Skip to contentSkip to conversation

Question on beacon-miss handling.

DISCUSSION Ben Greear 4 participants 10 messages Jul 5, 2014 – Jul 7, 2014 Qualcomm Atheros ATH10K wireless driver

lore
Share & tools
lore

Ben Greear

author
#0 lore raw

While poking around in the firmware, I notice that ath10k driver only requests a few vdevs to be supported by the beacon-miss firmware logic, though the driver supposedly supports more vdevs than that...

Any idea how this is supposed to work with more vdevs than bmiss_offload_max_vdev?

Maybe it is best to just disable beacon-miss logic in firmware entirely and let the driver/host handle it?

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
#1 lore raw

Where is it in the firmware again?

The beacon miss stuff is a little .. odd. Mostly because there's only one (and maybe two) timers for it and they're a pain in the ass to configure correctly. Having it scale (correctly) to more than two vdevs may be a little challenging.

-a

Quoted 19 lines · Ben Greear

While poking around in the firmware, I notice that ath10k driver only requests a few vdevs to be supported by the beacon-miss firmware logic, though the driver supposedly supports more vdevs than that...

Any idea how this is supposed to work with more vdevs than bmiss_offload_max_vdev?

Maybe it is best to just disable beacon-miss logic in firmware entirely and let the driver/host handle it?

Thanks,
Ben

--
Ben Greear <[email protected]>
Candela Technologies Inc http://www.candelatech.com

Quoted 4 lines

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

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

author
#2 lore raw
Quoted 1 line · Adrian Chadd

Where is it in the firmware again?

wlan_swbmiss_offload.c

Quoted 4 lines

The beacon miss stuff is a little .. odd. Mostly because there's only one (and maybe two) timers for it and they're a pain in the ass to configure correctly. Having it scale (correctly) to more than two vdevs may be a little challenging.

I've seen other funny stuff where an AP will crash hard(-ish?) and the associated stations take a long time to figure out their AP is dead. Could be related to this maybe..or maybe was something else. I have not investigated closely.

I'd be happy to have all the beacon miss stuff move to the driver..seems a waste of limited resources in the firmware.

Thanks,
Ben

Quoted 36 lines

-a

Quoted 19 lines · Ben Greear

While poking around in the firmware, I notice that ath10k driver only requests a few vdevs to be supported by the beacon-miss firmware logic, though the driver supposedly supports more vdevs than that...

Any idea how this is supposed to work with more vdevs than bmiss_offload_max_vdev?

Maybe it is best to just disable beacon-miss logic in firmware entirely and let the driver/host handle it?

Thanks,
Ben

--
Ben Greear <[email protected]>
Candela Technologies Inc http://www.candelatech.com

Quoted 4 lines

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

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

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
#3 lore raw
Quoted 14 lines · Ben Greear
Quoted 2 lines · Adrian Chadd

Where is it in the firmware again?

wlan_swbmiss_offload.c

Quoted 4 lines

The beacon miss stuff is a little .. odd. Mostly because there's only one (and maybe two) timers for it and they're a pain in the ass to configure correctly. Having it scale (correctly) to more than two vdevs may be a little challenging.

Quoted 7 lines

I've seen other funny stuff where an AP will crash hard(-ish?) and the associated stations
take a long time to figure out their AP is dead. Could be related to this maybe..or maybe was something else. I have not investigated closely.

I'd be happy to have all the beacon miss stuff move to the driver..seems a waste of limited resources in the firmware.

Think about it in terms of "minimising how much the host is woken up."

That's what the target market of the STA firmware for this chip is thinking about.

-a

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

#4 lore raw
Quoted 4 lines · Adrian Chadd

Think about it in terms of "minimising how much the host is woken up."

That's what the target market of the STA firmware for this chip is thinking about.

I don't think beacon filtering is enabled for ath10k, which means all beacons are sent to mac80211 anyway. HW/FW connection monitoring (in STA mode) is also not enabled in ath10k, so mac80211's beacon loss/miss infrastructure is already used, no ?

Sujith

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

#5 lore raw
Quoted 10 lines · Sujith Manoharan
Quoted 4 lines · Adrian Chadd

Think about it in terms of "minimising how much the host is woken up."

That's what the target market of the STA firmware for this chip is thinking about.

I don't think beacon filtering is enabled for ath10k, which means all beacons are sent to mac80211 anyway. HW/FW connection monitoring (in STA mode) is also not enabled in ath10k, so mac80211's beacon loss/miss infrastructure is already used, no ?

Sure, but my response was just describing the "why" behind why the firmware would want to be doing beacon offload processing.

-a

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

author
#6 lore raw
In reply to Sujith Manoharan
Quoted 10 lines · Sujith Manoharan
Quoted 4 lines · Adrian Chadd

Think about it in terms of "minimising how much the host is woken up."

That's what the target market of the STA firmware for this chip is thinking about.

I don't think beacon filtering is enabled for ath10k, which means all beacons are sent to mac80211 anyway. HW/FW connection monitoring (in STA mode) is also not enabled in ath10k, so mac80211's beacon loss/miss infrastructure is already used, no ?

Maybe the driver should make this 0 instead of 2 then?

hw.h:#define TARGET_10X_BMISS_OFFLOAD_MAX_VDEV 2

That would save resources in the firmware.

Thanks,
Ben

Quoted 3 lines

Sujith

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
#7 lore raw
Quoted 3 lines · Ben Greear

Maybe the driver should make this 0 instead of 2 then?
hw.h:#define TARGET_10X_BMISS_OFFLOAD_MAX_VDEV 2
That would save resources in the firmware.

Yes, it would. ath10k doesn't seem to be handling roam events too, so maybe that can be set to 0 - not sure if there are plans to properly enable and use roaming functionality that is already in the firmware.

Sujith

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

author
#8 lore raw
Quoted 9 lines · Sujith Manoharan
Quoted 3 lines · Ben Greear

Maybe the driver should make this 0 instead of 2 then?
hw.h:#define TARGET_10X_BMISS_OFFLOAD_MAX_VDEV 2
That would save resources in the firmware.

Yes, it would. ath10k doesn't seem to be handling roam events too, so maybe that can be set to 0 - not sure if there are plans to properly enable and use roaming functionality that is already in the firmware.

I have set both to zero when using my firmware, and so far, everything seems to be working fine.

I am tempted to just compile the roaming and beacon-miss code entirely out of the firmware to save a bit of memory, but I guess that might possibly break some driver, so if I do, I will have to add a new firmware image variant...

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
#9 lore raw
In reply to Sujith Manoharan
Quoted 9 lines · Sujith Manoharan
Quoted 3 lines · Ben Greear

Maybe the driver should make this 0 instead of 2 then?
hw.h:#define TARGET_10X_BMISS_OFFLOAD_MAX_VDEV 2
That would save resources in the firmware.

Yes, it would. ath10k doesn't seem to be handling roam events too, so maybe that can be set to 0 - not sure if there are plans to properly enable and use roaming functionality that is already in the firmware.

That is correct. ath10k does not currently support hw/fw conn monitoring. Things may change in future as we upgrade to support 10.2 firmware, but AFAIK there's no direct plan to use this offloads. And regarding the initialization of BMISS/ROAM_OFFLOADs, yes looks like this can be erased to save some firmware memory.

-Bartosz

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