Re: [PATCH] fat: Refactor shortname parsing

["Steven J. Magnani" <steve@digidescorp.com>]

On Mon, 2012-07-02 at 22:40 +0900, OGAWA Hirofumi wrote: 
> "Steven J. Magnani" <steve@digidescorp.com> writes:
> 
> > On Sat, 2012-06-30 at 05:09 +0900, OGAWA Hirofumi wrote: 
> >> OGAWA Hirofumi <hirofumi@mail.parknet.co.jp> writes:
> >> 
> >> >
> >> > 			if (is_vfat)
> >> > 				ptname[i++] = fat_tolower(!nocase, c);
> >> 
> >> Of course, if (!is_vfat). Sorry.
> >
> > I agree that the nocase logic is confusing, but I'm pretty sure this
> > change would break the code. 
> 
> I might be wrong here though, which place is broken with it?
> 
> > 'nocase' is always zero for vfat, which does not recognize that option.
> > For msdos, it is zero by default, and 1 if the 'nocase' option was
> > specified.
> > In all cases it is necessary to copy *something* to ptname.
> 
> Looks like, we overwrite bufname by converted bufuname in is_vfat case?
> 
> 	if (isvfat) {
> 		bufuname[j] = 0x0000;
> 		i = fat_uni_to_x8(sb, bufuname, bufname, sizeof(bufname));
> 	}

True, but with the change you suggest we lose the incrementing of 'i',
which likely would cause an infinite loop for 1:1 Unicode conversions.

Steve