Re: intel graphics loosing msi interrupt on subsequent starts

From: Alexey Shuvaev <shuvaev_at_physik.uni-wuerzburg.de>
Date: Tue, 5 May 2009 00:31:48 +0200
On Mon, May 04, 2009 at 02:05:41PM -0500, Robert Noland wrote:
> On Mon, 2009-05-04 at 20:40 +0200, Alexey Shuvaev wrote:
> > Hello all!
> > 
> > Sorry if it is already reported...
> > 
> > I have recently upgaded X server from 1.4  to 1.6
> > (ports from ~23.01.2009 -> 03.05.2009).
> > On the first start everything works ok (including [glx]gears).
> > On the second and subsequent starts everything looks working but jerky,
> > I need to move mouse around to get some windows redrawn,
> > [glx]gears print warning message about not getting vblank interrupts
> > and outputs something about 1-2 frames per 5 seconds.
> > 
> > The inspectation with vmstat -i has shown that the card generates
> > interrupts only during the first start of X server.
> > 
> > If I set hw.pci.enable_msi="0" everything is working fine (start X server
> > multiple times, switch consoles, [glx]gears, number of irq16-s
> > is increasing, ...).
> 
> irq16 is likely shared, so this may or may not be true...  At least in
> some cases, I was seeing the interrupt handler processing events when
> some other device on the shared interrupt fired.  Interrupts on Intel
> have been a real pain.  There is a drm specific tuneable for disabling
> msi hw.drm.msi=0.  I have a patch which I'm currently using on my g45
> that overhauls the way that we handle interrupts.  
> 
> http://people.freebsd.org/~rnoland/drm-intel-050209.patch
> 
> Reports have been mixed with this, but it is working for me...
> 
So far works for me too...
irq256: vgapci0                     7759         21

Thanks! :)
Alexey.
Received on Mon May 04 2009 - 20:32:04 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:47 UTC