linux-kernel - Re: [PATCH 5/6] usb: musb: gadget: implement send

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]

Date:   Wed, 31 Oct 2018 19:26:20 -0400
From:   Paul Elder <paul.elder@...asonboard.com>
To:     Bin Liu <b-liu@...com>, laurent.pinchart@...asonboard.com,
        kieran.bingham@...asonboard.com, gregkh@...uxfoundation.org,
        linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org,
        balbi@...nel.org, stern@...land.harvard.edu, rogerq@...com
Subject: Re: [PATCH 5/6] usb: musb: gadget: implement send_response

Hi Bin,

On Thu, Oct 11, 2018 at 11:07:46AM -0500, Bin Liu wrote:
> Hi,
> 
> On Tue, Oct 09, 2018 at 10:49:02PM -0400, Paul Elder wrote:
> > This patch implements a mechanism to signal the MUSB driver to reply to
> > a control OUT request with STALL or ACK.
> > 
> > Signed-off-by: Paul Elder <paul.elder@...asonboard.com>
> > Reviewed-by: Laurent Pinchart <laurent.pinchart@...asonboard.com>
> 
> The patch looks good to me, here is my Acked-by:
> 
> Acked-by: Bin Liu <b-liu@...com>
> 
> but I am unable to test this patch set - the current Greg's usb-next
> tree gives deadlock error between composite_disconnect() and
> usb_function_deactivate() when loading g_webcam.ko on BeagleboneBlack.
> The error happens before applying this patch set.

We don't use g_webcam anymore, because it doesn't offer the flexibility
that configfs does (for example, only one function can be configured with
g_webcam). There are no features that g_webcam offers that configfs doesn't.

I was unable to reproduce the deadlock that you mention on Greg's
usb-next tree. Which commit were you on?
I did, however, get the deadlock that you mention upon *killing* the
userspace application providing the stream, not when loading g_webcam.ko.

Here is a sample script for setting up a UVC gadget through configfs
(I haven't tested this exact script; I extracted the functional
components):

CONFIGFS="/sys/kernel/config"
GADGET="$CONFIGFS/usb_gadget"
VID="0x0525"
PID="0xa4a2"
SERIAL="0123456789"
MANUF=$(hostname)
PRODUCT="UVC Gadget"
UDC=`ls /sys/class/udc`

cd $GADGET/g1
echo $VID > idVendor
echo $PID > idProduct

mkdir -p strings/0x409
echo $SERIAL > strings/0x409/serialnumber
echo $MANUF > strings/0x409/manufacturer
echo $PRODUCT > strings/0x409/product

mkdir configs/c.1
mkdir configs/c.1/strings/0x409

# create uvc
CONFIG="configs/c.1"
FUNCTION="uvc.0"

mkdir functions/$FUNCTION

# create frame 640x360 uncompressed
WIDTH=640
HEIGHT=360

wdir=functions/$FUNCTION/streaming/uncompressed/u/${HEIGHT}p

mkdir $wdir
echo $WIDTH > $wdir/wWidth
echo $HEIGHT > $wdir/wHeight
echo $(( $WIDTH * $HEIGHT * 2 )) > $wdir/dwMaxVideoFrameBufferSize
cat <<EOF > $wdir/dwFrameInterval
666666
100000
5000000
EOF
# end create frame

mkdir functions/$FUNCTION/streaming/header/h
cd functions/$FUNCTION/streaming/header/h
ln -s ../../uncompressed/u
cd ../../class/fs
ln -s ../../header/h
cd ../../class/hs
ln -s ../../header/h
cd ../../../control
mkdir header/h
ln -s header/h class/fs
ln -s header/h class/ss
cd ../../../

# Set the packet size: uvc gadget max size is 3k...
echo 3072 > functions/$FUNCTION/streaming_maxpacket
echo 2048 > functions/$FUNCTION/streaming_maxpacket
echo 1024 > functions/$FUNCTION/streaming_maxpacket

ln -s functions/$FUNCTION configs/c.1
# end create uvc

echo $UDC > UDC


Thanks,

Paul