00:14
dansan has joined #openwrt-devel
01:34
jeff___m has joined #openwrt-devel
02:20
parazyd has quit [Ping timeout: 480 seconds]
02:22
hanetzer has joined #openwrt-devel
03:00
tSYS has quit [Quit: *squeak*]
03:00
tSYS has joined #openwrt-devel
03:34
skynet2 has quit [Ping timeout: 480 seconds]
04:08
rua has quit [Quit: Leaving.]
05:05
gch981213 has joined #openwrt-devel
05:30
gch981213 has joined #openwrt-devel
06:12
torv is now known as Guest14814
06:12
torv has joined #openwrt-devel
06:14
Guest14814 has quit [Ping timeout: 480 seconds]
06:14
torv has quit [Remote host closed the connection]
06:15
torv has joined #openwrt-devel
08:28
jeff___m has quit [Remote host closed the connection]
08:32
jeff___m has joined #openwrt-devel
08:40
f00b4r0 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
08:41
f00b4r0 has joined #openwrt-devel
08:42
jeff___m has quit [Ping timeout: 480 seconds]
09:06
<
mrkiko >
Ok, definitely worrysome situation with Belking RT3200
09:07
<
mrkiko >
Wondering if my units boot again
10:26
robimarko has joined #openwrt-devel
10:32
<
rmilecki >
robimarko: Ansuel: i've a new problem:
10:32
<
rmilecki >
[ 27.209194] Aquantia AQR113C mdio-bus:08: aqr107_wait_reset_complete failed: -110
10:32
<
rmilecki >
it comes from aqr107_config_init()
10:33
<
rmilecki >
the most common reason for that error is missing firmware (not running)
10:33
<
robimarko >
That error means that the PHY is not ready
10:33
<
rmilecki >
it's the same case here BUT the difference is why firmware is not running
10:34
<
rmilecki >
in my case aqr107_config_init() was called BEFORE aqr107_probe() return
10:34
<
rmilecki >
is there some race in PHY subsystem?
10:34
<
robimarko >
Are you trying initramfs or flashed image?
10:34
<
robimarko >
I have a similar issue when flashed
10:34
<
robimarko >
When initramfs is booted it all works fine
10:34
<
robimarko >
But flashed one of my 2 AQR-s is not found
10:35
<
robimarko >
And the other returns -110
10:35
<
rmilecki >
uh, so more debugging ;)
10:35
<
robimarko >
I have been scratching my head for a while against that
10:35
<
robimarko >
But havent got time to properly debug it
10:58
cation has joined #openwrt-devel
10:59
<
rmilecki >
i think it's a matter of how much time it takes to load aquantia firmware
11:00
<
rmilecki >
probing first PHY takes quite some time but it still happens before probing Ethernet driver/hardware
11:00
<
rmilecki >
now, probing 2nd Ethernet core hpapens during probing 2nd aquantia PHY
11:00
<
rmilecki >
and that seems like a problem
11:04
<
robimarko >
I am having a probing race or something as well
11:05
<
robimarko >
Cause, in my case the second AQR is not even identified as AQR
11:05
<
robimarko >
It only gets identified way later
11:05
<
robimarko >
Its all somehow racing
11:09
<
robimarko >
Ansuel: Any chance you can review the ipq60xx PR from 8dev?
11:59
goliath has joined #openwrt-devel
12:37
rsalvaterra has joined #openwrt-devel
12:41
torv has quit [Quit: torv]
12:43
torv has joined #openwrt-devel
13:06
torv has quit [Remote host closed the connection]
13:10
torv has joined #openwrt-devel
13:19
<
Ansuel >
robimarko looking at the mtk code i notice there is some comments for gangload and daisychain
13:19
<
Ansuel >
i wonder if our problem is exactly that
13:20
rua has joined #openwrt-devel
13:20
<
Ansuel >
maybe the problem is present only with case where we have more than one aquantia phy ?
13:21
<
Ansuel >
for ipq60xx i recently received a router with that so i can also test it but what is the current state of it? I remember there were still some thing to polish ?
13:25
rua has quit [Remote host closed the connection]
13:26
rua has joined #openwrt-devel
13:39
<
robimarko >
Ansuel: Those are for when you have AQR-s connected to one another
13:40
<
robimarko >
Ansuel: There is still stuff to polish but the PR in its state is usable
13:41
<
Ansuel >
btw random idea is there some regs to check the version of the tz fw?
13:42
<
robimarko >
Ansuel: You can get the TZ FW string via SMEM
13:42
<
robimarko >
But that can happen only way later
13:42
<
Ansuel >
loved if psci: dont advertise OSI support for IPQ6018 could have some better handling (what did they say upstream?)
13:43
<
robimarko >
I did not submit it upstream as its really not upstreamable
13:43
<
robimarko >
Only way would be to add a new DT property to advertise the quirk
13:44
<
Ansuel >
yes was thinking just about that
13:44
<
robimarko >
I see the 8devices PR as a good starting point for now
13:45
<
robimarko >
There is stuff missing for sure, I only gave it one round of reviews
13:45
<
robimarko >
Mantas found the missing clock so that avoids the whole mess of including dozens of clocks just to get wifi working
13:45
<
robimarko >
And he fixed the USB3 port
13:46
<
Ansuel >
yep yep just asserting the situation
13:46
<
robimarko >
He finally pushed me to start working on the SoC
13:46
<
Ansuel >
what i currently hate is the fact that the wifi fw still crash... that is really strange
13:46
<
Ansuel >
(with the new fw versions)
13:47
<
robimarko >
Yeah, I agree
13:47
<
Ansuel >
we should really check if that happens with a qsdk mac80211
13:47
<
robimarko >
Though we are completely missing the 2.9 FW series for it
13:47
<
robimarko >
2.6 and 2.7 were broken on ipq807x as well
13:49
<
Ansuel >
mhhh the problem was with BDF right?
13:49
<
Ansuel >
or I'm wrong?
13:49
skynet2 has joined #openwrt-devel
13:53
<
robimarko >
We dont know, as 2.9 magically worked again
13:54
<
Ansuel >
well I remeber 2.6 and 2.7 working only on qnap
13:54
<
Ansuel >
with other legacy device being broken
13:55
<
Ansuel >
do you have by chance a dts for for the linksys mr7350 ?
13:55
<
Ansuel >
we should be able to check for fw support by testing if an upstream BDF works with the new fw
13:59
<
robimarko >
AX6 was completely broken before 2.9
14:00
<
Ansuel >
let me do some comparison maybe the same problem with ipq807x is also present on ipq60xx
14:01
<
Ansuel >
for sure them not publishing fw for 2 years is very bad... maybe us adding support in openwrt will change things up ?
14:01
<
robimarko >
I hope it does, cause QSDK FW for IPQ807x and IPQ60xx is pretty much the same
14:02
<
robimarko >
They always have the same versions
14:02
<
robimarko >
No idea why they dont update the public versions for IPQ60xx
14:03
<
Ansuel >
I will try to send a mail to kavlo
14:05
<
Ansuel >
interesting is that i have a gl-axt1800 that reported to correctly work with 2.7
14:06
<
Ansuel >
after some fixup
14:06
<
Ansuel >
but his crash was with ailed to extract regulatory info from received event
14:06
<
robimarko >
That basically means that BDF has no regulatory info
14:09
<
Ansuel >
well i really need to check this... would be ideal to use 2.7 and just fix the BDF
14:09
<
robimarko >
I actually have the 8devices Mango DVK
14:09
<
robimarko >
So, I will eventually try it on the device
14:09
<
Ansuel >
last question do you have the serial pinout of the linksys?
14:11
<
Ansuel >
curious what the other pins are used (unpopulated on my board)
14:12
<
robimarko >
That should be JTAG (If you are talking about the dual row header)
14:14
rua has quit [Quit: Leaving.]
14:16
<
Ansuel >
btw at the end i manage to find the cause for the arm decompressor and it seems upstream they liked the fix
14:16
<
robimarko >
That is great to hear, what is the idea?
14:17
<
Ansuel >
with DTB_COMPAT mem_start call to validate (and fix the address) is skipped as it could be augumented by ATAGs but it was never called again
14:17
<
Ansuel >
resulting in ignoring completely the memory node data in the appended DTB
14:18
<
Ansuel >
with this and a less drastic patch to ignore MEM ATAGs AUTO_Z can work correctly and the data from DTB is used
14:18
<
robimarko >
That is great to hear
14:20
<
robimarko >
None of the commits have description
14:20
<
robimarko >
Though the code actually looks OK
14:21
<
Ansuel >
been like 4 email i'm sending to lui offering for help... they always answer with "We will review internally"
14:21
<
Ansuel >
small details that net-next is closed (or isn't?)
14:21
<
robimarko >
Its closed
14:22
<
Ansuel >
they will get angry again i guess
14:22
<
robimarko >
Think they open tommorow, but lack of commit messages is no go
14:22
<
robimarko >
But with this series, IPQ5018 is basically usable and has better networking driver than anything
14:22
<
Ansuel >
finally... i have to send the last 3 series for at803x so we can finally detach everything from ssdk (aside from aqr)
14:23
<
robimarko >
Perfect
14:23
<
Ansuel >
btw they said they will "consider" working on an upstream version of EDMAv1
14:23
<
robimarko >
Sure, I will hold my breath for it
14:23
<
Ansuel >
after EDMAv2 is upstream
14:27
<
robimarko >
Look at the IPQ807x UNIPHY diagram
14:27
<
Ansuel >
yep was the first thing i notice
14:27
<
Ansuel >
more stuff to detach from ssdk
14:27
<
robimarko >
One day
14:28
<
Ansuel >
they added only ipq50xx stuff
14:28
<
Ansuel >
am i wrong...
14:28
<
robimarko >
Yeah, only ipq50xx
14:28
<
Ansuel >
also ipq60xx is the same ?
14:28
<
robimarko >
ipq60xx is basically the same uniphy as ipq807x
14:29
<
robimarko >
Minus the
14:29
<
robimarko >
*Minus the UNIPHY2
14:30
<
mrkiko >
and is there a plan by them tocatc up for the lost performance with nss in the whole plan? Or am I missing something here?
14:30
<
robimarko >
And the UNIPHY1 is SGMII+ max, no 10G
14:31
<
Ansuel >
mrkiko nope... their plan with new ipq95xx was trash nss and move to correct PPE
14:31
<
robimarko >
mrkiko: Nothing can replace that perfmormance, but we are finally seeing some upstreaming of basic wired networking components here
14:32
<
mrkiko >
thanks. Very nice. I wouldn't have expected to see this happening.
14:45
rua has joined #openwrt-devel
15:16
<
robimarko >
Ansuel: Well, QCA geniouses made the UNIPHY driver stupid
15:16
<
robimarko >
They just hardcode values in the DTS
15:17
<
robimarko >
It is not aware of the required MAC mode
15:17
<
robimarko >
So, the frequency is just hardcoded in the DTS
15:17
<
robimarko >
So, its not suitable for on the fly MAC mode changes like for example happens when AQR-s are used
15:22
<
Ansuel >
maybe they want to expand it later?
15:35
jeff___m has joined #openwrt-devel
15:36
<
robimarko >
Doubt that, it would make sense to do it now
15:36
<
robimarko >
But that would require wiring the driver into phylink as a proper PCS one
15:40
<
robimarko >
What happened?
15:40
<
Ansuel >
no i mean it way too much work
15:41
<
robimarko >
It shouldnt be as there are existing PCS drivers doing exactly that
15:41
<
Ansuel >
CBT U-Boot ver: 2.3.01 ([IPQ6018].[SPF11.CSU1].[code version])
15:41
<
Ansuel >
very new uboot feel strange
15:41
<
robimarko >
Whats strange, it hasnt really changed functionally in ages
15:42
<
robimarko >
I am using SPF12.0 CSU1 for a board we support, had to enable lots of things to make it usable
15:42
<
robimarko >
U-Boot without long help is so weird
15:43
<
Ansuel >
mhhh should i flash the image on the mr?
15:44
<
robimarko >
It should work, but I havent tested it
15:44
jeff___m has quit [Ping timeout: 480 seconds]
15:46
<
Ansuel >
Unknown command 'tftpboot' - try 'help' .... usual linksys
15:46
<
robimarko >
Use dhcp
15:46
<
Ansuel >
nha for some reason they always rename tftpboot with tftp
15:48
<
Ansuel >
can confirm the image works
15:48
<
robimarko >
There is one thing missing from the PR, CPUFreq NVMEM needs to be backported
15:49
<
Ansuel >
mhhh maybe we can for now polish it a bit and set is as disabled to suspend buildbot image generation
15:50
<
robimarko >
Yes, I would just clean the obviously wrong things
15:50
<
robimarko >
And merge it marked as source
15:50
<
robimarko >
Then we can add couple of boards and fixup things along the way
15:50
<
robimarko >
And then its easy to remove the source marking
15:50
<
robimarko >
Otherwise it will drag on for a while
15:51
<
robimarko >
I have couple of boards and there will be a good number of users
15:51
<
Ansuel >
Wrong Image Format for bootm command
15:51
<
Ansuel >
ERROR: can't get kernel image!
15:51
<
Ansuel >
(treid sysupgrade from initramfs)
15:52
<
robimarko >
Honestly, no idea if sysupgrade was even wired properly
15:53
<
robimarko >
rmilecki: any news on the u-boot env parser being able to parse MAC-s from custom cell names?
15:53
<
Ansuel >
ubimkvol: error!: UBI device does not have free logical eraseblocks
15:54
<
robimarko >
Check with ubinfo
15:54
<
robimarko >
Linksys probably has QSDK UBI volumes that are not removed by default
15:54
<
Ansuel >
guess some fixup are needed in sysupgrad
15:55
<
robimarko >
If its just UBI volumes, you can find examples of doing that
15:55
<
robimarko >
Like I said, havent tried flashing as it was just a dev device
15:56
<
Ansuel >
btw news for the rootfs options thing?
15:57
<
robimarko >
I rebased my stuff, but havent retested
15:57
<
robimarko >
Its been on the TODO list for a while
15:57
<
Ansuel >
boot_part=2 nice same dual partition way
16:06
<
Ansuel >
[e]Failed to get erase block status
16:06
<
Ansuel >
MHHHHHHHHHHHHHHHHH
16:07
mentalow has quit [Quit: :]]
16:08
mentalow has joined #openwrt-devel
16:10
<
Ansuel >
well kernel partition is 8mb....
16:15
rua has quit [Quit: Leaving.]
16:17
rua has joined #openwrt-devel
16:29
<
Ansuel >
robimarko do you have the BDF ?
16:29
<
Ansuel >
with the upstream one it seems to work
16:30
<
Ansuel >
ok no actually it crashed bad on me connecting to it
16:30
<
Ansuel >
NOC_error.c:474 NOCError: FATAL ERRORparam0 :zero, param1 :zero, param2 :zero.
16:36
jeff___m has joined #openwrt-devel
16:44
jeff___m has quit [Ping timeout: 480 seconds]
17:23
jeff___m has joined #openwrt-devel
17:26
<
Ansuel >
ok it doesn't seems to be a problem with BDF
17:26
<
Ansuel >
but more or instability?
17:31
jeff___m has quit [Ping timeout: 480 seconds]
17:38
jeff___m has joined #openwrt-devel
17:46
jeff___m has quit [Ping timeout: 480 seconds]
18:27
jeff___m has joined #openwrt-devel
18:27
<
robimarko >
I did not want to submit it until it was all working
18:27
<
Ansuel >
well from what i can see 2.7 crash sure but correctly recovers
18:27
<
Ansuel >
the BDF works correctly
18:28
<
Ansuel >
I also sent a mail to kalle
18:38
<
Ansuel >
I'm temted to merge patches and ess dtsi to have the ball rolling for ipq60xx
18:39
<
Ansuel >
but no device or anything
18:39
<
robimarko >
I am fine with that, but I would drop all of the placeholder scripts he included
18:39
<
robimarko >
I think I even commented regarding that
18:39
<
robimarko >
Those should be added as the devices need them
18:40
<
Ansuel >
for sure i have to edit some commits have to see
18:44
<
robimarko >
I am all for adding the target, we can then fixup individual compoments easily
19:50
jeff___m has quit [Remote host closed the connection]
21:58
skynet2 has quit [Quit: Leaving]
22:04
skynet2 has joined #openwrt-devel
23:23
goliath has quit [Quit: SIGSEGV]
23:41
torv has quit [Remote host closed the connection]
23:41
torv has joined #openwrt-devel
23:47
torv has quit [Remote host closed the connection]
23:47
torv has joined #openwrt-devel
23:54
gch981213 has joined #openwrt-devel