Netgear R7800 exploration (IPQ8065, QCA9984)

johnnysl · January 27, 2017, 11:18am

seeing another issue with the ath10k driver. Anyone else seen this as well?:
(or know where to properly let the maintainers know?)

[432063.937009] ath10k_pci 0001:01:00.0: received unexpected tx_fetch_ind event: in push mode
[444459.011125] ath10k_pci 0000:01:00.0: firmware crashed! (uuid 3ba8fc67-fd2a-48dd-99a7-78ee3bab0a54)
[444459.011238] ath10k_pci 0000:01:00.0: qca9984/qca9994 hw1.0 target 0x01000000 chip_id 0x00000000 sub 168c:cafe
[444459.019025] ath10k_pci 0000:01:00.0: kconfig debug 0 debugfs 1 tracing 0 dfs 1 testmode 1
[444459.035241] ath10k_pci 0000:01:00.0: firmware ver 10.4-3.3-00102 api 5 features no-p2p,mfp,peer-flow-ctrl,btcoex-param crc32 6120b2da
[444459.037475] ath10k_pci 0000:01:00.0: board_file api 1 bmi_id N/A crc32 5f733fb2
[444459.049410] ath10k_pci 0000:01:00.0: htt-ver 2.2 wmi-op 6 htt-op 4 cal file max-sta 512 raw 0 hwcrypto 1
[444459.058989] ath10k_pci 0000:01:00.0: firmware register dump:
[444459.066443] ath10k_pci 0000:01:00.0: [00]: 0x0000000A 0x00000000 0x0099F301 0x00000000
[444459.072143] ath10k_pci 0000:01:00.0: [04]: 0x00000000 0x00000000 0x00000000 0x00000000
[444459.079955] ath10k_pci 0000:01:00.0: [08]: 0x00000000 0x00000000 0x00000000 0x00000000
[444459.087941] ath10k_pci 0000:01:00.0: [12]: 0x00000000 0x00000000 0x00000000 0x00000000
[444459.095928] ath10k_pci 0000:01:00.0: [16]: 0x00000000 0x00000000 0x00000000 0x0099F301
[444459.103933] ath10k_pci 0000:01:00.0: [20]: 0x00000000 0x00401B40 0x00000000 0x00000000
[444459.111901] ath10k_pci 0000:01:00.0: [24]: 0x00000000 0x00000000 0x00000000 0x00000000
[444459.119887] ath10k_pci 0000:01:00.0: [28]: 0x00000000 0x00000000 0x00000000 0x00000000
[444459.127871] ath10k_pci 0000:01:00.0: [32]: 0x00000000 0x00000000 0x00000000 0x00000000
[444459.135859] ath10k_pci 0000:01:00.0: [36]: 0x00000000 0x00000000 0x00000000 0x00000000
[444459.143845] ath10k_pci 0000:01:00.0: [40]: 0x00000000 0x00000000 0x00000000 0x00000000
[444459.151829] ath10k_pci 0000:01:00.0: [44]: 0x00000000 0x00000000 0x00000000 0x00000000
[444459.159817] ath10k_pci 0000:01:00.0: [48]: 0x00000000 0x00000000 0x00000000 0x00000000
[444459.167802] ath10k_pci 0000:01:00.0: [52]: 0x00000000 0x00000000 0x00000000 0x00000000
[444459.175788] ath10k_pci 0000:01:00.0: [56]: 0x00000000 0x00000000 0x00000000 0x00000000
[444459.287312] ieee80211 phy0: Hardware restart was requested
[444461.763924] ath10k_pci 0000:01:00.0: device successfully recovered

hnyman · January 27, 2017, 11:43am

[quote="johnnysl, post:21, topic:285"]
seeing another issue with the ath10k driver. Anyone else seen this as well?:(or know where to properly let the maintainers know?)
[/quote]Haven't seen it on my device.

The correct place to report it is in
https://bugs.lede-project.org/

dissent1 · January 27, 2017, 2:53pm

Try this https://github.com/dissent1/r7800/commit/9edae9438b5bf7792434fa8ade24307aa7637ea5

johnnysl · January 27, 2017, 4:08pm

Thanks, will have a look

johnnysl · January 27, 2017, 4:10pm

ehm.i thought this would be an ath10k issue, requiring feedback directly to the blob developers.

will test the new blob that dissent1 pointed out first, if that fixes it, we can do a simple pull request.

komawoyo · January 27, 2017, 5:14pm

You may also look at this: https://github.com/paul-chambers/netgear-r7800/tree/eeac2e10190f6f45e32e4c7012c4babc351898d8/package/qca-ssdk if you are interested in Hardware NAT

johnnysl · January 28, 2017, 7:58pm

Ok, replaced the ath10k Makefile with yours, but the new ath10k firmware failed to load on boot. Have currently swapped back to default Lede build again.

dissent1 · January 29, 2017, 8:00am

Weird stuff, I'm testing this new firmware for 3-4 days already without issues. dd-wrt has also switched to this one, because it has 80+80 mode fixed.

johnnysl · January 29, 2017, 2:56pm

Quite strange, dmesg showed that the driver failed to load the firmware, as if it was the wrong type or something.
only thing i did was a rm Makefile and a wget Makefile from your github and recompile.

dissent1 · January 30, 2017, 11:42am

What are your wireless settings?

johnnysl · January 30, 2017, 12:01pm

That doesn't matter as it already fails during the boot phase when the driver gets loaded.

but european country,
2.4ghz: channel 6, 17dbm, 20mhz
5ghz: channel 120, 20dbm, 80mhz AC

Will try to test some more later this week.

dissent1 · January 30, 2017, 1:44pm

On working router try to manually replace latest firmware-5.bin (3.4-00068) and board-2.bin in /lib/firmware/ath10k/... folder and reboot

https://source.codeaurora.org/quic/qsdk/oss/firmware/ath10k-firmware/tree/ath10k/QCA9984/hw1.0

Have you tried other channels, country? Maybe it has smth to do with dfs feature failing during firmware initialisation?

johnnysl · January 30, 2017, 4:42pm

Think i found the issue. a wget of the makefile gives me a file with 2400 lines somehow, where i should get 302 lines.
so i've now manually cut and paste the differences, and recompiled. Lets see how that works.

hnyman · January 30, 2017, 4:53pm

You can also download any Github commit by just adding ".patch" at the end:

wget https://github.com/dissent1/r7800/commit/9edae9438b5bf7792434fa8ade24307aa7637ea5.patch

You can apply that patch easily. No need for manual editing...

enri · February 1, 2017, 3:47am

@hnyman
in choosing your new router, what factor make you choose R7800 over WRT1900ACS?
is it related to ath10 over mlwifi?

i am still able to send back my wrt... seems your R7800 have more fun

Edit: Do you have any figure about USB 3 write speed? Are they the same like the small net builder mentioned, 55MBps?

hnyman · February 1, 2017, 7:04am

I havenot tested wrt1900 series so no real comparison on details. You can read more of my device selection from

enri · February 1, 2017, 7:16am

Thanks

I have read that post as well, and come up with a conclusion I should go for R7800.

Asking the previous questions is just a final check to see if there is any glithces...

Talking about the USB, it should be fast. Just curious on real world first hand speed.

johnnysl · February 1, 2017, 9:49am

Thanks for the tip.

The newer firmware now works great.
Any chance we could still get this newer firmware into Lede 17.01?

johnnysl · February 1, 2017, 6:19pm

so the newer firmware is now in trunk, but over the day i still see some cosmetic errors that would be fixed by this new firmware:

[ 20.864003] ath10k_pci 0001:01:00.0: firmware ver 10.4-3.4-00068 api 5 features no-p2p,mfp,peer-flow-ctrl,btcoex-param crc32 fb5bee97
[ 6742.486101] ath10k_pci 0001:01:00.0: received unexpected tx_fetch_ind event: in push mode
[ 6742.486151] ath10k_pci 0001:01:00.0: received unexpected tx_fetch_ind event: in push mode
[ 9799.748258] ath10k_pci 0001:01:00.0: received unexpected tx_fetch_ind event: in push mode
[ 9799.748329] ath10k_pci 0001:01:00.0: received unexpected tx_fetch_ind event: in push mode
[ 9799.755500] ath10k_pci 0001:01:00.0: received unexpected tx_fetch_ind event: in push mode
[ 9799.763722] ath10k_pci 0001:01:00.0: received unexpected tx_fetch_ind event: in push mode

guess they still have some work to do

dissent1 · February 1, 2017, 6:41pm

Could you apply this to your tree
https://git.lede-project.org/?p=lede/nbd/staging.git;a=commit;h=f2a75b5b5867910dbe14bf7366fd9c4ebd0880d2