Re: [PATCH v5 11/11] thp, vmstat: implement HZP_ALLOC and HZP_ALLOC_FAILED events

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 11/14/2012 03:41 PM, David Rientjes wrote:
> 
> Nobody is going to know what hzp_ is, sorry.  It's better to be more 
> verbose and name them what they actually are: THP_ZERO_PAGE_ALLOC and 
> THP_ZERO_PAGE_ALLOC_FAILED.  But this would assume we want to lazily 
> allocate them, which I disagree with hpa about.
> 

You want to permanently sit on 2 MiB of memory on all systems?  That
being an obvious nonstarter, then you end up having to make some kind of
static control, with all the problems that entails (if Linus had not set
his foot down on tunables a long time ago, we today would have had a
Linux mm which only performed well if you manually set hundreds or
thousands of parameters) you either have lazy allocation or you go with
the virtual huge zero page solution and just accept that either is going
to perform poorly under some set of pathological circumstances.

	-hpa

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>


[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux]     [Linux OMAP]     [Linux MIPS]     [ECOS]     [Asterisk Internet PBX]     [Linux API]