poor performance with W2CBW003 (Marvell 8686)

poor performance with W2CBW003 (Marvell 8686)

[Andreas Kemnade]

Hi,

on the GTA04 (www.gta04.org) 
which has a W2CBW003 chip for wifi connected to a DM3730CBP CPU. data
rates are quite low using the libertas driver from kernel 3.2-3.4 
(https://github.com/neilbrown/gta04/branches)
even near an accesspoint. That seems not depend on the encryption
settings. Rates are around 100KByte/s. iwconfig shows 5.5Mb/s if it is
not connected to an accesspoint with specially configured rates (but
still the throughput is low).

In contrary data rates are high (>1MByte/s) with a kernel based on
the 2.6.32  BeagleBoard XM hw-validation kernel
https://github.com/goldelico/gta04-kernel
That behaviour was reported by several people.

I tried also the libertas-tf driver with kernel 3.2 using the
instructions here:
http://corysohrakoff.wordpress.com/2011/09/13/enabling-wifi-ap-mode-on-a-gumstix-overo/
That was also slow.

Any hints?

Greetings
Andreas Kemnade

Re: poor performance with W2CBW003 (Marvell 8686)

[Dan Williams]

On Sat, 2012-06-09 at 15:03 +0200, Andreas Kemnade wrote:
> Hi,
> 
> on the GTA04 (www.gta04.org) 
> which has a W2CBW003 chip for wifi connected to a DM3730CBP CPU. data
> rates are quite low using the libertas driver from kernel 3.2-3.4 
> (https://github.com/neilbrown/gta04/branches)
> even near an accesspoint. That seems not depend on the encryption
> settings. Rates are around 100KByte/s. iwconfig shows 5.5Mb/s if it is
> not connected to an accesspoint with specially configured rates (but
> still the throughput is low).
> 
> In contrary data rates are high (>1MByte/s) with a kernel based on
> the 2.6.32  BeagleBoard XM hw-validation kernel
> https://github.com/goldelico/gta04-kernel
> That behaviour was reported by several people.
> 
> I tried also the libertas-tf driver with kernel 3.2 using the
> instructions here:
> http://corysohrakoff.wordpress.com/2011/09/13/enabling-wifi-ap-mode-on-a-gumstix-overo/
> That was also slow.
> 
> Any hints?

GSPI or SDIO?  What SD host controller?  Does the controller support
4-bit mode correctly?  What block sizes does it support?  Most often the
issue is with crappy  host controllers that need workarounds or quirks
to correctly support the higher speeds that the driver can do.  The
benchmark I always used was that on a laptop with a Ricoh SD controller,
I could easily pull megabits per second even 15 feet away.

(that's not to say there aren't inefficiencies in the driver, just that
we know that given a good SDHC and a good SDHC driver, that the libertas
driver itself is not the bottleneck).

Dan

Re: poor performance with W2CBW003 (Marvell 8686)

[Andreas Kemnade]

Hi,

On Tue, 2012-06-12 at 14:18 -0500, Dan Williams wrote:
> On Sat, 2012-06-09 at 15:03 +0200, Andreas Kemnade wrote:
> > Hi,
> > 
> > on the GTA04 (www.gta04.org) 
> > which has a W2CBW003 chip for wifi connected to a DM3730CBP CPU. data
> > rates are quite low using the libertas driver from kernel 3.2-3.4 
> > (https://github.com/neilbrown/gta04/branches)
> > even near an accesspoint. That seems not depend on the encryption
> > settings. Rates are around 100KByte/s. iwconfig shows 5.5Mb/s if it is
> > not connected to an accesspoint with specially configured rates (but
> > still the throughput is low).
> > 
> > In contrary data rates are high (>1MByte/s) with a kernel based on
> > the 2.6.32  BeagleBoard XM hw-validation kernel
> > https://github.com/goldelico/gta04-kernel
> > That behaviour was reported by several people.
> > 
> > I tried also the libertas-tf driver with kernel 3.2 using the
> > instructions here:
> > http://corysohrakoff.wordpress.com/2011/09/13/enabling-wifi-ap-mode-on-a-gumstix-overo/
> > That was also slow.
> > 
> > Any hints?
> 
> GSPI or SDIO?  What SD host controller?  Does the controller support
It is the host controller integrated into the SoC (the DM3730 from TI).
There is
a level shifter (the 74avca406) in between.

> 4-bit mode correctly?  What block sizes does it support?  Most often the
> issue is with crappy  host controllers that need workarounds or quirks
> to correctly support the higher speeds that the driver can do.  The
> benchmark I always used was that on a laptop with a Ricoh SD controller,
> I could easily pull megabits per second even 15 feet away.
>
4 bit is supported.
The hardware does not impose unresolvable limitations, the 2.6.32 kernel
gives high performance. But since that kernel is not mainstream, it is
very hard to find out which patch gives the slowdown. git bisect
is not very useful in that situation.
SD memory card reading speed is always ok (on another of the integrated
sd host controllers).

I'm citing from the 
AM/DM37x Multimedia Device
Silicon Revision 1.x
Version O
Technical Reference Manual from TI page 3396
from http://www.ti.com/product/dm3730
here:

>The known limitations are as follows:
>· No built-in hardware support for error correction codes (ECC). See
>the Multimedia Card System
>   Specification, v4.2, and the SD Memory Card Specifications, v2.0,
>for details about ECC.
>· The maximum block size defined in the SD Memory Card Specifications,
>v2.0, that the host driver can
>   read and write to the buffer in the host controller is 2048 bytes. 
> MMC supports a maximum block size
>   of 1024 bytes. Up to 512 byte transfers, the buffer in MMC is
>considered as a double buffering with
>   ping-pong management; half of the buffer can be written while the
>other part is read. For 512 to 1024
>   byte transfers, the entire buffer is dedicated to the transfer (read
>only or write only).


>The differences between the MMC/SD/SDIO host controllers and a
>standard SD host controller are
>defined by the SD Card Specification, Part A2, SD Host Controller
>Standard Specification, v1.00, as
>follows:
>· The MMC/SD/SDIO host controllers support MMC cards.
>· The MMC/SD/SDIO host controller is defined as a DMA slave device. A
>standard SD host controller is
>    defined as a DMA master controller that can start and stop a DMA
>transfer. MMC/SD/SDIO host
>    controllers support DMA transfers through slave DMA requests.
>· The clock divider in MMC/SD/SDIO host controller supports a wider
>range of frequency than specified
>    in the SD Memory Card Specifications, v2.0. The MMC/SD/SDIO host
>controller supports odd and
>    even clock ratio.
>· The MMC/SD/SDIO host controller supports configurable busy time-out.

Maybe you can name the limitation which might be problematic here.

Greetings
Andreas Kemnade

Re: poor performance with W2CBW003 (Marvell 8686)

[hong zhang]

> 4 bit is supported.
> The hardware does not impose unresolvable limitations, the
> 2.6.32 kernel
> gives high performance. But since that kernel is not
> mainstream, it is
> very hard to find out which patch gives the slowdown. git
> bisect
> is not very useful in that situation.
> SD memory card reading speed is always ok (on another of the
> integrated
> sd host controllers).

Have you tested receive throughput? Any difference between TX and Rx throughput? 

Re: poor performance with W2CBW003 (Marvell 8686)

[Andreas Kemnade]

[-- Attachment #1: Type: text/plain, Size: 687 bytes --]

Hi,

On Wed, 13 Jun 2012 10:06:54 -0700 (PDT)
hong zhang <henryzhang62@yahoo.com> wrote:

> > 4 bit is supported.
> > The hardware does not impose unresolvable limitations, the
> > 2.6.32 kernel
> > gives high performance. But since that kernel is not
> > mainstream, it is
> > very hard to find out which patch gives the slowdown. git
> > bisect
> > is not very useful in that situation.
> > SD memory card reading speed is always ok (on another of the
> > integrated
> > sd host controllers).
> 
> Have you tested receive throughput? Any difference between TX and Rx throughput? 
> 
I tested both.
No significant difference with 3.x.

Greetings
Andreas Kemnade

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: poor performance with W2CBW003 (Marvell 8686)

[hong zhang]

You have to identify problem caused by 
SDIO host controller or wifi driver.

If driver is working well in x86 with Roch SDIO, that means wifi driver is ok. 

You have to find out
1. SDIO interface if using Interrupt or PIO
2. 1 bit or 4 bit
3. TI BSP SDIO controller 

--- On Wed, 6/13/12, Andreas Kemnade <andreas@kemnade.info> wrote:

> From: Andreas Kemnade <andreas@kemnade.info>
> Subject: Re: poor performance with W2CBW003 (Marvell 8686)
> To: "hong zhang" <henryzhang62@yahoo.com>
> Cc: "Dan Williams" <dcbw@redhat.com>, linux-wireless@vger.kernel.org, dsd@laptop.org, libertas-dev@lists.infradead.org
> Date: Wednesday, June 13, 2012, 12:42 PM
> Hi,
> 
> On Wed, 13 Jun 2012 10:06:54 -0700 (PDT)
> hong zhang <henryzhang62@yahoo.com>
> wrote:
> 
> > > 4 bit is supported.
> > > The hardware does not impose unresolvable
> limitations, the
> > > 2.6.32 kernel
> > > gives high performance. But since that kernel is
> not
> > > mainstream, it is
> > > very hard to find out which patch gives the
> slowdown. git
> > > bisect
> > > is not very useful in that situation.
> > > SD memory card reading speed is always ok (on
> another of the
> > > integrated
> > > sd host controllers).
> > 
> > Have you tested receive throughput? Any difference
> between TX and Rx throughput? 
> > 
> I tested both.
> No significant difference with 3.x.
> 
> Greetings
> Andreas Kemnade
> 

Re: poor performance with W2CBW003 (Marvell 8686)

[Andreas Kemnade]

[-- Attachment #1: Type: text/plain, Size: 633 bytes --]

Hi,

On Wed, 13 Jun 2012 11:39:08 -0700 (PDT)
hong zhang <henryzhang62@yahoo.com> wrote:

> You have to identify problem caused by 
> SDIO host controller or wifi driver.
> 
> If driver is working well in x86 with Roch SDIO, that means wifi driver is ok. 
> 
> You have to find out
> 1. SDIO interface if using Interrupt or PIO

Hmm, very interesting point, of course, there are interrupts. But after reading
the commit logs in the patched 2.6.32 kernel I see there is a patch enabling
asynchronous *sdio* interrupts. That looks clearly suspicious. 

> 2. 1 bit or 4 bit

Hmm, 4bit.

Greetings
Andreas Kemnade

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: poor performance with W2CBW003 (Marvell 8686)

[hong zhang]

bit mode and interrupt or PIO are configured in SDIO hoster side.
Not in driver side.

--- On Thu, 6/14/12, Andreas Kemnade <andreas@kemnade.info> wrote:

> From: Andreas Kemnade <andreas@kemnade.info>
> Subject: Re: poor performance with W2CBW003 (Marvell 8686)
> To: "hong zhang" <henryzhang62@yahoo.com>
> Cc: "Dan Williams" <dcbw@redhat.com>, linux-wireless@vger.kernel.org, dsd@laptop.org, libertas-dev@lists.infradead.org
> Date: Thursday, June 14, 2012, 12:45 AM
> Hi,
> 
> On Wed, 13 Jun 2012 11:39:08 -0700 (PDT)
> hong zhang <henryzhang62@yahoo.com>
> wrote:
> 
> > You have to identify problem caused by 
> > SDIO host controller or wifi driver.
> > 
> > If driver is working well in x86 with Roch SDIO, that
> means wifi driver is ok. 
> > 
> > You have to find out
> > 1. SDIO interface if using Interrupt or PIO
> 
> Hmm, very interesting point, of course, there are
> interrupts. But after reading
> the commit logs in the patched 2.6.32 kernel I see there is
> a patch enabling
> asynchronous *sdio* interrupts. That looks clearly
> suspicious. 
> 
> > 2. 1 bit or 4 bit
> 
> Hmm, 4bit.
> 
> Greetings
> Andreas Kemnade
> 

Re: poor performance with W2CBW003 (Marvell 8686)

[hong zhang]

Actually, you have to check SDIO driver done by TI.
There are 5 SDIO pins need to check. 4 data pins and one clock pin.

Double check the SDIO pins muxs and clock rate using TI reference manual.

By performance issues.
1. bit modes 1 or 4
2. interrupt/PIO
3. SDIO clock rate

Command 52 and 53 are important.
See this link for SDIO reference.
http://www.sandisk.com/Assets/File/OEM/Manuals/SD_SDIO_specsv1.pdf

--- On Thu, 6/14/12, hong zhang <henryzhang62@yahoo.com> wrote:

> From: hong zhang <henryzhang62@yahoo.com>
> Subject: Re: poor performance with W2CBW003 (Marvell 8686)
> To: "Andreas Kemnade" <andreas@kemnade.info>
> Cc: "Dan Williams" <dcbw@redhat.com>, linux-wireless@vger.kernel.org, dsd@laptop.org, libertas-dev@lists.infradead.org
> Date: Thursday, June 14, 2012, 11:11 AM
> bit mode and interrupt or PIO are
> configured in SDIO hoster side.
> Not in driver side.
> 
> --- On Thu, 6/14/12, Andreas Kemnade <andreas@kemnade.info>
> wrote:
> 
> > From: Andreas Kemnade <andreas@kemnade.info>
> > Subject: Re: poor performance with W2CBW003 (Marvell
> 8686)
> > To: "hong zhang" <henryzhang62@yahoo.com>
> > Cc: "Dan Williams" <dcbw@redhat.com>,
> linux-wireless@vger.kernel.org,
> dsd@laptop.org, libertas-dev@lists.infradead.org
> > Date: Thursday, June 14, 2012, 12:45 AM
> > Hi,
> > 
> > On Wed, 13 Jun 2012 11:39:08 -0700 (PDT)
> > hong zhang <henryzhang62@yahoo.com>
> > wrote:
> > 
> > > You have to identify problem caused by 
> > > SDIO host controller or wifi driver.
> > > 
> > > If driver is working well in x86 with Roch SDIO,
> that
> > means wifi driver is ok. 
> > > 
> > > You have to find out
> > > 1. SDIO interface if using Interrupt or PIO
> > 
> > Hmm, very interesting point, of course, there are
> > interrupts. But after reading
> > the commit logs in the patched 2.6.32 kernel I see
> there is
> > a patch enabling
> > asynchronous *sdio* interrupts. That looks clearly
> > suspicious. 
> > 
> > > 2. 1 bit or 4 bit
> > 
> > Hmm, 4bit.
> > 
> > Greetings
> > Andreas Kemnade
> > 
> 
> _______________________________________________
> libertas-dev mailing list
> libertas-dev@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/libertas-dev
> 

Re: poor performance with W2CBW003 (Marvell 8686)

[Andreas Kemnade]

[-- Attachment #1: Type: text/plain, Size: 690 bytes --]

Hi,

On Thu, 14 Jun 2012 09:11:06 -0700 (PDT)
hong zhang <henryzhang62@yahoo.com> wrote:

> bit mode and interrupt or PIO are configured in SDIO hoster side.
> Not in driver side.
> 
yes, I wrote the last email just to say that I have found something to look at.
That is of course in the host driver. So the thing I did was reading
commit logs of the 2.6.32 kernel working well.

The only thing which puzzles me is the 5.5Mb/s data rate iwconfig shows.
After playing with the host side driver, it also shows higher rates. But
why does host driver affects *that* data rate of link qualitiy is good?

Thanks for all your ideas where to look at.

Greetings
Andreas Kemnade

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: poor performance with W2CBW003 (Marvell 8686)

[hong zhang]

To measure throughput, should not check iwconfig that is not accurate.

1. use wpa_supplicant to associate with AP
2. use iperf to measure throughput using UDP for Rx and Tx.

for SDIO host, command 52 doing register configuration. For rate, command 53 handles data rate. as sdio spec says clock rate is 0 to 100kHz means low speed mode. clock 0 to 25MHz means high speed mode.

Max throughput = clock-rate*num-of-data-line
for 4-bit, high clock,
max-throughput=25M*4=100Mbits/sec.
for 1-bit,low rate
max-throughput=100K*1=100Kbits/sec.



--- On Fri, 6/15/12, Andreas Kemnade <andreas@kemnade.info> wrote:

> From: Andreas Kemnade <andreas@kemnade.info>
> Subject: Re: poor performance with W2CBW003 (Marvell 8686)
> To: "hong zhang" <henryzhang62@yahoo.com>
> Cc: "Dan Williams" <dcbw@redhat.com>, linux-wireless@vger.kernel.org, dsd@laptop.org, libertas-dev@lists.infradead.org
> Date: Friday, June 15, 2012, 1:18 AM
> Hi,
> 
> On Thu, 14 Jun 2012 09:11:06 -0700 (PDT)
> hong zhang <henryzhang62@yahoo.com>
> wrote:
> 
> > bit mode and interrupt or PIO are configured in SDIO
> hoster side.
> > Not in driver side.
> > 
> yes, I wrote the last email just to say that I have found
> something to look at.
> That is of course in the host driver. So the thing I did was
> reading
> commit logs of the 2.6.32 kernel working well.
> 
> The only thing which puzzles me is the 5.5Mb/s data rate
> iwconfig shows.
> After playing with the host side driver, it also shows
> higher rates. But
> why does host driver affects *that* data rate of link
> qualitiy is good?
> 
> Thanks for all your ideas where to look at.
> 
> Greetings
> Andreas Kemnade
> 

Re: poor performance with W2CBW003 (Marvell 8686)

[Dan Williams]

On Fri, 2012-06-15 at 08:18 +0200, Andreas Kemnade wrote:
> Hi,
> 
> On Thu, 14 Jun 2012 09:11:06 -0700 (PDT)
> hong zhang <henryzhang62@yahoo.com> wrote:
> 
> > bit mode and interrupt or PIO are configured in SDIO hoster side.
> > Not in driver side.
> > 
> yes, I wrote the last email just to say that I have found something to look at.
> That is of course in the host driver. So the thing I did was reading
> commit logs of the 2.6.32 kernel working well.
> 
> The only thing which puzzles me is the 5.5Mb/s data rate iwconfig shows.
> After playing with the host side driver, it also shows higher rates. But
> why does host driver affects *that* data rate of link qualitiy is good?

As said, that's the WiFi air link rate which is determined by the
firmware based on the link quality with the AP, on things like RSSI,
missed beacons, TX/RX errors, etc.  That's not the SDIO data rate which
is completely independent.

Dan

Re: poor performance with W2CBW003 (Marvell 8686)

[Martin Chaplet]

Hi Andreas,

I'm facing same issue on a 3.0.X kernel.
I've tried to merge some patches to support SDIO irq on 3.0 kernel but without
success.

Interrupts are triggered and CPU load is greatly improved but performance is not
here (1 Mbps up, 3 Mbps down)

Did you actually get a working "SDIO irq" patch on a recent kernel or did you
finally use 2.6.32 one ?


Thanks,

Martin