Michael Biebl | 21 Nov 19:18
Favicon

Bug#506458: /etc/udev/rules.d/90-hal.rules missing '@' in socket line

David Madore wrote:
> Package: hal
> Version: 0.5.11-6
> 
> The file /etc/udev/rules.d/90-hal.rules contains the line
> 
> RUN+="socket:/org/freedesktop/hal/udev_event"
> 
> Obviously the /org/freedesktop/hal/udev_event Unix domain socket does
> not exist nor is it supposed to.  However, the socket of the same name
> exists in the abstract domain (with hald listening to it).  According
> to the udev manual page, to connect to an abstract socket, an '@' sign
> should start the name of the socket.  So I suggest rewriting the file
> as:
> 
> RUN+="socket:@/org/freedesktop/hal/udev_event"
> 
> so that udevd will actually talk to hald.
> 
> (I'm baffled at how this could have gone unnoticed for so long...  How
> can hal even work if udev doesn't forward uevent information to it?)
> 

I guess the reason why nobody noticed anything simply is, that it just works atm.

Maybe udev checks for both, a real socket and a abstract namespace socket, when
no '@' is given.

Martin?

(Continue reading)

Michael Biebl | 23 Nov 18:17
Favicon

Bug#506458: /etc/udev/rules.d/90-hal.rules missing '@' in socket line

Michael Biebl wrote:
> David Madore wrote:
>> Package: hal
>> Version: 0.5.11-6
>>
>> The file /etc/udev/rules.d/90-hal.rules contains the line
>>
>> RUN+="socket:/org/freedesktop/hal/udev_event"
>>
>> Obviously the /org/freedesktop/hal/udev_event Unix domain socket does
>> not exist nor is it supposed to.  However, the socket of the same name
>> exists in the abstract domain (with hald listening to it).  According
>> to the udev manual page, to connect to an abstract socket, an '@' sign
>> should start the name of the socket.  So I suggest rewriting the file
>> as:
>>
>> RUN+="socket:@/org/freedesktop/hal/udev_event"
>>
>> so that udevd will actually talk to hald.
>>
>> (I'm baffled at how this could have gone unnoticed for so long...  How
>> can hal even work if udev doesn't forward uevent information to it?)
>>
> 
> I guess the reason why nobody noticed anything simply is, that it just works atm.
> 
> Maybe udev checks for both, a real socket and a abstract namespace socket, when
> no '@' is given.
> 
> Martin?
(Continue reading)


Gmane