On Wed, Aug 07, 2013 at 07:40:15PM +0000, Christoph Lameter wrote:
> On Wed, 7 Aug 2013, Kent Overstreet wrote:
> 
> > I was breaking it apart because I was using this_cpu elsewhere too - for
> > the bitmap of which cpus have non empty freelists.
> 
> this_cpu can be retrieved with smp_processor_id().
> 
> > Or is this_cpu_ptr() doing something smarter than per_cpu_ptr(ptr,
> > smp_processer_id())? There's so many variants I'm not 100% sure they're
> > the same.
> 
> Yes it is. It uses a sepecial register that contains the offset of this
> cpus per cpu area instead of going through the table of all processor
> offsets. Its less code.

Alright, well here's a fixup patch - untested for the moment though.

One thing that was bugging me - I was never able to figure out for sure
if smp_processor_id() returns a number in the range [0, nr_cpu_ids), at
least I couldn't find where it was documented - could you tell me if
that's true?

>From e2b8016de49c28c0ccbe7849d7254f005c7e2e77 Mon Sep 17 00:00:00 2001
From: Kent Overstreet <[email protected]>
Date: Wed, 7 Aug 2013 12:52:58 -0700
Subject: [PATCH] idr: Use this_cpu_ptr() for percpu_ida


diff --git a/lib/idr.c b/lib/idr.c
index fb374c3..320ffea 100644
--- a/lib/idr.c
+++ b/lib/idr.c
@@ -748,12 +748,10 @@ int percpu_ida_alloc(struct percpu_ida *pool, gfp_t gfp)
        DEFINE_WAIT(wait);
        struct percpu_ida_cpu *tags;
        unsigned long flags;
-       unsigned this_cpu;
        int tag;
 
        local_irq_save(flags);
-       this_cpu = smp_processor_id();
-       tags = per_cpu_ptr(pool->tag_cpu, this_cpu);
+       tags = this_cpu_ptr(pool->tag_cpu);
 
        /* Fastpath */
        tag = alloc_local_tag(pool, tags);
@@ -782,7 +780,8 @@ int percpu_ida_alloc(struct percpu_ida *pool, gfp_t gfp)
                if (tags->nr_free) {
                        tag = tags->freelist[--tags->nr_free];
                        if (tags->nr_free)
-                               set_bit(this_cpu, pool->cpus_have_tags);
+                               set_bit(smp_processor_id(),
+                                       pool->cpus_have_tags);
                }
 
                spin_unlock(&pool->ida.lock);
@@ -794,8 +793,7 @@ int percpu_ida_alloc(struct percpu_ida *pool, gfp_t gfp)
                schedule();
 
                local_irq_save(flags);
-               this_cpu = smp_processor_id();
-               tags = per_cpu_ptr(pool->tag_cpu, this_cpu);
+               tags = this_cpu_ptr(pool->tag_cpu);
        }
 
        finish_wait(&pool->wait, &wait);
@@ -814,13 +812,12 @@ void percpu_ida_free(struct percpu_ida *pool, unsigned 
tag)
 {
        struct percpu_ida_cpu *tags;
        unsigned long flags;
-       unsigned nr_free, this_cpu;
+       unsigned nr_free;
 
        BUG_ON(tag >= pool->nr_tags);
 
        local_irq_save(flags);
-       this_cpu = smp_processor_id();
-       tags = per_cpu_ptr(pool->tag_cpu, this_cpu);
+       tags = this_cpu_ptr(pool->tag_cpu);
 
        spin_lock(&tags->lock);
        tags->freelist[tags->nr_free++] = tag;
@@ -829,7 +826,8 @@ void percpu_ida_free(struct percpu_ida *pool, unsigned tag)
        spin_unlock(&tags->lock);
 
        if (nr_free == 1) {
-               set_bit(this_cpu, pool->cpus_have_tags);
+               set_bit(smp_processor_id(),
+                       pool->cpus_have_tags);
                wake_up(&pool->wait);
        }
 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Reply via email to