linux-kernel - Re: [PATCH 5.4 17/78] HID: Fix slab-out-of-bounds read in hid_field

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200205093226.GC1164405@kroah.com>
Date:   Wed, 5 Feb 2020 09:32:26 +0000
From:   Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To:     peter enderborg <peter.enderborg@...y.com>
Cc:     linux-kernel@...r.kernel.org,
        Alan Stern <stern@...land.harvard.edu>,
        Jiri Kosina <jkosina@...e.cz>, stable@...r.kernel.org,
        syzbot+09ef48aa58261464b621@...kaller.appspotmail.com
Subject: Re: [PATCH 5.4 17/78] HID: Fix slab-out-of-bounds read in
 hid_field_extract (Broken!)

On Wed, Feb 05, 2020 at 08:12:27AM +0100, peter enderborg wrote:
> On 1/14/20 11:00 AM, Greg Kroah-Hartman wrote:
> > From: Alan Stern <stern@...land.harvard.edu>
> >
> > commit 8ec321e96e056de84022c032ffea253431a83c3c upstream.
> >
> > The syzbot fuzzer found a slab-out-of-bounds bug in the HID report
> > handler.  The bug was caused by a report descriptor which included a
> > field with size 12 bits and count 4899, for a total size of 7349
> > bytes.
> >
> > The usbhid driver uses at most a single-page 4-KB buffer for reports.
> > In the test there wasn't any problem about overflowing the buffer,
> > since only one byte was received from the device.  Rather, the bug
> > occurred when the HID core tried to extract the data from the report
> > fields, which caused it to try reading data beyond the end of the
> > allocated buffer.
> >
> > This patch fixes the problem by rejecting any report whose total
> > length exceeds the HID_MAX_BUFFER_SIZE limit (minus one byte to allow
> > for a possible report index).  In theory a device could have a report
> > longer than that, but if there was such a thing we wouldn't handle it
> > correctly anyway.
> >
> > Reported-and-tested-by: syzbot+09ef48aa58261464b621@...kaller.appspotmail.com
> > Signed-off-by: Alan Stern <stern@...land.harvard.edu>
> > CC: <stable@...r.kernel.org>
> > Signed-off-by: Jiri Kosina <jkosina@...e.cz>
> > Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
> >
> > ---
> >  drivers/hid/hid-core.c |    6 ++++++
> >  1 file changed, 6 insertions(+)
> >
> > --- a/drivers/hid/hid-core.c
> > +++ b/drivers/hid/hid-core.c
> > @@ -288,6 +288,12 @@ static int hid_add_field(struct hid_pars
> >  	offset = report->size;
> >  	report->size += parser->global.report_size * parser->global.report_count;
> >  
> > +	/* Total size check: Allow for possible report index byte */
> > +	if (report->size > (HID_MAX_BUFFER_SIZE - 1) << 3) {
> > +		hid_err(parser->device, "report is too long\n");
> > +		return -1;
> > +	}
> > +
> >  	if (!parser->local.usage_index) /* Ignore padding fields */
> >  		return 0;
> >  
> >
> >
> >
> This patch breaks Elgato StreamDeck.

Does that mean the device is broken with a too-large of a report?

Is it broken in Linus's tree?  If so, can you work with the HID
developers to fix it there so we can backport the fix to all stable
trees?

thanks,

greg k-h