signal/ptrace: Don't leak unitialized kernel memory with PTRACE_PEEK_SIGINFO
authorEric W. Biederman <ebiederm@xmission.com>
Tue, 28 May 2019 23:46:37 +0000 (18:46 -0500)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 19 Jun 2019 06:18:00 +0000 (08:18 +0200)
commit662b831dde6160096091a1a9560281dfd450c68f
tree3069ac7f0f26ef287ec991f6670232518c84688f
parent54a20289cbfbf7e617e5a57dde2d75570bab45ad
signal/ptrace: Don't leak unitialized kernel memory with PTRACE_PEEK_SIGINFO

[ Upstream commit f6e2aa91a46d2bc79fce9b93a988dbe7655c90c0 ]

Recently syzbot in conjunction with KMSAN reported that
ptrace_peek_siginfo can copy an uninitialized siginfo to userspace.
Inspecting ptrace_peek_siginfo confirms this.

The problem is that off when initialized from args.off can be
initialized to a negaive value.  At which point the "if (off >= 0)"
test to see if off became negative fails because off started off
negative.

Prevent the core problem by adding a variable found that is only true
if a siginfo is found and copied to a temporary in preparation for
being copied to userspace.

Prevent args.off from being truncated when being assigned to off by
testing that off is <= the maximum possible value of off.  Convert off
to an unsigned long so that we should not have to truncate args.off,
we have well defined overflow behavior so if we add another check we
won't risk fighting undefined compiler behavior, and so that we have a
type whose maximum value is easy to test for.

Cc: Andrei Vagin <avagin@gmail.com>
Cc: stable@vger.kernel.org
Reported-by: syzbot+0d602a1b0d8c95bdf299@syzkaller.appspotmail.com
Fixes: 84c751bd4aeb ("ptrace: add ability to retrieve signals without removing from a queue (v4)")
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
kernel/ptrace.c