Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 16 Dec 2006 14:44:26 +0300
From:      Andrey Chernov <ache@FreeBSD.org>
To:        Robert Watson <rwatson@FreeBSD.org>
Cc:        current@FreeBSD.org
Subject:   Re: sysv_ipc.c broken in v1.30 (was Re: sysvshm appearse broken in -current)
Message-ID:  <20061216114426.GA7735@nagual.pp.ru>
In-Reply-To: <20061216112117.P72986@fledge.watson.org>
References:  <20061216055903.GA2712@nagual.pp.ru> <20061216111656.GA7501@nagual.pp.ru> <20061216112117.P72986@fledge.watson.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Dec 16, 2006 at 11:25:55AM +0000, Robert Watson wrote:
> Yes, you can find the details in kern/106078.
> 
> The thrust of the problem is that applications apparently pass access mode 
> arguments to shmget() in situations other than file creation, which isn't 
> documented in the spec.  I've been doing a bit of on-and-off research on 
> this, but need to do some more before I'm ready to change our 
> implementation to simply ignore the argument.  I hope to look at it again 
> this week sometime; it's unclear to me what applications are trying to 
> accomplish with the mode field in the non-IPC_CREAT case, and none of the 
> man pages and documentation I've found on various UNIX systems to date 
> suggest anything in particular.

See t-shm.c code in either dk-milter or dkim-milter to gather the sample 
of operation. Those test written in way to be passed in all sysv ipc 
conformant machines. Which isn't our FreeBSD now :(

I think removing that old code is the root of the problem:

                 * Always permit the creator/owner to update the object
                 * protections regardless of whether the object mode
                 * permits it.
                 */
                if (mode & IPC_M)
                        return (0);

I.e. old code not even check for IPC_W or IPC_R in case of IPC_M presense.

Moreover, old code allows _anything_ for suser:

        if ((mode & perm->mode) != mode) {
                if (suser(td) != 0)
                        return (EACCES);
	}

> >On Sat, Dec 16, 2006 at 08:59:03AM +0300, Andrey Chernov wrote:
> >>It seems shm is broken in very recent -current.
> >>Trying to build dkim-milter or dk-milter port (from root, of course) I
> >>got:
> >>
> >>./t-shm
> >>shmget: Permission denied
> >>^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> >>shminit failed: Permission denied
> >>1..bad! t-shm.c:260 r == 0
> >>add -DSM_CONF_SHM=0 to confENVDEF in devtools/Site/site.config.m4
> >>and start over.
> >>0 of 1 tests completed successfully
> >>*** 1 error in test! ***
> >>2..bad! t-shm.c:177 cnt <= MAX_CNT
> >>add -DSM_CONF_SHM=0 to confENVDEF in devtools/Site/site.config.m4
> >>and start over.
> >>4 of 5 tests completed successfully
> >>*** 1 error in test! ***
> >
> >
> >-- 
> >http://ache.pp.ru/
> >


-- 
http://ache.pp.ru/



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20061216114426.GA7735>