Received-SPF: pass (google.com: domain of linux-wireless-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Sender: Larry Finger <larry.finger@gmail.com>
Message-ID: <7106ab11-197d-e0d1-f54a-f56d438a022d@lwfinger.net>
Date:   Tue, 16 May 2023 13:31:48 -0500
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.11.0
Subject: Re: Driver for rtw8723ds
Content-Language: en-US
To:     Martin Blumenstingl <martin.blumenstingl@googlemail.com>,
        =?UTF-8?Q?Jernej_=c5=a0krabec?= <jernej.skrabec@gmail.com>
Cc:     linux-wireless <linux-wireless@vger.kernel.org>
References: <72a8eeb1-c91c-80a7-5a09-1b7963e0996b@lwfinger.net>
 <3162376.5fSG56mABF@jernej-laptop>
 <cee4e4f8-78e9-1bf7-c1a9-8ae5c662c785@lwfinger.net>
 <1855371.CQOukoFCf9@jernej-laptop>
 <CAFBinCCjzLH7ZZJ9Siw1Lp49qneO+qRJp-g8rGwAYuRwNj2Oyw@mail.gmail.com>
 <CAFBinCAD-7Mi06z-RGVREDQ3bYsDUOyDkBgoNW5mDLuKM8H7sg@mail.gmail.com>
From:   Larry Finger <Larry.Finger@lwfinger.net>
In-Reply-To: <CAFBinCAD-7Mi06z-RGVREDQ3bYsDUOyDkBgoNW5mDLuKM8H7sg@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk

On 5/16/23 05:22, Martin Blumenstingl wrote:
> On Mon, May 15, 2023 at 11:11 PM Martin Blumenstingl
> <martin.blumenstingl@googlemail.com> wrote:
>>
>> Hi Jernej,
>>
>> On Mon, May 15, 2023 at 10:37 PM Jernej Škrabec
>> <jernej.skrabec@gmail.com> wrote:
>> [...]
>>>> With aggregation disabled, we still get "Invalid RX packet size!" messages. I am
>>>> changing the statement to log (curr_pkt_len + pkt_desc_sz) > rx_len. I will let
>>>> you know when the OP responds.
>>>
>>> Yeah, I saw. I just find another possible reason, which fits nicely in current
>>> situation. Vendor driver parses drv_info_sz and shift fields only if packet
>>> is normal, e.g. not c2h type. However, rtw88 always parses those fields. It's
>>> possible that they have some value which should be ignored on 8723ds. I
>>> appended another patch to test.
>> I tried that patch and it didn't work for me (I can't get the card to
>> assoc to my AP with that patch).
>> Additionally I tried a simplified version (attached) and it didn't work.
>>
>> I'm out of time for today though so I cannot continue testing.
> I took time during my lunch break for some more experiments and came
> up with the attached patch (the vendor driver does something similar
> and I only noticed that after I observed some rtw_rx_pkt_stat with
> pkt_len being zero).
> It survived 30 minutes of uptime, updating my system and several
> iperf3 runs (in both directions).
> iperf results:
> - RX: 48 Mbit/s
> - TX: 33 Mbit/s
> 
> And to be clear, those results are with:
> - the word IO bugfix
> - the initial two patches from this series
> - Larry's addition of the second RTL8723DS SDIO ID
> - the attached patch

Martin,

Please send me a copy of the version of sdio.c that works. It seems likely that 
I got mine all messed up as the OP at GitHub is getting lots of warnings from 
net/mac80211/rx.c:803.

I though I followed all the patches you and Jernej sent, but I must have messed 
something up.

Thanks,

Larry