Package: libmtp-common
Version: 1.1.17-2
Severity: important
See my original bug report on Ubuntu:
https://bugs.launchpad.net/bugs/1863239
Hi,
When testing HPLIP I found out that CUPS backends can access my printer
only when they are running as root, when they are running as the special
user lp, as it is usually the case, they cannot access the printer.
So I tried to find out why and saw that the /dev/bus/usb/*/* device file
for the printer has group ownership "audio" and not "lp":
till@till-x1yoga:~/ubuntu/hplip/focal/debian/hplip-3.19.12+dfsg0$ ll
/dev/bus/usb/*/*
crw-rw-r-- 1 root root 189, 0 Feb 11 14:17 /dev/bus/usb/001/001
crw-rw-r-- 1 root root 189, 2 Feb 11 14:17 /dev/bus/usb/001/003
crw-rw-r-- 1 root root 189, 3 Feb 11 14:17 /dev/bus/usb/001/004
crw-rw-r-- 1 root plugdev 189, 4 Feb 14 12:23 /dev/bus/usb/001/005
crw-rw-r-- 1 root root 189, 5 Feb 11 14:17 /dev/bus/usb/001/006
crw-rw----+ 1 root audio 189, 62 Feb 14 12:37 /dev/bus/usb/001/063
crw-rw-r-- 1 root root 189, 128 Feb 11 14:17 /dev/bus/usb/002/001
crw-rw-r-- 1 root root 189, 130 Feb 13 09:38 /dev/bus/usb/002/003
till@till-x1yoga:~/ubuntu/hplip/focal/debian/hplip-3.19.12+dfsg0$
The printer is the device /dev/bus/usb/001/063:
till@till-x1yoga:~/ubuntu/hplip/focal/debian/hplip-3.19.12+dfsg0$ lsusb
Bus 002 Device 003: ID 0b95:1790 ASIX Electronics Corp. AX88179 Gigabit
Ethernet
Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
Bus 001 Device 005: ID 138a:0097 Validity Sensors, Inc.
Bus 001 Device 004: ID 04f2:b5ce Chicony Electronics Co., Ltd Integrated
Camera
Bus 001 Device 003: ID 8087:0a2b Intel Corp.
Bus 001 Device 006: ID 056a:50b7 Wacom Co., Ltd Pen and multitouch sensor
Bus 001 Device 063: ID 03f0:7a12 HP, Inc HP OfficeJet Pro 8730
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
till@till-x1yoga:~/ubuntu/hplip/focal/debian/hplip-3.19.12+dfsg0$
This turned out to be a bug in the UDEV rules.
The file /lib/udev/rules.d/69-libmtp.rules of the libmtp-common assigns
"audio" group ownership to devices which are supposed to be audio or
video players and allow uploading files to them via USB using the MTP
protocol.
First it includes some devices explicitly and then it lists thousands of
supported devices. In the end there is some rule for passing a wide
range of devices through an auto-probing:
---------
# Autoprobe vendor-specific, communication and PTP devices
ENV{ID_MTP_DEVICE}!="1", ENV{MTP_NO_PROBE}!="1",
ENV{COLOR_MEASUREMENT_DEVICE}!="1", ENV{ID_GPHOTO}!="1",
ENV{libsane_matched}!="yes", ATTR{bDeviceClass}=="00|02|06|ef|ff",
PROGRAM="mtp-probe /sys$env{DEVPATH} $attr{busnum} $attr{devnum}",
RESULT=="1", SYMLINK+="libmtp-%k", MODE="660", GROUP="audio",
ENV{ID_MTP_DEVICE}="1", ENV{ID_MEDIA_PLAYER}="1"
---------
And this auto-probing tests positive on my printer:
till@till-x1yoga:~$ /lib/udev/mtp-probe
/sys/devices/pci0000:00/0000:00:14.0/usb1/1-1 001 063
1
till@till-x1yoga:~$
Path taken from the output of "sudo udevadm monitor -e", blob with
"ID_MEDIA_PLAYER" in it.
The device path of the printer I have taken from the blob in the udevadm
output (attached to previous comment) which also contains
"ID_MEDIA_PLAYER=1". So mtp-probe identifies my printer as a media
player and assigns the device file to the "audio" group.
I assume that this happens to most or even all HP printers, so an
exclusion of only my device via Vendor and Product ID would not be the
correct solution.
There was already a measure against wrongly identifying HP printers as
media players, but it is a rather dirty workaround which does not work
any more (therefore this bug). The rule in the end of 69-libmtp.rules
checks the absence of the env variable libsane_matched and this variable
is set for all HP printers by HPLIP. First, this rule fails miserably if
HPLIP is not installed, and I cannot imagine that the libmtp package
depends on HPLIP only to identify unsupported devices. Also the libmtp
rules are applied for both "add" and "bind" actions, whereas the rules
of HPLIP (56-hpmud.rules) are only applied for "add" and so the bug
happens on a "bind" action, here the HPLIP rules do not set said env
variable and so the libmtp rules probe the HP printers.
One can theoretically work around this problem by mucking with the UDEV
rules of HPLIP, but this is a REALLY DIRTY workaround, so please DO NOT
add an hplip task to this bug report.
In addition, HPLIP will not be installed by default any more in the not
too far future, as prnting and scanning will get snapped. Also we want
printer driver Snaps (Printer Applications) not to run as root if
possible, so we need to be sure that USB printer device files always
belong to the group "lp" for all printer manufacturers and without HPLIP.
This patch would fix the bug:
----------
--- /lib/udev/rules.d/69-libmtp.rules~ 2020-02-11 13:06:23.000000000 +0100
+++ /lib/udev/rules.d/69-libmtp.rules 2020-03-19 21:01:48.696689026 +0100
@@ -26,6 +26,10 @@
ATTR{idVendor}=="0471", ATTR{idProduct}=="083f", GOTO="libmtp_rules_end"
# DUALi NFC readers
ATTR{idVendor}=="1db2", ATTR{idProduct}=="060*", GOTO="libmtp_rules_end"
+# HP printers
+ATTR{idVendor}=="03f0", ENV{ID_USB_INTERFACES}=="*:0701??:*|*:ffcc00:",
GOTO="libmtp_rules_end"
+# Printers
+ENV{ID_USB_INTERFACES}=="*:0701??:*", GOTO="libmtp_rules_end"
# Creative ZEN Vision
ATTR{idVendor}=="041e", ATTR{idProduct}=="411f", SYMLINK+="libmtp-%k",
MODE="660", GROUP="audio", ENV{ID_MTP_DEVICE}="1", ENV{ID_MEDIA_PLAYER}="1"
# Creative Portable Media Center
----------
It adds two rules to skip printers. The second skips all kinds of
printers by checking whether the standard interface of a printer is
present in the USB device. The first skips HP printers, it is the same
rule as HPLIP uses. I have added it because it covers a second
interface, which is probably of some weird proprietary HP printer
devices. For my printer the bug goes away with this patch, also with any
one of the two rules commented out.
I think one can assume that there is no printer available which doubles
as a media player.
As the file 69-libmtp.rules is not explicitly contained in the package
but gets generated, the generator needs to get patched. An appropriate
patch is attached.
Could you apply this patch to also fix this bug for Debian and to let
the fixed package get synced in Ubuntu?
Thanks in advance
Till
Description: On UDEV discovery do not let printers get auto-probed
Devices which are not explicitly known as supported by libmtp gwt auto-probed
by mtp-probe. Unfortunately, many devices which are not audio or media players
test positive on the auto-probing and so their USB device files get "audio"
group ownership. making them inaccessible for the sub system which is actually
responsible for them. Therefore there are also rules for skipping some
devices.
.
This patch adds skipping rules for printers, as they have to belong to the
"lp" group for CUPS or Printer Applications to be able to access them. The
extra rule for HP printers covers some weird, proprietary devices from HP,
it is copied from HPLIP
Ubuntu-bug: https://bugs.launchpad.net/bugs/1863239
Author: Till Kamppeter <till.kamppe...@gmail.com>
--- a/util/mtp-hotplug.c
+++ b/util/mtp-hotplug.c
@@ -163,6 +163,10 @@
printf("ATTR{idVendor}==\"0471\", ATTR{idProduct}==\"083f\", GOTO=\"libmtp_rules_end\"\n");
printf("# DUALi NFC readers\n");
printf("ATTR{idVendor}==\"1db2\", ATTR{idProduct}==\"060*\", GOTO=\"libmtp_rules_end\"\n");
+ printf("# HP printers\n");
+ printf("ATTR{idVendor}==\"03f0\", ENV{ID_USB_INTERFACES}==\"*:0701??:*|*:ffcc00:\", GOTO=\"libmtp_rules_end\"\n");
+ printf("# Printers\n");
+ printf("ENV{ID_USB_INTERFACES}==\"*:0701??:*\", GOTO=\"libmtp_rules_end\"\n");
break;
case style_udev_old:
printf("# UDEV-style hotplug map for libmtp\n");
