LFS Book 6.3 - udev rule 25 is creating wrong firewire device

[mladen]

Hi. I successfully made a LFS system from Book 6.3. After that i installed the libraw1394 and dvgrab to capture video from DV camera. dvgrab and testlibraw did not work with error: no raw1394 device found ( I had raw 1394 module loaded ) I found out that udev is creating raw1394 device in /dev/raw/ instead in /dev . After commenting out line KERNEL == raw[0-9] ... from udev rules 25 everything is working as expected. Conclusion: udev rules 25 is creating raw disk device in /dev/raw instead of raw1394 device in /dev .

Keep up good work.

mladen

[bryan@linuxfromscratch.org]

The rule's match is wrong, yes (it's matching both raw-disk-I/O devices and Firewire devices). But I don't think that removing the rule entirely is the correct solution, as /dev/raw/ is supposed to be where the raw-disk-I/O nodes go. I think a better solution is to figure out an extra key to use so the rule only matches raw disk devices, instead of both disk and Firewire.

Or, for the development book, to move to the udev-provided ruleset (we need to do that anyway before 7.0). I assume they have this fixed; let me go check...

Yes, it looks like the udev sample ruleset has a SUBSYSTEM=="raw" added to that KERNEL=="raw[0-9]*" rule (probably to prevent the Firewire match that you're seeing). Does it fix the problem if you recreate that rule, and add the SUBSYSTEM match to it?

[bryan@linuxfromscratch.org]

I believe this is fixed in r8545, since we finally moved to upstream's rules.