Re: make test fails (Kernel 2.6.12.3)

First of all, thanks for your help on this.

This is a quad opteron (x86_64) machine, so that makes sense. I am 
assuming by your response that maybe enbd isn't ported to this, can this 
be done, or would it take a lot of modifications?

For completeness, uname -a:

Linux 2.6.12.3 #2 SMP Mon Aug 8 07:47:07 PDT 2005 x86_64 x86_64 x86_64 
GNU/Linux

GCC:
gcc version 3.4.3 20050227 (Red Hat 3.4.3-22.1)

Peter T. Breuer wrote:
> "Also sprach Robert Scussel:"
> 
>>Before posting the only thing I can think of for proof, I looked at 
>>nbd/enbd-maketest and I thought it interesting that the load () function 
>>doesn't load .ko modules, only .o files
> 
> 
> It doesn't do any loading at all. You have to load the modules
> yourself beforehand.
> 
> 
> 
>>and since my enbd and 
>>enbd_ioctl modules are .ko files, they aren't loaded correctly.
> 
> 
> They aren't loaded at all - unless you have an entry in modprobe.conf,
> of course.
> 
> 
>>I didn't 
>>find a enbd_bufferwr.ko or .o at all, so I don't know if my build went 
>>bad somewhere or what.
> 
> 
> It didn't.
> 
> 
>>I redid it from scratch:
> 
> 
> No need ...
> 
> 
>>-DDEBUG=0 -o file.o -c /var/tmp/nbd-2.4.32/nbd/file.c
>>/var/tmp/nbd-2.4.32/nbd/file.c: In function `getsize1':
>>/var/tmp/nbd-2.4.32/nbd/file.c:405: warning: long long int format, long 
>>unsigned int arg (arg 4)
> 
> 
> That's strange! What architecture are you on? SOmething like that can
> provoke a crash! That line is:
> 
>            MSG("set size of fd %d to %Ld\n", fd, *res);
> 
> and res is declared in the argument of the function:
> 
>            getsize1(struct nbd_file * self, int fd, u64 *res) {
> 
> 
> so either your compiler is broken or you have a machine on which long
> unsigned int is a u64!
> 
> 
> Which?
> 
> Peter


-- 
Robert Scussel
1024D/BAF70959/0036 B19E 86CE 181D 0912  5FCC 92D8 1EA1 BAF7 0959