Re: [ECOS] Block device IO

[Savin Zlobec <savin at elatec dot si>]

David N. Welton wrote:

> Savin Zlobec <savin@elatec.si> writes:
>
>  
>
> > Now I'am thinking about a third layer to do the position ->
> > block/offset translation.  Maybe it would be cleaner to put it into
> > a library since the only way I see to set block numbers is trough
> > set_config interface (?) .
> >    
>
>  
>
> > Is there any similar effort wich will be contributed back to eCos ?
> >    
>
> I wrote the following simplistic code:
>
> int bytes2sectors (char *oldbuf, unsigned int bytelen, unsigned int inpos,
> 		   char **newbuf, unsigned int *sectorlen, unsigned int *outpos)
> {
>    int sector_offset = 0;
>
>    sector_offset = inpos - (*outpos * SECTOR_SIZE);
>    *outpos = inpos / SECTOR_SIZE;
>    *sectorlen = ((bytelen + sector_offset) / SECTOR_SIZE) + 1;
>    *newbuf = malloc(*sectorlen * SECTOR_SIZE);
>    if (*newbuf == NULL) {
> 	diag_printf("Error malloc'ing %d bytes in bytes2sectors\n",
> 		    *sectorlen * SECTOR_SIZE);
> 	return -1;
>    }
>    return 0;
> }
>
> int sector2bytes (char *newbuf, unsigned int newlen, unsigned int newpos,
> 		  char *oldbuf, unsigned int *oldlen, unsigned int oldpos)
> {
>    unsigned int sector_offset;
>
>    sector_offset = oldpos - (newpos * SECTOR_SIZE);
>    memcpy(oldbuf, newbuf + sector_offset, *oldlen);
>    free(newbuf);
> }
>
> Basically the idea is that you hand it a buffer that is sized in
> bytes, and it hands you back a buffer that is big enough to handle the
> whole thing, rounded up to blocks.
>  
I have written a caching layer to do the block memory management.
Now I just need a simple byte access - my FAT16 needs it and I would
like my code to be in sync as much as possible with the "eCos tendencies".

I'am thinking about doing something like:

cyg_io_set_config(disk_h, CYG_IO_SET_CONFIG_DISK_CURRENT_BLOCK,  &block, 
&l);
cyg_io_bread(disk_h, buf,  &len,  pos);

len and pos would be in bytes and in case of current position being 
greater than block size the
current block number would be adjusted.


savin



--
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss