Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-09 Thread Eric Dumazet
On Mon, Jan 9, 2017 at 9:45 AM, Michal Hocko wrote: > What about those non-default configurations. Do they really want to > invoke the OOM killer rather than fallback to the vmalloc? In our case, we use 4096 slots per fq, so that is a 16KB memory allocation. And these allocations happen right af

Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-09 Thread Michal Hocko
On Mon 09-01-17 08:00:16, Eric Dumazet wrote: > On Mon, Jan 9, 2017 at 2:22 AM, Michal Hocko wrote: > > > > the changelog doesn't mention it but this, unlike other kvmalloc > > conversions is not without functional changes. The kmalloc part > > will be weaker than it is with the original code for

Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-09 Thread Eric Dumazet
On Mon, Jan 9, 2017 at 2:22 AM, Michal Hocko wrote: > > the changelog doesn't mention it but this, unlike other kvmalloc > conversions is not without functional changes. The kmalloc part > will be weaker than it is with the original code for !costly (<64kB) > requests, because we are enforcing __G

Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-09 Thread Michal Hocko
On Fri 06-01-17 17:19:44, Michal Hocko wrote: [...] > From 8eadf8774daecdd6c4de37339216282a16920458 Mon Sep 17 00:00:00 2001 > From: Michal Hocko > Date: Fri, 6 Jan 2017 17:03:31 +0100 > Subject: [PATCH] net: use kvmalloc rather than open coded variant > > fq_alloc_node, alloc_netdev_mqs and neti

Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-06 Thread Vlastimil Babka
On 01/06/2017 06:08 PM, Eric Dumazet wrote: > On Fri, Jan 6, 2017 at 8:55 AM, Vlastimil Babka wrote: >> On 01/06/2017 05:48 PM, Eric Dumazet wrote: >>> On Fri, Jan 6, 2017 at 8:31 AM, Vlastimil Babka wrote: >>> I wonder what's that cause of the penalty (when accessing the vmapped a

Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-06 Thread Eric Dumazet
On Fri, Jan 6, 2017 at 8:55 AM, Vlastimil Babka wrote: > On 01/06/2017 05:48 PM, Eric Dumazet wrote: >> On Fri, Jan 6, 2017 at 8:31 AM, Vlastimil Babka wrote: >> >>> >>> I wonder what's that cause of the penalty (when accessing the vmapped >>> area I suppose?) Is it higher risk of collisions cach

Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-06 Thread Vlastimil Babka
On 01/06/2017 05:48 PM, Eric Dumazet wrote: > On Fri, Jan 6, 2017 at 8:31 AM, Vlastimil Babka wrote: > >> >> I wonder what's that cause of the penalty (when accessing the vmapped >> area I suppose?) Is it higher risk of collisions cache misses within the >> area, compared to consecutive physical

Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-06 Thread Eric Dumazet
On Fri, Jan 6, 2017 at 8:48 AM, Eric Dumazet wrote: > On Fri, Jan 6, 2017 at 8:31 AM, Vlastimil Babka wrote: > >> >> I wonder what's that cause of the penalty (when accessing the vmapped >> area I suppose?) Is it higher risk of collisions cache misses within the >> area, compared to consecutive p

Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-06 Thread Eric Dumazet
On Fri, Jan 6, 2017 at 8:31 AM, Vlastimil Babka wrote: > > I wonder what's that cause of the penalty (when accessing the vmapped > area I suppose?) Is it higher risk of collisions cache misses within the > area, compared to consecutive physical adresses? I believe tests were done with 48 fq qdis

Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-06 Thread Vlastimil Babka
On 01/06/2017 04:39 PM, Eric Dumazet wrote: > On Fri, Jan 6, 2017 at 7:20 AM, Michal Hocko wrote: >> >> Hi Eric, >> I am currently checking kmalloc with vmalloc fallback users and convert >> them to a new kvmalloc helper [1]. While I am adding a support for >> __GFP_REPEAT to kvmalloc [2] I was wo

Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-06 Thread Michal Hocko
On Fri 06-01-17 17:07:43, Michal Hocko wrote: > On Fri 06-01-17 07:39:14, Eric Dumazet wrote: > > On Fri, Jan 6, 2017 at 7:20 AM, Michal Hocko wrote: > > > > > > Hi Eric, > > > I am currently checking kmalloc with vmalloc fallback users and convert > > > them to a new kvmalloc helper [1]. While I

Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-06 Thread Michal Hocko
On Fri 06-01-17 07:39:14, Eric Dumazet wrote: > On Fri, Jan 6, 2017 at 7:20 AM, Michal Hocko wrote: > > > > Hi Eric, > > I am currently checking kmalloc with vmalloc fallback users and convert > > them to a new kvmalloc helper [1]. While I am adding a support for > > __GFP_REPEAT to kvmalloc [2] I

Re: __GFP_REPEAT usage in fq_alloc_node

2017-01-06 Thread Eric Dumazet
On Fri, Jan 6, 2017 at 7:20 AM, Michal Hocko wrote: > > Hi Eric, > I am currently checking kmalloc with vmalloc fallback users and convert > them to a new kvmalloc helper [1]. While I am adding a support for > __GFP_REPEAT to kvmalloc [2] I was wondering what is the reason to use > __GFP_REPEAT in

__GFP_REPEAT usage in fq_alloc_node

2017-01-06 Thread Michal Hocko
Hi Eric, I am currently checking kmalloc with vmalloc fallback users and convert them to a new kvmalloc helper [1]. While I am adding a support for __GFP_REPEAT to kvmalloc [2] I was wondering what is the reason to use __GFP_REPEAT in fq_alloc_node in the first place. c3bd85495aef ("pkt_sched: fq: