[coreboot] lar directory handling patch

[Peter Stuge]

On Mon, Feb 25, 2008 at 01:47:59PM -0700, Myles Watson wrote:
>  				strncpy(fullname, name, MAX_PATH);
> +				strncpy(fullpathname, pathname, MAX_PATH);

--8<-- strcpy(3)
The strncpy() function is similar, except that at most n bytes  of  src
are  copied.  Warning: If there is no null byte among the first n bytes
of src, the string placed in dest will not be null terminated.
-->8--

One fix is to simply memset() fullname and fullpathname at the start
of the function - I don't like assuming that heap variables are
zero initialized.


>  				if (name[(strlen(name)) - 1] != '/') {
>  					strncat(fullname, "/", MAX_PATH);
>  				}
>  
> +				if (name[(strlen(pathname)) - 1] != '/') {
> +					strncat(fullpathname, "/", MAX_PATH);
> +				}
> +
>  				strncat(fullname, namelist[n]->d_name,
>  					MAX_PATH);
>  
> -				add_files(fullname);
> +				strncat(fullpathname, namelist[n]->d_name,
> +					MAX_PATH);
> +
> +				add_files(fullname,fullpathname,thisalgo);

This algorithm protects against overflow, but I would like if it also
raised an error when MAX_PATH isn't big enough.


> +int add_files(const char *name, const char * pathname_in, 
> +		const enum compalgo algo_in)
..
> +		ret = lar_process_name((char*)name, &filename, &pathname, &thisalgo);

Why is this cast needed? Does lar_process_name() modify name? If not
please fix it's prototype so no cast is needed.


> +	/*printf("%s: %s (%s:%s)\n",__FUNCTION__,name,filename,pathname);*/

Is there some debug functionality in lar for these?


>  	if (elfparse() && iself(ptr)) {
> -		output_elf_segments(lar, pathname, ptr, size, thisalgo);
> +		output_elf_segments(lar, file->pathname, ptr, size, file->algo);

How does this work with file->algo vs. zeroes compression?


//Peter