Re: [PATCH] RDMA/cm: Add min length checks to user structure copies

From: Jason Gunthorpe <jgg@nvidia.com>
To: Leon Romanovsky <leonro@mellanox.com>
Cc: <linux-rdma@vger.kernel.org>,
	<syzbot+086ab5ca9eafd2379aa6@syzkaller.appspotmail.com>,
	<syzbot+7446526858b83c8828b2@syzkaller.appspotmail.com>
Subject: Re: [PATCH] RDMA/cm: Add min length checks to user structure copies
Date: Fri, 24 Jul 2020 10:52:44 -0300	[thread overview]
Message-ID: <20200724135244.GK2021234@nvidia.com> (raw)
In-Reply-To: <20200724134545.GB5479@unreal>

On Fri, Jul 24, 2020 at 04:45:45PM +0300, Leon Romanovsky wrote:
> On Fri, Jul 24, 2020 at 10:19:29AM -0300, Jason Gunthorpe wrote:
> > These are missing throughout ucma, it harmlessly copies garbage from
> > userspace, but in this new code which uses min to compute the copy length
> > it can result in uninitialized stack memory. Check for minimum length at
> > the very start.
> >
> >   BUG: KMSAN: uninit-value in ucma_connect+0x2aa/0xab0 drivers/infiniband/core/ucma.c:1091
> >   CPU: 0 PID: 8457 Comm: syz-executor069 Not tainted 5.8.0-rc5-syzkaller #0
> >   Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
> >   Call Trace:
> >    __dump_stack lib/dump_stack.c:77 [inline]
> >    dump_stack+0x1df/0x240 lib/dump_stack.c:118
> >    kmsan_report+0xf7/0x1e0 mm/kmsan/kmsan_report.c:121
> >    __msan_warning+0x58/0xa0 mm/kmsan/kmsan_instr.c:215
> >    ucma_connect+0x2aa/0xab0 drivers/infiniband/core/ucma.c:1091
> >    ucma_write+0x5c5/0x630 drivers/infiniband/core/ucma.c:1764
> >    do_loop_readv_writev fs/read_write.c:737 [inline]
> >    do_iter_write+0x710/0xdc0 fs/read_write.c:1020
> >    vfs_writev fs/read_write.c:1091 [inline]
> >    do_writev+0x42d/0x8f0 fs/read_write.c:1134
> >    __do_sys_writev fs/read_write.c:1207 [inline]
> >    __se_sys_writev+0x9b/0xb0 fs/read_write.c:1204
> >    __x64_sys_writev+0x4a/0x70 fs/read_write.c:1204
> >    do_syscall_64+0xb0/0x150 arch/x86/entry/common.c:386
> >    entry_SYSCALL_64_after_hwframe+0x44/0xa9
> >
> > Fixes: 34e2ab57a911 ("RDMA/ucma: Extend ucma_connect to receive ECE parameters")
> > Fixes: 0cb15372a615 ("RDMA/cma: Connect ECE to rdma_accept")
> > Reported-by: syzbot+086ab5ca9eafd2379aa6@syzkaller.appspotmail.com
> > Reported-by: syzbot+7446526858b83c8828b2@syzkaller.appspotmail.com
> > Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
> >  drivers/infiniband/core/ucma.c | 4 ++++
> >  1 file changed, 4 insertions(+)
> >
> > diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c
> > index 5b87eee8ccc8b6..d03dacaef78805 100644
> > +++ b/drivers/infiniband/core/ucma.c
> > @@ -1084,6 +1084,8 @@ static ssize_t ucma_connect(struct ucma_file *file, const char __user *inbuf,
> >  	size_t in_size;
> >  	int ret;
> >
> > +	if (in_len < offsetofend(typeof(cmd), reserved))
> > +		return -EINVAL;
> 
> This check wasn't before the patches citied in Fixes lines. This is why
> I didn't add them while extended ucma_*.

It wasn't a bug before.

> >  	in_size = min_t(size_t, in_len, sizeof(cmd));
> >  	if (copy_from_user(&cmd, inbuf, in_size))
> >  		return -EFAULT;

Because this used to be:

       if (copy_from_user(&cmd, inbuf, sizeof(cmd)))

Which always completely filled the stack memory.

Jason