Skip to contentSkip to conversation

Anyone else see host machine lockups?

DISCUSSION Ben Greear 3 participants 5 messages Apr 18, 2014 – Apr 22, 2014 Qualcomm Atheros ATH10K wireless driver

lore
Share & tools
lore

Ben Greear

author
#0 lore raw

We can reliably lock up a host system (no splats, no mention of firmware crashes, system just dies) when we bring up or down a bunch of vifs in a system with 3 radios in a PC.

Anyone else seeing similar?

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

Heh, no, but back in the day this'd happen if the DMA engine wasn't shut down in the right order. The NIC would start DMAing crap all over random memory.

I wonder if that's happening here.

-a

Quoted 14 lines · Ben Greear

We can reliably lock up a host system (no splats, no mention of firmware crashes, system just dies) when we bring up or down a bunch of vifs in a system with 3 radios in a PC.

Anyone else seeing similar?

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 5 lines · Adrian Chadd

Heh, no, but back in the day this'd happen if the DMA engine wasn't shut down in the right order. The NIC would start DMAing crap all over random memory.

I wonder if that's happening here.

Maybe, but I would hope that such a problem would at least often crash the OS with some strange crashes as opposed to just silent host lockup?

Thanks,
Ben

Quoted 25 lines

-a

Quoted 18 lines · Ben Greear

We can reliably lock up a host system (no splats, no mention of firmware crashes, system just dies) when we bring up or down a bunch of vifs in a system with 3 radios in a PC.

Anyone else seeing similar?

Thanks,
Ben

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

_______________________________________________
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
In reply to Ben Greear
Quoted 5 lines · Ben Greear

We can reliably lock up a host system (no splats, no mention of firmware crashes, system just dies) when we bring up or down a bunch of vifs in a system with 3 radios in a PC.

Anyone else seeing similar?

From what I understand this seems to be faulty CUS hardware design. The problem occurs when trying to perform cold reset on target device which sometimes leads to voltage drops/spikes which in turn breaks PCI-E link. Some host controllers apparently can't deal with it and hang the whole machine (e.g. Thinkpad T430), while others just lose connectivity with the device and need re-plugging (e.g. Dell E6420) or display a splat and need to be restarted before given PCI-E slot can be used again (e.g. AP135's "Data bus error").

Michal

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

author
#4 lore raw
Quoted 7 lines · Michal Kazior
Quoted 5 lines · Ben Greear

We can reliably lock up a host system (no splats, no mention of firmware crashes, system just dies) when we bring up or down a bunch of vifs in a system with 3 radios in a PC.

Anyone else seeing similar?

Quoted 8 lines
Quoted 1 line

From what I understand this seems to be faulty CUS hardware design.

The problem occurs when trying to perform cold reset on target device which sometimes leads to voltage drops/spikes which in turn breaks PCI-E link. Some host controllers apparently can't deal with it and hang the whole machine (e.g. Thinkpad T430), while others just lose connectivity with the device and need re-plugging (e.g. Dell E6420) or display a splat and need to be restarted before given PCI-E slot can be used again (e.g. AP135's "Data bus error").

I see this on WLE900VX NICs as well, so it's not just a CUS issue.

And, I don't think it is just cold-reset, because I do not see any console messages about attempting a cold reset.

Thanks,
Ben

Quoted 4 lines

Michal

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