[mythtv-users] Long delays with remote occasionally

[Adam Skinner]

> > On Tue, Jun 9, 2009 at 5:15 AM, Adam Skinner<kingmoffa at gmail.com> >
> > wrote:
> > Upgrading to the new nvidia drivers didn't really solve anything.
> >
> > My problem is mostly when moving around the menus and when more disc
> > activity is happening.
> >
> > Looking at:
> >
> > http://www.lirc.org/html/technical.html#bugs
> >
> > Im wondering if its an interrupt issue.
> >
> > I have a pair of SATA drives in the box - using hdparm -u1 /dev/sdX
> > results in the following error msg:
> >
> > HDIO_GET_UNMASKINTR failed: Inappropriate ioctl for device
> >
> > Anoyone know how to unmask (or get the status of) an interrupt for a
> > SATA drive?
> >
> 
> 
> 
> There are three causes.  Josh solved it and it is documented in the
> middle
> of this thread:
> http://www.gossamer-threads.com/lists/engine?do=post_view_flat;post=347546;page=3;sb=post_latest_reply;so=ASC;mh=25;list=mythtv
> 
> There is a setting, a version of fixes, and a patch that Josh wrote.
> 


Thanks for the info -  parts 1 and 2 dont seem applicable (running
latest last week 0.21 svn fixes with nvidia use events). 

Part 3 using Josh's patch seems interesting...


> I'm fairly certain that's not related.  I had this issue in a big
> way...with several minute pauses occasionally.  In my case the big
> factor was the video probing corrected by Josh's patch.  I have a
> serial receiver as well.
> 
> I can say almost for sure that it's _not_ related to lirc, as in my
> case and every other I've heard, the keyboard was blocked as well.
> About everyone seems to have noticed it specifically after upgrading
> to xorg 1.5...I've always suspected that that's related.
> 
> As far as the the hdparm unmask interrupt thing, I used to use that on
> my backend IDE drives (under Gentoo).  I recently upgraded my machines
> to the 2.6.28 kernel and switched IDE drivers from the old IDE drivers
> to the new libata drivers (the new SATA/PATA drivers).  They don't
> support that interrupt unmask at all, and I don't think they plan to.
> I believe that was only a factor in very old hardware.  As a matter of
> fact they currently don't support 32 bit mode either, though that's
> not considered important either.
> 
> I can tell you for sure that my backend performance is _noticeably_
> better since that kernel upgrade to the new drivers, so I doubt that
> interrupt, or anything having to do with serial lirc specifically, is
> your problem.
> 
> 
Thanks for the info - was clutching as straws with interrupt unmask as I
kinda had a feeling it wasn't used anymore.

When things do get slow navigating the menus with my lirc serial remote
- I can use my iPhone that talks to to the frontend via an app using
telnet. This is so much more responsive. Unsure about response with
keyboard as I dont have one hooked up often and its usually bluetooth
ifs so.

I have a few FEs - they all use lirc serial with the same home brew
receiver (from ebay) and remote (hp media centre) with no problems. I
will switch the receiver and remote for known good ones just to be safe
- but not expecting anything.