On Thu, Sep 10, 2015 at 6:26 AM, Gabriele Paoloni <gabriele.paoloni@xxxxxxxxxx> wrote: >> From: Bjorn Helgaas [mailto:bhelgaas@xxxxxxxxxx] >> I think there was a separate patch for the pcie-spear13xx.c bug, >> wasn't there? At least, I don't see that fix in *this* patch. >> > > From your previous comments: > ******** > *2) Pratyush, this one's for you :) spear13xx_pcie_establish_link() > * calls dw_pcie_cfg_read() and dw_pcie_cfg_write() several times like > * this: > * > * dw_pcie_cfg_read(pp->dbi_base, exp_cap_off + PCI_EXP_DEVCTL, 4, ...) > * dw_pcie_cfg_write(pp->dbi_base, exp_cap_off + PCI_EXP_DEVCTL, 4, ...) > ******** > > I think that the implementation of dw_pcie_cfg_read() and > dw_pcie_cfg_write() that I proposed in this patch is compatible with > the previous calls in pcie-spear13xx.c. So the calls in pcie-spear13xx.c > were buggy for the previous implementations of the functions; with > the reworked functions there is no need to modify the function calls > in pcie-spear13xx.c...do you agree? Oh, I see, I didn't read it closely enough, but you're probably right. What I would probably do is apply Pratyush's fix first, then apply your patch on top. That way we can do the spear13xx bug fix by itself (which should be a fairly obviously-correct patch), then your interface change (which would be extended to touch all the calls including spear13xx and be similarly obviously correct). I think that would make it a bit easier for future changelog archaeologists. Bjorn -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
