Anyone have luck with WLE900VX?

Anyone have luck with WLE900VX?

[Ben Greear]

I managed to get a few WLE900VX, and I've seen some funny things so far.

First, with the official firmware 10.1.467.1-1 it immediately crashes on startup.
I don't have that firmware source, so I don't know why it crashes.
I wasn't even trying to create more than the default wlanX station.

With my firmware (based on 10.1.389), it boots up OK and I can create
and associate 32 station vifs.

Both NICs that I have tried so far report the same mostly bogus
MAC:  00:03:07:12:34:56

The NIC's spec sheet claims it is 3x3, but 'iw phy phyX info' shows only
2 streams supported for both the 2.4Ghz n band and the 5Ghz /ac band.

When stations associate (with ath9k AP in this case), throughput is
very bad..tx rate-control rate is stuck at 6Mbps, and rx rate is normally around
300Mbps.  (Rates are as reported by OS/driver, I did not try sniffing
the air..but total goodput throughput was in line with that bad tx rate.)

RX signal is all over the place, even though AP and STA are close by.

This is using over-the-air connection.

If anyone else has any experience with these NICs I'm interested to
hear about it...

Thanks,
Ben


-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Yeoh Chun-Yeow]

WLE900VX is dual band card, right? 11ac should work on 5GHz. Is the FW
good enough to fall back to 11n when operating in 2.4GHz?

I still wait my card to come so haven't tested yet.

----
Chun-Yeow

On Sat, Feb 22, 2014 at 9:05 AM, Ben Greear <greearb@candelatech.com> wrote:
> I managed to get a few WLE900VX, and I've seen some funny things so far.
>
> First, with the official firmware 10.1.467.1-1 it immediately crashes on startup.
> I don't have that firmware source, so I don't know why it crashes.
> I wasn't even trying to create more than the default wlanX station.
>
> With my firmware (based on 10.1.389), it boots up OK and I can create
> and associate 32 station vifs.
>
> Both NICs that I have tried so far report the same mostly bogus
> MAC:  00:03:07:12:34:56
>
> The NIC's spec sheet claims it is 3x3, but 'iw phy phyX info' shows only
> 2 streams supported for both the 2.4Ghz n band and the 5Ghz /ac band.
>
> When stations associate (with ath9k AP in this case), throughput is
> very bad..tx rate-control rate is stuck at 6Mbps, and rx rate is normally around
> 300Mbps.  (Rates are as reported by OS/driver, I did not try sniffing
> the air..but total goodput throughput was in line with that bad tx rate.)
>
> RX signal is all over the place, even though AP and STA are close by.
>
> This is using over-the-air connection.
>
> If anyone else has any experience with these NICs I'm interested to
> hear about it...
>
> Thanks,
> Ben
>
>
> --
> Ben Greear <greearb@candelatech.com>
> Candela Technologies Inc  http://www.candelatech.com
>
>
> _______________________________________________
> ath10k mailing list
> ath10k@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/ath10k

_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Adrian Chadd]

On 21 February 2014 17:47, Yeoh Chun-Yeow <yeohchunyeow@gmail.com> wrote:
> WLE900VX is dual band card, right? 11ac should work on 5GHz. Is the FW
> good enough to fall back to 11n when operating in 2.4GHz?
>
> I still wait my card to come so haven't tested yet.


.. sounds like it doesn't have any OTP on the NIC.

Is it perhaps designed to mate to a board with calibration data on the
SoC board, rather than NIC?


-a

> ----
> Chun-Yeow
>
> On Sat, Feb 22, 2014 at 9:05 AM, Ben Greear <greearb@candelatech.com> wrote:
>> I managed to get a few WLE900VX, and I've seen some funny things so far.
>>
>> First, with the official firmware 10.1.467.1-1 it immediately crashes on startup.
>> I don't have that firmware source, so I don't know why it crashes.
>> I wasn't even trying to create more than the default wlanX station.
>>
>> With my firmware (based on 10.1.389), it boots up OK and I can create
>> and associate 32 station vifs.
>>
>> Both NICs that I have tried so far report the same mostly bogus
>> MAC:  00:03:07:12:34:56
>>
>> The NIC's spec sheet claims it is 3x3, but 'iw phy phyX info' shows only
>> 2 streams supported for both the 2.4Ghz n band and the 5Ghz /ac band.
>>
>> When stations associate (with ath9k AP in this case), throughput is
>> very bad..tx rate-control rate is stuck at 6Mbps, and rx rate is normally around
>> 300Mbps.  (Rates are as reported by OS/driver, I did not try sniffing
>> the air..but total goodput throughput was in line with that bad tx rate.)
>>
>> RX signal is all over the place, even though AP and STA are close by.
>>
>> This is using over-the-air connection.
>>
>> If anyone else has any experience with these NICs I'm interested to
>> hear about it...
>>
>> Thanks,
>> Ben
>>
>>
>> --
>> Ben Greear <greearb@candelatech.com>
>> Candela Technologies Inc  http://www.candelatech.com
>>
>>
>> _______________________________________________
>> ath10k mailing list
>> ath10k@lists.infradead.org
>> http://lists.infradead.org/mailman/listinfo/ath10k
>
> _______________________________________________
> ath10k mailing list
> ath10k@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/ath10k

_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Ben Greear]

On 02/21/2014 05:47 PM, Yeoh Chun-Yeow wrote:
> WLE900VX is dual band card, right? 11ac should work on 5GHz. Is the FW
> good enough to fall back to 11n when operating in 2.4GHz?
>
> I still wait my card to come so haven't tested yet.

It appears to be dual band, and it was working at least somewhat
with my /n AP.  I was on the 5-ghz band though..haven't tested 2.4.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Adrian Chadd]

Check the firmware source; the MAC is read out of caldata and I
believe the default is (random three digits) + 12:34:56

So maybe the board didn't reset fully and the OTP or EEPROM didn't
load? You should be able to read the OTP data via the memory mapped
registers exposed to the host, as long as you know how to read OTP.
:-P


-a


On 22 February 2014 11:20, Ben Greear <greearb@candelatech.com> wrote:
> On 02/21/2014 05:47 PM, Yeoh Chun-Yeow wrote:
>>
>> WLE900VX is dual band card, right? 11ac should work on 5GHz. Is the FW
>> good enough to fall back to 11n when operating in 2.4GHz?
>>
>> I still wait my card to come so haven't tested yet.
>
>
> It appears to be dual band, and it was working at least somewhat
> with my /n AP.  I was on the 5-ghz band though..haven't tested 2.4.
>
>
> Thanks,
> Ben
>
> --
> Ben Greear <greearb@candelatech.com>
> Candela Technologies Inc  http://www.candelatech.com
>
>
> _______________________________________________
> ath10k mailing list
> ath10k@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/ath10k

_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Ben Greear]

On 02/21/2014 07:13 PM, Adrian Chadd wrote:
> On 21 February 2014 17:47, Yeoh Chun-Yeow <yeohchunyeow@gmail.com> wrote:
>> WLE900VX is dual band card, right? 11ac should work on 5GHz. Is the FW
>> good enough to fall back to 11n when operating in 2.4GHz?
>>
>> I still wait my card to come so haven't tested yet.
>
>
> .. sounds like it doesn't have any OTP on the NIC.
>
> Is it perhaps designed to mate to a board with calibration data on the
> SoC board, rather than NIC?

It appears to be a standard stand-alone mini-pci-e NIC.  Possibly firmware changes
are needed to make it able to read the OTP or whatever similar thing holds
that information on this NIC.

Maybe the upstream .467 firmware crashed because of some changes I made to ath10k
driver, so I'll try backing those out next week and will test again.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Ben Greear]

On 02/22/2014 11:24 AM, Adrian Chadd wrote:
> Check the firmware source; the MAC is read out of caldata and I
> believe the default is (random three digits) + 12:34:56
>
> So maybe the board didn't reset fully and the OTP or EEPROM didn't
> load? You should be able to read the OTP data via the memory mapped
> registers exposed to the host, as long as you know how to read OTP.
> :-P

Thanks, I'll go poking at that on Monday.  I have never messed with
OTP before, but perhaps I'll get lucky :)

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Michal Kazior]

On 22 February 2014 02:05, Ben Greear <greearb@candelatech.com> wrote:
> I managed to get a few WLE900VX, and I've seen some funny things so far.

[...]

> When stations associate (with ath9k AP in this case), throughput is
> very bad..tx rate-control rate is stuck at 6Mbps, and rx rate is normally around
> 300Mbps.  (Rates are as reported by OS/driver, I did not try sniffing
> the air..but total goodput throughput was in line with that bad tx rate.)

Unfortunately firmware doesn't provide a sane way to get tx rate for
tx status reporting so ath10k reports whatever a bunch of zeroes means
for given mode/band.


Michał

_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Kalle Valo]

Ben Greear <greearb@candelatech.com> writes:

> I managed to get a few WLE900VX, and I've seen some funny things so far.

Do you have any more info what type of board it is? Is it CUS223, XBNNN
(N being a decimal) or something else?

What's the chip id?

> First, with the official firmware 10.1.467.1-1 it immediately crashes on startup.
> I don't have that firmware source, so I don't know why it crashes.
> I wasn't even trying to create more than the default wlanX station.

Is that with stock ath10k driver or does ath10k contain any custom
changes?

> With my firmware (based on 10.1.389), it boots up OK and I can create
> and associate 32 station vifs.
>
> Both NICs that I have tried so far report the same mostly bogus
> MAC:  00:03:07:12:34:56
>
> The NIC's spec sheet claims it is 3x3, but 'iw phy phyX info' shows only
> 2 streams supported for both the 2.4Ghz n band and the 5Ghz /ac band.

Hmm, I wonder if otp.bin fails. Can you check what result code you get
from otp.bin? Add something like this to ath10k_download_and_run_otp():

	ret = ath10k_bmi_execute(ar, address, &exec_param);
        ath10k_dbg(ATH10K_DBG_BOOT, "boot otp execute result %d\n", exec_param);

-- 
Kalle Valo

_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Ben Greear]

On 02/24/2014 08:17 AM, Kalle Valo wrote:
> Ben Greear <greearb@candelatech.com> writes:
> 
>> I managed to get a few WLE900VX, and I've seen some funny things so far.
> 
> Do you have any more info what type of board it is? Is it CUS223, XBNNN
> (N being a decimal) or something else?

http://www.compex.com.sg/Datasheets/WLE900VX_Dsv1.0.0.pdf

I was told the ref design, but cannot remember it..will ask
again.  It is not a high power design like the CUS223, evidently.

> What's the chip id?

05:00.0 Network controller: Atheros Communications Inc. Device 003c
	Flags: bus master, fast devsel, latency 0, IRQ 18
	Memory at f4400000 (64-bit, non-prefetchable) [size=2M]
	Expansion ROM at f4600000 [disabled] [size=64K]
	Capabilities: [40] Power Management version 2
	Capabilities: [50] MSI: Enable- Count=1/8 Maskable+ 64bit-
	Capabilities: [70] Express Endpoint, MSI 00
	Capabilities: [100] Advanced Error Reporting
	Capabilities: [140] Virtual Channel
	Capabilities: [160] Device Serial Number 00-00-00-00-00-00-00-00
	Kernel driver in use: ath10k_pci


>> First, with the official firmware 10.1.467.1-1 it immediately crashes on startup.
>> I don't have that firmware source, so I don't know why it crashes.
>> I wasn't even trying to create more than the default wlanX station.
> 
> Is that with stock ath10k driver or does ath10k contain any custom
> changes?

It had custom changes, and at least one patch that might cause
trouble:  The one that doubled the number of key cache entries that
I posted a few days ago.

I can try with all that backed out.

>> With my firmware (based on 10.1.389), it boots up OK and I can create
>> and associate 32 station vifs.
>>
>> Both NICs that I have tried so far report the same mostly bogus
>> MAC:  00:03:07:12:34:56
>>
>> The NIC's spec sheet claims it is 3x3, but 'iw phy phyX info' shows only
>> 2 streams supported for both the 2.4Ghz n band and the 5Ghz /ac band.
> 
> Hmm, I wonder if otp.bin fails. Can you check what result code you get
> from otp.bin? Add something like this to ath10k_download_and_run_otp():
> 
> 	ret = ath10k_bmi_execute(ar, address, &exec_param);
>         ath10k_dbg(ATH10K_DBG_BOOT, "boot otp execute result %d\n", exec_param);


Will do, thanks!

Ben


-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Ben Greear]

On 02/24/2014 08:35 AM, Ben Greear wrote:
> On 02/24/2014 08:17 AM, Kalle Valo wrote:

>>> First, with the official firmware 10.1.467.1-1 it immediately crashes on startup.
>>> I don't have that firmware source, so I don't know why it crashes.
>>> I wasn't even trying to create more than the default wlanX station.
>>
>> Is that with stock ath10k driver or does ath10k contain any custom
>> changes?
> 
> It had custom changes, and at least one patch that might cause
> trouble:  The one that doubled the number of key cache entries that
> I posted a few days ago.
> 
> I can try with all that backed out.

Ok, with my ath10k changes backed out, the .467.1-1 firmware appears
to come up OK and the MAC address shows up right, NIC claims 3x3, etc.

Same driver software, on my .389 firmware, and MAC is not right, etc.

So, it seems my firmware is not supporting this NIC :(

Also, please do not apply that key-cache patch I posted a few days
ago...looks like upstream firmware may not like it.

Thanks,
Ben


-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Kalle Valo]

Ben Greear <greearb@candelatech.com> writes:

> On 02/24/2014 08:17 AM, Kalle Valo wrote:
>> Ben Greear <greearb@candelatech.com> writes:
>> 
>>> I managed to get a few WLE900VX, and I've seen some funny things so far.
>> 
>> Do you have any more info what type of board it is? Is it CUS223, XBNNN
>> (N being a decimal) or something else?
>
> http://www.compex.com.sg/Datasheets/WLE900VX_Dsv1.0.0.pdf
>
> I was told the ref design, but cannot remember it..will ask
> again.  It is not a high power design like the CUS223, evidently.

No need to ask, the PDF even mentions it:

"Qualcomm-Atheros QCA9880 Version 2, XB140 Reference Design"

XB140 is a dual band 3x3 board.

>> What's the chip id?
>
> 05:00.0 Network controller: Atheros Communications Inc. Device 003c
> 	Flags: bus master, fast devsel, latency 0, IRQ 18
> 	Memory at f4400000 (64-bit, non-prefetchable) [size=2M]
> 	Expansion ROM at f4600000 [disabled] [size=64K]
> 	Capabilities: [40] Power Management version 2
> 	Capabilities: [50] MSI: Enable- Count=1/8 Maskable+ 64bit-
> 	Capabilities: [70] Express Endpoint, MSI 00
> 	Capabilities: [100] Advanced Error Reporting
> 	Capabilities: [140] Virtual Channel
> 	Capabilities: [160] Device Serial Number 00-00-00-00-00-00-00-00
> 	Kernel driver in use: ath10k_pci

I was actually referring to the internal chip id, not PCI id, which is a
good way to check what chip it really is:

static int ath10k_core_check_chip_id(struct ath10k *ar)
{
	u32 hw_revision = MS(ar->chip_id, SOC_CHIP_ID_REV);

	ath10k_dbg(ATH10K_DBG_BOOT, "boot chip_id 0x%08x hw_revision 0x%x\n",
		   ar->chip_id, hw_revision);

-- 
Kalle Valo

_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Ben Greear]

On 02/24/2014 09:08 AM, Kalle Valo wrote:

> I was actually referring to the internal chip id, not PCI id, which is a
> good way to check what chip it really is:
> 
> static int ath10k_core_check_chip_id(struct ath10k *ar)
> {
> 	u32 hw_revision = MS(ar->chip_id, SOC_CHIP_ID_REV);
> 
> 	ath10k_dbg(ATH10K_DBG_BOOT, "boot chip_id 0x%08x hw_revision 0x%x\n",
> 		   ar->chip_id, hw_revision);
> 

Ahh, like this?

ath10k: boot chip_id 0x043202ff hw_revision 0x2


I do not see any OTP errors when loading my firmware, at least not
as seen from the ath10k driver.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Ben Greear]

On 02/24/2014 08:17 AM, Kalle Valo wrote:
> Ben Greear <greearb@candelatech.com> writes:

>> Both NICs that I have tried so far report the same mostly bogus
>> MAC:  00:03:07:12:34:56
>>
>> The NIC's spec sheet claims it is 3x3, but 'iw phy phyX info' shows only
>> 2 streams supported for both the 2.4Ghz n band and the 5Ghz /ac band.
> 
> Hmm, I wonder if otp.bin fails. Can you check what result code you get
> from otp.bin? Add something like this to ath10k_download_and_run_otp():
> 
> 	ret = ath10k_bmi_execute(ar, address, &exec_param);
>         ath10k_dbg(ATH10K_DBG_BOOT, "boot otp execute result %d\n", exec_param);

I tried the standard .389 firmware that mine is based on, and I see similar non-working
behaviour, so it does not appear to be a bug I added to my firmware.

I also notice that the otp binary in the .467.1-1 firmware is a fair bit bigger
than what is included in the .389 firmware.

So, probably I need fixes to otp (at least).

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Ben Greear]

On 02/22/2014 11:24 AM, Adrian Chadd wrote:
> Check the firmware source; the MAC is read out of caldata and I
> believe the default is (random three digits) + 12:34:56
> 
> So maybe the board didn't reset fully and the OTP or EEPROM didn't
> load? You should be able to read the OTP data via the memory mapped
> registers exposed to the host, as long as you know how to read OTP.
> :-P

Looks like my problem is just old firmware that does not have the
EEPROM reference tables needed by the WLE900VX.

But, if I hack on the OTP logic and make it use a reference table
for the reference design the NIC claims (XB140), then the NIC does
show up as 3x3, and gets proper MAC, etc.

This is a nasty hack, but maybe it will hold me over until I
can get access to a more proper solution.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: Anyone have luck with WLE900VX?

[Adrian Chadd]

... which default table does it say it wants?


-a

On 24 February 2014 15:47, Ben Greear <greearb@candelatech.com> wrote:
> On 02/22/2014 11:24 AM, Adrian Chadd wrote:
>> Check the firmware source; the MAC is read out of caldata and I
>> believe the default is (random three digits) + 12:34:56
>>
>> So maybe the board didn't reset fully and the OTP or EEPROM didn't
>> load? You should be able to read the OTP data via the memory mapped
>> registers exposed to the host, as long as you know how to read OTP.
>> :-P
>
> Looks like my problem is just old firmware that does not have the
> EEPROM reference tables needed by the WLE900VX.
>
> But, if I hack on the OTP logic and make it use a reference table
> for the reference design the NIC claims (XB140), then the NIC does
> show up as 3x3, and gets proper MAC, etc.
>
> This is a nasty hack, but maybe it will hold me over until I
> can get access to a more proper solution.
>
> Thanks,
> Ben
>
> --
> Ben Greear <greearb@candelatech.com>
> Candela Technologies Inc  http://www.candelatech.com
>

_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k