Re: [PATCH v2 1/1] ath10k: add option for chip-id based BDF selection

[Doug Anderson <dianders@chromium.org>]

Hi,

On Tue, Nov 24, 2020 at 1:19 AM Rakesh Pillai <pillair@codeaurora.org> wrote:
>
> > -----Original Message-----
> > From: Doug Anderson <dianders@chromium.org>
> > Sent: Tuesday, November 24, 2020 6:27 AM
> > To: Abhishek Kumar <kuabhs@chromium.org>
> > Cc: Kalle Valo <kvalo@codeaurora.org>; Rakesh Pillai
> > <pillair@codeaurora.org>; LKML <linux-kernel@vger.kernel.org>; ath10k
> > <ath10k@lists.infradead.org>; Brian Norris <briannorris@chromium.org>;
> > linux-wireless <linux-wireless@vger.kernel.org>; David S. Miller
> > <davem@davemloft.net>; Jakub Kicinski <kuba@kernel.org>; netdev
> > <netdev@vger.kernel.org>
> > Subject: Re: [PATCH v2 1/1] ath10k: add option for chip-id based BDF
> > selection
> >
> > Hi,
> >
> > On Thu, Nov 12, 2020 at 12:09 PM Abhishek Kumar <kuabhs@chromium.org>
> > wrote:
> > >
> > > In some devices difference in chip-id should be enough to pick
> > > the right BDF. Add another support for chip-id based BDF selection.
> > > With this new option, ath10k supports 2 fallback options.
> > >
> > > The board name with chip-id as option looks as follows
> > > board name 'bus=snoc,qmi-board-id=ff,qmi-chip-id=320'
> > >
> > > Tested-on: WCN3990 hw1.0 SNOC WLAN.HL.3.2.2-00696-QCAHLSWMTPL-1
> > > Tested-on: QCA6174 HW3.2 WLAN.RM.4.4.1-00157-QCARMSWPZ-1
> > > Signed-off-by: Abhishek Kumar <kuabhs@chromium.org>
> > > ---
> > >
> > > (no changes since v1)
> >
> > I think you need to work on the method you're using to generate your
> > patches.  There are most definitely changes since v1.  You described
> > them in your cover letter (which you don't really need for a singleton
> > patch) instead of here.
> >
> >
> > > @@ -1438,12 +1439,17 @@ static int
> > ath10k_core_create_board_name(struct ath10k *ar, char *name,
> > >         }
> > >
> > >         if (ar->id.qmi_ids_valid) {
> > > -               if (with_variant && ar->id.bdf_ext[0] != '\0')
> > > +               if (with_additional_params && ar->id.bdf_ext[0] != '\0')
> > >                         scnprintf(name, name_len,
> > >                                   "bus=%s,qmi-board-id=%x,qmi-chip-id=%x%s",
> > >                                   ath10k_bus_str(ar->hif.bus),
> > >                                   ar->id.qmi_board_id, ar->id.qmi_chip_id,
> > >                                   variant);
> > > +               else if (with_additional_params)
> > > +                       scnprintf(name, name_len,
> > > +                                 "bus=%s,qmi-board-id=%x,qmi-chip-id=%x",
> > > +                                 ath10k_bus_str(ar->hif.bus),
> > > +                                 ar->id.qmi_board_id, ar->id.qmi_chip_id);
> >
> > I believe this is exactly opposite of what Rakesh was requesting.
> > Specifically, he was trying to eliminate the extra scnprintf() but I
> > think he still agreed that it was a good idea to generate 3 different
> > strings.  I believe the proper diff to apply to v1 is:
> >
> > https://crrev.com/c/255643

Wow, I seem to have deleted the last digit from my URL.  Should have been:

https://crrev.com/c/2556437

> >
> > -Doug
>
> Hi Abhishek/Doug,
>
> I missed on reviewing this change. Also I agree with Doug that this is not the change I was looking for.
>
> The argument "with_variant" can be renamed to "with_extra_params". There is no need for any new argument to this function.
> Case 1: with_extra_params=0,  ar->id.bdf_ext[0] = 0             ->   The default name will be used (bus=snoc,qmi_board_id=0xab)
> Case 2: with_extra_params=1,  ar->id.bdf_ext[0] = 0             ->   bus=snoc,qmi_board_id=0xab,qmi_chip_id=0xcd
> Case 3: with_extra_params=1,  ar->id.bdf_ext[0] = "xyz"      ->   bus=snoc,qmi_board_id=0xab,qmi_chip_id=0xcd,variant=xyz
>
> ar->id.bdf_ext[0] depends on the DT entry for variant field.

I'm confused about your suggestion.  Maybe you can help clarify.  Are
you suggesting:

a) Only two calls to ath10k_core_create_board_name()

I'm pretty sure this will fail in some cases.  Specifically consider
the case where the device tree has a "variant" defined but the BRD
file only has one entry for (board-id) and one for (board-id +
chip-id) but no entry for (board-id + chip-id + variant).  If you are
only making two calls then I don't think you'll pick the right one.

Said another way...

If the device tree has a variant:
1. We should prefer a BRD entry that has board-id + chip-id + variant
2. If #1 isn't there, we should prefer a BRD entry that has board-id + chip-id
3. If #1 and #2 aren't there we fall back to a BRD entry that has board-id.

...without 3 calls to ath10k_core_create_board_name() we can't handle
all 3 cases.


b) Three calls to ath10k_core_create_board_name() but the caller
manually whacks "ar->id.bdf_ext[0]" for one of the calls

This doesn't look like it's a clean solution, but maybe I'm missing something.


-Doug