Rik van Riel wrote:
> 
> Hi,
> 
> The attached patch seems to fix all the reported deadlock
> problems with the new VM. Basically they could be grouped
> into 2 categories:
> 
> 1) __GFP_IO related locking issues
> 2) something sleeps on a free/clean/inactive page goal
>    that isn't worked towards
> 
> The patch has survived some heavy stresstesting on both
> SMP and UP machines. I hope nobody will be able to find
> a way to still crash this one ;)
> 
> A second change is a more dynamic free memory target
> (now freepages.high + inactive_target / 3), this seems
> to help a little bit in some loads.
> 

Hi,

I ran lmbench on test9-pre7 with and without the patch.

Test machine was a slow medium memory UP box:
Cx586@120Mhz, no optimizations, 56M

I still experience instability on this machine with both the
patched and vanilla kernel. It usually takes the form of
sudden total lockups, but on occasion I have seen oops +
panic at boot.

This summary doesn't show any performance advantage to the
patch, but the detailed plots show that memory access
latency degrades more gracefully wrt array size.

For bandwidth, I'm only including the summary. Vanilla is
listed first in each entry. Full lmbench report files on
request.

Tom