Re: [OpenWrt-Devel] [PATCH 2/3] ipq-wifi: select board-2.bin at runtime

[Matthew McClintock]

On Thu, Jan 11, 2018 at 11:36 AM, Christian Lamparter
 wrote:
> On Thursday, January 11, 2018 7:44:33 AM CET Mathias Kresin wrote:
>> 11.01.2018 07:23, Mathias Kresin:
>> > 10.01.2018 11:15, Thomas Hebb:
>> >> Currently, we solve the problem of different IPQ4019 boards needing
>> >> different sets of Wi-Fi calibration data (board-2.bin) by using an
>> >> ipq-wifi-* package to overwrite board-2.bin in the filesystem. This
>> >> presents a problem when we need the same image to support multiple
>> >> boards, as we do for the upcoming Netgear EX6100v2 and EX6150v2
>> >> support.
>> >
>> > I don't get what kind of issue do you have with this approach. Why not
>> > just add ipq-wifi-ex6100v2 to the DEVICE_PACKAGES of your EX6150v2.
>> >
>> > I'm quite sure I miss here something. Which either means it is to early
>> > for me or the commit message doesn't provide the whole picture.
>>
>> After reviewing your Netgear EX6100v2/EX6150v2 support patch I got your
>> issue. One image for two boards where the distinction of the board is
>> done in userspace and each board requires a different board-2.bin.
>
> Wait! It is possible to do this now:
> Sven Eckelmann added support for extracting the variant from the DT.
> "ath10k: search DT for qcom,ath10k-calibration-variant" [0]
>
> So, if Thomas splits the EX6100v2 and EX6150v2 image. He could have
> two separate dts. This way, each device can have its unique
> "qcom,ath10k-calibration-variant" and both can share a single
> board-2.bin.
>
> @Thomas, it's possible to get the EX6100v2/EX6150v2 board.bin into
> the official board-2.bin. The process is explained on the linux-wireless
> wiki:
> 
> It would be great if you could post the raw board.bin files there
> too.

There are also tools to repackage here:
https://github.com/qca/qca-swiss-army-knife

So you could make your own with just the calibration ones you were
interested in (to save space?) or just add your missing ones to a
custom board-2.bin.

-M
___
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel

Re: [OpenWrt-Devel] OpenWrt or LEDE to support industry device; BoardFarm testing

[Matthew McClintock]

On Wed, Aug 9, 2017 at 1:01 AM, Pfeuffer, Dieter  wrote:
> Hello Matthew,
>
> I read about the BoardFarm testing framwework in this mail:
> Date: Wed, 26 Jul 2017 16:15:52 +
> From: Shelly Coen 
> Subject: [OpenWrt-Devel] Notes for last weeks prplwrt meeting, and a
> reminder about tomorrow's meeting
>
>>* board farm - use to automatically test trunk and latest releases on a 
>>variety of platforms — need resources to execute
>
> Now, I found a git repo with further information:
> https://github.com/qca/boardfarm :
> Boardfarm is an open-source framework that was developed at Qualcomm to 
> automate testing of OpenWrt routers and other devices.
t farm where all the targets run?

If you do try that and have questions feel free to reach out.

-M
___
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel

Re: [OpenWrt-Devel] OpenWrt or LEDE to support industry device; BoardFarm testing

[Matthew McClintock]

On Tue, Aug 8, 2017 at 3:03 AM, Pfeuffer, Dieter  wrote:
> Today, I have read about the BoardFarm testing framework.
> 5. It is possible to run the tests locally in our company or is there one 
> global test farm where all the targets run?
> We have already a Jenkins server running and I could provide a NM50 for 
> testing.
> 6. Where can I found more information about the BoardFarm testing framework?

Which test framework are you talking about?

-M
___
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel

Re: [OpenWrt-Devel] Is "using gpio to simulate I2C bus" robust on AR9331?

[Matthew McClintock]

On Wed, May 3, 2017 at 8:19 PM, Xuebing Wang  wrote:
> Hi community,
>
> I am using Atheros AR9331 + OpenWRT Chaos Calmer 15.05 on a commercial
> product.
>
> I've used I2C on many projects, all with dedicated I2C controllers. I've
> never used GPIOs to simulate I2C on Linux.
>
> I may be wrong, I am tending to think that there is no way to generate good
> I2C waveforms by using 2 GPIOs (SCL/SDA), except that we disable all
> interrupts and maybe using high precision timer when outputing SCL/SDA?
>
> Is "using gpio to simulate I2C bus" robust on AR9331?


What speeds are you trying to achieve? There is already a driver in
tree for bitbanged gpio.

-M
___
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel

[OpenWrt-Devel] ubus network restart issue?

[Matthew McClintock]

All,

I've seen a pretty easy to reproduce ubus issue related to network
restart. I've tested this on various releases include the recently
release LEDE branch as well as OpenWrt CC. I've also tested on more
than one target and with various delays including as long as 2
minutes.

Before:

root@lede:/# ubus list
dhcp
file
iwinfo
log
network
network.device
network.interface
network.interface.lan
network.interface.loopback
network.interface.wan
network.interface.wan6
network.wireless
service
session
system
uci

root@lede:/# while true; do ubus call network restart; sleep 10; ubus list | gre
p network || { echo network failed; ubus list; break; }; done

[ snip - always about 6-7 iterations of the while loop ]

network failed
dhcp
file
iwinfo
log
service
session
system
uci

The end result is that we're unable to make ubus calls to all the
missing objects and we're stuck in this state getting an 'Object not
found' message if we try to operate on the missing items. Running
/etc/init.d/network restart will recover from this state.

Any one have ideas? Any ideas on where to start?

-M
___
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel

Re: [OpenWrt-Devel] [PATCH] ar71xx: add support for TRENDnet TEW-823DRU

[Matthew McClintock]

On Oct 5, 2016 12:09 AM, "Mathias Kresin"  wrote:
>
> 05.10.2016 06:36, Matthew McClintock:
>
>> On Sun, Sep 11, 2016 at 1:02 AM, Cezary Jackiewicz
>>  wrote:
>>>
>>> Dnia 2016-09-11, o godz. 00:05:18
>>> Matthew McClintock  napisał(a):
>>>
>>>> On Sat, Dec 12, 2015 at 1:53 AM, Cezary Jackiewicz
>>>>  wrote:
>>>>>
>>>>> Add support for TRENDnet TEW-823DRU:
>>>>> - cpu: QCA9558 @720MHz
>>>>> - flash: 16MB
>>>>> - ram: 256MB
>>>>> - 4+1 RJ45 100/1000Mbps
>>>>> - wifi:  QCA9558 (bgn) and QCA9880-BR4A (ac)
>>>>> - 1x USB 2.0
>>>>>
>>>>> Signed-off-by: Cezary Jackiewicz 
>>>>
>>>>
>>>>
>>>> Can you confirm the serial port info is correct here?
>>>> https://wiki.openwrt.org/toh/trendnet/tew-823dru
>>>
>>>
>>> Yes, four pin hole on the left.
>>
>>
>> FWIW, I also found that the u-boot command saveenv overwrites the
>> first sector containing Linux so don't run that command.
>
>
> Which indicates the partition layout is wrong:
>
>
tew823dru_mtdlayout=mtdparts=spi0.0:192k(u-boot)ro,64k(nvram)ro,15296k(firmware),192k(lang)ro,512k(my-dlink)ro,64k(mac)ro,64k(art)ro
>
> Without having knowledge about the board, I guess the partition layout
has to be:
>
> spi0.0:256k(u-boot)ro,64k(nvram)...
>
> Side note: I would be really surprised if a TRENDnet devices has a
my-dlink partition. I rather guess the lang and my-dlink partition are not
required for OpenWrt and the flash space can be used for the firmware
partition instead.
>
> Mathias

What's there works. I think it's right because normally you can't change
the bootcmd. So we're in a state were you simply can't save the uboot env.

Thought the my-dlink stuff was weird but the partitions at least are there.
Not sure if removing them is recoverable for switch back to stock for
example.

-M
___
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel

Re: [OpenWrt-Devel] [PATCH] ar71xx: add support for TRENDnet TEW-823DRU

[Matthew McClintock]

On Sun, Sep 11, 2016 at 1:02 AM, Cezary Jackiewicz
 wrote:
> Dnia 2016-09-11, o godz. 00:05:18
> Matthew McClintock  napisał(a):
>
>> On Sat, Dec 12, 2015 at 1:53 AM, Cezary Jackiewicz
>>  wrote:
>> > Add support for TRENDnet TEW-823DRU:
>> > - cpu: QCA9558 @720MHz
>> > - flash: 16MB
>> > - ram: 256MB
>> > - 4+1 RJ45 100/1000Mbps
>> > - wifi:  QCA9558 (bgn) and QCA9880-BR4A (ac)
>> > - 1x USB 2.0
>> >
>> > Signed-off-by: Cezary Jackiewicz 
>>
>>
>> Can you confirm the serial port info is correct here?
>> https://wiki.openwrt.org/toh/trendnet/tew-823dru
>
> Yes, four pin hole on the left.

FWIW, I also found that the u-boot command saveenv overwrites the
first sector containing Linux so don't run that command.

-M
___
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel

Re: [OpenWrt-Devel] [PATCH] ar71xx: add support for TRENDnet TEW-823DRU

[Matthew McClintock]

On Sat, Dec 12, 2015 at 1:53 AM, Cezary Jackiewicz
 wrote:
> Add support for TRENDnet TEW-823DRU:
> - cpu: QCA9558 @720MHz
> - flash: 16MB
> - ram: 256MB
> - 4+1 RJ45 100/1000Mbps
> - wifi:  QCA9558 (bgn) and QCA9880-BR4A (ac)
> - 1x USB 2.0
>
> Signed-off-by: Cezary Jackiewicz 


Can you confirm the serial port info is correct here?
https://wiki.openwrt.org/toh/trendnet/tew-823dru

-M
___
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel

Re: [OpenWrt-Devel] [PATCH RFC 10/10] ipq806x: wrap ubi in itb

[Matthew McClintock]

On 8/21/15, 6:43 AM, "openwrt-devel on behalf of Jonas Gorski" 
 wrote:

> On 21.08.2015 13:39, John Crispin wrote:
>> 
>> 
>> On 21/08/2015 13:28, Jonas Gorski wrote:
>>> On 21.08.2015 13:24, John Crispin wrote:
 
 
 On 21/08/2015 13:17, Jonas Gorski wrote:
> The stock QSDK expects the ubi to be wrapped in an itb image, so
> provide one for easy sysupgrade from QSDK.
> 
> FIXME: currently does not work? Seems to flash fine, but ubi cannot
> be mounted by OpenWrt and needs to be flashed a second time.
 
 how do you flash it the second time ? using the same process ?
>>> 
>>> Booted a ramdisk OpenWrt kernel, then flashed the ubi with ubiformat. Which 
>>> didn't seem to differ from what the preinstalled QSDK did, but somehow 
>>> works.
>>> 
>>> Jonas
>>> 
>> 
>> does sysupgrade work from owrt->owrt ?
> 
> Yes, even without the last two patches. The last two are only for "easy" 
> sysupgrade from QSDK (in theory). Therefore also as RFC.

The following patch is somehow causing the issue:

target/linux/generic/patches-3.18/494-mtd-ubi-add-EOF-marker-support.patch

Still investigating further.

[3.295677] UBI: auto-attach mtd11
[3.295698] UBI: attaching mtd11 to ubi0
[3.334563] UBI: EOF marker found, PEBs from 29 will be erased
[3.334725] UBI: scanning is finished
[3.339293] UBI error: ubi_read_volume_table: the layout volume was not found
[3.343386] UBI error: ubi_attach_mtd_dev: failed to attach mtd11, error -22
[3.350150] UBI error: ubi_auto_attach: cannot attach mtd11

But with the above patch removed the board will boot when flashed from u-boot.

-M
___
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel