standard queue implementation?

standard queue implementation?

[Eli Carter]

All,

I did a quick look in include/linux for a standard implementation of an
array-based circular queue, but I didn't see one.

I was thinking something that could be declared, allocated, and then
used with an addq and a removeq.  A deallocator would also be good.

Is there such a beast in the kernel?  If not, it seems that having
something like this would reduce the potential for bugs.

Thoughts?

Eli 
-----------------------.   No wonder we didn't get this right first time
Eli Carter             |      through. It's not really all that horribly 
eli.carter(at)inet.com `- complicated, but the _details_ kill you. Linus

Re: standard queue implementation?

[george anzinger]

Eli Carter wrote:
> 
> All,
> 
> I did a quick look in include/linux for a standard implementation of an
> array-based circular queue, but I didn't see one.
> 
> I was thinking something that could be declared, allocated, and then
> used with an addq and a removeq.  A deallocator would also be good.
> 
> Is there such a beast in the kernel?  If not, it seems that having
> something like this would reduce the potential for bugs.
> 
> Thoughts?
> 
Are you possibly looking for include/linux/list.h ?

Routines to build and manager doubly linked circular lists.

George

Re: standard queue implementation?

[Eli Carter]

george anzinger wrote:
> 
> Eli Carter wrote:
> >
> > All,
> >
> > I did a quick look in include/linux for a standard implementation of an
> > array-based circular queue, but I didn't see one.
> >
> > I was thinking something that could be declared, allocated, and then
> > used with an addq and a removeq.  A deallocator would also be good.
> >
> > Is there such a beast in the kernel?  If not, it seems that having
> > something like this would reduce the potential for bugs.
> >
> > Thoughts?
> >
> Are you possibly looking for include/linux/list.h ?
> 
> Routines to build and manager doubly linked circular lists.

I've seen that, but no.  I want a queue of pointers, and the queue
can/should be of fixed length.  (I don't want to deal with
allocating/deallocating nodes for this... I just need something simple.)

For now, I'll just write my own, but I may try submitting it as a
"kernel library" type thing later on.

Thanks,

Eli
-----------------------.   No wonder we didn't get this right first time
Eli Carter             |      through. It's not really all that horribly 
eli.carter(at)inet.com `- complicated, but the _details_ kill you. Linus