[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <51BCC7CB.8060005@schaufler-ca.com>
Date: Sat, 15 Jun 2013 13:00:11 -0700
From: Casey Schaufler <casey@...aufler-ca.com>
To: Tomasz Stanislawski <t.stanislaws@...sung.com>
CC: linux-security-module@...r.kernel.org, m.szyprowski@...sung.com,
kyungmin.park@...sung.com, r.krypa@...sung.com,
linux-kernel@...r.kernel.org,
Casey Schaufler <casey@...aufler-ca.com>
Subject: Re: [RFC 4/5] security: smack: add kmem_cache for smack_rule allocations
On 6/13/2013 8:29 AM, Tomasz Stanislawski wrote:
> On ARM, sizeof(struct smack_rule)==20. Allocation by kmalloc() uses a
> 32-byte-long chunk to allocate 20 bytes. Just ask ksize(). It means that 40%
> of memory is simply wasted for padding bytes.
>
> The problem is fixed in this patch by using kmem_cache. The cache allocates
> struct smack_rule using 24-byte-long chunks according to ksize(). This reduces
> amount of used memory by 25%.
I'm not opposed to this change, but could I see some performance
numbers to justify it? In particular, I'm concerned about the rules
load impact.
> Signed-off-by: Tomasz Stanislawski <t.stanislaws@...sung.com>
> ---
> security/smack/smack.h | 3 +++
> security/smack/smack_lsm.c | 11 ++++++++++-
> security/smack/smackfs.c | 2 +-
> 3 files changed, 14 insertions(+), 2 deletions(-)
>
> diff --git a/security/smack/smack.h b/security/smack/smack.h
> index 8ad3095..38ba673 100644
> --- a/security/smack/smack.h
> +++ b/security/smack/smack.h
> @@ -233,6 +233,9 @@ extern struct mutex smack_known_lock;
> extern struct list_head smack_known_list;
> extern struct list_head smk_netlbladdr_list;
>
> +/* Cache for fast and thrifty allocations */
> +extern struct kmem_cache *smack_rule_cache;
> +
> extern struct security_operations smack_ops;
>
> /*
> diff --git a/security/smack/smack_lsm.c b/security/smack/smack_lsm.c
> index d52c780..7aa696a 100644
> --- a/security/smack/smack_lsm.c
> +++ b/security/smack/smack_lsm.c
> @@ -3564,6 +3564,9 @@ static __init void init_smack_known_list(void)
> list_add(&smack_known_web.list, &smack_known_list);
> }
>
> +/* KMEM caches for fast and thrifty allocations */
> +struct kmem_cache *smack_rule_cache;
> +
> /**
> * smack_init - initialize the smack system
> *
> @@ -3577,10 +3580,16 @@ static __init int smack_init(void)
> if (!security_module_enable(&smack_ops))
> return 0;
>
> + smack_rule_cache = KMEM_CACHE(smack_rule, 0);
> + if (!smack_rule_cache)
> + return -ENOMEM;
> +
> tsp = new_task_smack(smack_known_floor.smk_known,
> smack_known_floor.smk_known, GFP_KERNEL);
> - if (tsp == NULL)
> + if (tsp == NULL) {
> + kmem_cache_destroy(smack_rule_cache);
> return -ENOMEM;
> + }
>
> printk(KERN_INFO "Smack: Initializing.\n");
>
> diff --git a/security/smack/smackfs.c b/security/smack/smackfs.c
> index e8c57f3..c08b1ec 100644
> --- a/security/smack/smackfs.c
> +++ b/security/smack/smackfs.c
> @@ -217,7 +217,7 @@ static int smk_set_access(struct smack_parsed_rule *srp,
> }
>
> if (found == 0) {
> - sp = kzalloc(sizeof(*sp), GFP_KERNEL);
> + sp = kmem_cache_zalloc(smack_rule_cache, GFP_KERNEL);
> if (sp == NULL) {
> rc = -ENOMEM;
> goto out;
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists