On Fri, Jul 22, 2022 at 09:53:07AM +0200, Arnd Bergmann wrote: > On Thu, Jul 21, 2022 at 11:58 PM Stafford Horne <shorne@xxxxxxxxx> wrote: > > On Thu, Jul 21, 2022 at 12:37:33PM -0500, Bjorn Helgaas wrote: > > > > > Thanks, you are right, I think some of the earlier functions may have needed it, > > which is why I had it earlier. But now that we have removed those we should be > > able to remove this. > > > > That said, I think some of the architecture includes could also be removed. On > > OpenRISC we are able to get away with only having the global asm-generic/pci.h > > so we don't need a wrapper pci.h header at all. > > > > However, I don't have everything setup to build all of those architectures so I > > was being a bit conservative to remove headers. I'll see what I can do in the > > next version. > > No need to worry about linux/types.h, this is pretty much included everywhere > already and neither the risk of extraneous includes or missing ones is > something to worry about. I'd just remove it if you do another respin, or > we can leave it in if Bjorn wants to just apply the v5 version. I will respin a v6 as we didn't get a reply on this. Bjorn are you planning to apply the series before the upcoming merge window? Originally I was thinking to merge these with my OpenRISC PCI support patches as OpenRISC depends on this. However, the series is getting a bit more involved for OpenRISC only. Once you get it merged I will like to rebase my OpenRISC PCI support patches on you branch. Hopefully, we can make it by the merge window. Also, Palmer has a conflict with the series and his RISC-V for-next branch. He would like to have a tag or branch to be able to merge into is branch to resolve the conflict. -Stafford
