[PATCH] Convert Intel 82371AB/EB/MB boards to TINY_BOOTBLOCK

[Uwe Hermann]

See patch.

Uwe.

[PATCH] Convert Intel 82371AB/EB/MB boards to TINY_BOOTBLOCK

[Stefan Reinauer]

On 10/7/10 8:52 AM, Uwe Hermann wrote:
> v4_i82371eb_bootblock.patch
>
> Convert all Intel 82371AB/EB/MB based boards to TINY_BOOTBLOCK.
>
> Also:
>
> Unfortunately Intel 440BX + 82371AB/EB/MB boards can have their ISA device
> on various PCI bus:device.function locations.
> Examples we encountered: 00:07.0, 00:04.0, or 00:14.0.
>
> Thus, instead of hardcoding PCI bus:device.function numbers such as
> PCI_DEV(0, 7, 0), we now simply find the ISA device via PCI IDs, which
> works the same on all boards.
>
> As an additional benefit this patch also gets rid of one .c file include
> in romstage.c.
>
> Abuild-tested.
>
> Signed-off-by: Uwe Hermann

Acked-by: Stefan Reinauer 
with one comment:
> Index: src/southbridge/intel/i82371eb/i82371eb_enable_rom.c
>
> --- src/southbridge/intel/i82371eb/i82371eb_enable_rom.c	(Revision 5917)
> +++ src/southbridge/intel/i82371eb/i82371eb_enable_rom.c	(Arbeitskopie)
> -static void i82371eb_enable_rom(device_t dev)
> +static void i82371eb_enable_rom(void)
>  {
>    u16 reg16;
>
> device_t dev;
>
> /*
> * Note: The Intel 82371AB/EB/MB ISA device can be on different
> * PCI bus:device.function locations on different boards.
> * Examples we encountered: 00:07.0, 00:04.0, or 00:14.0.
> * But scanning for the PCI IDs (instead of hardcoding
> * bus/device/function numbers) works on all boards.
> */
>
> dev = pci_locate_device(PCI_ID(PCI_VENDOR_ID_INTEL,
> 		       PCI_DEVICE_ID_INTEL_82371AB_ISA), 0);

Does it make sense to use pci_locate_device_on_bus() here instead? Well,
maybe not because we start on bus 0 and the device is always there.
Maybe it's kind of too philosophical to distinguish.

Stefan

[PATCH] Convert Intel 82371AB/EB/MB boards to TINY_BOOTBLOCK

[Uwe Hermann]

On Thu, Oct 07, 2010 at 09:20:58AM -0700, Stefan Reinauer wrote:
> Acked-by: Stefan Reinauer

Thanks, r5918.
> > dev = pci_locate_device(PCI_ID(PCI_VENDOR_ID_INTEL,
> > 		       PCI_DEVICE_ID_INTEL_82371AB_ISA), 0);
>
> Does it make sense to use pci_locate_device_on_bus() here instead? Well,
> maybe not because we start on bus 0 and the device is always there.
> Maybe it's kind of too philosophical to distinguish.

Hm, dunno. Maybe it saves some microseconds during boot or makes the
image slightly smaller? In that case it might make sense, but otherwise
it probably doesn't matter much.

Uwe.