Slow X startup and Hammering /proc/pci

[Thomas Leonard]

On Sat, 03 Jun 2006 15:21:11 +0300, Daniel Stone wrote:
> On Fri, Jun 02, 2006 at 08:28:33PM +0100, Thomas Leonard wrote:
[...]
>> Here are four traces, showing each combination of patches:
>> 
>> http://www.ecs.soton.ac.uk/~tal/traces/
>> 
>> (ignore the last two points on each graph, as that just shows when I hit
>> ctrl-alt-backspace to exit!)
>> 
>> Summary:
>> 
>> Original: 5.99 s
>> With bugzilla patch: 5.07 s
>> With my change: 4.31 s
>> With both: 3.84 s
> 
> http://people.freedesktop.org/~daniels/xorg-pci-speedup.diff is extremely
> nasty (lots of mangling between os-support/bus, and the OS subdirs), but
> it does the trick.  Does it still give you the same performance boost?

No, because it still calls pciFindNext() on the last iteration of the loop...
adding a break makes it work:

	if (idx == xf86OSMaxPciDevs)
		break;
	tag = pciFindNext();


-- 
Dr Thomas Leonard		http://rox.sourceforge.net
GPG: 9242 9807 C985 3C07 44A6  8B9A AE07 8280 59A5 3CC1