Possible issue with tifm_7xx1 and interrupt latency

Possible issue with tifm_7xx1 and interrupt latency

[Jonathan Woithe]

Although somewhat circumstancial, during the debugging of a FFADO issue with
a user it seemed possible that the tifm_7xx1 module may have been switching
interrupts off for long periods of time.  For a full history of the problem
please refer to

  http://subversion.ffado.org/ticket/250

Unfortunately the original reporter no longer has access to the hardware
concerned and the report itself dates from 2 years ago (I kept forgetting
to report it).  However, it may be useful none-the-less.

Please CC me any replies.

Thanks
  jonathan

Re: Possible issue with tifm_7xx1 and interrupt latency

[Alex Dubov]

----- Original Message -----
> From: Jonathan Woithe <jwoithe@just42.net>
> 
> Although somewhat circumstancial, during the debugging of a FFADO issue with
> a user it seemed possible that the tifm_7xx1 module may have been switching
> interrupts off for long periods of time.  For a full history of the problem
> please refer to
> 
>   http://subversion.ffado.org/ticket/250
> 
> Unfortunately the original reporter no longer has access to the hardware
> concerned and the report itself dates from 2 years ago (I kept forgetting
> to report it).  However, it may be useful none-the-less.

Unfortunately, I don't have access to that hardware either.

However, there are really only few operations in that driver between "irqsave"
and "irqrestore", unless PIO mode is used (which may, ostensibly, cause this
sort of interrupt delays). On the other hand, PIO mode there is purely intended
for troubleshooting purposes and can only be enabled through special module
parameter by an adventurous user.

> 
> Please CC me any replies.
> 
> Thanks
>   jonathan

Regards,
Alex