Intel video driver question - cards second function

[Jesse Barnes]

On Saturday, September 15, 2007 10:21:39 am Jedrzej Solecki wrote:
> Hello,
> I am reading the xfree86 video Intel driver source code and I have one
> question (maybe stupid - I'm just a newbie:)).
> I am asking about this piece of code (src/i810_driver.c:I810Probe()):
>
>  /*
>   * Mobile platforms may have both function 0 and 1 active, but they
>   * are handled as a single entity.  To make sure that the function 1
>   * entity isn't assigned to a screen, check for and claim it here
>   * first.
>   *
>   * XXX If function 1's resources are ever needed, they'll need to be
>   * added to the screen and marked active.
>   */
>  for (ppPci = VideoInfo; ppPci != NULL && *ppPci != NULL; ppPci++) {
>     if ((*ppPci)->vendor == PCI_VENDOR_INTEL &&
>         (*ppPci)->func == 1) {
>        for (id = I810PciChipsets; id->PCIid != -1; id++) {
>           if (id->PCIid == (*ppPci)->chipType) {
>              /* Claim slot */
>              if (xf86CheckPciSlot((*ppPci)->bus, (*ppPci)->device,
>                                   (*ppPci)->func)) {
>                 xf86ClaimPciSlot((*ppPci)->bus, (*ppPci)->device,
>                                  (*ppPci)->func, drv, id->PCIid,
>                                  NULL, FALSE);
>              }
>              break;
>           }
>        }
>     }
>  }
>
> I have a notebook with the 945GM controller with active functions 0
> and 1. I would like to ask if this code:
>
>              /* Claim slot */
>              if (xf86CheckPciSlot((*ppPci)->bus, (*ppPci)->device,
>                                   (*ppPci)->func)) {
>                 xf86ClaimPciSlot((*ppPci)->bus, (*ppPci)->device,
>                                  (*ppPci)->func, drv, id->PCIid,
>                                  NULL, FALSE);
>
> should be executed in this case.
> If yes then something is wrong because it doesn't:)
> With the first function (PCI:0:2:0) this condition:
> (*ppPci)->func == 1)
> is not met and that's OK.
> For the second funtion (1) this condition:
> if (id->PCIid == (*ppPci)->chipType)
> is never met because I810PciChipsets does not contain the
> "(*ppPci)->chipType" value. There is only "PCI_CHIP_I945_GM
> 0x27A2" defined in the common.h .
> But the second function's (PCI:0:2:1) chipType value is 0x27A6 which
> is not defined in the common.h.
> I would like to ask if this is not there on purpose and I
> misunderstood this part of code or if there is something wrong:)
>
> lspci -n:
> 00:02.0 0300: 8086:27a2 (rev 03)
> 00:02.1 0380: 8086:27a6 (rev 03)

Your analysis correct, and the code is broken. :)  But the bug is harmless:  
the second device is a vestige of dual head Windows support, and isn't really 
useful.  Eventually we'll probably remove the above code and the 
corresponding warning messages about subfunctions not being claimed...

Jesse