lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 12 Feb 2024 21:29:53 +0800
From: Chengming Zhou <zhouchengming@...edance.com>
To: Nhat Pham <nphamcs@...il.com>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
 Johannes Weiner <hannes@...xchg.org>, Yosry Ahmed <yosryahmed@...gle.com>,
 linux-mm@...ck.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/2] mm/zswap: change zswap_pool kref to percpu_ref

On 2024/2/12 05:21, Nhat Pham wrote:
> On Sun, Feb 11, 2024 at 5:58 AM Chengming Zhou
> <zhouchengming@...edance.com> wrote:
>>
>> All zswap entries will take a reference of zswap_pool when
>> zswap_store(), and drop it when free. Change it to use the
>> percpu_ref is better for scalability performance.
>>
>> Testing kernel build in tmpfs with memory.max=2GB
>> (zswap shrinker and writeback enabled with one 50GB swapfile).
>>
>>         mm-unstable  zswap-global-lru
>> real    63.20        63.12
>> user    1061.75      1062.95
>> sys     268.74       264.44
>>
>> Signed-off-by: Chengming Zhou <zhouchengming@...edance.com>
>> ---
>>  mm/zswap.c | 30 +++++++++++++++++++++---------
>>  1 file changed, 21 insertions(+), 9 deletions(-)
>>
>> diff --git a/mm/zswap.c b/mm/zswap.c
>> index 7668db8c10e3..afb31904fb08 100644
>> --- a/mm/zswap.c
>> +++ b/mm/zswap.c
>> @@ -173,7 +173,7 @@ struct crypto_acomp_ctx {
>>  struct zswap_pool {
>>         struct zpool *zpools[ZSWAP_NR_ZPOOLS];
>>         struct crypto_acomp_ctx __percpu *acomp_ctx;
>> -       struct kref kref;
>> +       struct percpu_ref ref;
>>         struct list_head list;
>>         struct work_struct release_work;
>>         struct hlist_node node;
>> @@ -303,6 +303,7 @@ static void zswap_update_total_size(void)
>>  /*********************************
>>  * pool functions
>>  **********************************/
>> +static void __zswap_pool_empty(struct percpu_ref *ref);
>>
>>  static struct zswap_pool *zswap_pool_create(char *type, char *compressor)
>>  {
>> @@ -356,13 +357,18 @@ static struct zswap_pool *zswap_pool_create(char *type, char *compressor)
>>         /* being the current pool takes 1 ref; this func expects the
>>          * caller to always add the new pool as the current pool
>>          */
>> -       kref_init(&pool->kref);
>> +       ret = percpu_ref_init(&pool->ref, __zswap_pool_empty,
>> +                             PERCPU_REF_ALLOW_REINIT, GFP_KERNEL);
>> +       if (ret)
>> +               goto ref_fail;
>>         INIT_LIST_HEAD(&pool->list);
>>
>>         zswap_pool_debug("created", pool);
>>
>>         return pool;
>>
>> +ref_fail:
>> +       cpuhp_state_remove_instance(CPUHP_MM_ZSWP_POOL_PREPARE, &pool->node);
>>  error:
>>         if (pool->acomp_ctx)
>>                 free_percpu(pool->acomp_ctx);
>> @@ -435,8 +441,8 @@ static void __zswap_pool_release(struct work_struct *work)
>>
>>         synchronize_rcu();
>>
>> -       /* nobody should have been able to get a kref... */
>> -       WARN_ON(kref_get_unless_zero(&pool->kref));
> 
> Do we no longer care about this WARN? IIUC, this is to catch someone
> still holding a reference to the pool at release time, which sounds
> like a bug. I think we can simulate the similar behavior with:

Ok, I thought it has already been put to 0 when we're here, so any tryget
will fail. But keeping this WARN_ON() is also fine to me, will keep it.

Thanks.

> 
> WARN_ON(percpu_ref_tryget(&pool->ref))
> 
> no? percpu_ref_tryget() should fail when the refcnt goes back down to
> 0. Then we can do percpu_ref_exit() as well.
> 
>> +       /* nobody should have been able to get a ref... */
>> +       percpu_ref_exit(&pool->ref);
>>
>>         /* pool is now off zswap_pools list and has no references. */
>>         zswap_pool_destroy(pool);
>> @@ -444,11 +450,11 @@ static void __zswap_pool_release(struct work_struct *work)
>>
>>  static struct zswap_pool *zswap_pool_current(void);
>>
>> -static void __zswap_pool_empty(struct kref *kref)
>> +static void __zswap_pool_empty(struct percpu_ref *ref)
>>  {
>>         struct zswap_pool *pool;
>>
>> -       pool = container_of(kref, typeof(*pool), kref);
>> +       pool = container_of(ref, typeof(*pool), ref);
>>
>>         spin_lock(&zswap_pools_lock);
>>
>> @@ -467,12 +473,12 @@ static int __must_check zswap_pool_get(struct zswap_pool *pool)
>>         if (!pool)
>>                 return 0;
>>
>> -       return kref_get_unless_zero(&pool->kref);
>> +       return percpu_ref_tryget(&pool->ref);
>>  }
>>
>>  static void zswap_pool_put(struct zswap_pool *pool)
>>  {
>> -       kref_put(&pool->kref, __zswap_pool_empty);
>> +       percpu_ref_put(&pool->ref);
>>  }
>>
>>  static struct zswap_pool *__zswap_pool_current(void)
>> @@ -602,6 +608,12 @@ static int __zswap_param_set(const char *val, const struct kernel_param *kp,
>>
>>         if (!pool)
>>                 pool = zswap_pool_create(type, compressor);
>> +       else {
>> +               /* Resurrect percpu_ref to percpu mode. */
>> +               percpu_ref_resurrect(&pool->ref);
>> +               /* Drop the ref from zswap_pool_find_get(). */
>> +               zswap_pool_put(pool);
>> +       }
>>
>>         if (pool)
>>                 ret = param_set_charp(s, kp);
>> @@ -640,7 +652,7 @@ static int __zswap_param_set(const char *val, const struct kernel_param *kp,
>>          * or the new pool we failed to add
>>          */
>>         if (put_pool)
>> -               zswap_pool_put(put_pool);
>> +               percpu_ref_kill(&put_pool->ref);
>>
>>         return ret;
>>  }
>>
>> --
>> b4 0.10.1
> 
> The rest of the code looks solid to me FWIW. Number seems to indicate
> this is a good idea as well.

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ