[PATCH] missing checks in exec_permission_light()

[PATCH] missing checks in exec_permission_light()

[Alexander Viro]

	Missing checks added...

diff -urN C12-0/fs/namei.c C12-current/fs/namei.c
--- C12-0/fs/namei.c	Tue Apr 30 20:23:38 2002
+++ C12-current/fs/namei.c	Tue Apr 30 23:37:15 2002
@@ -324,6 +324,12 @@
 	if (mode & MAY_EXEC)
 		return 0;
 
+	if ((inode->i_mode & S_IXUGO) && capable(CAP_DAC_OVERRIDE))
+		return 0;
+
+	if (S_ISDIR(inode->i_mode) && capable(CAP_DAC_READ_SEARCH))
+		return 0;
+
 	return -EACCES;
 }
 

Re: [PATCH] missing checks in exec_permission_light()

[Daniel Pittman]

On Tue, 30 Apr 2002, Alexander Viro wrote:
> 	Missing checks added...
> 
> diff -urN C12-0/fs/namei.c C12-current/fs/namei.c
> --- C12-0/fs/namei.c	Tue Apr 30 20:23:38 2002
> +++ C12-current/fs/namei.c	Tue Apr 30 23:37:15 2002
> @@ -324,6 +324,12 @@
>  	if (mode & MAY_EXEC)
>  		return 0;
> 
> +	if ((inode->i_mode & S_IXUGO) && capable(CAP_DAC_OVERRIDE))
> +		return 0;
> +
> +	if (S_ISDIR(inode->i_mode) && capable(CAP_DAC_READ_SEARCH))
> +		return 0;
> +
>  	return -EACCES;
>  }

Looking at this it seems that it would explain the odd set of errors I
got reported during bootup under 2.5.12 -- a set of "permission denied"
errors from find(1) where, under 2.5.6, none had occurred.

These were on directories that are not owner by root, with the process
running as root.

I can look deeper into the problem, though, if you don't think that this
is the cause of it.

Regards,
        Daniel

-- 
It is easier to build strong children than to repair broken men.
        -- Frederick Douglas

Re: [PATCH] missing checks in exec_permission_light()

[Pavel Machek]

Hi!

> +	if (S_ISDIR(inode->i_mode) && capable(CAP_DAC_READ_SEARCH))
> +		return 0;

Is this right? This means that root can do cat /, no? That does not
seem like expected behaviour.
									Pavel
-- 
(about SSSCA) "I don't say this lightly.  However, I really think that the U.S.
no longer is classifiable as a democracy, but rather as a plutocracy." --hpa

Re: [PATCH] missing checks in exec_permission_light()

[Alexander Viro]

On Fri, 3 May 2002, Pavel Machek wrote:

> Hi!
> 
> > +	if (S_ISDIR(inode->i_mode) && capable(CAP_DAC_READ_SEARCH))
> > +		return 0;
> 
> Is this right? This means that root can do cat /, no? That does not
> seem like expected behaviour.

1) it's permission(..., MAY_EXEC)
2) in any case, root _can_ open "/" with O_RDONLY.  Always could.  That's
what you do for ls /, after all - open(2) followed by getdents(2).  Now,
read(2) will fail (check what ->read() for directories is set to), but
that has nothing to permission checks.

Re: [PATCH] missing checks in exec_permission_light()

[Pavel Machek]

Hi!

> > > +	if (S_ISDIR(inode->i_mode) && capable(CAP_DAC_READ_SEARCH))
> > > +		return 0;
> > 
> > Is this right? This means that root can do cat /, no? That does not
> > seem like expected behaviour.
> 
> 1) it's permission(..., MAY_EXEC)

Okay, sorry.
								Pavel
-- 
Casualities in World Trade Center: ~3k dead inside the building,
cryptography in U.S.A. and free speech in Czech Republic.

Re: [PATCH] missing checks in exec_permission_light()

[pmenage]

In article <0C01A29FBAE24448A792F5C68F5EA47D27C6C7@nasdaq.ms.ensim.com>,
you write:
>Hi!
>
>> +	if (S_ISDIR(inode->i_mode) && capable(CAP_DAC_READ_SEARCH))
>> +		return 0;
>
>Is this right? This means that root can do cat /, no? That does not
>seem like expected behaviour.
>									Pavel

exec_permission_lite() is only used for MAY_EXEC checks.

Paul