creatordev / openwrt Goto Github PK

@mtusnio -
Copying files using scp/wget over Ethernet seems to max up the bandwith at about 2.2 MB/s. Additionally, running iperf3 over UDP nets the following results:

Local connection (CI40 to PC)
iperf3 -c 192.168.0.1 -i 1 -u -b 200M
Connecting to host 192.168.0.1, port 5201
[ 4] local 192.168.0.2 port 49873 connected to 192.168.0.1 port 5201
[ ID] Interval Transfer Bandwidth Total Datagrams
[ 4] 0.00-1.00 sec 1.90 MBytes 15.9 Mbits/sec 243
[ 4] 1.00-2.00 sec 2.02 MBytes 17.0 Mbits/sec 259
[ 4] 2.00-3.00 sec 2.08 MBytes 17.4 Mbits/sec 266
[ 4] 3.00-4.00 sec 2.09 MBytes 17.5 Mbits/sec 267
[ 4] 4.00-5.00 sec 2.08 MBytes 17.4 Mbits/sec 266
[ 4] 5.00-6.00 sec 2.09 MBytes 17.5 Mbits/sec 267
[ 4] 6.00-7.00 sec 2.08 MBytes 17.4 Mbits/sec 266
[ 4] 7.00-8.00 sec 2.09 MBytes 17.5 Mbits/sec 267
[ 4] 8.00-9.00 sec 2.09 MBytes 17.5 Mbits/sec 267
[ 4] 9.00-10.00 sec 2.09 MBytes 17.5 Mbits/sec 267

[ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
[ 4] 0.00-10.00 sec 20.6 MBytes 17.3 Mbits/sec 0.762 ms 10/2634 (0.38%)
[ 4] Sent 2634 datagrams
iperf Done.

Iperf3 server
iperf3 -c bouygues.testdebit.info -i 1 -u -b 200M
Connecting to host bouygues.testdebit.info, port 5201
[ 4] local 10.40.24.6 port 60806 connected to 89.84.127.53 port 5201
[ ID] Interval Transfer Bandwidth Total Datagrams
[ 4] 0.00-1.00 sec 1.86 MBytes 15.6 Mbits/sec 238
[ 4] 1.00-2.00 sec 2.12 MBytes 17.8 Mbits/sec 271
[ 4] 2.00-3.00 sec 2.14 MBytes 18.0 Mbits/sec 274
[ 4] 3.00-4.00 sec 2.09 MBytes 17.5 Mbits/sec 267
[ 4] 4.00-5.00 sec 2.10 MBytes 17.6 Mbits/sec 269
[ 4] 5.00-6.00 sec 2.13 MBytes 17.9 Mbits/sec 273
[ 4] 6.00-7.00 sec 2.09 MBytes 17.6 Mbits/sec 268
[ 4] 7.00-8.00 sec 2.09 MBytes 17.5 Mbits/sec 267
[ 4] 8.00-9.00 sec 2.07 MBytes 17.4 Mbits/sec 265
[ 4] 9.00-10.00 sec 2.09 MBytes 17.6 Mbits/sec 268

[ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
[ 4] 0.00-10.00 sec 20.8 MBytes 17.4 Mbits/sec 0.775 ms 0/2660 (0%)
[ 4] Sent 2660 datagrams
iperf Done.

@abhijit-mahajani-imgtec -
As per openwrt#95,
Remove target/linux/pistachio/creator-platform-default.config from openwrt-upstream

https://forum.creatordev.io/t/wifi-ap-radio-settings-broken/210/15

"Luci shows RX/TX bitrates of only 6Mbits/48Mbits and 20MHz bandwidth when using AC VHT80 or VHT40 modes."

Task is to verify against user settings + on another distro to identify cause.

@abhijit-mahajani-imgtec -
It would be better to put target/linux/pistachio/base-files/lib/netifd/proto/lowpan.sh content into an extra package for example into the iwpan package, this will probably also work on other devices and should be used there.

Often OpenWrt does not build due to external servers dieing or old tars being deleted before we have updated usually giving something similar to the below log:

Download failed.
--2016-11-15 14:48:14-- http://mirror2.openwrt.org/sources/ca-certificates_20160104.tar.xz
Resolving mirror2.openwrt.org (mirror2.openwrt.org)... 46.4.11.11
Connecting to mirror2.openwrt.org (mirror2.openwrt.org)|46.4.11.11|:80... connected.
HTTP request sent, awaiting response... 404 Not Found
2016-11-15 14:48:14 ERROR 404: Not Found.

The majority of these can be avoid using the LOCAL_MIRROR option of openwrt and hosting a copy of the downloads on our servers.

@Ham22 -
As soon as you try to enable kgdb you get a kernel panic with these options enabled:
CONFIG_DEBUG_KERNEL=y
CONFIG_KGDB=y
CONFIG_KGDB_SERIAL_CONSOLE=y
CONFIG_SERIAL_KGDB_NMI=n
CONFIG_KGDB_TESTS=n
CONFIG_KGDB_LOW_LEVEL_TRAP=n
CONFIG_KGDB_KDB=n
root@OpenWrt:/usr/bin# echo ttyS1 > /sys/module/kgdboc/parameters/kgdboc
[61669.361461] CPU 0 Unable to handle kernel paging request at virtual address 36314352, epc == 36314353, ra == 842e4d90
[61669.373481] Oops1:
[61669.376039] CPU: 0 PID: 649 Comm: ash Not tainted 4.1.13 #13
[61669.382359] task: 8fbceaa0 ti: 8f942000 task.ti: 8f942000
[61669.388375] $ 0 : 00000000 00000043 36314353 00000001
[61669.394244] $ 4 : 8ec14664 00000001 00000000 0000002c
[61669.400099] $ 8 : 0a310000 8ed62b86 00000010 00000001
[61669.405954] $12 : 7f983210 00000000 00000000 776fd320
[61669.411848] $16 : 00000000 8ec14664 00000000 8ec14200
[61669.417777] $20 : 84850ee4 00000004 84783d6c 00000020
[61669.423708] $24 : 00000010 84047df8
[61669.429636] $28 : 8f942000 8f943d48 004631c0 842e4d90
[61669.435573] Hi : 00000000
[61669.438812] Lo : 00000000
[61669.442117] epc : 36314353 0x36314353
[61669.446474] ra : 842e4d90 uart_poll_init+0xfc/0x12c
[61669.452229] Status: 11000403 KERNEL EXL IE
[61669.456975] Cause : 10800008
[61669.460212] BadVA : 36314352
[61669.463451] PrId : 0001a120 (MIPS interAptiv (multi))
[61669.469203] Modules linked in: pppoe ppp_async iptable_nat pppox ppp_generic pistachio nf_nat_ipv4 nf_conntrack_ipv6 nf_conntrack_ipv4 ipt_REJECT ipt_MASQUERADE xt_time xt_tcpudp xt_state xt_nat xt_multiport xt_mark xt_mac xt_limit xt_id xt_conntrack xt_comment xt_TCPMSS xt_REDIRECT xt_LOG xt_CT snd_soc_tpa6130a2 slhc sc16is7xx pistachio_internal_dac pistachio_event_timer_module nhc_udp nhc_routing nhc_mobility nhc_ipv6 nhc_hop nhc_fragment nhc_dest nf_reject_ipv4 nf_nat_redirect nf_nat_masquerade_ipv4 nf_nat nf_log_ipv4 nf_defrag_ipv6 nf_defrag_ipv4 nf_conntrack_rtcache nf_conntrack iptable_raw iptable_mangle iptable_filter ip_tables img_spdif_out img_spdif_in img_parallel_out img_i2s_out img_i2s_in ca8210sm mac802154 ieee802154_6lowpan ieee802154_socket ieee802154 6lowpan snd_soc_core i2c_dev ledtrig_heartbeat ip6t_REJECT nf_reject_ipv6 nf_log_ipv6 nf_log_common ip6table_raw ip6table_mangle ip6table_filter ip6_tables x_tables uccp420wlan snd_pcm_dmaengine snd_compress snd_pcm_oss snd_mixer_oss snd_pcm snd_timer snd_rawmidi snd_seq_device snd_hwdep snd soundcore dns_resolver leds_gpio
[61669.578077] Process ash (pid: 649, threadinfo=8f942000, task=8fbceaa0, tls=776fd614)
[61669.586728] Stack : 8f5ff330 8f5ff410 847644c0 8ee14100 00000001 00000000 00002580 00000008
0000006e 0000006e 8f40df38 8f3db900 84780000 00000001 8f943dd0 842d56e4
8f058000 840faa4c 00002301 00000001 00000001 84850ee9 00000000 84850000
84780000 84850000 847845ac 8efcc20c 00000030 842eed50 8f943e60 00000000
8ee14100 8f943ef4 00000000 840ee8a4 8f058000 8ed62b80 8f9ec394 84780000
...
[61669.627077] Call Trace:
[61669.630019] [<842d56e4>] tty_find_polling_driver+0x134/0x20c
[61669.636406] [<840faa4c>] dput+0xe0/0x314
[61669.640865] [<842eed50>] configure_kgdboc+0xdc/0x1a0
[61669.646525] [<840ee8a4>] path_cleanup+0x30/0x58
[61669.651740] [<84047e7c>] param_attr_store+0x84/0xbc
[61669.657234] [<840dfa08>] __kmalloc+0x4c/0x13c
[61669.662186] [<84009adc>] mutex_lock+0x14/0x90
[61669.667158] [<84145010>] kernfs_fop_write+0x12c/0x1a8
[61669.672939] [<840e4ad0>] __vfs_write+0x30/0x104
[61669.678101] [<840e7344>] __sb_start_write+0xd4/0x120
[61669.683737] [<840e5160>] rw_verify_area+0xb8/0x150
[61669.689192] [<840e53d0>] vfs_write+0xbc/0x168
[61669.694140] [<84031c08>] SyS_wait4+0xb4/0xe8
[61669.698979] [<84102c50>] __fdget_pos+0x14/0x60
[61669.704019] [<840e5c78>] SyS_write+0x60/0xc4
[61669.708842] [<8402fb3c>] child_wait_callback+0x0/0x68
[61669.714553] [<84002ba8>] syscall_common+0x30/0x54
[61669.719876]
[61669.721563]
Code: (Bad address in epc)
[61669.729900]
[61669.731676] --[ end trace 3ad2abe44f2eea86 ]--

@Ham22 -
If you run 'make kernel_menuconfig' and then save the results without changing anything the build of the kernel then fails due to wrong default config. This makes changing anything rather difficult with manual changes required that have to take into account any config dependencies.

It looks like every option affected by this is causing a problem because that option is added in a patch in target/linux/pistachio meaning that scripts/kconfig.pl doesn't know about it. Abhijit Mahajani should the patches be touching the config files in linux itself? How are the kernel config files in openwrt related to the kernel config files in its own tree, does it just ignore them?
Possible fix is to remove any config file stuff in the patches and instead put all of that directly into pistachio/config-4.1 or possibly even just do that latter of the two.
This is fixed in #51

@abhijit-mahajani-imgtec -
Our Ci40 port from https://github.com/Creatordev/openwrt has to be upstreamed to https://github.com/openwrt/openwrt
However it seems that we need to create a new fork of github openwrt and add our changes for pistachio port there.

@nikhil-zinjurde-imgtec
Having channels for Wi-Fi and 6LoWPAN on Ci40 close to each other can lead to Wi-Fi stamping on 6LoWPAN and suppressing it completely.
First thing to do is to advise developers through our documentation notes about the possible interference problems (as covered in CKT-140)
Secondly, it would be very useful to have an event or a warning when someone sets the channels of Wi-Fi and 6LoWPAN too close. This way we could make developers aware about the preventions to take and potentially reduce the issues or queries being raised on Creator forums

It would be useful for headless demos if the user could tell which stage of the boot sequence the device is at by some sort of LED change. For example, all LEDs might be turned on by the bootloader, then one LED might turn off when the kernel is loaded, another for init started, another once LWM2M daemons are running.
At the minimum, a single LED that indicates the device is ready would be really useful, as the demoer can wait for the LED before proceeding.

@abhijit-mahajani-imgtec

Acceptance criteria:
Bluetooth interface initialises and a bluetooth scan works.

Add script/pipeline to promote builds to the release server i.e. automate the current manual process https://imghub.imgtec.org/display/IM/Tagging+Releases .

I want to be able to specify an existing build in our CI system, press a button and all the process required to promote it to an official release on http://downloads.creatordev.io/ to happen.

Add Imagination's Pistachio SoC and (Ci40) Marduk platform support into https://github.com/openwrt/openwrt
As a first stage, just add basic ethernet support i.e. as router as per openwrt standards.
and then later on lowpan, bluetooth, wifi etc remaining RF support.

The CI system does some magic to make things build meaning a normal user can't build it without knowing this magic. Remove the magic where possible or at least minimise the delta introduced.

@MattAtkinson -
Report from the Creator Forum:
https://forum.creatordev.io/t/green-led-of-ethernet-has-no-activity/109

There are no valid defaults set in the configs for a few packages:
VGA text console (VGA_CONSOLE) [Y/n/?] (NEW)
Enable Scrollback Buffer in System RAM (VGACON_SOFT_SCROLLBACK) [N/y/?] (NEW)
Socket and Networking Security Hooks (SECURITY_NETWORK) [Y/n/?] y
XFRM (IPSec) Networking Security Hooks (SECURITY_NETWORK_XFRM) [N/y/?] (NEW)

@francois-berder-imgtec -
We need hci, hciconfig and hcitool to run test_bluetooth and test_ble
gatttool cannot be added because of license incompatibility.

bluez depends on glib2 which depends on iconv.
The current build of glib2 is broken and a pull request exists to fix the issue: openwrt/packages#2425. However, this pull request is now closed.
Delaying this task until build of glib2 is fixed.

It seems that glib2 is compiling for host then for target.
iconv-full does not have host configuration in feed. But iconv is compiled for host. Hence, during the configuration of glib2 it uses header and libraries from iconv instead of iconv-full.

See my comments here: openwrt/packages#2322.

Incremental approach for adding bluetooth support into openwrt and upstream it.

Incremental approach for adding wifi support into openwrt and upstream it.
Basic wifi support is already there in CreatorDev/openwrt on ci40 branch and for upstreaming we need to add on top of openwrt#201

@nikhil-zinjurde-imgtec -
We should make developers of Ci40 and Creator Kit aware of the reasons for having Wi-Fi and 6LoWPAN on channels far away from each other. We should also mention the default channel (i.e. 26) being used for our demo applications

@abhijit-mahajani-imgtec -
Though we have added a workaround for replace opkg by opkg-smime, we need to fix it properly since we should not change the default openwrt behaviour i.e. DEFAULT_PACKAGES from include/target.mk should not be touched.
Issue is
when opkg and opkg-smime are both enabled, opkg.conf getting copied from opkg-unsigned and not from opkg-smime. so need to fix such that when opkg-smime is enabled it should copy opkg-smime.conf as opkg.conf into target file system.

Sometimes when booting the networking settings take about 2 minutes to load. During that time the system makes multiple attempts at enabling both ethernet and wifi shown by proddata and uccp debug messages in the logs e.g:
img/uccp420wlan/MCP_LOADER.ldr is loaded
img/uccp420wlan/MAC_LOADER.ldr is loaded
Initialising Proddata
Deinitialising Proddata
Deinitialising DeviceData
Deinitialising UserOTPAccess

Eventually it brings up eth and wlan only (i.e. no lowpan), however seems to continuously switch between up & down states forever if an ethernet cable is plugged in.

@abhijit-mahajani-imgtec -
Last looked at mid June 2016

Add one boardfarm as a node to Jenkins and add the existing boardfarm command to the jenkinsfile.
Publish Junit report file.

needs to deal with the parametised builds

@Ham22 -
ci40 openwrt uses a lot of defaults that may not necessarily be the best option for us.

• We should discuss which libc to use but glibc is probably better for us as we have strict size limits.
• We should probably configure it to use our codescape sdk toolchain (this would likely be an external prebuilt so would speed up the build massively)

@abhijit-mahajani-imgtec -
As per openwrt#95.,
Changes done in modules.mk for nhc_fragment should be patched to kmod-6lowpan separately and send a PR for that only.

@abhijit-mahajani-imgtec -
We are using get_mac/set_mac functions written in shell scripts using proddata apis. But it seems that OpenWrt has different way for generating mac addresses. As suggested by Hauke in openwrt#95
please do not use these hacks to set the mac address, have a look at how it is solved here: target/linux/lantiq/base-files/etc/board.d/02_network
so it seems there are two tasks needed to be done:

1. Port WinBond OTP driver in u-boot so that it can read mac addresses from OTP and have ethaddr, wifiaddr, wifiapaddr, lowpanaddr, btaddr etc in uEnv partition. (https://issues.ba.imgtec.org/browse/CKT-444)
2. Add 02_network scripts, 06_set_iface_mac to set the mac addresses and network accordingly. This can be done in present task.

@Ham22 -
During bring up could not get the PP line to actually assert presence. Had to always use the SW presence mechanism.

@abhijit-mahajani-imgtec
prof on the creator forum confirmed that the startup issues he's having affect both 0.9.4 and 0.10.4:
Paste from https://forum.creatordev.io/t/wifi-ap-radio-settings-broken/210:

I've got a problem with the WiFi AP configuration, as soon as I make any change to the radio configuration from the defaults WiFI doesn't ever recover and I'll only receive those messages in the system log:

[ 422.867171] img/uccp420wlan/MCP_LOADER.ldr is loaded
[ 422.946554] img/uccp420wlan/MAC_LOADER.ldr is loaded
[ 424.524722] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
[ 434.265328] img/uccp420wlan/MCP_LOADER.ldr is loaded
[ 434.348674] img/uccp420wlan/MAC_LOADER.ldr is loaded
[ 435.944295] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
[ 447.741800] img/uccp420wlan/MCP_LOADER.ldr is loaded
[ 447.821123] img/uccp420wlan/MAC_LOADER.ldr is loaded
[ 449.404827] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
To reproduce I do a factory reset, then create an AP with the default settings, verify that the AP works, then switch the Operating Mode from "Legacy" to "AC" and from that point on (even when switching back to "Legacy") the AP doesn't work anymore. The webinterface shows "Wireless is disabled or not associated" as status.

Incremental approach for adding 6lowpan support into openwrt and upstream it.
Basic 6lowpan support is already there in CreatorDev/openwrt on ci40 branch and for upstreaming we need to add on top of openwrt#201
And lowpan configuration needs to be added in wpan package or proper openwrt way.

@abhijit-mahajani-imgtec -
As per openwrt#95,
The wifi driver should better use package/kernel/mac80211/ It is only possible to have one mac80211 and cfg80211 subsystem loaded into the kernel at the time and OpenWrt uses mac80211 from backports in package/kernel/mac80211/ . You can either add some patches to package/kernel/mac80211/ or add a separate package like package/kernel/mwlwifi/Makefile

All PRs fail until rebuilt due to default parameters not being set until a build a been run.

@francois-berder-imgtec -
Inserting the UART click on mikrobus 1, sending data overt UART does not switch on the RX led.

@abhijit-mahajani-imgtec -
As per the openwrt#95,
there are review comments regarding WiFi/mac80211.sh
1.If we are changing in original mac80211.sh, then just propose a patch with the delta, rather than keeping a copy as target/linux/pistachio/base-files/lib/netifd/wireless/mac80211.sh
2.target/linux/pistachio/base-files/etc/config/wireless should get auto-generated by mac80211.sh at first boot when this devices gets detected.

@abhijit-mahajani-imgtec -
As per openwrt#95, regarding target/linux/pistachio/base-files/etc/asound.conf
Doesn't this file belong to the asound package?
Then you can have install script similarly like we have in uboot-env.

TPM reset is connected to a gpio rather than system reset. We don't currently toggle this meaning that the tpm doesn't get reset on a reboot. The booloader should toggle it to make sure.

Is the fact that the tpm reset line is connected to a Danube gpio a security risk. Need to look into it as not sure but the spec states that you should not be able to re-initialise the chip if the host device is not initialising itself.

From current tip OpenWrt, if you run "./scripts/feeds update -a" you get lots of messages about syntax errors. It seems like everything works correctly, but it looks like it's broken.

Cloning into './feeds/packages'...
remote: Counting objects: 3722, done.
remote: Compressing objects: 100% (3132/3132), done.
remote: Total 3722 (delta 143), reused 3007 (delta 91), pack-reused 0
Receiving objects: 100% (3722/3722), 2.29 MiB | 838.00 KiB/s, done.
Resolving deltas: 100% (143/143), done.
Checking connectivity... done.
Create index file './feeds/packages.index' 
/home/img/creator/toolchain/openwrt/scripts/getver.sh: line 22: v0.10.5: syntax error: invalid arithmetic operator (error token is ".10.5")
/home/img/creator/toolchain/openwrt/scripts/getver.sh: line 22: v0.10.5: syntax error: invalid arithmetic operator (error token is ".10.5")
/home/img/creator/toolchain/openwrt/scripts/getver.sh: line 22: v0.10.5: syntax error: invalid arithmetic operator (error token is ".10.5")
Collecting package info: done
/home/img/creator/toolchain/openwrt/scripts/getver.sh: line 22: v0.10.5: syntax error: invalid arithmetic operator (error token is ".10.5")
Collecting target info: done```

@Ham22 -
Add a reference verified boot implementation to the ci40 openwrt code.
Note to make use of @shraddha's work done for the caskeid project

Add kmods for tpm

@Ham22 -
Have added the code required to get the tpm running but I had to use SW PP in order to get it working as HW PP doesn't seem to assert when I pull the PIN to 3v3.

@abhijit-mahajani-imgtec -
Pistachio porting into Openwrt requires hundreds of kernel patches, since not all are available on 4.4.14 upstream openwrt kernel. Currently there are 130 odd patches. Need to assess what all patches can be squashed such that patches per feature can be added. bug fixes can be squashed in appropriate commits.
also there have been some suggestions that /etc/config/network, etc changes which are overriding openwrt defaults will not get accepted by openwrt upstream, also the feeds.conf.default changes will not get accepted. so need to have those commits/changes removed while sending upstream pull request.
This task is created to track all such changes which need to be removed or changed from CreatorDev/openwrt.

@francois-berder-imgtec -

It should not have that many devices in the /dev folder

root@OpenWrt:/# ls /dev/
adsp ptyq4 ptyz5 ttyr6
ashmem ptyq5 ptyz6 ttyr7
audio ptyq6 ptyz7 ttyr8
bus ptyq7 ptyz8 ttyr9
console ptyq8 ptyz9 ttyra
cpu_dma_latency ptyq9 ptyza ttyrb
dsp ptyqa ptyzb ttyrc
full ptyqb ptyzc ttyrd
hwrng ptyqc ptyzd ttyre
i2c-0 ptyqd ptyze ttyrf
i2c-1 ptyqe ptyzf ttys0
iio:device0 ptyqf random ttys1
img-atu ptyr0 rfkill ttys2
input ptyr1 root ttys3
kmsg ptyr2 rtc0 ttys4
log ptyr3 sda ttys5
mapper ptyr4 sda1 ttys6
mem ptyr5 shm ttys7
memory_bandwidth ptyr6 snd ttys8
mixer ptyr7 spidev0.2 ttys9
mtd0 ptyr8 spidev0.3 ttysa
mtd0ro ptyr9 tpm0 ttysb
mtd1 ptyra tty ttysc
mtd1ro ptyrb ttyS0 ttysd
mtd2 ptyrc ttyS1 ttyse
mtd2ro ptyrd ttySC0 ttysf
mtd3 ptyre ttySC1 ttyt0
mtd3ro ptyrf ttya0 ttyt1
mtd4 ptys0 ttya1 ttyt2
mtd4ro ptys1 ttya2 ttyt3
mtd5 ptys2 ttya3 ttyt4
mtd5ro ptys3 ttya4 ttyt5
mtdblock0 ptys4 ttya5 ttyt6
mtdblock1 ptys5 ttya6 ttyt7
mtdblock2 ptys6 ttya7 ttyt8
mtdblock3 ptys7 ttya8 ttyt9
mtdblock4 ptys8 ttya9 ttyta
mtdblock5 ptys9 ttyaa ttytb
network_latency ptysa ttyab ttytc
network_throughput ptysb ttyac ttytd
null ptysc ttyad ttyte
ppp ptysd ttyae ttytf
ptmx ptyse ttyaf ttyu0
ptp0 ptysf ttyb0 ttyu1
pts ptyt0 ttyb1 ttyu2
ptya0 ptyt1 ttyb2 ttyu3
ptya1 ptyt2 ttyb3 ttyu4
ptya2 ptyt3 ttyb4 ttyu5
ptya3 ptyt4 ttyb5 ttyu6
ptya4 ptyt5 ttyb6 ttyu7
ptya5 ptyt6 ttyb7 ttyu8
ptya6 ptyt7 ttyb8 ttyu9
ptya7 ptyt8 ttyb9 ttyua
ptya8 ptyt9 ttyba ttyub
ptya9 ptyta ttybb ttyuc
ptyaa ptytb ttybc ttyud
ptyab ptytc ttybd ttyue
ptyac ptytd ttybe ttyuf
ptyad ptyte ttybf ttyv0
ptyae ptytf ttyc0 ttyv1
ptyaf ptyu0 ttyc1 ttyv2
ptyb0 ptyu1 ttyc2 ttyv3
ptyb1 ptyu2 ttyc3 ttyv4
ptyb2 ptyu3 ttyc4 ttyv5
ptyb3 ptyu4 ttyc5 ttyv6
ptyb4 ptyu5 ttyc6 ttyv7
ptyb5 ptyu6 ttyc7 ttyv8
ptyb6 ptyu7 ttyc8 ttyv9
ptyb7 ptyu8 ttyc9 ttyva
ptyb8 ptyu9 ttyca ttyvb
ptyb9 ptyua ttycb ttyvc
ptyba ptyub ttycc ttyvd
ptybb ptyuc ttycd ttyve
ptybc ptyud ttyce ttyvf
ptybd ptyue ttycf ttyw0
ptybe ptyuf ttyd0 ttyw1
ptybf ptyv0 ttyd1 ttyw2
ptyc0 ptyv1 ttyd2 ttyw3
ptyc1 ptyv2 ttyd3 ttyw4
ptyc2 ptyv3 ttyd4 ttyw5
ptyc3 ptyv4 ttyd5 ttyw6
ptyc4 ptyv5 ttyd6 ttyw7
ptyc5 ptyv6 ttyd7 ttyw8
ptyc6 ptyv7 ttyd8 ttyw9
ptyc7 ptyv8 ttyd9 ttywa
ptyc8 ptyv9 ttyda ttywb
ptyc9 ptyva ttydb ttywc
ptyca ptyvb ttydc ttywd
ptycb ptyvc ttydd ttywe
ptycc ptyvd ttyde ttywf
ptycd ptyve ttydf ttyx0
ptyce ptyvf ttye0 ttyx1
ptycf ptyw0 ttye1 ttyx2
ptyd0 ptyw1 ttye2 ttyx3
ptyd1 ptyw2 ttye3 ttyx4
ptyd2 ptyw3 ttye4 ttyx5
ptyd3 ptyw4 ttye5 ttyx6
ptyd4 ptyw5 ttye6 ttyx7
ptyd5 ptyw6 ttye7 ttyx8
ptyd6 ptyw7 ttye8 ttyx9
ptyd7 ptyw8 ttye9 ttyxa
ptyd8 ptyw9 ttyea ttyxb
ptyd9 ptywa ttyeb ttyxc
ptyda ptywb ttyec ttyxd
ptydb ptywc ttyed ttyxe
ptydc ptywd ttyee ttyxf
ptydd ptywe ttyef ttyy0
ptyde ptywf ttyp0 ttyy1
ptydf ptyx0 ttyp1 ttyy2
ptye0 ptyx1 ttyp2 ttyy3
ptye1 ptyx2 ttyp3 ttyy4
ptye2 ptyx3 ttyp4 ttyy5
ptye3 ptyx4 ttyp5 ttyy6
ptye4 ptyx5 ttyp6 ttyy7
ptye5 ptyx6 ttyp7 ttyy8
ptye6 ptyx7 ttyp8 ttyy9
ptye7 ptyx8 ttyp9 ttyya
ptye8 ptyx9 ttypa ttyyb
ptye9 ptyxa ttypb ttyyc
ptyea ptyxb ttypc ttyyd
ptyeb ptyxc ttypd ttyye
ptyec ptyxd ttype ttyyf
ptyed ptyxe ttypf ttyz0
ptyee ptyxf ttyq0 ttyz1
ptyef ptyy0 ttyq1 ttyz2
ptyp0 ptyy1 ttyq2 ttyz3
ptyp1 ptyy2 ttyq3 ttyz4
ptyp2 ptyy3 ttyq4 ttyz5
ptyp3 ptyy4 ttyq5 ttyz6
ptyp4 ptyy5 ttyq6 ttyz7
ptyp5 ptyy6 ttyq7 ttyz8
ptyp6 ptyy7 ttyq8 ttyz9
ptyp7 ptyy8 ttyq9 ttyza
ptyp8 ptyy9 ttyqa ttyzb
ptyp9 ptyya ttyqb ttyzc
ptypa ptyyb ttyqc ttyzd
ptypb ptyyc ttyqd ttyze
ptypc ptyyd ttyqe ttyzf
ptypd ptyye ttyqf ubi_ctrl
ptype ptyyf ttyr0 urandom
ptypf ptyz0 ttyr1 usbmon0
ptyq0 ptyz1 ttyr2 usbmon1
ptyq1 ptyz2 ttyr3 watchdog
ptyq2 ptyz3 ttyr4 watchdog0
ptyq3 ptyz4 ttyr5 zero

@Ham22 -
Netboot is very useful for quick debug cycles when you are making constant changes to the kernel or anything else for that matter.
But it doesn't want to just work so needs some effort put in.

Note saw this patch 0098-pistachio-enable-NFS-client-support.patch in our patch list

@francois-berder-imgtec -
Defining all 8 gpios in mmc driver, although only 4 gpios are used, caused some issues during prodtest.
exporting gpio23 (sdhost data6) and configuring it as output confuses the driver and the emmc partition is not displayed anymore in fdisk -l output.
Both emmc and sdcard fails it some gpio are configured as output (gpio23, gpio21)
This can be seen if running the sdcard test then the gpio on GATE.

@Ham22 -
The pistachio makefiles for openwrt seem to have some issues:

• They do a lot of stuff that is already there in the generic makefiles - we just don't use them
• They confuse ubi and ubifs differences
• They are doing things that aren't used needed