On Sat, Jan 23, 2010 at 06:37:35PM +, Mindaugas Rasiukevicius wrote:
>
> Rationale of this change: there would be a single allocation for LWP and
> PCB, which I think would reduce some memory fragmentation, as well as some
> allocation overhead, especially on architectures with 16k-pages or mo
On Sat, Jan 23, 2010 at 12:43:10PM -0800, Matt Thomas wrote:
>
> On Jan 23, 2010, at 1:21 AM, David Laight wrote:
>
> > On Sat, Jan 23, 2010 at 05:42:44PM +1100, matthew green wrote:
> >>
> >> what is the purpose of this change?
>
> To coalesce the number of allocations needed for lwp creation.
On Jan 23, 2010, at 1:21 AM, David Laight wrote:
> On Sat, Jan 23, 2010 at 05:42:44PM +1100, matthew green wrote:
>>
>> what is the purpose of this change?
To coalesce the number of allocations needed for lwp creation.
>> struct lwp is approx 700-1000 bytes on our platforms.
>> that's a signif
David Laight wrote:
> >
> > struct lwp is approx 700-1000 bytes on our platforms.
> > that's a significant chunk to remove from kernel stacks isn't it?
>
> My thoughts exactly
Right, e.g. on amd64, struct lwp is 900+ and struct pcb is 600+ bytes.
I do not think it takes too much from kerne
On Sat, Jan 23, 2010 at 05:42:44PM +1100, matthew green wrote:
>
> what is the purpose of this change?
>
> struct lwp is approx 700-1000 bytes on our platforms.
> that's a significant chunk to remove from kernel stacks isn't it?
My thoughts exactly
There are also lurking ideas to avoid nee
what is the purpose of this change?
struct lwp is approx 700-1000 bytes on our platforms.
that's a significant chunk to remove from kernel stacks isn't it?
.mrg.
Matt Thomas wrote:
> If we are going to get rid of swappable uareas, then we should go all
> the way.
>
> <...>
>
> The separate lwp pool should go away and the lwp should just be placed
> at the bottom of the uarea followed by the pcb. l_addr should be
> renamed l_pcb.
>
> <...>
Here i