Skip to contentSkip to conversation

Anyone brought up 9984 NIC on x86-64?

DISCUSSION Ben Greear 4 participants 7 messages Jul 13, 2016 Qualcomm Atheros ATH10K wireless driver

lore
Share & tools
lore

Ben Greear

author
#0 lore raw

I added a few patches to my 4.7 kernel to enable ath10k to load for 9984, but it immediately crashes due to IOMMU error when trying to allocate memory.

I'm building the latest linux-ath tree now and will test that when it is done, but I am curious to know if anyone has successfully booted 9984 on x86-64?

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
Quoted 11 lines · Ben Greear

I added a few patches to my 4.7 kernel to enable ath10k to load for 9984, but it immediately crashes due to IOMMU error when trying to allocate memory.

I'm building the latest linux-ath tree now and will test that when it is done, but I am curious to know if anyone has successfully booted 9984 on x86-64?

Thanks,
Ben

Today's stock linux-ath and stock 9984 firmware with 9984 NIC crashes on bootup as well :(

How completely frustrating!

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
#2 lore raw
In reply to Ben Greear
Quoted 8 lines

I added a few patches to my 4.7 kernel to enable ath10k to load for 9984, but it immediately crashes due to IOMMU error when trying to allocate memory.

I'm building the latest linux-ath tree now and will test that when it is done, but I am curious to know if anyone has successfully booted 9984 on x86-64?

Ben,

Are you seeing IOMMU error even after wtih your DMA_BIDIRECTIONAL change? Could you please try with

https://patchwork.kernel.org/patch/9175029/

But increase WMI_MAX_MEM_CHUNK_SIZE to 512.

-Rajkumar

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

author
#3 lore raw
Quoted 11 lines · Rajkumar
Quoted 8 lines

I added a few patches to my 4.7 kernel to enable ath10k to load for 9984, but it immediately crashes due to IOMMU error when trying to allocate memory.

I'm building the latest linux-ath tree now and will test that when it is done, but I am curious to know if anyone has successfully booted 9984 on x86-64?

Ben,

Are you seeing IOMMU error even after wtih your DMA_BIDIRECTIONAL change? Could you please try with

My BIDIRECTIONAL patch is in my 4.7, and that did not fix the issue.

It is more for the IOMMU access failures that I saw and the redhat user reported.

I would appreciate it if you could roll that patch into your testing set and make sure you don't see regressions on your systems. That way maybe it can go upstream.

For reference, this is today's un-patched linux-ath:

DMA: Out of SW-IOMMU space for 770224 bytes at device 0000:05:00.0

Kernel panic - not syncing: DMA: Random memory could be DMA read

CPU: 5 PID: 288 Comm: kworker/u16:3 Not tainted 4.7.0-rc7-wt-ath+ #34
Hardware name: To be filled by O.E.M. To be filled by O.E.M./ChiefRiver, BIOS 4.6.5 06/07/2013
Workqueue: ath10k_aux_wq ath10k_wmi_event_service_ready_work [ath10k_core]
 0000000000000000 ffff8800d3edfbf0 ffffffff8140abd1 ffffffff81c88b78
 ffff8800d3edfc80 ffff8800d3edfc70 ffffffff811ea260 ffff880000000008
 ffff8800d3edfc80 ffff8800d3edfc18 ffffffff811ea591 ffff88021e34db68
Call Trace:
 [<ffffffff8140abd1>] dump_stack+0x63/0x82
 [<ffffffff811ea260>] panic+0xca/0x216
 [<ffffffff811ea591>] ? printk+0x43/0x4b
 [<ffffffff814382a2>] swiotlb_map_page+0x1a2/0x1d0
 [<ffffffffa07ea1a9>] ath10k_wmi_event_service_ready_work+0x3f9/0x750 [ath10k_core]
 [<ffffffff8113667a>] ? dequeue_entity+0x26a/0xba0
 [<ffffffff8111aebd>] process_one_work+0x15d/0x4b0
 [<ffffffff8111b253>] worker_thread+0x43/0x4e0
 [<ffffffff8111b210>] ? process_one_work+0x4b0/0x4b0
 [<ffffffff81120a34>] kthread+0xc4/0xe0
 [<ffffffff81853e9f>] ret_from_fork+0x1f/0x40
 [<ffffffff81120970>] ? kthread_worker_fn+0x180/0x180
Kernel Offset: disabled
---[ end Kernel panic - not syncing: DMA: Random memory could be DMA read

I tried with your patch set to 512 block size, but that fails with similar DMA error and reboots.

Your original patch with 256 block size boots up the firmware OK on my system.

I guess I'll do more testing with your original 256k chunk patch and see how that works.

Thanks,
Ben

Quoted 12 lines

https://patchwork.kernel.org/patch/9175029/

But increase WMI_MAX_MEM_CHUNK_SIZE to 512.

-Rajkumar

_______________________________________________
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
#4 lore raw
Quoted 7 lines · Ben Greear

For reference, this is today's un-patched linux-ath:

DMA: Out of SW-IOMMU space for 770224 bytes at device 0000:05:00.0

Kernel panic - not syncing: DMA: Random memory could be DMA read

Hi,

So I looked at this code in a bit more detail. Ideally for large requests you shouldn't be asking for separate bounce buffers, you'd just allocate memory that was within the 32 bit physmem boundary so you don't /need/ a bounce buffer.

That's likely what's tripping things up.

Now, on freebsd I'd just request memory that was under 32 bit and the memory allocator would fail the allocation if it wasn't available. Bounce buffers get used for general system things like mbufs (our skbs), disk buffers, etc that can come from any memory and need a temporary mapping. This kind of stuff isn't temporary.

So to save me having to dig into what the linux kernel allocator provides (since I'm really a freebsd developer at heart!), what's the "right" way to maintain large, non-ephermal memory allocation that may have to do DMA? eg, like descriptor rings, etc.

-adrian

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

author
#5 lore raw
In reply to Ben Greear
Quoted 50 lines · Ben Greear
Quoted 11 lines · Rajkumar
Quoted 8 lines

I added a few patches to my 4.7 kernel to enable ath10k to load for 9984, but it immediately crashes due to IOMMU error when trying to allocate memory.

I'm building the latest linux-ath tree now and will test that when it is done, but I am curious to know if anyone has successfully booted 9984 on x86-64?

Ben,

Are you seeing IOMMU error even after wtih your DMA_BIDIRECTIONAL change? Could you please try with

My BIDIRECTIONAL patch is in my 4.7, and that did not fix the issue.

It is more for the IOMMU access failures that I saw and the redhat user reported.

I would appreciate it if you could roll that patch into your testing set and make sure you don't see regressions on your systems. That way maybe it can go upstream.

For reference, this is today's un-patched linux-ath:

DMA: Out of SW-IOMMU space for 770224 bytes at device 0000:05:00.0

Kernel panic - not syncing: DMA: Random memory could be DMA read

CPU: 5 PID: 288 Comm: kworker/u16:3 Not tainted 4.7.0-rc7-wt-ath+ #34
Hardware name: To be filled by O.E.M. To be filled by O.E.M./ChiefRiver, BIOS 4.6.5 06/07/2013
Workqueue: ath10k_aux_wq ath10k_wmi_event_service_ready_work [ath10k_core]
  0000000000000000 ffff8800d3edfbf0 ffffffff8140abd1 ffffffff81c88b78
  ffff8800d3edfc80 ffff8800d3edfc70 ffffffff811ea260 ffff880000000008
  ffff8800d3edfc80 ffff8800d3edfc18 ffffffff811ea591 ffff88021e34db68
Call Trace:
  [<ffffffff8140abd1>] dump_stack+0x63/0x82
  [<ffffffff811ea260>] panic+0xca/0x216
  [<ffffffff811ea591>] ? printk+0x43/0x4b
  [<ffffffff814382a2>] swiotlb_map_page+0x1a2/0x1d0
  [<ffffffffa07ea1a9>] ath10k_wmi_event_service_ready_work+0x3f9/0x750 [ath10k_core]
  [<ffffffff8113667a>] ? dequeue_entity+0x26a/0xba0
  [<ffffffff8111aebd>] process_one_work+0x15d/0x4b0
  [<ffffffff8111b253>] worker_thread+0x43/0x4e0
  [<ffffffff8111b210>] ? process_one_work+0x4b0/0x4b0
  [<ffffffff81120a34>] kthread+0xc4/0xe0
  [<ffffffff81853e9f>] ret_from_fork+0x1f/0x40
  [<ffffffff81120970>] ? kthread_worker_fn+0x180/0x180
Kernel Offset: disabled
---[ end Kernel panic - not syncing: DMA: Random memory could be DMA read
Quoted 8 lines

I tried with your patch set to 512 block size, but that fails with similar DMA error and reboots.

Your original patch with 256 block size boots up the firmware OK on my system.

I guess I'll do more testing with your original 256k chunk patch and see how that works.

I just realized that I was booting with "intel_iommu=off", that is probably why I had to enable the 256k chunk patch.

And FYI, at least my 9984 firmware appears to blow up when passed multiple chunks...I'm still trying to figure out exactly why.

That said, would be nice if ath10k worked on systems w/out intel-iommu too...

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
#6 lore raw
In reply to Manoharan, Rajkumar

Am 13.07.2016 um 19:26 schrieb Manoharan, Rajkumar:

Quoted 8 lines
Quoted 8 lines

I added a few patches to my 4.7 kernel to enable ath10k to load for 9984, but it immediately crashes due to IOMMU error when trying to allocate memory.

I'm building the latest linux-ath tree now and will test that when it is done, but I am curious to know if anyone has successfully booted 9984 on x86-64?

Quoted 7 lines

Ben,

Are you seeing IOMMU error even after wtih your DMA_BIDIRECTIONAL change? Could you please try with

https://patchwork.kernel.org/patch/9175029/

But increase WMI_MAX_MEM_CHUNK_SIZE to 512.

while debugging i found out that it will allocate up to 800 kb here on 99xx chipsets. truncating this value in any way will definitly cause out of bound access conditions. so i assume the code is wrong. the firmware requests the size, so it should get it. otherwise unexpected behaviour is guaranteed

Sebastian

Quoted 8 lines

-Rajkumar

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

Signature · 17 lines
Mit freundlichen Grüssen / Regards

Sebastian Gottschall / CTO

NewMedia-NET GmbH - DD-WRT
Firmensitz:  Berliner Ring 101, 64625 Bensheim
Registergericht: Amtsgericht Darmstadt, HRB 25473
Geschäftsführer: Peter Steinhäuser, Christian Scheele
http://www.dd-wrt.com
email: [email protected]
Tel.: +496251-582650 / Fax: +496251-5826565


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