Re: Worst case performance of up()
From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Date: 2006-12-05 21:59:07
From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Date: 2006-12-05 21:59:07
quoted
If ppc_md.power_save is NULL, then the idle loop does not touch the flag word of the idle thread, and everything works.Good to know, there might be other cases of performance issues due to touching that flag in a loop...
More specifically... that means we have the same performance problem on SMP PowerMacs (dual G4s & Xserve G4s)... on those machines, there's a firmware property telling us not to use NAP mode and we do so by clearing powersave_nap. Paul, do you reckon the best solution here is to clear ppc_md.power_save when detecting an SMP powermac ? (Or the flush-on-lock property) Ben.