Re: [RFC PATCH V5] mm readahead: Fix readahead fail for no local memory and limit readahead pages

[Raghavendra K T <raghavendra.kt@linux.vnet.ibm.com>]

On 02/11/2014 03:05 AM, David Rientjes wrote:
> On Mon, 10 Feb 2014, Raghavendra K T wrote:
>
>> So I understood that you are suggesting implementations like below
>>
>> 1) I do not have problem with the below approach, I could post this in
>> next version.
>> ( But this did not include 4k limit Linus mentioned to apply)
>>
>> unsigned long max_sane_readahead(unsigned long nr)
>> {
>>          unsigned long local_free_page;
>>          int nid;
>>
>>          nid = numa_mem_id();
>>
>>          /*
>>           * We sanitize readahead size depending on free memory in
>>           * the local node.
>>           */
>>          local_free_page = node_page_state(nid, NR_INACTIVE_FILE)
>>                            + node_page_state(nid, NR_FREE_PAGES);
>>          return min(nr, local_free_page / 2);
>> }
>>
>> 2) I did not go for below because Honza (Jan Kara) had some
>> concerns for 4k limit for normal case, and since I am not
>> the expert, I was waiting for opinions.
>>
>> unsigned long max_sane_readahead(unsigned long nr)
>> {
>>          unsigned long local_free_page, sane_nr;
>>          int nid;
>>
>>          nid = numa_mem_id();
>> 	/* limit the max readahead to 4k pages */
>> 	sane_nr = min(nr, MAX_REMOTE_READAHEAD);
>>
>>          /*
>>           * We sanitize readahead size depending on free memory in
>>           * the local node.
>>           */
>>          local_free_page = node_page_state(nid, NR_INACTIVE_FILE)
>>                            + node_page_state(nid, NR_FREE_PAGES);
>>          return min(sane_nr, local_free_page / 2);
>> }
>>
>
> I have no opinion on the 4KB pages, either of the above is just fine.
>

I was able to test (1) implementation on the system where readahead 
problem occurred. Unfortunately it did not help.

Reason seem to be that CONFIG_HAVE_MEMORYLESS_NODES dependency of
numa_mem_id(). The PPC machine I am facing problem has topology like
this:

numactl -H
---------
available: 2 nodes (0-1)
node 0 cpus: 0 1 2 3 4 5 6 7 12 13 14 15 16 17 18 19 20 21 22 23 24 25
...
node 0 size: 0 MB
node 0 free: 0 MB
node 1 cpus: 8 9 10 11 32 33 34 35 ...
node 1 size: 8071 MB
node 1 free: 2479 MB
node distances:
node   0   1
   0:  10  20
   1:  20  10

So it seems numa_mem_id() does not help for all the configs..
Am I missing something ?

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