Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1694035 > unrolled thread
| Started by | Tim Chen <tim.c.chen@linux.intel.com> |
|---|---|
| First post | 2017-07-22 00:50 +0200 |
| Last post | 2017-07-24 19:00 +0200 |
| Articles | 3 — 2 participants |
Back to article view | Back to linux.kernel
This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by
below is the oldest one visible, not the original post.
[PATCH 2/2] mm/swap: Remove lock_initialized flag from swap_slots_cache Tim Chen <tim.c.chen@linux.intel.com> - 2017-07-22 00:50 +0200
Re: [PATCH 2/2] mm/swap: Remove lock_initialized flag from swap_slots_cache "Huang\, Ying" <ying.huang@intel.com> - 2017-07-24 04:20 +0200
Re: [PATCH 2/2] mm/swap: Remove lock_initialized flag from swap_slots_cache Tim Chen <tim.c.chen@linux.intel.com> - 2017-07-24 19:00 +0200
| From | Tim Chen <tim.c.chen@linux.intel.com> |
|---|---|
| Date | 2017-07-22 00:50 +0200 |
| Subject | [PATCH 2/2] mm/swap: Remove lock_initialized flag from swap_slots_cache |
| Message-ID | <u5OUa-2dk-9@gated-at.bofh.it> |
We will only reach the lock initialization code
in alloc_swap_slot_cache when the cpu's swap_slots_cache's slots
have not been allocated and swap_slots_cache has not been initialized
previously. So the lock_initialized check is redundant and unnecessary.
Remove lock_initialized flag from swap_slots_cache to save memory.
Reported-by: Wenwei Tao <wenwei.tww@alibaba-inc.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
---
include/linux/swap_slots.h | 1 -
mm/swap_slots.c | 9 ++++-----
2 files changed, 4 insertions(+), 6 deletions(-)
diff --git a/include/linux/swap_slots.h b/include/linux/swap_slots.h
index 6ef92d1..a75c30b 100644
--- a/include/linux/swap_slots.h
+++ b/include/linux/swap_slots.h
@@ -10,7 +10,6 @@
#define THRESHOLD_DEACTIVATE_SWAP_SLOTS_CACHE (2*SWAP_SLOTS_CACHE_SIZE)
struct swap_slots_cache {
- bool lock_initialized;
struct mutex alloc_lock; /* protects slots, nr, cur */
swp_entry_t *slots;
int nr;
diff --git a/mm/swap_slots.c b/mm/swap_slots.c
index 4c5457c..c039e6c 100644
--- a/mm/swap_slots.c
+++ b/mm/swap_slots.c
@@ -140,11 +140,10 @@ static int alloc_swap_slot_cache(unsigned int cpu)
if (cache->slots || cache->slots_ret)
/* cache already allocated */
goto out;
- if (!cache->lock_initialized) {
- mutex_init(&cache->alloc_lock);
- spin_lock_init(&cache->free_lock);
- cache->lock_initialized = true;
- }
+
+ mutex_init(&cache->alloc_lock);
+ spin_lock_init(&cache->free_lock);
+
cache->nr = 0;
cache->cur = 0;
cache->n_ret = 0;
--
2.9.4
[toc] | [next] | [standalone]
| From | "Huang\, Ying" <ying.huang@intel.com> |
|---|---|
| Date | 2017-07-24 04:20 +0200 |
| Message-ID | <u6B8u-79T-9@gated-at.bofh.it> |
| In reply to | #1694035 |
Hi, Tim,
Tim Chen <tim.c.chen@linux.intel.com> writes:
> We will only reach the lock initialization code
> in alloc_swap_slot_cache when the cpu's swap_slots_cache's slots
> have not been allocated and swap_slots_cache has not been initialized
> previously. So the lock_initialized check is redundant and unnecessary.
> Remove lock_initialized flag from swap_slots_cache to save memory.
Is there a race condition with CPU offline/online when preempt is enabled?
CPU A CPU B
----- -----
get_swap_page()
get cache[B], cache[B]->slots != NULL
preempted and moved to CPU A
be offlined
be onlined
alloc_swap_slot_cache()
mutex_lock(cache[B]->alloc_lock)
mutex_init(cache[B]->alloc_lock) !!!
The cache[B]->alloc_lock will be reinitialized when it is still held.
Best Regards,
Huang, Ying
> Reported-by: Wenwei Tao <wenwei.tww@alibaba-inc.com>
> Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
> ---
> include/linux/swap_slots.h | 1 -
> mm/swap_slots.c | 9 ++++-----
> 2 files changed, 4 insertions(+), 6 deletions(-)
>
> diff --git a/include/linux/swap_slots.h b/include/linux/swap_slots.h
> index 6ef92d1..a75c30b 100644
> --- a/include/linux/swap_slots.h
> +++ b/include/linux/swap_slots.h
> @@ -10,7 +10,6 @@
> #define THRESHOLD_DEACTIVATE_SWAP_SLOTS_CACHE (2*SWAP_SLOTS_CACHE_SIZE)
>
> struct swap_slots_cache {
> - bool lock_initialized;
> struct mutex alloc_lock; /* protects slots, nr, cur */
> swp_entry_t *slots;
> int nr;
> diff --git a/mm/swap_slots.c b/mm/swap_slots.c
> index 4c5457c..c039e6c 100644
> --- a/mm/swap_slots.c
> +++ b/mm/swap_slots.c
> @@ -140,11 +140,10 @@ static int alloc_swap_slot_cache(unsigned int cpu)
> if (cache->slots || cache->slots_ret)
> /* cache already allocated */
> goto out;
> - if (!cache->lock_initialized) {
> - mutex_init(&cache->alloc_lock);
> - spin_lock_init(&cache->free_lock);
> - cache->lock_initialized = true;
> - }
> +
> + mutex_init(&cache->alloc_lock);
> + spin_lock_init(&cache->free_lock);
> +
> cache->nr = 0;
> cache->cur = 0;
> cache->n_ret = 0;
[toc] | [prev] | [next] | [standalone]
| From | Tim Chen <tim.c.chen@linux.intel.com> |
|---|---|
| Date | 2017-07-24 19:00 +0200 |
| Subject | Re: [PATCH 2/2] mm/swap: Remove lock_initialized flag from swap_slots_cache |
| Message-ID | <u6OS5-7yA-7@gated-at.bofh.it> |
| In reply to | #1694404 |
On 07/23/2017 07:15 PM, Huang, Ying wrote: > Hi, Tim, > > Tim Chen <tim.c.chen@linux.intel.com> writes: > >> We will only reach the lock initialization code >> in alloc_swap_slot_cache when the cpu's swap_slots_cache's slots >> have not been allocated and swap_slots_cache has not been initialized >> previously. So the lock_initialized check is redundant and unnecessary. >> Remove lock_initialized flag from swap_slots_cache to save memory. > > Is there a race condition with CPU offline/online when preempt is enabled? > > CPU A CPU B > ----- ----- > get_swap_page() > get cache[B], cache[B]->slots != NULL > preempted and moved to CPU A > be offlined > be onlined > alloc_swap_slot_cache() > mutex_lock(cache[B]->alloc_lock) > mutex_init(cache[B]->alloc_lock) !!! > > The cache[B]->alloc_lock will be reinitialized when it is still held. Looks like for this case the lock_initialized flag is still needed to prevent such races and prevent re-initialization of taken locks. Okay, let's scrap patch 2. Thanks. Tim
[toc] | [prev] | [standalone]
Back to top | Article view | linux.kernel
csiph-web