e820.c : why char* ? (2nd try)

e820.c : why char* ? (2nd try)

[PUCCETTI Armand]

In function

sanitize_e820_map(struct e820entry * biosmap, char * pnr_map)

of file e820.c, the last argument is treated as an integer, but when 
compiling with gcc
it belongs to [-128,127].

Inside the body of that function, however, we use that argument as an 
int to fill up a table of a different
size, namely:

static struct change_member *change_point [256]
static struct change_member change_point_list [256]
static int __init sanitize_e820_map(struct e820entry * biosmap, char * 
pnr_map)
{
...
old_nr = *pnr_map;
...
for (i=0; i < 2*old_nr; i++) {
       change_point[i] = &change_point_list[i];
...
}

After the loop, change_point is not completely filled-up, as 
change_point [254] and change_point [255]
are never filled nor used.

Is it voluntary that pnr_map gets typed and used like this?

Armand

Re: e820.c : why char* ? (2nd try)

[Ian Campbell]

On Thu, 2006-06-29 at 11:36 +0200, PUCCETTI Armand wrote:
> In function
> 
> sanitize_e820_map(struct e820entry * biosmap, char * pnr_map)
> 
> of file e820.c, the last argument is treated as an integer, but when 
> compiling with gcc it belongs to [-128,127].

This is not a Xen specific issue. On native the value comes from the
BIOS and is a single byte, hence the type. Xen uses the same function.

I don't think it is possible for there to ever be more than 127 entries
in an e820 map so other than potentially wasting a couple of entries in
the tables I don't see a problem.

Ian.