search for: nr_free

...cling through them every > + * time we steal is a bit easier and more or less equivalent: > + */ > + unsigned cpu_last_stolen; > + > + /* For sleeping on allocation failure */ > + wait_queue_head_t wait; > + > + /* > + * Global freelist - it's a stack where nr_free points to the > + * top > + */ > + unsigned nr_free; > + unsigned *freelist; > + } ____cacheline_aligned_in_smp; Why the ____cacheline_aligned_in_smp? > +}; > > ... > > + > +/* Percpu IDA */ > + > +/* > + * Number of tags we move between the perc...

...cling through them every > + * time we steal is a bit easier and more or less equivalent: > + */ > + unsigned cpu_last_stolen; > + > + /* For sleeping on allocation failure */ > + wait_queue_head_t wait; > + > + /* > + * Global freelist - it's a stack where nr_free points to the > + * top > + */ > + unsigned nr_free; > + unsigned *freelist; > + } ____cacheline_aligned_in_smp; Why the ____cacheline_aligned_in_smp? > +}; > > ... > > + > +/* Percpu IDA */ > + > +/* > + * Number of tags we move between the perc...

...)), + * we want to pick a cpu at random. Cycling through them every + * time we steal is a bit easier and more or less equivalent: + */ + unsigned cpu_last_stolen; + + /* For sleeping on allocation failure */ + wait_queue_head_t wait; + + /* + * Global freelist - it's a stack where nr_free points to the + * top + */ + unsigned nr_free; + unsigned *freelist; + } ____cacheline_aligned_in_smp; +}; + +int percpu_ida_alloc(struct percpu_ida *pool, gfp_t gfp); +void percpu_ida_free(struct percpu_ida *pool, unsigned tag); + +void percpu_ida_destroy(struct percpu_ida *pool); +int per...

From: Nicholas Bellinger <nab at linux-iscsi.org> Hi folks, This is an updated series for adding tag pre-allocation support of target fabric descriptor memory, utilizing Kent's latest per-cpu ida bits here, along with Christoph Lameter's latest comments: [PATCH 04/10] idr: Percpu ida http://marc.info/?l=linux-kernel&m=137160026006974&w=2 The first patch is a

From: Nicholas Bellinger <nab at linux-iscsi.org> Hi folks, This is an updated series for adding tag pre-allocation support of target fabric descriptor memory, utilizing Kent's latest per-cpu ida bits here, along with Christoph Lameter's latest comments: [PATCH 04/10] idr: Percpu ida http://marc.info/?l=linux-kernel&m=137160026006974&w=2 The first patch is a

On Fri, 16 Aug 2013, Nicholas A. Bellinger wrote: > + spinlock_t lock; Remove the spinlock. > + unsigned nr_free; > + unsigned freelist[]; > +}; > + > +static inline void move_tags(unsigned *dst, unsigned *dst_nr, > + unsigned *src, unsigned *src_nr, > + unsigned nr) > +{ > + *src_nr -= nr; > + memcpy(dst + *dst_nr, src + *src_nr, sizeof(unsigned) * nr); > + *dst_...

On Fri, 16 Aug 2013, Nicholas A. Bellinger wrote: > + spinlock_t lock; Remove the spinlock. > + unsigned nr_free; > + unsigned freelist[]; > +}; > + > +static inline void move_tags(unsigned *dst, unsigned *dst_nr, > + unsigned *src, unsigned *src_nr, > + unsigned nr) > +{ > + *src_nr -= nr; > + memcpy(dst + *dst_nr, src + *src_nr, sizeof(unsigned) * nr); > + *dst_...

...we steal is a bit easier and more or less equivalent: > > + */ > > + unsigned cpu_last_stolen; > > + > > + /* For sleeping on allocation failure */ > > + wait_queue_head_t wait; > > + > > + /* > > + * Global freelist - it's a stack where nr_free points to the > > + * top > > + */ > > + unsigned nr_free; > > + unsigned *freelist; > > + } ____cacheline_aligned_in_smp; > > Why the ____cacheline_aligned_in_smp? It's separating the RW stuff that isn't always touched from the RO stuff that...

...> > > + * cpus_have_tags > > > + * > > > + * global lock held and irqs disabled, don't need percpu lock > > > + */ > > > + prepare_to_wait(&pool->wait, &wait, TASK_UNINTERRUPTIBLE); > > > + > > > + if (!tags->nr_free) > > > + alloc_global_tags(pool, tags); > > > + if (!tags->nr_free) > > > + steal_tags(pool, tags); > > > + > > > + if (tags->nr_free) { > > > + tag = tags->freelist[--tags->nr_free]; > > > + if (tags->nr_free)...

...> > > + * cpus_have_tags > > > + * > > > + * global lock held and irqs disabled, don't need percpu lock > > > + */ > > > + prepare_to_wait(&pool->wait, &wait, TASK_UNINTERRUPTIBLE); > > > + > > > + if (!tags->nr_free) > > > + alloc_global_tags(pool, tags); > > > + if (!tags->nr_free) > > > + steal_tags(pool, tags); > > > + > > > + if (tags->nr_free) { > > > + tag = tags->freelist[--tags->nr_free]; > > > + if (tags->nr_free)...

Removing the percpu_ida code nets over 400 lines of removal. It's not as spectacular as deleting an entire architecture, but it's still a worthy reduction in lines of code. Untested due to lack of hardware and not understanding how to set up a target platform. Changes from v1: - Fixed bugs pointed out by Jens in iscsit_wait_for_tag() - Abstracted out tag freeing as requested by Bart

Removing the percpu_ida code nets over 400 lines of removal. It's not as spectacular as deleting an entire architecture, but it's still a worthy reduction in lines of code. Untested due to lack of hardware and not understanding how to set up a target platform. Changes from v1: - Fixed bugs pointed out by Jens in iscsit_wait_for_tag() - Abstracted out tag freeing as requested by Bart

From: Matthew Wilcox <mawilcox at microsoft.com> This is a pretty rough-and-ready conversion of the target drivers from using percpu_ida to sbitmap. It compiles; I don't have a target setup, so it's completely untested. I haven't tried to do anything particularly clever here, so it's possible that, for example, the wait queue in iscsi_target_util could be more clever, like

...021c 100644 --- a/lib/idr.c +++ b/lib/idr.c @@ -1178,7 +1178,13 @@ EXPORT_SYMBOL(ida_init); #define IDA_PCPU_SIZE ((IDA_PCPU_BATCH_MOVE * 3) / 2) struct percpu_ida_cpu { + /* + * Even though this is percpu, we need a lock for tag stealing by remote + * CPUs: + */ spinlock_t lock; + + /* nr_free/freelist form a stack of free IDs */ unsigned nr_free; unsigned freelist[]; }; @@ -1209,21 +1215,21 @@ static inline void steal_tags(struct percpu_ida *pool, unsigned cpus_have_tags, cpu = pool->cpu_last_stolen; struct percpu_ida_cpu *remote; - for (cpus_have_tags = bitmap_weight(...

..._tags > > > > + * > > > > + * global lock held and irqs disabled, don't need percpu lock > > > > + */ > > > > + prepare_to_wait(&pool->wait, &wait, TASK_UNINTERRUPTIBLE); > > > > + > > > > + if (!tags->nr_free) > > > > + alloc_global_tags(pool, tags); > > > > + if (!tags->nr_free) > > > > + steal_tags(pool, tags); > > > > + > > > > + if (tags->nr_free) { > > > > + tag = tags->freelist[--tags->nr_free]; > > &...

...;> + struct list_head *curr; >>> + struct page_info_item *info; >>> + >>> + if (zone_is_empty(zone)) >>> + return 0; >>> + >>> + spin_lock_irqsave(&zone->lock, flags); >>> + >>> + if (*pos + zone->free_area[order].nr_free > size) >>> + return -ENOSPC; >> >> Urg, so this won't partially fill? So, what the nr_free pages limit where we no >> longer fit in the kmalloc()'d buffer where this simply won't work? > > Yes. My initial implementation is partially fill, it'...

...;> + struct list_head *curr; >>> + struct page_info_item *info; >>> + >>> + if (zone_is_empty(zone)) >>> + return 0; >>> + >>> + spin_lock_irqsave(&zone->lock, flags); >>> + >>> + if (*pos + zone->free_area[order].nr_free > size) >>> + return -ENOSPC; >> >> Urg, so this won't partially fill? So, what the nr_free pages limit where we no >> longer fit in the kmalloc()'d buffer where this simply won't work? > > Yes. My initial implementation is partially fill, it'...

...> +{ > + unsigned long pfn, flags; > + unsigned int t; > + struct list_head *curr; > + struct page_info_item *info; > + > + if (zone_is_empty(zone)) > + return 0; > + > + spin_lock_irqsave(&zone->lock, flags); > + > + if (*pos + zone->free_area[order].nr_free > size) > + return -ENOSPC; Urg, so this won't partially fill? So, what the nr_free pages limit where we no longer fit in the kmalloc()'d buffer where this simply won't work? > + for (t = 0; t < MIGRATE_TYPES; t++) { > + list_for_each(curr, &zone->free_area[or...

...> +{ > + unsigned long pfn, flags; > + unsigned int t; > + struct list_head *curr; > + struct page_info_item *info; > + > + if (zone_is_empty(zone)) > + return 0; > + > + spin_lock_irqsave(&zone->lock, flags); > + > + if (*pos + zone->free_area[order].nr_free > size) > + return -ENOSPC; Urg, so this won't partially fill? So, what the nr_free pages limit where we no longer fit in the kmalloc()'d buffer where this simply won't work? > + for (t = 0; t < MIGRATE_TYPES; t++) { > + list_for_each(curr, &zone->free_area[or...

...t; > > > of these. > > > That wouldn't be a good choice for us. If we check how the regular > > > allocation works, there are many many things we need to consider when pages > > > are allocated to users. > > > For example, we need to take care of the nr_free > > > counter, we need to check the watermark and perform the related actions. > > > Also the folks working on arch_alloc_page to monitor page allocation > > > activities would get a surprise..if page allocation is allowed to work in > > > this way. > > &gt...