Hi Geert-san,
> From: Geert Uytterhoeven, Sent: Tuesday, July 14, 2020 5:42 PM
>
> Hi Prabhakar,
>
> On Tue, Jul 14, 2020 at 10:30 AM Lad, Prabhakar
> <prabhakar.csengg@xxxxxxxxx> wrote:
> > On Tue, Jul 14, 2020 at 9:09 AM Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> wrote:
> > > On Mon, Jul 13, 2020 at 11:35 PM Lad Prabhakar
> > > <prabhakar.mahadev-lad.rj@xxxxxxxxxxxxxx> wrote:
> > > > From: Marian-Cristian Rotariu <marian-cristian.rotariu.rb@xxxxxxxxxxxxxx>
> > > >
> > > > Add support for RZ/G2H (R8A774E1) SoC IPMMUs.
> > > >
> > > > Signed-off-by: Marian-Cristian Rotariu <marian-cristian.rotariu.rb@xxxxxxxxxxxxxx>
> > > > Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@xxxxxxxxxxxxxx>
> > >
> > > Thanks for your patch!
> > >
> > > > --- a/drivers/iommu/ipmmu-vmsa.c
> > > > +++ b/drivers/iommu/ipmmu-vmsa.c
> > > > @@ -751,6 +751,7 @@ static const struct soc_device_attribute soc_rcar_gen3[] = {
> > > > static const struct soc_device_attribute soc_rcar_gen3_whitelist[] = {
> > > > { .soc_id = "r8a774b1", },
> > > > { .soc_id = "r8a774c0", },
> > > > + { .soc_id = "r8a774e1", },
> > >
> > > Adding an entry to soc_rcar_gen3_whitelist[] doesn't do anything, unless
> > > you also add the same entry to soc_rcar_gen3[].
> > >
> > I think the comment "For R-Car Gen3 use a white list to opt-in slave
> > devices." is misleading. Booting through the kernel I do see iommu
> > groups (attached is the logs).
>
> Indeed. Without an entry in soc_rcar_gen3[], the IPMMU is enabled
> unconditionally, and soc_rcar_gen3_whitelist[] is ignored.
> That's why you want an entry in both, unless you have an R-Car Gen3
> SoC where the IPMMU works correctly with all slave devices present.
> Perhaps soc_rcar_gen3[] should be renamed to soc_rcar_gen3_greylist[]
> (or soc_rcar_gen3_maybelist[]) to make this clear?
I think so (we should rename it).
> > Also the recent patch to add
> > "r8a77961" just adds to soc_rcar_gen3_whitelist.
>
> Oops, commit 17fe16181639801b ("iommu/renesas: Add support for r8a77961")
> did it wrong, too.
Thank you for the point it out. We should add r8a77961 to the soc_rcar_gen3[].
However, I don't know why I could not realize this issue...
So, I investigated this a little and then, IIUC, glob_match() which
soc_device_match() uses seems to return true, if *pat = "r8a7796" and *str = "r8a77961".
Best regards,
Yoshihiro Shimoda
