[ath9k-devel] post kernel 4.3.x regressions in ath9k_htc

[ath9k-devel] post kernel 4.3.x regressions in ath9k_htc

[UsuarioAnonimo]

I've addressed my concerns previously to the dev mailing list. My case 
as stated previously is the following:

"My primary conntection to the internet has been the Alfa AWUS036NHA 
wireless adapter which uses the Atheros AR9271 802.11n chipset. The 
applicable linux driver/module and firmware is ath9k_htc. Ever since 
the upgrade from linux kernel 4.3.x to 4.4.x the throughput of the 
adapter chipset has been severely affected making the adapter basically 
unusable."

After packet sniffing with Wireshark I'm not sure it's a throughput 
issue, but rather a latency issue with dup_acks and spurious 
retransmissions resulting fairly quickly in malformed packets and 
reconnects as the throughput drops to basically zero. I wish I knew how 
to analyze packets better and make the connection to driver performance.

I'm still convinced the regressions were introduced between kernel 
4.3.x and 4.4.x. To that end I have done some investigating. It looks 
like there were a lot of changes to ath9k between versions 4.3 and 4.4 
that affect the AR9271 chipset. The changes seem to have to do with 
initialization and calling the firmware, and calling the development 
firmware.

One such change is:

An extra option is added in ath9k/htc_drv_init.c
   41 int htc_use_dev_fw = 0;
   42 module_param_named(use_dev_fw, htc_use_dev_fw, int, 0444);
   43 MODULE_PARM_DESC(use_dev_fw, "Use development FW version");

But the big changes are in these two files, hif_usb.c and hif_usb.h 
(diff output b/t vers. 4.3 and 4.4):
http://lxr.free-electrons.com/diff/drivers/net/wireless/ath/ath9k/hif_usb.c?v=4.3;diffval=4.4;diffvar=v
http://lxr.free-electrons.com/diff/drivers/net/wireless/ath/ath9k/hif_usb.h?v=4.3;diffval=4.4;diffvar=v

I was wondering if I'm on to something and if anyone with any knowledge 
would mind double-checking the changes signaled in the diff files to 
see if there could be a regression lurking in there.

Also, I am happy to run any tests and provide output and log files from 
my various pre-4.3.x kernel OS intalls, which still function perfectly 
well, and the 4.4.x and later OS installs, which are basically 
unusable. They start of fast, but, within a matter of 30 seconds or so, 
gum up and result in 100% packet loss.

Thank you very much for your attention and consideration.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20160831/6d688c4d/attachment.htm 

[ath9k-devel] post kernel 4.3.x regressions in ath9k_htc

[Oleksij Rempel]

Why do you starting describing same issue in a new email tread?

please go back to:
[ath9k-devel] ath9k_htc kernel driver regression affecting throughput


Am 31.08.2016 um 21:00 schrieb UsuarioAnonimo:
> I've addressed my concerns previously to the dev mailing list. My case
> as stated previously is the following:
> 
> "My primary conntection to the internet has been the Alfa AWUS036NHA
> wireless adapter which uses the Atheros AR9271 802.11n chipset. The
> applicable linux driver/module and firmware is ath9k_htc. Ever since the
> upgrade from linux kernel 4.3.x to 4.4.x the throughput of the adapter
> chipset has been severely affected making the adapter basically unusable."
> 
> After packet sniffing with Wireshark I'm not sure it's a throughput
> issue, but rather a latency issue with dup_acks and spurious
> retransmissions resulting fairly quickly in malformed packets and
> reconnects as the throughput drops to basically zero. I wish I knew how
> to analyze packets better and make the connection to driver performance.
> 
> I'm still convinced the regressions were introduced between kernel 4.3.x
> and 4.4.x. To that end I have done some investigating. It looks like
> there were a lot of changes to ath9k between versions 4.3 and 4.4 that
> affect the AR9271 chipset. The changes seem to have to do with
> initialization and calling the firmware, and calling the development
> firmware.
> 
> One such change is:
> 
> An extra option is added in ath9k/htc_drv_init.c
>    41 int htc_use_dev_fw = 0;
>    42 module_param_named(use_dev_fw, htc_use_dev_fw, int, 0444);
>    43 MODULE_PARM_DESC(use_dev_fw, "Use development FW version");
> 
> But the big changes are in these two files, hif_usb.c and hif_usb.h
> (diff output b/t vers. 4.3 and 4.4):
> http://lxr.free-electrons.com/diff/drivers/net/wireless/ath/ath9k/hif_usb.c?v=4.3;diffval=4.4;diffvar=v
> http://lxr.free-electrons.com/diff/drivers/net/wireless/ath/ath9k/hif_usb.h?v=4.3;diffval=4.4;diffvar=v
> 
> I was wondering if I'm on to something and if anyone with any knowledge
> would mind double-checking the changes signaled in the diff files to see
> if there could be a regression lurking in there.
> 
> Also, I am happy to run any tests and provide output and log files from
> my various pre-4.3.x kernel OS intalls, which still function perfectly
> well, and the 4.4.x and later OS installs, which are basically unusable.
> They start of fast, but, within a matter of 30 seconds or so, gum up and
> result in 100% packet loss.
> 
> Thank you very much for your attention and consideration.
> 
> 
> _______________________________________________
> ath9k-devel mailing list
> ath9k-devel at lists.ath9k.org
> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
> 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 213 bytes
Desc: OpenPGP digital signature
Url : http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20160901/b555e49b/attachment.pgp