Re: process creation time increases linearly with shmem

[Parag Warudkar]

On Thu, 2005-08-25 at 16:22 +0200, Andi Kleen wrote:
> But I'm not sure it's a good idea in all cases. Would need a lot of 
> benchmarking  at least.
> 
> -Andi
> 

Exactly - one problem is that this forces all of the hugetlb users to go
the lazy faulting way. This is more or less similar to the original
problem the fork() forces everything to be mapped and some apps don't
like it. Same way, some apps may not want hugetlb pages to be all
pre-mapped. 

That's why I was alluding towards having the user specify MAP_SHARED|
MAP_LAZY or something to that tune and then have fork() honor it. So
people who want all things pre-mapped will not specify MAP_LAZY, just
MAP_SHARED. 

Now I don't even know if above is possible and workable for all
scenarios but that's why I was asking.. :)

Parag

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/