2013-04-12 19:16:35

by Jeff Layton

[permalink] [raw]
Subject: [PATCH] vfs: fix audit_inode call in O_CREAT case of do_last

Jiri reported a regression in auditing of open(..., O_CREAT) syscalls.
In older kernels, creating a file with open(..., O_CREAT) created
audit_name records that looked like this:

type=PATH msg=audit(1360255720.628:64): item=1 name="/abc/foo" inode=138810 dev=fd:00 mode=0100640 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
type=PATH msg=audit(1360255720.628:64): item=0 name="/abc/" inode=138635 dev=fd:00 mode=040750 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0

...in recent kernels though, they look like this:

type=PATH msg=audit(1360255402.886:12574): item=2 name=(null) inode=264599 dev=fd:00 mode=0100640 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
type=PATH msg=audit(1360255402.886:12574): item=1 name=(null) inode=264598 dev=fd:00 mode=040750 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
type=PATH msg=audit(1360255402.886:12574): item=0 name="/abc/foo" inode=264598 dev=fd:00 mode=040750 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0

Richard bisected to determine that the problems started with commit
bfcec708, but the log messages have changed with some later
audit-related patches.

The problem is that this audit_inode call is passing in the parent of
the dentry being opened, but audit_inode is being called with the parent
flag false. This causes later audit_inode and audit_inode_child calls to
match the wrong entry in the audit_names list.

This patch simply sets the flag to properly indicate that this inode
represents the parent. With this, the audit_names entries are back to
looking like they did before.

Cc: <[email protected]> # v3.7+
Cc: Richard Guy Briggs <[email protected]>
Reported-by: Jiri Jaburek <[email protected]>
Signed-off-by: Jeff Layton <[email protected]>
---
fs/namei.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/namei.c b/fs/namei.c
index 57ae9c8..85e40d1 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -2740,7 +2740,7 @@ static int do_last(struct nameidata *nd, struct path *path,
if (error)
return error;

- audit_inode(name, dir, 0);
+ audit_inode(name, dir, LOOKUP_PARENT);
error = -EISDIR;
/* trailing slashes? */
if (nd->last.name[nd->last.len])
--
1.7.1


2013-04-12 21:39:26

by Jeff Layton

[permalink] [raw]
Subject: Re: [PATCH] vfs: fix audit_inode call in O_CREAT case of do_last

On Fri, 12 Apr 2013 15:16:32 -0400
Jeff Layton <[email protected]> wrote:

> Jiri reported a regression in auditing of open(..., O_CREAT) syscalls.
> In older kernels, creating a file with open(..., O_CREAT) created
> audit_name records that looked like this:
>
> type=PATH msg=audit(1360255720.628:64): item=1 name="/abc/foo" inode=138810 dev=fd:00 mode=0100640 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
> type=PATH msg=audit(1360255720.628:64): item=0 name="/abc/" inode=138635 dev=fd:00 mode=040750 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
>
> ...in recent kernels though, they look like this:
>
> type=PATH msg=audit(1360255402.886:12574): item=2 name=(null) inode=264599 dev=fd:00 mode=0100640 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
> type=PATH msg=audit(1360255402.886:12574): item=1 name=(null) inode=264598 dev=fd:00 mode=040750 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
> type=PATH msg=audit(1360255402.886:12574): item=0 name="/abc/foo" inode=264598 dev=fd:00 mode=040750 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
>
> Richard bisected to determine that the problems started with commit
> bfcec708, but the log messages have changed with some later
> audit-related patches.
>
> The problem is that this audit_inode call is passing in the parent of
> the dentry being opened, but audit_inode is being called with the parent
> flag false. This causes later audit_inode and audit_inode_child calls to
> match the wrong entry in the audit_names list.
>
> This patch simply sets the flag to properly indicate that this inode
> represents the parent. With this, the audit_names entries are back to
> looking like they did before.
>
> Cc: <[email protected]> # v3.7+
> Cc: Richard Guy Briggs <[email protected]>
> Reported-by: Jiri Jaburek <[email protected]>
> Signed-off-by: Jeff Layton <[email protected]>
> ---
> fs/namei.c | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/fs/namei.c b/fs/namei.c
> index 57ae9c8..85e40d1 100644
> --- a/fs/namei.c
> +++ b/fs/namei.c
> @@ -2740,7 +2740,7 @@ static int do_last(struct nameidata *nd, struct path *path,
> if (error)
> return error;
>
> - audit_inode(name, dir, 0);
> + audit_inode(name, dir, LOOKUP_PARENT);
> error = -EISDIR;
> /* trailing slashes? */
> if (nd->last.name[nd->last.len])

BTW, there are a couple of other calls to audit_inode earlier in this
function (the LAST_BIND and LAST_ROOT cases). I suspect those also need
similar treatment, but I've left them alone for now since I didn't have
a clear way to verify it. Let me know if you'd like me to respin this
with those changed as well.

Thanks,
--
Jeff Layton <[email protected]>

2013-04-17 03:02:29

by Richard Guy Briggs

[permalink] [raw]
Subject: Re: [PATCH] vfs: fix audit_inode call in O_CREAT case of do_last

On Fri, Apr 12, 2013 at 03:16:32PM -0400, Jeff Layton wrote:
> Jiri reported a regression in auditing of open(..., O_CREAT) syscalls.
> In older kernels, creating a file with open(..., O_CREAT) created
> audit_name records that looked like this:
>
> type=PATH msg=audit(1360255720.628:64): item=1 name="/abc/foo" inode=138810 dev=fd:00 mode=0100640 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
> type=PATH msg=audit(1360255720.628:64): item=0 name="/abc/" inode=138635 dev=fd:00 mode=040750 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
>
> ...in recent kernels though, they look like this:
>
> type=PATH msg=audit(1360255402.886:12574): item=2 name=(null) inode=264599 dev=fd:00 mode=0100640 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
> type=PATH msg=audit(1360255402.886:12574): item=1 name=(null) inode=264598 dev=fd:00 mode=040750 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
> type=PATH msg=audit(1360255402.886:12574): item=0 name="/abc/foo" inode=264598 dev=fd:00 mode=040750 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:default_t:s0
>
> Richard bisected to determine that the problems started with commit
> bfcec708, but the log messages have changed with some later
> audit-related patches.
>
> The problem is that this audit_inode call is passing in the parent of
> the dentry being opened, but audit_inode is being called with the parent
> flag false. This causes later audit_inode and audit_inode_child calls to
> match the wrong entry in the audit_names list.
>
> This patch simply sets the flag to properly indicate that this inode
> represents the parent. With this, the audit_names entries are back to
> looking like they did before.

This patch fixes the problem for me.

Tested-by: Richard Guy Briggs <[email protected]>

> Cc: <[email protected]> # v3.7+
> Cc: Richard Guy Briggs <[email protected]>
> Reported-by: Jiri Jaburek <[email protected]>
> Signed-off-by: Jeff Layton <[email protected]>
> ---
> fs/namei.c | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/fs/namei.c b/fs/namei.c
> index 57ae9c8..85e40d1 100644
> --- a/fs/namei.c
> +++ b/fs/namei.c
> @@ -2740,7 +2740,7 @@ static int do_last(struct nameidata *nd, struct path *path,
> if (error)
> return error;
>
> - audit_inode(name, dir, 0);
> + audit_inode(name, dir, LOOKUP_PARENT);
> error = -EISDIR;
> /* trailing slashes? */
> if (nd->last.name[nd->last.len])
> --
> 1.7.1
>
> --
> Linux-audit mailing list
> Linux-a[email protected]
> https://www.redhat.com/mailman/listinfo/linux-audit

- RGB

--
Richard Guy Briggs <[email protected]>
Senior Software Engineer
AMER ENG Base Operating Systems
Remote, Canada, Ottawa
Voice: 1.647.777.2635
Internal: (81) 32635