[ENBD] 2.4.32 write & m_r problem

[Peter T. Breuer]

"Also sprach Arne Wiebalck:"
> There are also some warnings that we may delete on this occasion:

Thanks - that one I have done here ...

> /home/wiebalck/nbd-2.4.32/nbd/shmem.c: In function `init_shmem':
> /home/wiebalck/nbd-2.4.32/nbd/shmem.c:389: warning: assignment from
> incompatible pointer type
> 
> in shemem.h we have:
> 
>  int (*wait_seqno_timeout)(struct nbd_shmem *data, unsigned seqno,
>       unsigned mdelay);
>       ^^^^^^^^
> 
> but in shmem.c we assign
>  maybe_wait_for_seqno (struct nbd_shmem *data, unsigned seqno,
>                        int mdelay)
>                        ^^^

Done.

> In file included from /home/wiebalck/nbd-2.4.32/nbd/sem.c:16:
> /home/wiebalck/nbd-2.4.32/nbd/cliserv.h: In function `ntohll':
> /home/wiebalck/nbd-2.4.32/nbd/cliserv.h:136: warning: implicit declaration
> of function `ntohl'
> 
> I see that you commented out the
> 
> #include <netinet/in.h>  ... ?

It's probably better to include it in sem.c. Done.

> ------
> 
> /home/wiebalck/nbd-2.4.32/nbd/sem.c: In function `sem_wait_timeout':
> /home/wiebalck/nbd-2.4.32/nbd/sem.c:63: warning: implicit declaration of
> function `mygettimeofday'
> 
> a missing #include "time.h"

Yes. Got that.

> 
> replacing file.c or enbd-server.c from the different falvours of 2.4.32
> did not work, btw. one did not compile properly, one did not run. I'll
> look into it.

OK - I am waiting on results. I had a class just now.

I was working on the locking for 2.6 kernels. They seem to have
some problem with fcntl on shared memory, so I was changing the
wait on the lock to have a timeout.

As soon as we find the present problem, I'll issue 2.4.32, as I don't
want any more bouts of debugging rather than development.

Peter