linux-security-module.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Mateusz Guzik <mjguzik@gmail.com>
To: john.johansen@canonical.com
Cc: paul@paul-moore.com, jmorris@namei.org, serge@hallyn.com,
	linux-kernel@vger.kernel.org, apparmor@lists.ubuntu.com,
	linux-security-module@vger.kernel.org, Neeraj.Upadhyay@amd.com,
	Mateusz Guzik <mjguzik@gmail.com>
Subject: [PATCH] apparmor: try to avoid refing the label in apparmor_file_open
Date: Thu, 20 Jun 2024 15:15:24 +0200	[thread overview]
Message-ID: <20240620131524.156312-1-mjguzik@gmail.com> (raw)

It can be done in the common case.

A 24-way open1_processes from will-it-scale (separate file open) shows:
  29.37%  [kernel]           [k] apparmor_file_open
  26.84%  [kernel]           [k] apparmor_file_alloc_security
  26.62%  [kernel]           [k] apparmor_file_free_security
   1.32%  [kernel]           [k] clear_bhb_loop

apparmor_file_open is eliminated from the profile with the patch.

Throughput (ops/s):
before:	6092196
after:	8309726 (+36%)

Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>
---

I think this is a worthwhile touch up regardless of what happens with
label refcouting in the long run. It does not of course does not fully
fix the problem.

I concede the naming is not consistent with other stuff in the file and
I'm not going to argue about it -- happy to name it whatever as long as
the problem is sorted out.

Am I missing something which makes the approach below not work to begin
with?

 security/apparmor/include/cred.h | 20 ++++++++++++++++++++
 security/apparmor/lsm.c          |  5 +++--
 2 files changed, 23 insertions(+), 2 deletions(-)

diff --git a/security/apparmor/include/cred.h b/security/apparmor/include/cred.h
index 58fdc72af664..7265d2f81dd5 100644
--- a/security/apparmor/include/cred.h
+++ b/security/apparmor/include/cred.h
@@ -63,6 +63,26 @@ static inline struct aa_label *aa_get_newest_cred_label(const struct cred *cred)
 	return aa_get_newest_label(aa_cred_raw_label(cred));
 }
 
+static inline struct aa_label *aa_get_newest_cred_label_condref(const struct cred *cred,
+								bool *needput)
+{
+	struct aa_label *l = aa_cred_raw_label(cred);
+
+	if (unlikely(label_is_stale(l))) {
+		*needput = true;
+		return aa_get_newest_label(l);
+	}
+
+	*needput = false;
+	return l;
+}
+
+static inline void aa_put_label_condref(struct aa_label *l, bool needput)
+{
+	if (unlikely(needput))
+		aa_put_label(l);
+}
+
 /**
  * aa_current_raw_label - find the current tasks confining label
  *
diff --git a/security/apparmor/lsm.c b/security/apparmor/lsm.c
index 2cea34657a47..4bf87eac4a56 100644
--- a/security/apparmor/lsm.c
+++ b/security/apparmor/lsm.c
@@ -461,6 +461,7 @@ static int apparmor_file_open(struct file *file)
 	struct aa_file_ctx *fctx = file_ctx(file);
 	struct aa_label *label;
 	int error = 0;
+	bool needput;
 
 	if (!path_mediated_fs(file->f_path.dentry))
 		return 0;
@@ -477,7 +478,7 @@ static int apparmor_file_open(struct file *file)
 		return 0;
 	}
 
-	label = aa_get_newest_cred_label(file->f_cred);
+	label = aa_get_newest_cred_label_condref(file->f_cred, &needput);
 	if (!unconfined(label)) {
 		struct mnt_idmap *idmap = file_mnt_idmap(file);
 		struct inode *inode = file_inode(file);
@@ -494,7 +495,7 @@ static int apparmor_file_open(struct file *file)
 		/* todo cache full allowed permissions set and state */
 		fctx->allow = aa_map_file_to_perms(file);
 	}
-	aa_put_label(label);
+	aa_put_label_condref(label, needput);
 
 	return error;
 }
-- 
2.43.0


             reply	other threads:[~2024-06-20 13:15 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-06-20 13:15 Mateusz Guzik [this message]
2024-06-20 16:26 ` [PATCH] apparmor: try to avoid refing the label in apparmor_file_open John Johansen
2024-06-20 16:41   ` Mateusz Guzik
2024-06-20 17:08     ` John Johansen

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20240620131524.156312-1-mjguzik@gmail.com \
    --to=mjguzik@gmail.com \
    --cc=Neeraj.Upadhyay@amd.com \
    --cc=apparmor@lists.ubuntu.com \
    --cc=jmorris@namei.org \
    --cc=john.johansen@canonical.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-security-module@vger.kernel.org \
    --cc=paul@paul-moore.com \
    --cc=serge@hallyn.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).