Re: [ckrm-tech] [PATCH 3/3][RFC] Containers: Pagecache controller reclaim

Re: [ckrm-tech] [PATCH 3/3][RFC] Containers: Pagecache controller reclaim

[Shane]

Anyone else have trouble fitting this patch ???.
I see a later version today, but not markedly different from this
mornings (Aus time). Initially I thought I had the first version, prior
to Balbir's RSS controller V2 re-write, but apparently not.
Kernel 2.6.20.1

Had to toss it away so I could do some base line testing - I'll redo the
build and see where the mis-matches are.

Shane ...

Re: [ckrm-tech] [PATCH 3/3][RFC] Containers: Pagecache controller reclaim

[Vaidyanathan Srinivasan]

Shane wrote:
> Anyone else have trouble fitting this patch ???.
> I see a later version today, but not markedly different from this
> mornings (Aus time). Initially I thought I had the first version, prior
> to Balbir's RSS controller V2 re-write, but apparently not.
> Kernel 2.6.20.1

Hi Shane,

I did post the same patch again today since the previous post
yesterday did not showup on LKML.  I have not changed the version
since it is the same patch.

Next time around i will explicitly mention that this is the same patch
posted again.

> Had to toss it away so I could do some base line testing - I'll redo the
> build and see where the mis-matches are.

Please let me know if so see any problem running the patch.  The
patches are against 2.6.20 only since dependent patches are at that level.

--Vaidy

Re: [ckrm-tech] [PATCH 3/3][RFC] Containers: Pagecache controller reclaim

[Shane]

On Tue, 2007-03-06 at 16:43 +0530, Vaidyanathan Srinivasan wrote:
> 
> Please let me know if so see any problem running the patch.  The
> patches are against 2.6.20 only since dependent patches are at that level.

My problem - a bad copy of the patch. It patches o.k.
However, it fails to compile vmscan. This looks a bit dodgy;

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

@@ -1470,11 +1494,13 @@ unsigned long shrink_all_memory(unsigned
        int pass;
        struct reclaim_state reclaim_state;
        struct scan_control sc = {
-               .gfp_mask = GFP_KERNEL,
+               .gfp_mask = GFdefined(CONFIG_CONTAINER_PAGECACHE_ACCT)
+P_KERNEL,

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

I deleted what looks like an over-enthusiastic "copy-and-paste", and it
compiled o.k.
Testing continues.

Shane ...

Re: [ckrm-tech] [PATCH 3/3][RFC] Containers: Pagecache controller reclaim

[Vaidyanathan Srinivasan]

Shane wrote:
> On Tue, 2007-03-06 at 16:43 +0530, Vaidyanathan Srinivasan wrote:
>> Please let me know if so see any problem running the patch.  The
>> patches are against 2.6.20 only since dependent patches are at that level.
> 
> My problem - a bad copy of the patch. It patches o.k.
> However, it fails to compile vmscan. This looks a bit dodgy;
> 
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 
> @@ -1470,11 +1494,13 @@ unsigned long shrink_all_memory(unsigned
>         int pass;
>         struct reclaim_state reclaim_state;
>         struct scan_control sc = {
> -               .gfp_mask = GFP_KERNEL,
> +               .gfp_mask = GFdefined(CONFIG_CONTAINER_PAGECACHE_ACCT)
> +P_KERNEL,
> 
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 
> I deleted what looks like an over-enthusiastic "copy-and-paste", and it
> compiled o.k.
> Testing continues.

OOPS!! How did it get there!  That is certainly some random mouse
click.  Thanks for pointing that out.

--Vaidy