Skip to contentSkip to conversation

Repeatable crash of CT firmware (at least).

DISCUSSION Ben Greear 1 participant 2 messages Aug 6, 2015 – Aug 11, 2015 Qualcomm Atheros ATH10K wireless driver

lore
Share & tools
lore

Ben Greear

author
#0 lore raw

In case someone has time an interest, I would be curious if upstream firmware/kernels see this problem. Unfortunately, it takes a bit of setup to do this test (namely, constant tx source, and if doing this with something like ath9k, then you might need a variable attenuator).

Possibly just a very loud constant tx source would reproduce the bug as well, and possibly it would happen in AP mode, etc.

Channel 1, HT40

Configure ath10k in station mode, associate with ath10k AP, start transmitting frames (about 4kpps tx, 1kpps rx in my case)

Add a constant-transmit noise source at around -65 db on the same channel. This should cause ath10k station (and AP) to go silent.

Shortly after, the station system's ath10k firmware crashes due to assert in the firmware CE logic.

I know that a similar CE assert can happen if the host does bad things with the tx descriptors, so I am suspicious that is the case here. For instance, maybe firmware is still trying to transmit the frame when host tears down the station vdev??

Of course, this could all be due to bugs in CT firmware and/or my kernel patches...

Thanks,
Ben

Signature · 8 lines
Ben Greear <[email protected]>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k
author
#1 lore raw

As an update, stock upstream kernels and multiple different official firmware versions crash as well.

I've sent more details off to those that might be able to fix things.

Thanks,
Ben

Quoted 30 lines · Ben Greear

In case someone has time an interest, I would be curious if upstream firmware/kernels see this problem. Unfortunately, it takes a bit of setup to do this test (namely, constant tx source, and if doing this with something like ath9k, then you might need a variable attenuator).

Possibly just a very loud constant tx source would reproduce the bug as well, and possibly it would happen in AP mode, etc.

Channel 1, HT40

Configure ath10k in station mode, associate with ath10k AP, start transmitting frames (about 4kpps tx, 1kpps rx in my case)

Add a constant-transmit noise source at around -65 db on the same channel. This should cause ath10k station (and AP) to go silent.

Shortly after, the station system's ath10k firmware crashes due to assert in the firmware CE logic.

I know that a similar CE assert can happen if the host does bad things with the tx descriptors, so I am suspicious that is the case here. For instance, maybe firmware is still trying to transmit the frame when host tears down the station vdev??

Of course, this could all be due to bugs in CT firmware and/or my kernel patches...

Thanks,
Ben

Signature · 8 lines
Ben Greear <[email protected]>
Candela Technologies Inc  http://www.candelatech.com


_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k