p
> Xen - xen-4.1.0-rc6
> Linux 2.6.32.7
>
> Sergey.
>


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