On Wed, Feb 15, 2023 at 11:29:00AM +0530, Basavaraj Natikar wrote:
>
> On 2/15/2023 11:19 AM, Mika Westerberg wrote:
> > On Tue, Feb 14, 2023 at 02:01:52PM -0600, Sanjay R Mehta wrote:
> >> void tb_check_quirks(struct tb_switch *sw)
> >> {
> >> + struct tb_switch *root_sw = sw->tb->root_switch;
> >> int i;
> >>
> >> for (i = 0; i < ARRAY_SIZE(tb_quirks); i++) {
> >> const struct tb_quirk *q = &tb_quirks[i];
> >>
> >> - if (q->hw_vendor_id && q->hw_vendor_id != sw->config.vendor_id)
> >> + if (q->hw_vendor_id && (q->hw_vendor_id != sw->config.vendor_id &&
> >> + q->hw_vendor_id != root_sw->config.vendor_id))
> > Again, why you need to change this?
> >
> > If you have the AMD host router device ID in the list the quirk applies
> > and that makes the driver skip enabling CL states for that link. It
> > should not matter if we enable CL states in the deeper parts of the
> > topology (even if we actually do not at the moment) because that's
> > completely different link, right.
>
> Thank you Mike for point it out then [PATCH v4] <https://lore.kernel.org/all/1676402030-4653-1-git-send-email-Sanju.Mehta@xxxxxxx/#r> patch changes solves the purpose.
>
> [PATCH v4] thunderbolt: Add quirk to disable CLx <https://lore.kernel.org/all/1676402030-4653-1-git-send-email-Sanju.Mehta@xxxxxxx/#r>
> https://lore.kernel.org/all/1676402030-4653-1-git-send-email-Sanju.Mehta@xxxxxxx/
>
> Please review the changes
Indeed v4 looks good. I just skipped it because there appeared several
versions of the patch in my inbox overnight so picked the last one for
review ;-)
