Re: [osrc-patches] [PATCH] x86: Add PCI IDs for AMD Barcelona PCI devices

[Jesse Barnes]

On Monday, July 14, 2008 2:15 am Robert Richter wrote:
> On 11.07.08 22:08:37, Arjan van de Ven wrote:
> > On Sat, 12 Jul 2008 06:56:15 +0200
> >
> > Ingo Molnar <mingo@xxxxxxx> wrote:
> > > * Robert Richter <robert.richter@xxxxxxx> wrote:
> > > > Ingo, what about this patch. Could you apply it somewhere to the
> > > > tip tree? I will then fix all hardcoded device entries in the code.
> > >
> > > i suspect we could do it if the include/linux/pci_ids.h modification
> > > is fine with Jesse - it appears the maintenance policy right now is
> > > for everyone to add to include/linux/pci_ids.h on an as-needed basis:
> > >
> > > $ git-log-line linus..linux-next include/linux/pci_ids.h
> > >
> > > # 1126de5: Merge commit 'mmc/next'
> > > # bd3b052: Merge commit 'galak/powerpc-next'
> > > # edf0e24: powerpc/85xx: Add support for MPC8536DS
> > > # 34f80b0: bnx2x: Add support for BCM57711 HW
> > > # d3bca0e: sdhci: support JMicron secondary interface
> > > # 4ae127d: Merge branch 'master' of
> > > master.kernel.org:/pub/scm/linux/kernel/git/ # da65b53e4: Merge
> > > branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/ #
> > > da57e6983: tg3: Add 5785 ASIC revision
> > >
> > > Which would normally go fine and not create conflicts because the
> > > modifications are distributed randomly over that file.
> > >
> > > I'd suggest for you to send it together with the clean up patches, so
> > > that the context is clear.
>
> Ok. Will do it that way.
>
> > actually lately for most cases it seems the plan isn't to add to
> > pci_ids.h, but just do the PCI ID directly. It's not like a #define
> > adds any kind of information.
> > Vendor ID's otoh still happen (but obviously AMD is there since
> > forever.)
>
> Since the usage of these CPU device ids is spread over the whole
> kernel it makes sence to define it at a single point in
> pci_ids.h. This differs to device drivers that use only a single file
> with all the code, and thus, global defines are not necessary.

That's the rule I've been following too.  If a ID is just used in one place, 
like a driver, just keep the ID definition there (if you define it at all).  
But if it's used in multiple places around the tree, add a #define to 
pci_ids.h.

Thanks,
Jesse

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/