Re: [patch] rudimetary fallocate support for xfs_io

[Nathan Scott <nscott@aconex.com>]

On Thu, 2008-05-22 at 16:10 +1000, David Chinner wrote:
> 
> +#if defined(__i386__)
> +#define __NR_fallocate 324
> +#elif defined(__x86_64__)
> +#define __NR_fallocate 285
> +#elif defined(__ia64__)
> +#define __NR_fallocate 1303
> +#endif
> +
> +static int
> +fallocate(int fd, int cmd, off64_t start, off64_t len)
> +{
> +       return syscall(__NR_fallocate, fd, cmd, start, len);
> +}

This will cause xfsprogs build breakage on any architecture
other than the above three; check out the approach taken in
xfs-cmds/attr/libattr/syscalls.c to avoid this (give ENOSYS
for the unknown archs).

cheers.

-- 
Nathan