Received-SPF: pass (google.com: domain of linux-wireless-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH V2 3/6] brcmfmac: reserve 2 credits for host tx control
 path
To:     Wright Feng <wright.feng@cypress.com>,
        Chi-Hsien Lin <chi-hsien.lin@cypress.com>,
        Kalle Valo <kvalo@codeaurora.org>
Cc:     linux-wireless@vger.kernel.org, brcm80211-dev-list@broadcom.com,
        brcm80211-dev-list@cypress.com,
        Arend van Spriel <arend.vanspriel@broadcom.com>,
        Franky Lin <franky.lin@broadcom.com>,
        Hante Meuleman <hante.meuleman@broadcom.com>,
        Amar Shankar <amsr@cypress.com>,
        Jia-Shyr Chuang <joseph.chuang@cypress.com>
References: <20200610152106.175257-1-chi-hsien.lin@cypress.com>
 <20200610152106.175257-4-chi-hsien.lin@cypress.com>
 <f0910f96-1d23-daf1-b517-363e59bff105@gmail.com>
 <875z9yr7lg.fsf@codeaurora.org>
 <ffcf55cc-b27d-78ec-ff4f-e7efa3087712@gmail.com>
 <87a6zapbqp.fsf@codeaurora.org>
 <0923f45f-03e3-a870-3d2a-35f982482e1a@cypress.com>
 <bed18d45-6119-53da-2b9c-d98d121c8298@gmail.com>
 <a7aaf6ac-d6df-ab57-4939-474b542f007c@cypress.com>
From:   Dmitry Osipenko <digetx@gmail.com>
Message-ID: <112956e1-df38-0a35-c2f0-8d9e1568d3c5@gmail.com>
Date:   Tue, 11 Aug 2020 17:54:00 +0300
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.10.0
MIME-Version: 1.0
In-Reply-To: <a7aaf6ac-d6df-ab57-4939-474b542f007c@cypress.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 8bit
Sender: linux-wireless-owner@vger.kernel.org
Precedence: bulk

11.08.2020 11:35, Wright Feng пишет:
..
> Hi Dmitry,
> We have a fix for this issue. But we got the different test result
> numbers from yours, so would you help us verify the same with
> following patch in your setup?
> With your confirmation, I can make sure we see the same issue.
> 
> ---
>  drivers/net/wireless/broadcom/brcm80211/brcmfmac/sdio.c | 9 +++++++--
>  1 file changed, 7 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/wireless/broadcom/brcm80211/brcmfmac/sdio.c
> b/drivers/net/wireless/broadcom/brcm80211/brcmfmac/sdio.c
> index e8712ad3ac45..ea07bb1bfe27 100644
> --- a/drivers/net/wireless/broadcom/brcm80211/brcmfmac/sdio.c
> +++ b/drivers/net/wireless/broadcom/brcm80211/brcmfmac/sdio.c
> @@ -664,8 +664,13 @@ static void pkt_align(struct sk_buff *p, int len,
> int align)
>  /* To check if there's window offered */
>  static bool data_ok(struct brcmf_sdio *bus)
>  {
> -    /* Reserve TXCTL_CREDITS credits for txctl */
> -    return (bus->tx_max - bus->tx_seq) > TXCTL_CREDITS &&
> +    u8 tx_rsv = 0;
> +
> +    /* Reserve TXCTL_CREDITS credits for txctl when it is ready to send */
> +    if (bus->ctrl_frame_stat)
> +        tx_rsv = TXCTL_CREDITS;
> +
> +    return (bus->tx_max - bus->tx_seq) > tx_rsv &&
>             ((bus->tx_max - bus->tx_seq) & 0x80) == 0;
>  }

Hello, Wright!

I tried this patch and it doesn't fix the problem.

Could you please tell what firmware you're using?

I'm using stock firmware that comes from the linux-firmware package and
this NVRAM [1].

[1]
https://github.com/grate-driver/linux-firmware/blob/master/brcm/brcmfmac4329-sdio.acer%2Cpicasso.txt

 brcmfmac: brcmf_fw_alloc_request: using brcm/brcmfmac4329-sdio for chip
BCM4329/3
 brcmfmac: brcmf_c_process_clm_blob: no clm_blob available (err=-2),
device may have limited channels available
 brcmfmac: brcmf_c_preinit_dcmds: Firmware: BCM4329/3 wl0: Sep  2 2011
14:48:19 version 4.220.48

It also interesting that you're getting 40Mbit/s. I never managed to get
that speed on 4329 using upstream brcmfmac driver, but was getting it
using downstream BCMDHD driver. At some point I even tried to figure out
what's the difference which makes the upstream driver to have
significantly lower throughput than downstream and had older BCMDHD
ported to a recent upstream kernel, unfortunately I couldn't spot
anything obvious. Having BCMDHD and brcmfmac back-to-back on the same
hardware, the same kernel and using same firmware (IIRC), the BCMDHD was
more than x2 faster. I also remember that I had to enforce HT mode on AP
in order to get speeds over 15Mbit/s using BCMDHD, which made me to
conclude that upstream brcmfmac driver just doesn't support that HT mode
for 4329, but now I'm seeing yours iperf results and it makes me wonder..