On Fri, 19 Oct 2012 08:20:49 -0400
Jeff Layton <jlayton@xxxxxxxxxx> wrote:
> We want to allow unused idmap keys to eventually be purged from the
> cache. Implement that by simply setting a timeout on the key whenever we
> use it. We use the same default timeout as the NFS idmap cache, but add
> a module parm to make it tunable.
>
> Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx>
> ---
> fs/cifs/cifsacl.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/fs/cifs/cifsacl.c b/fs/cifs/cifsacl.c
> index 12d70ee..0c0a594 100644
> --- a/fs/cifs/cifsacl.c
> +++ b/fs/cifs/cifsacl.c
> @@ -26,6 +26,7 @@
> #include <linux/string.h>
> #include <linux/keyctl.h>
> #include <linux/key-type.h>
> +#include <linux/moduleparam.h>
> #include <keys/user-type.h>
> #include "cifspdu.h"
> #include "cifsglob.h"
> @@ -44,6 +45,11 @@ static const struct cifs_sid sid_user = {1, 2 , {0, 0, 0, 0, 0, 5}, {} };
>
> static const struct cred *root_cred;
>
> +static unsigned int cifs_idmap_cache_timeout = 600;
> +module_param(cifs_idmap_cache_timeout, uint, 0644);
> +MODULE_PARM_DESC(cifs_idmap_cache_timeout, "Number of seconds that ID mappings "
> + "stay in cache. (default=600)");
> +
> static int
> cifs_idmap_key_instantiate(struct key *key, struct key_preparsed_payload *prep)
> {
> @@ -214,6 +220,7 @@ id_to_sid(unsigned int cid, uint sidtype, struct cifs_sid *ssid)
> goto out_key_put;
> }
> cifs_copy_sid(ssid, (struct cifs_sid *)sidkey->payload.data);
> + key_set_timeout(sidkey, cifs_idmap_cache_timeout);
> out_key_put:
> key_put(sidkey);
> out_revert_creds:
> @@ -272,6 +279,7 @@ sid_to_id(struct cifs_sb_info *cifs_sb, struct cifs_sid *psid,
> else
> fgid = (gid_t)sidkey->payload.value;
>
> + key_set_timeout(sidkey, cifs_idmap_cache_timeout);
> out_key_put:
> key_put(sidkey);
> out_revert_creds:
Now that I've given this some thought, I think we might be better off
removing the above patch and instead having the userspace program set
the timeout on the keys.
It's a pity that older versions of the program didn't do that, but no
one has reported a problem with that so far...
Thoughts?
--
Jeff Layton <jlayton@xxxxxxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe linux-cifs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
