MIME-Version: 1.0
References: <3B9D4DB2-D2CD-44FE-817A-F6EA8A0AD734@gmail.com>
 <CAFBinCBWyiObgbyuT2Xc=FnoTTTUBu3sePpybdrocVZr7SVDWw@mail.gmail.com>
 <7df7bdcad3474791ae69f73f8a761173@realtek.com> <9E1F8017-3F54-416D-9BD3-DF2F1C625094@gmail.com>
 <e4404fcb-69f6-b4c0-7683-69101bbeaa96@lexina.in>
In-Reply-To: <e4404fcb-69f6-b4c0-7683-69101bbeaa96@lexina.in>
From:   Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Date:   Thu, 23 Feb 2023 22:00:08 +0100
Message-ID: <CAFBinCDFDv3RJQnydU8_ryFwzT2bevQsciGTsvotjsft4fVmXg@mail.gmail.com>
Subject: Re: Bug/Problem with rtl_bt/rtl8822cs_config.bin
To:     Vyacheslav <adeep@lexina.in>
Cc:     Christian Hewitt <christianshewitt@gmail.com>,
        Max Chou <max.chou@realtek.com>,
        Hilda Wu <hildawu@realtek.com>,
        Josh Boyer <jwboyer@kernel.org>,
        Linux Firmware <linux-firmware@kernel.org>,
        Bluez mailing list <linux-bluetooth@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Hello Vyacheslav,

On Thu, Feb 23, 2023 at 3:14 PM Vyacheslav <adeep@lexina.in> wrote:
[...]
> > The bug report stimulated Martin=E2=80=99s maintainer memory over unfin=
ished improvements
> > to the Amlogic UART used in the g12a and newer SoC families (we are tes=
ting with
> > g12b and sm1 devices) resulting in this series being submitted last nig=
ht:
> >
> > https://patchwork.kernel.org/project/linux-amlogic/list/?series=3D72417=
2
>
> However, this does not fix the problem with rtl8822cs BT on gxl/axg chips=
ets
My understanding is that Meson GXBB and newer can also use XTAL along
with the UART controller's internal divider - without any additional
pre-divider like "XTAL divided by 2" or "XTAL divided by 3".
Using XTAL directly (24MHz) also makes it possible to achieve 1500000
baud without any jitter.

The problem with the Meson UART driver however is that using XTAL
directly is not implemented.
Yu Tu from Amlogic worked on patches last year. v6 of his series made
it into -next briefly, but got reverted due to issues. v7 then
stalled: [0] and only the bare minimum (enabling "XTAL divided by 2")
for the S4 SoC made it upstream in the end.

My suggestion is:
- let's get my small patchset upstreamed which at least fixes the
1500000 baud rate on G12A (and newer) SoCs
- then work on resurrecting Yu Tu's patchset and implement "XTAL
without any pre-dividers" support on top of that. Common Clock
Framework will then choose the correct parent depending on the SoC
generation, hopefully allowing arbitrary baud rates in the end - with
no jitter/as little jitter as possible.

What do you think?


Best regards,
Martin


[0] https://lore.kernel.org/all/20220225073922.3947-1-yu.tu@amlogic.com/