Short experiment with libudev to support media controller concept
Hans,
The inline source file at the end of this post is a small program I used
to play with libudev to see if it would complement the media controller
concept (as you suspected it would).
Documentation on the libudev calls is here:
http://www.kernel.org/pub/linux/utils/kernel/hotplug/libudev/
The test program I wrote takes a (type,major,minor) tuple and lists the
device node and device symlinks as fetched by libudev.
My test setup was a little strange since libudev is no longer maintained
separately but is bundled in with udev. On my Fedora 9 system I have
udev v124 (Fedora 9 stock) and libudev v143 (custom built from the udev
143 source).
Here's some output:
$ ./finddev -c -M 1 -m 3
Requested device: type 'c', major 1, minor 3
Device directory path: '/dev'
Device node: '/dev/null'
Device link: '/dev/XOR'
$ ls -al /dev/ | grep '[ /]null'
crw-rw-rw- 1 root root 1, 3 2009-07-04 08:34 null
lrwxrwxrwx 1 root root 4 2009-07-04 08:34 X0R -> null
lrwxrwxrwx 1 root root 4 2009-07-04 08:34 XOR -> null
(Hmmm, not perfect for /dev/null)
$ ./finddev -b -M 11 -m 0
Requested device: type 'b', major 11, minor 0
Device directory path: '/dev'
Device node: '/dev/sr0'
Device link: '/dev/scd0'
Device link: '/dev/disk/by-path/pci-:00:14.1-scsi-1:0:0:0'
Device link: '/dev/cdrom'
Device link: '/dev/cdrw'
$ ls -alR /dev/* | grep '[ /]sr0'
lrwxrwxrwx 1 root root 3 2009-07-04 08:34 /dev/cdrom -> sr0
lrwxrwxrwx 1 root root 3 2009-07-04 08:34 /dev/cdrw -> sr0
lrwxrwxrwx 1 root root 3 2009-07-04 08:34 /dev/scd0 -> sr0
brw-rw+ 1 root disk11, 0 2009-07-04 08:34 /dev/sr0
lrwxrwxrwx 1 root root 9 2009-07-04 08:34 pci-:00:14.1-scsi-1:0:0:0 ->
../../sr0
(OK for the CDROM drive.)
$ ./finddev -c -M 81 -m 9
Requested device: type 'c', major 81, minor 9
Device directory path: '/dev'
Device node: '/dev/video0'
Device link: '/dev/video'
$ ls -alR /dev/* | grep '[ /]video0'
lrwxrwxrwx 1 root root 6 2009-07-04 08:34 /dev/video -> video0
crw-rw+ 1 root root81, 9 2009-07-04 08:34 /dev/video0
(OK for video nodes)
$ ./finddev -c -M 116 -m 6
Requested device: type 'c', major 116, minor 6
Device directory path: '/dev'
Device node: '/dev/snd/pcmC0D0p'
$ ls -alR /dev/* | grep '[ /]pcmC0D0p'
crw-rw+ 1 root root 116, 6 2009-07-04 13:43 pcmC0D0p
(OK for ALSA PCM stream nodes).
Do you have any other particular questions about libudev's capabilities?
Regards,
Andy
/*
* finddev.c - Find a device node given (type,major,minor) using libudev
* Copyright (C) 2009 Andy Walls
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License along
* with this program; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
*
* Compilation:
* gcc -o finddev finddev.c -Wall -ludev
*
* Example invocation (/dev/null on my system):
* ./finddev -c -M 1 -m 3
*/
#include
#include
#include
#include
#include
#include
struct parsed_args {
int major;
int minor;
char type;
};
int parse_args(int argc, char *argv[], struct parsed_args *args)
{
int c, ret;
ret = 0;
args->major = args->minor = -1;
args->type = 'c';
while ((c = getopt(argc, argv, "bcM:m:")) != -1) {
switch (c) {
case 'b': args->type = 'b'; break;
case 'c': args->type = 'c'; break;
case 'M': args->major = atoi(optarg); break;
case 'm': args->minor = atoi(optarg); break;
default: ret = -1; break;
}
}
if (ret || args->major == -1 || args->minor == -1) {
fprintf(stderr, "Usage: %s [-b|-c] -M major -m minor\n",
argv[0]);
ret = -1;
}
return ret;
}
int main(int argc, char *argv[])
{
struct parsed_args args;
dev_t devnum;
struct udev *udev;
struct udev_device *udev_device;
const char *s;
struct udev_list_entry *udev_list_entry;
if (parse_args(argc,argv, &args))
exit(1);
printf("Requested device: type '%c', major %d, minor %d\n",
args.type, args.major, args.minor);
devnum = makedev(args.major, args.minor);
udev = udev_new();
if (udev == NULL) {
fpri
Re: Short experiment with libudev to support media controller concept
On Saturday 04 July 2009 19:52:15 Andy Walls wrote: > Hans, > > The inline source file at the end of this post is a small program I used > to play with libudev to see if it would complement the media controller > concept (as you suspected it would). > > Documentation on the libudev calls is here: > > http://www.kernel.org/pub/linux/utils/kernel/hotplug/libudev/ > > The test program I wrote takes a (type,major,minor) tuple and lists the > device node and device symlinks as fetched by libudev. > > My test setup was a little strange since libudev is no longer maintained > separately but is bundled in with udev. On my Fedora 9 system I have > udev v124 (Fedora 9 stock) and libudev v143 (custom built from the udev > 143 source). > > Here's some output: > > $ ./finddev -c -M 1 -m 3 > Requested device: type 'c', major 1, minor 3 > Device directory path: '/dev' > Device node: '/dev/null' > Device link: '/dev/XOR' > > $ ls -al /dev/ | grep '[ /]null' > crw-rw-rw- 1 root root 1, 3 2009-07-04 08:34 null > lrwxrwxrwx 1 root root 4 2009-07-04 08:34 X0R -> null > lrwxrwxrwx 1 root root 4 2009-07-04 08:34 XOR -> null > > (Hmmm, not perfect for /dev/null) > > > $ ./finddev -b -M 11 -m 0 > Requested device: type 'b', major 11, minor 0 > Device directory path: '/dev' > Device node: '/dev/sr0' > Device link: '/dev/scd0' > Device link: '/dev/disk/by-path/pci-:00:14.1-scsi-1:0:0:0' > Device link: '/dev/cdrom' > Device link: '/dev/cdrw' > > $ ls -alR /dev/* | grep '[ /]sr0' > lrwxrwxrwx 1 root root 3 2009-07-04 08:34 /dev/cdrom -> sr0 > lrwxrwxrwx 1 root root 3 2009-07-04 08:34 /dev/cdrw -> sr0 > lrwxrwxrwx 1 root root 3 2009-07-04 08:34 /dev/scd0 -> sr0 > brw-rw+ 1 root disk11, 0 2009-07-04 08:34 /dev/sr0 > lrwxrwxrwx 1 root root 9 2009-07-04 08:34 pci-:00:14.1-scsi-1:0:0:0 -> > ../../sr0 > > (OK for the CDROM drive.) > > > $ ./finddev -c -M 81 -m 9 > Requested device: type 'c', major 81, minor 9 > Device directory path: '/dev' > Device node: '/dev/video0' > Device link: '/dev/video' > > $ ls -alR /dev/* | grep '[ /]video0' > lrwxrwxrwx 1 root root 6 2009-07-04 08:34 /dev/video -> video0 > crw-rw+ 1 root root81, 9 2009-07-04 08:34 /dev/video0 > > (OK for video nodes) > > > $ ./finddev -c -M 116 -m 6 > Requested device: type 'c', major 116, minor 6 > Device directory path: '/dev' > Device node: '/dev/snd/pcmC0D0p' > > $ ls -alR /dev/* | grep '[ /]pcmC0D0p' > crw-rw+ 1 root root 116, 6 2009-07-04 13:43 pcmC0D0p > > (OK for ALSA PCM stream nodes). > > > Do you have any other particular questions about libudev's capabilities? Hi Andy, This looks very promising. Can you try a few things like adding new symlinks for video devices in the udev config file, or renaming the video0 node to, say, mpeg0? If finddev still returns the right nodes, then all that a media controller needs to do is to export the major and minor numbers for each node. That's exactly what I'm hoping for. Note that I will not have a lot of time to work in v4l-dvb for the next 10 to 14 days as I have visitors this week and will be traveling abroad next week. Thanks! Hans -- Hans Verkuil - video4linux developer - sponsored by TANDBERG Telecom -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: Short experiment with libudev to support media controller concept
On Sat, 2009-07-04 at 13:52 -0400, Andy Walls wrote: > Hans, > > The inline source file at the end of this post is a small program I used > to play with libudev to see if it would complement the media controller > concept (as you suspected it would). > > Documentation on the libudev calls is here: > > http://www.kernel.org/pub/linux/utils/kernel/hotplug/libudev/ > > The test program I wrote takes a (type,major,minor) tuple and lists the > device node and device symlinks as fetched by libudev. > > My test setup was a little strange since libudev is no longer maintained > separately but is bundled in with udev. On my Fedora 9 system I have > udev v124 (Fedora 9 stock) and libudev v143 (custom built from the udev > 143 source). > > Here's some output: > > $ ./finddev -c -M 81 -m 9 > Requested device: type 'c', major 81, minor 9 > Device directory path: '/dev' > Device node: '/dev/video0' > Device link: '/dev/video' > > $ ls -alR /dev/* | grep '[ /]video0' > lrwxrwxrwx 1 root root 6 2009-07-04 08:34 /dev/video -> video0 > crw-rw+ 1 root root81, 9 2009-07-04 08:34 /dev/video0 > > (OK for video nodes) > > > $ ./finddev -c -M 116 -m 6 > Requested device: type 'c', major 116, minor 6 > Device directory path: '/dev' > Device node: '/dev/snd/pcmC0D0p' > > $ ls -alR /dev/* | grep '[ /]pcmC0D0p' > crw-rw+ 1 root root 116, 6 2009-07-04 13:43 pcmC0D0p > > (OK for ALSA PCM stream nodes). > > > Do you have any other particular questions about libudev's capabilities? OK, so more tests. Two Major groups: with udev still running and with udev not running (killed after I log in). Case 1: - udev running: - Adding a manual symlink # cd /dev # ln -s video0 mpeg0 Result: finddev using libudev doesn't find the manual symlink Case 2: - udev running - A manual mknod # cd /dev # mknod mpeg0 c 81 9 Result: finddev doesn't find the manually created device node Case 3: - same as case 2, but manually delete /dev/video0 Result: finddev reports /dev/video0 ! :( Case 4: - same as case 3 - add to 50-udev-default.rules: KERNEL=="video[0-9]*", SYMLINK+="mpeg%n" - Reload udev rules udevadm control --reload_rules udevadm trigger --subsystem-match=video4linux Result: findddev reports the new '/dev/mpeg0' symlink for /dev/video0 as well as the the '/dev/video' synmlink and the '/dev/video0' device node. :) Case 5: - same as case 3 - add to 50-udev-default.rules: KERNEL=="video[0-9]*", NAME="mpeg%n" - Reload udev rules udevadm control --reload_rules udevadm trigger --subsystem-match=video4linux Result: SELinux gripes at me because HAL is allowed to acces the attributes of /dev/mpeg*. :) finddev reports the new /dev/mpeg0 and the /dev/video symlink to it and they both exist in the filesystem. :) Case 6: - after case 5 - kill udevd. :) Result: finddev finds /dev/mpeg0 and the /dev/video symlink Case 7: - after case 6: - manually remove /dev/mpeg0 and /dev/video Result: finddev reports /dev/mpeg0 and the /dev/video symlink. :? Apparently libudev uses what's in the udev database and /sys but doesn't look in the /dev directory for manual actions, even when udevd is dead. Regards, Andy > Regards, > Andy -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: Short experiment with libudev to support media controller concept
On Saturday 11 July 2009 21:19:46 Andy Walls wrote: > On Sat, 2009-07-04 at 13:52 -0400, Andy Walls wrote: > > Hans, > > > > The inline source file at the end of this post is a small program I > > used to play with libudev to see if it would complement the media > > controller concept (as you suspected it would). > > > > Documentation on the libudev calls is here: > > > > http://www.kernel.org/pub/linux/utils/kernel/hotplug/libudev/ > > > > The test program I wrote takes a (type,major,minor) tuple and lists the > > device node and device symlinks as fetched by libudev. > > > > My test setup was a little strange since libudev is no longer > > maintained separately but is bundled in with udev. On my Fedora 9 > > system I have udev v124 (Fedora 9 stock) and libudev v143 (custom built > > from the udev 143 source). > > > > Here's some output: > > > > > > $ ./finddev -c -M 81 -m 9 > > Requested device: type 'c', major 81, minor 9 > > Device directory path: '/dev' > > Device node: '/dev/video0' > > Device link: '/dev/video' > > > > $ ls -alR /dev/* | grep '[ /]video0' > > lrwxrwxrwx 1 root root 6 2009-07-04 08:34 /dev/video -> > > video0 crw-rw+ 1 root root81, 9 2009-07-04 08:34 /dev/video0 > > > > (OK for video nodes) > > > > > > $ ./finddev -c -M 116 -m 6 > > Requested device: type 'c', major 116, minor 6 > > Device directory path: '/dev' > > Device node: '/dev/snd/pcmC0D0p' > > > > $ ls -alR /dev/* | grep '[ /]pcmC0D0p' > > crw-rw+ 1 root root 116, 6 2009-07-04 13:43 pcmC0D0p > > > > (OK for ALSA PCM stream nodes). > > > > > > Do you have any other particular questions about libudev's > > capabilities? > > OK, so more tests. Two Major groups: with udev still running and with > udev not running (killed after I log in). > > Case 1: > > - udev running: > - Adding a manual symlink ># cd /dev ># ln -s video0 mpeg0 > > Result: finddev using libudev doesn't find the manual symlink > > > Case 2: > > - udev running > - A manual mknod ># cd /dev ># mknod mpeg0 c 81 9 > > Result: finddev doesn't find the manually created device node > > > Case 3: > - same as case 2, but manually delete /dev/video0 > > Result: finddev reports /dev/video0 ! :( That's not surprising as cases 1-3 by-pass udev, so libudev knows nothing about those changes. > > > Case 4: > - same as case 3 > - add to 50-udev-default.rules: > KERNEL=="video[0-9]*", SYMLINK+="mpeg%n" > - Reload udev rules > udevadm control --reload_rules > udevadm trigger --subsystem-match=video4linux > > Result: findddev reports the new '/dev/mpeg0' symlink for /dev/video0 as > well as the the '/dev/video' synmlink and the '/dev/video0' device > node. :) Nice! > > Case 5: > - same as case 3 > - add to 50-udev-default.rules: > KERNEL=="video[0-9]*", NAME="mpeg%n" > - Reload udev rules > udevadm control --reload_rules > udevadm trigger --subsystem-match=video4linux > > Result: SELinux gripes at me because HAL is allowed to acces the > attributes of /dev/mpeg*. :) > finddev reports the new /dev/mpeg0 and the /dev/video symlink to it and > they both exist in the filesystem. :) Nice again! > > Case 6: > - after case 5 > - kill udevd. :) > > Result: finddev finds /dev/mpeg0 and the /dev/video symlink Nice test :-) and nice result. > Case 7: > - after case 6: > - manually remove /dev/mpeg0 and /dev/video > > Result: finddev reports /dev/mpeg0 and the /dev/video symlink. :? Similar to cases 1-3: you bypass udev so libudev won't know about it. > > > Apparently libudev uses what's in the udev database and /sys but doesn't > look in the /dev directory for manual actions, even when udevd is dead. That makes sense as udev is basically in charge of the /dev/ contents and making manual /dev/ changes while udev is running defeats the purpose of udev. I'm really happy finddev finds the symlinks as well. This means that the major and minor numbers are all that is needed in order to reliably find the right device nodes in /dev. Thanks! Hans > > Regards, > Andy > > > Regards, > > Andy > > -- > To unsubscribe from this list: send the line "unsubscribe linux-media" in > the body of a message to majord...@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- Hans Verkuil - video4linux developer - sponsored by TANDBERG -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
