Difference wait_event_interruptible and interruptible_wait_on

Difference wait_event_interruptible and interruptible_wait_on

[Hendrik Wiese]

Hello,

I'm porting a device driver from 2.2.14 to 2.6.7 and I got some problems 
doing this...

one of them is the following:
I know that a call to interruptible_wait_on puts a process into sleep 
state and that wait_event_interruptible does the same. But the 
difference is that wait_event_interruptible needs a condition to pass to 
wake up the processes. I do not need that mechanism since I wake up the 
processes at other places inside my driver with wake_up_interruptible 
calls. So how do I get a function similar to interruptible_wait_on where 
no condition is needed using kernel 2.6?

Thanks a lot and please CC me, 'cos I haven't subscribed to the LKML yet.

Kind regards,
Hendrik

Re: Difference wait_event_interruptible and interruptible_wait_on

[Davide Rossetti]

Hendrik Wiese wrote:

> Hello,
>
> I'm porting a device driver from 2.2.14 to 2.6.7 and I got some 
> problems doing this...
>
> one of them is the following:
> I know that a call to interruptible_wait_on puts a process into sleep 
> state and that wait_event_interruptible does the same. But the 
> difference is that wait_event_interruptible needs a condition to pass 
> to wake up the processes. I do not need that mechanism since I wake up 
> the processes at other places inside my driver with 
> wake_up_interruptible calls. So how do I get a function similar to 
> interruptible_wait_on where no condition is needed using kernel 2.6?

I did not check, so maybe I'm wrong, anyway the condition 
wait_event_interruptible is used to avoid the race window between the 
time you decide to call it and the time the process is actually made to 
sleep and ready to be woken up.
regards

Re: Difference wait_event_interruptible and interruptible_wait_on

[Hendrik Wiese]

Davide Rossetti wrote:

> Hendrik Wiese wrote:
>
>> Hello,
>>
>> I'm porting a device driver from 2.2.14 to 2.6.7 and I got some 
>> problems doing this...
>>
>> one of them is the following:
>> I know that a call to interruptible_wait_on puts a process into sleep 
>> state and that wait_event_interruptible does the same. But the 
>> difference is that wait_event_interruptible needs a condition to pass 
>> to wake up the processes. I do not need that mechanism since I wake 
>> up the processes at other places inside my driver with 
>> wake_up_interruptible calls. So how do I get a function similar to 
>> interruptible_wait_on where no condition is needed using kernel 2.6?
>
>
> I did not check, so maybe I'm wrong, anyway the condition 
> wait_event_interruptible is used to avoid the race window between the 
> time you decide to call it and the time the process is actually made 
> to sleep and ready to be woken up.
> regards
>
>
Thanks for your reply.

So what should such a condition be? What should be put there?

Re: Difference wait_event_interruptible and interruptible_wait_on

[Arjan van de Ven]

On Wed, 2004-11-24 at 14:42 +0100, Hendrik Wiese wrote:

> 
> So what should such a condition be? What should be put there?

depends on your code.. do you have an URL to your driver source so that
people can see what you're asking ?

Re: Difference wait_event_interruptible and interruptible_wait_on

[Hendrik Wiese]

Arjan van de Ven wrote:

>On Wed, 2004-11-24 at 14:42 +0100, Hendrik Wiese wrote:
>
>  
>
>>So what should such a condition be? What should be put there?
>>    
>>
>
>depends on your code.. do you have an URL to your driver source so that
>people can see what you're asking ?
>
>
>
>  
>
It is a proprietary driver developed for access to passive and active 
modules inside a vme-rack. Active modules can communicate with each 
other over an IP layer created by the driver on top of the VME 
architecture (dual port ram, Tundra Universe II PCI-VME Bridge) and on 
the other hand a device for communication between active and passive 
modules. I don't know if I may make the source code public, 'cos it is 
property of my employer. I have to communicate this. I know there is 
something called GPL and I also know what it means, but I fear my 
oppinion won't be heard by my employer. Well... it's worth a try.