Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 16 Oct 2016 13:27:51 +0000
From:      bugzilla-noreply@freebsd.org
To:        freebsd-bugs@FreeBSD.org
Subject:   [Bug 213527] [patch] [kernel] Rework on functions allocating credentials.
Message-ID:  <bug-213527-8-LTAypUH71I@https.bugs.freebsd.org/bugzilla/>
In-Reply-To: <bug-213527-8@https.bugs.freebsd.org/bugzilla/>
References:  <bug-213527-8@https.bugs.freebsd.org/bugzilla/>

next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D213527

Mateusz Guzik <mjg@FreeBSD.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|freebsd-bugs@FreeBSD.org    |mjg@FreeBSD.org
                 CC|                            |mjg@FreeBSD.org

--- Comment #2 from Mateusz Guzik <mjg@FreeBSD.org> ---
Hi.

The added argument indeed completes part of the task, but the patch is bugg=
y.

You consistently have:

newcred =3D crget(p->p_ucred->cr_agroups);
PROC_LOCK(p);

However, the stability of p_ucred is protected only with the proc lock held.
That is, by the time you read the address stored in p->p_ucred, the object
stored at that address can be freed. The crget cannot be moved inside becau=
se
crget can sleep in an unbound manner, while the lock in question disallows
that.

When dealing with the current process, you can cheat a little and use
td->td_ucred as a source for the number of groups.

Finally, I would argue crget() interface should be left as it is. Instead, a
new function (ncrget?) would be introduced and crget would become a wrapper
which uses the current default number of groups as an argument.

--=20
You are receiving this mail because:
You are the assignee for the bug.=



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-213527-8-LTAypUH71I>