mirror of https://lore.kernel.org/linux-amlogic/
 help / color / mirror / Atom feed
From: Christian Hewitt <christianshewitt@gmail.com>
To: Marc Gonzalez <marc.w.gonzalez@free.fr>
Cc: Martin Blumenstingl <martin.blumenstingl@googlemail.com>,
	Linux Wireless <linux-wireless@vger.kernel.org>,
	MMC <linux-mmc@vger.kernel.org>,
	AML <linux-amlogic@lists.infradead.org>,
	Neil Armstrong <neil.armstrong@linaro.org>,
	Ulf Hansson <ulf.hansson@linaro.org>,
	Pierre-Hugues Husson <phh@phh.me>
Subject: Re: [Performance regression] BCM4359/9 on S905X2
Date: Thu, 6 Apr 2023 21:07:26 +0400	[thread overview]
Message-ID: <F47757EE-02FA-4A0E-B3EF-54B59E7CD51C@gmail.com> (raw)
In-Reply-To: <74feca6f-4177-5645-6614-bc5f3dbf0be1@free.fr>

> On 6 Apr 2023, at 8:35 pm, Marc Gonzalez <marc.w.gonzalez@free.fr> wrote:
> 
> On 04/04/2023 23:06, Martin Blumenstingl wrote:
> 
>> Hi Marc,
> 
> Hello Martin :)
> 
>> On Tue, Apr 4, 2023 at 6:09 PM Marc Gonzalez wrote:
>> 
>>>                bus-width = <4>;
>>>                cap-sd-highspeed;
>>>                cap-mmc-highspeed;
>>>                max-frequency = <100000000>;
>> 
>> I would start by comparing the bus mode. You can get it from
>> /sys/kernel/debug/mmc2/ios
>> On the vendor kernel it should be in /sys/kernel/debug/sdio/ios (or
>> similar, I'm typing this from the top of my head).
>> It will give you insights on the clock and timing that has been
>> negotiated between the host and card.
>> 
>> From this information you can get the maximum bus speed, e.g. from [0]
>> Please note that any card will add overhead for communication, so bus
>> speed will not be equal to wifi throughput.
> 
> For the vendor kernel:
> 
> /sys/kernel/debug/sdio/sdio:0001/state:0x00000001
> /sys/kernel/debug/sdio/clock:200000000
> /sys/kernel/debug/sdio/ios:clock:               200000000 Hz
> /sys/kernel/debug/sdio/ios:actual clock:        199999997 Hz
> /sys/kernel/debug/sdio/ios:vdd:         21 (3.3 ~ 3.4 V)
> /sys/kernel/debug/sdio/ios:bus mode:    2 (push-pull)
> /sys/kernel/debug/sdio/ios:chip select: 0 (don't care)
> /sys/kernel/debug/sdio/ios:power mode:  1 (up)
> /sys/kernel/debug/sdio/ios:bus width:   2 (4 bits)
> /sys/kernel/debug/sdio/ios:timing spec: 6 (sd uhs SDR104)
> /sys/kernel/debug/sdio/ios:signal voltage:      1 (1.80 V)
> /sys/kernel/debug/sdio/ios:driver type: 0 (driver type B)
> 
> 
> For mainline:
> 
> /sys/kernel/debug/mmc2/mmc2:0001/state:0x00000001
> /sys/kernel/debug/mmc2/err_stats:# Command Timeout Occurred:     0
> /sys/kernel/debug/mmc2/err_stats:# Command CRC Errors Occurred:  0
> /sys/kernel/debug/mmc2/err_stats:# Data Timeout Occurred:        0
> /sys/kernel/debug/mmc2/err_stats:# Data CRC Errors Occurred:     0
> /sys/kernel/debug/mmc2/err_stats:# Auto-Cmd Error Occurred:      0
> /sys/kernel/debug/mmc2/err_stats:# ADMA Error Occurred:  0
> /sys/kernel/debug/mmc2/err_stats:# Tuning Error Occurred:        0
> /sys/kernel/debug/mmc2/err_stats:# CMDQ RED Errors:      0
> /sys/kernel/debug/mmc2/err_stats:# CMDQ GCE Errors:      0
> /sys/kernel/debug/mmc2/err_stats:# CMDQ ICCE Errors:     0
> /sys/kernel/debug/mmc2/err_stats:# Request Timedout:     0
> /sys/kernel/debug/mmc2/err_stats:# CMDQ Request Timedout:        0
> /sys/kernel/debug/mmc2/err_stats:# ICE Config Errors:    0
> /sys/kernel/debug/mmc2/err_stats:# Controller Timedout errors:   0
> /sys/kernel/debug/mmc2/err_stats:# Unexpected IRQ errors:        0
> /sys/kernel/debug/mmc2/err_state:0
> /sys/kernel/debug/mmc2/clock:100000000
> /sys/kernel/debug/mmc2/caps2:0x00040000
> /sys/kernel/debug/mmc2/caps:0x40040105
> /sys/kernel/debug/mmc2/ios:clock:               100000000 Hz
> /sys/kernel/debug/mmc2/ios:actual clock:        99999999 Hz
> /sys/kernel/debug/mmc2/ios:vdd:         21 (3.3 ~ 3.4 V)
> /sys/kernel/debug/mmc2/ios:bus mode:    2 (push-pull)
> /sys/kernel/debug/mmc2/ios:chip select: 0 (don't care)
> /sys/kernel/debug/mmc2/ios:power mode:  2 (on)
> /sys/kernel/debug/mmc2/ios:bus width:   2 (4 bits)
> /sys/kernel/debug/mmc2/ios:timing spec: 5 (sd uhs SDR50)
> /sys/kernel/debug/mmc2/ios:signal voltage:      1 (1.80 V)
> /sys/kernel/debug/mmc2/ios:driver type: 0 (driver type B)
> 
> 
> The clock is indeed running twice as fast on the vendor system.
> And in SDR104 vs SDR50 mode.
> 
> Adjusting mainline device tree...
> 
> 
> /sys/kernel/debug/mmc2/mmc2:0001/state:0x00000001
> /sys/kernel/debug/mmc2/err_stats:# Command Timeout Occurred:     0
> /sys/kernel/debug/mmc2/err_stats:# Command CRC Errors Occurred:  0
> /sys/kernel/debug/mmc2/err_stats:# Data Timeout Occurred:        0
> /sys/kernel/debug/mmc2/err_stats:# Data CRC Errors Occurred:     0
> /sys/kernel/debug/mmc2/err_stats:# Auto-Cmd Error Occurred:      0
> /sys/kernel/debug/mmc2/err_stats:# ADMA Error Occurred:  0
> /sys/kernel/debug/mmc2/err_stats:# Tuning Error Occurred:        0
> /sys/kernel/debug/mmc2/err_stats:# CMDQ RED Errors:      0
> /sys/kernel/debug/mmc2/err_stats:# CMDQ GCE Errors:      0
> /sys/kernel/debug/mmc2/err_stats:# CMDQ ICCE Errors:     0
> /sys/kernel/debug/mmc2/err_stats:# Request Timedout:     0
> /sys/kernel/debug/mmc2/err_stats:# CMDQ Request Timedout:        0
> /sys/kernel/debug/mmc2/err_stats:# ICE Config Errors:    0
> /sys/kernel/debug/mmc2/err_stats:# Controller Timedout errors:   0
> /sys/kernel/debug/mmc2/err_stats:# Unexpected IRQ errors:        0
> /sys/kernel/debug/mmc2/err_state:0
> /sys/kernel/debug/mmc2/clock:200000000
> /sys/kernel/debug/mmc2/caps2:0x00040000
> /sys/kernel/debug/mmc2/caps:0x40080105
> /sys/kernel/debug/mmc2/ios:clock:               200000000 Hz
> /sys/kernel/debug/mmc2/ios:actual clock:        199999997 Hz
> /sys/kernel/debug/mmc2/ios:vdd:         21 (3.3 ~ 3.4 V)
> /sys/kernel/debug/mmc2/ios:bus mode:    2 (push-pull)
> /sys/kernel/debug/mmc2/ios:chip select: 0 (don't care)
> /sys/kernel/debug/mmc2/ios:power mode:  2 (on)
> /sys/kernel/debug/mmc2/ios:bus width:   2 (4 bits)
> /sys/kernel/debug/mmc2/ios:timing spec: 6 (sd uhs SDR104)
> /sys/kernel/debug/mmc2/ios:signal voltage:      1 (1.80 V)
> /sys/kernel/debug/mmc2/ios:driver type: 0 (driver type B)
> 
> 
> # curl -o /dev/null http://192.168.1.254:8095/fixed/1G
>  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                 Dload  Upload   Total   Spent    Left  Speed
> 100 1024M  100 1024M    0     0  9747k      0  0:01:47  0:01:47 --:--:-- 9544k
> 
> With this new setting, WiFi throughput increases 20%
> (from 60 Mbps to 75 Mbps).
> 
> (Uggg, I've been using 10^9 for the amount transferred,
> but it looks to be 2^30 actually. Absolute speeds are
> actually 7% higher, but doesn't change the conclusion)
> 
> I also noticed that I reported 11 MB/s (88 Mbps) at the beginning
> of this thread. This would point to a performance regression caused
> by my defconfig & device tree changes :(
> 
> 
>>> vendor DTS has the following child node:
>>> 
>>>                sdio {
>>>                        pinname = "sdio";
>>>                        ocr_avail = <0x200080>; /**VDD voltage 3.3 ~ 3.4 */
>>>                        /* max_req_size = <0x20000>; */ /**128KB*/
>>>                        max_req_size = <0x400>;
>>>                        card_type = <3>;
>>>                        /* 3:sdio device(ie:sdio-wifi),
>>>                         * 4:SD combo (IO+mem) card
>>>                         */
>>>                        dmode = "pio";
>>>                };
>>> 
>>> Maybe the vendor kernel uses the above information to "boost"
>>> the performance of the SDIO-based WiFi adapter?
>> 
>> PIO is also what we support upstream with the
>> amlogic,dram-access-quirk; (which is enabled for &sd_emmc_a).
>> This suggests that the pinctrl trick that Neil mentioned is not used here.
>> 
>> I assume that the wifi driver on the vendor kernel is the brcmdhd (out
>> of tree) driver, while mainline uses brcmfmac.
> 
> Correct.
> hardware/wifi/broadcom/drivers/ap6xxx/bcmdhd.100.10.545.x
> vs
> drivers/net/wireless/broadcom/brcm80211/brcmfmac
> 
>> One idea that comes to my mind is to enable the
>> amlogic,dram-access-quirk; (and use pio mode on the vendor kernel) for
>> eMMC or SD card and then compare read/write speeds. If they are
>> similar-ish then the wifi performance difference is likely caused by
>> the wifi driver (or in the opposite case: if read/write speeds on
>> mainline with amlogic,dram-access-quirk; perform worse than the vendor
>> kernel with pio mode then it's likely that it's a meson-gx-mmc driver
>> limitation).
> 
> Thanks for the suggestion.
> 
> Are you aware whether someone has tried running the bcmdhd driver
> on mainline?

https://gitlab.baylibre.com/baylibre/amlogic/atv/linux/-/commits/narmstrong/v5.1/aml/integ-5.1-bcmdhd

IIRC not long after this ^ support for BCM4359 was added to brcmfmac so
using bcmdhd wasn’t pursued. Neil?

Christian

> Thanks for the test suggestion.
> 
>> From my work on the rtw88 SDIO wifi driver I can say that the
>> meson-gx-mmc driver can push (TX direction) at least 120Mbit/s.
>> I understand that this is half of what you get with the vendor kernel
>> - and that this is the wrong direction (you're testing RX while I'm
>> testing TX).
>> The point that I want to get across is: I think nobody has the one
>> answer why wifi performance is lower (personally I'm happy with having
>> it work at all, performance is second).
>> So it'll be a process to find the reason, and I think it requires
>> being creative due to the large amounts of different code (MMC driver,
>> wifi driver, additional patches...) between mainline and the vendor
>> kernel.
> 
> I really appreciate you sharing this insight.
> 
> Regards.
> 


_______________________________________________
linux-amlogic mailing list
linux-amlogic@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-amlogic

  reply	other threads:[~2023-04-06 17:07 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-03-21 10:40 Marc Gonzalez
2023-03-21 10:46 ` Jerome Brunet
2023-03-21 12:37   ` Performance lower than expected with " Marc Gonzalez
2023-03-21 13:54     ` Jerome Brunet
2023-03-21 16:48       ` Marc Gonzalez
     [not found]   ` <EEE9FD80-C106-4A9B-AA8D-5C151E540CFD@hewittfamily.org.uk>
2023-04-04 16:09     ` [Performance regression] " Marc Gonzalez
2023-04-04 21:06       ` Martin Blumenstingl
2023-04-06 16:35         ` Marc Gonzalez
2023-04-06 17:07           ` Christian Hewitt [this message]
2023-04-06 17:41             ` Neil Armstrong
2023-04-17 14:22           ` Marc Gonzalez
2023-03-21 11:45 ` Neil Armstrong

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=F47757EE-02FA-4A0E-B3EF-54B59E7CD51C@gmail.com \
    --to=christianshewitt@gmail.com \
    --cc=linux-amlogic@lists.infradead.org \
    --cc=linux-mmc@vger.kernel.org \
    --cc=linux-wireless@vger.kernel.org \
    --cc=marc.w.gonzalez@free.fr \
    --cc=martin.blumenstingl@googlemail.com \
    --cc=neil.armstrong@linaro.org \
    --cc=phh@phh.me \
    --cc=ulf.hansson@linaro.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®