Re: [RFC] Fix compilation failure of remote-fileio.c

[Daniel Jacobowitz <drow@mvista.com>]

On Sun, Dec 28, 2003 at 03:52:13PM +0200, Eli Zaretskii wrote:
> remote-fileio.c won't compile if `struct stat' doesn't have the
> `st_blocks' member.  The patch below fixes that, but I'm not sure it
> (and the associated patch to gdb/configure.in, which is not shown
> below) is the right fix.  Is it perhaps better not to use st_blocks at
> all, and instead compute the number of blocks as shown by the patch,
> for all platforms?

Probably not.  If it's correct for DJGPP then it's a bit of an
accident.

The reason it isn't correct in general:
                  blksize_t     st_blksize;  /* blocksize for filesystem I/O */
                  blkcnt_t      st_blocks;   /* number of blocks allocated */

It's not clear from those comments, but st_blksize is the "optimal IO
size":

       The value st_blocks gives the size of the file in 512-byte
       blocks.  (This may be smaller than st_size/512 e.g. when the
       file has holes.) The value st_blksize gives the "preferred"
       blocksize for efficient file system I/O.  (Writing to a file in
       smaller chunks may cause an inefficient read-modify-rewrite.)

So you probably want st->st_size / 512 instead.

> +#else
> +  remote_fileio_to_fio_ulong ((LONGEST) st->st_size / (LONGEST) st->st_blksize
> +			      + ((LONGEST) st->st_size % (LONGEST) st->st_blksize) != 0,
> +			      fst->fst_blocks);
> +#endif

A marginally more efficient way to write this is:
 ((LONGEST) st->st_size + st->st_blksize - 1) / st->st_blksize

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer