Hi, all.
2008/9/23 Stefan Roese sr@denx.de:
On Monday 22 September 2008, Andrew Dyer wrote:
I did a simple check.
Only powerpc and i386 and sh seem to use PCI. If powerpc does not have a problem, would you commit it?
Do I have your "go" (or at least no veto) to apply this patch?
(forgot to reply all on this, already sent to the O.P.)
I think this is going to break some boards/archs. From the top of the u-boot tree, if you do
find . -type f -exec grep pci_register_hose {} /dev/null ;
it shows all the source files that are registering a PCI bus. A quick look shows there are MIPS, ARM, x86 in that list at least, that aren't addressed by this change.
That's correct. This patch moves the original defines to the PPC specific header and introduces new defines in the SH specific header. All other ARCH's will not compile anymore when they try to include PCI support.
I understand that some macro of the PCI does not depend on only PPC as Andrew points it out. I am thinking other countermeasure.
So, NAK from me. An acceptable patch would be to move the original defines at least to all ARCH's already using PCI.
BTW: I fail to see what's really PPC specific about the pci code in question. Nobuhiro, could you please elaborate what's the problem here?
I wanted to remove CPU and baords of specific code from from PCI network driver. CONFIG_E500, DB64360 and DB64460 CPU is PPC. I thought that I could collect these with header files of PPC. Therefore I sent that patch in RFC.
Best regards, Nobuhiro