On Tue, Jan 18, 2022 at 10:13:14AM -0600, Bjorn Helgaas wrote:
> On Mon, Jan 17, 2022 at 11:03:55PM +0100, marek.vasut@xxxxxxxxx wrote:
> > + instr &= ~0xf;
> > + if ((instr == 0xf57ff060 || instr == 0xf3bf8f60) &&
> > + (pc == (u32)&rcar_pci_read_reg_workaround_start + 4)) {
> > + /*
> > + * If the instruction being executed was a read,
> > + * make it look like it read all-ones.
> > + */
> > + instr = *(unsigned long *)(pc - 4);
> > + reg = (instr >> 12) & 15;
> > +
> > + if ((instr & 0x0c100000) == 0x04100000) {
> > + if (instr & 0x00400000)
> > + val = 255;
> > + else
> > + val = -1;
>
> Can you please use PCI_SET_ERROR_RESPONSE() or something similar here
> to make this greppable?
I should have mentioned that PCI_SET_ERROR_RESPONSE() was added in the
current merge window, so it will appear in v5.17-rc1.
Bjorn
