[Intel-gfx] ResettRe: [Xen-devel] [v5][PATCH 0/5] xen: add Intel IGD passthrough support

Kay, Allen M allen.m.kay at intel.com
Sat Jul 19 02:27:21 CEST 2014


> For the MCH PCI registers that do need to be read - can you tell us which ones those are?

In qemu/hw/xen_pt_igd.c/igd_pci_read(), following MCH PCI config register reads are passthrough to the host HW.   Some of the registers are needed by Ironlake GFX driver which we probably can remove.  I did a trace recently on Broadwell,  the number of register accessed are even smaller (0, 2, 2c, 2e, 50, 52, a0, a4).  Given that we now have integrated MCH and GPU in the same socket, looks like driver can easily remove reads for offsets 0 - 0x2e.

		case 0x00:        /* vendor id */
		case 0x02:        /* device id */
		case 0x08:        /* revision id */
		case 0x2c:        /* sybsystem vendor id */
		case 0x2e:        /* sybsystem id */
		case 0x50:        /* SNB: processor graphics control register */
		case 0x52:        /* processor graphics control register */
		case 0xa0:        /* top of memory */
		case 0xb0:        /* ILK: BSM: should read from dev 2 offset 0x5c */
		case 0x58:        /* SNB: PAVPC Offset */
		case 0xa4:        /* SNB: graphics base of stolen memory */
		case 0xa8:        /* SNB: base of GTT stolen memory */

Allen

-----Original Message-----
From: Konrad Rzeszutek Wilk [mailto:konrad.wilk at oracle.com] 
Sent: Friday, July 18, 2014 6:45 AM
To: Kay, Allen M
Cc: Michael S. Tsirkin; Jesse Barnes; peter.maydell at linaro.org; xen-devel at lists.xensource.com; Ross Philipson; airlied at linux.ie; daniel.vetter at ffwll.ch; intel-gfx at lists.freedesktop.org; Kelly.Zytaruk at amd.com; qemu-devel at nongnu.org; Anthony Perard; Stefano Stabellini; anthony at codemonkey.ws; Paolo Bonzini; Zhang, Yang Z; Chen, Tiejun
Subject: Re: [Intel-gfx] ResettRe: [Xen-devel] [v5][PATCH 0/5] xen: add Intel IGD passthrough support

On Thu, Jul 17, 2014 at 05:37:12PM +0000, Kay, Allen M wrote:
> > That sounds great. Tiejun could you confirm that with windows driver guys too?
> 
> I believe windows driver can also assume specific CPU/PCH combos.  I will discuss this with native Windows driver guys.  Preferably, the same code path can be used for both native and virtualization cases to avoid frequent breakage as most developers and QA do not test new code changes in virtualization environment.
> 
> I have verified that Windows driver do not need to write to any MCH PCI registers on HSW/BDW so the PCI write function can be removed.  The  MCH PCI registers that need to be read, we are working with HW team to get them mirrored in GPU MMIO registers in future HW.
> 

For the MCH PCI registers that do need to be read - can you tell us which ones those are?

Thank you!
> Allen
> 
> -----Original Message-----
> From: Intel-gfx [mailto:intel-gfx-bounces at lists.freedesktop.org] On 
> Behalf Of Michael S. Tsirkin
> Sent: Thursday, July 03, 2014 1:28 PM
> To: Jesse Barnes
> Cc: peter.maydell at linaro.org; xen-devel at lists.xensource.com; Ross 
> Philipson; Konrad Rzeszutek Wilk; airlied at linux.ie; 
> daniel.vetter at ffwll.ch; intel-gfx at lists.freedesktop.org; 
> Kelly.Zytaruk at amd.com; qemu-devel at nongnu.org; Anthony Perard; Stefano 
> Stabellini; anthony at codemonkey.ws; Paolo Bonzini; Zhang, Yang Z; Chen, 
> Tiejun
> Subject: Re: [Intel-gfx] ResettRe: [Xen-devel] [v5][PATCH 0/5] xen: 
> add Intel IGD passthrough support
> 
> On Thu, Jul 03, 2014 at 12:09:28PM -0700, Jesse Barnes wrote:
> > On Thu, 3 Jul 2014 14:26:12 -0400
> > Konrad Rzeszutek Wilk <konrad.wilk at oracle.com> wrote:
> > 
> > > On Thu, Jul 03, 2014 at 10:32:12AM +0300, Michael S. Tsirkin wrote:
> > > > On Wed, Jul 02, 2014 at 12:23:37PM -0400, Konrad Rzeszutek Wilk wrote:
> > > > > On Wed, Jul 02, 2014 at 04:50:15PM +0200, Paolo Bonzini wrote:
> > > > > > Il 02/07/2014 16:00, Konrad Rzeszutek Wilk ha scritto:
> > > > > > >With this long thread I lost a bit context about the 
> > > > > > >challenges that exists. But let me try summarizing it here 
> > > > > > >- which will hopefully get some consensus.
> > > > > > >
> > > > > > >1). Fix IGD hardware to not use Southbridge magic addresses.
> > > > > > >    We can moan and moan but I doubt it is going to change.
> > > > > > 
> > > > > > There are two problems:
> > > > > > 
> > > > > > - Northbridge (i.e. MCH i.e. PCI host bridge) configuration 
> > > > > > space addresses
> > > > > 
> > > > > Right. So in  drivers/gpu/drm/i915/i915_dma.c:
> > > > > 1135 #define MCHBAR_I915 0x44                                                        
> > > > > 1136 #define MCHBAR_I965 0x48                     
> > > > > 
> > > > > 1147         int reg = INTEL_INFO(dev)->gen >= 4 ? MCHBAR_I965 : MCHBAR_I915;        
> > > > > 1152         if (INTEL_INFO(dev)->gen >= 4)                                          
> > > > > 1153                 pci_read_config_dword(dev_priv->bridge_dev, reg + 4, &temp_hi); 
> > > > > 1154         pci_read_config_dword(dev_priv->bridge_dev, reg, &temp_lo);             
> > > > > 1155         mchbar_addr = ((u64)temp_hi << 32) | temp_lo;                
> > > > > 
> > > > > and
> > > > > 
> > > > > 1139 #define DEVEN_REG 0x54                                                          
> > > > > 
> > > > > 1193         int mchbar_reg = INTEL_INFO(dev)->gen >= 4 ? MCHBAR_I965 : MCHBAR_I915; 
> > > > > 1202         if (IS_I915G(dev) || IS_I915GM(dev)) {                                  
> > > > > 1203                 pci_read_config_dword(dev_priv->bridge_dev, DEVEN_REG, &temp);  
> > > > > 1204                 enabled = !!(temp & DEVEN_MCHBAR_EN);                           
> > > > > 1205         } else {                                                                
> > > > > 1206                 pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp); 
> > > > > 1207                 enabled = temp & 1;                                             
> > > > > 1208         }                                                
> > > > > > 
> > > > > > - Southbridge (i.e. PCH i.e. ISA bridge) vendor/device ID; 
> > > > > > some versions of the driver identify it by class, some versions identify it by slot (1f.0).
> > > > > 
> > > > > Right, So in  drivers/gpu/drm/i915/i915_drv.c the giant 
> > > > > intel_detect_pch which sets the pch_type based on :
> > > > > 
> > > > >  432                 if (pch->vendor == PCI_VENDOR_ID_INTEL) {                       
> > > > >  433                         unsigned short id = pch->device & INTEL_PCH_DEVICE_ID_MASK;
> > > > >  434                         dev_priv->pch_id = id;                                  
> > > > >  435                                                                                 
> > > > >  436                         if (id == INTEL_PCH_IBX_DEVICE_ID_TYPE) { 
> > > > > 
> > > > > It checks for 0x3b00, 0x1c00, 0x1e00, 0x8c00 and 0x9c00.
> > > > > The INTEL_PCH_DEVICE_ID_MASK is 0xff00
> > > > > > 
> > > > > > To solve the first, make a new machine type, PIIX4-based, 
> > > > > > and pass through the registers you need.  The patch must 
> > > > > > document _exactly_ why the registers are safe to pass.  If 
> > > > > > they are not reserved on PIIX4, the patch must document what 
> > > > > > the same offsets mean on PIIX4, and why it's sensible to 
> > > > > > assume that firmware for virtual machine will not read/write them.  Bonus point for also documenting the same for Q35.
> > > > > 
> > > > > OK. They look to be related to setting up an MBAR , but I 
> > > > > don't understand why it is needed. Hopefully some of the i915 folks CC-ed here can answer.
> > > > 
> > > > In particular, I think setting up MBAR should move out of i915 
> > > > and become the bridge driver tweak on linux and windows.
> > > 
> > > That is an excellent idea.
> > > 
> > > However I am still curious to _why_ it has to be done in the first place.
> > 
> > The display part of the GPU is partly on the PCH, and it's possible 
> > to mix & match them a bit, so we have this detection code to figure 
> > things out.  In some cases, the PCH display portion may not even be 
> > present, so we look for that too.
> > 
> > Practically speaking, we could probably assume specific CPU/PCH 
> > combos, as I don't think they're generally mixed across generations, 
> > though SNB and IVB did have compatible sockets, so there is the 
> > possibility of mixing CPT and PPT PCHs, but those are register 
> > identical as far as the graphics driver is concerned, so even that should be safe.
> > 
> > Beyond that, the other MCH data we need to look at is mirrored into 
> > the GPU's MMIO space on current gens.  On older gens, we do need to 
> > poke at the memory controller a bit to get some info (see 
> > intel_setup_mchbar()), but that's not true of newer stuff.  Looks 
> > like we only short circuit that on VLV though; we could probably do 
> > it on
> > SNB+.
> 
> That sounds great. Tiejun could you confirm that with windows driver guys too?
> 
> > --
> > Jesse Barnes, Intel Open Source Technology Center
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx



More information about the Intel-gfx mailing list