Re: [PATCH] Change long to unsigned long in ptrace functions to fix sparse warnings

[Andrew Morton <akpm@linux-foundation.org>]

On Thu,  1 May 2008 16:19:41 -0700 (PDT)
Roland McGrath <roland@redhat.com> wrote:

> I don't know of any problems off hand, but I would testify that there
> aren't any niggling bits there.  Paranoia would suggest comparing object
> code before/after on each arch, since it should not change.
> 

oh gee, it's a ton of code to review, isn't it?  And it's _hard_ review :(

(reviews a bit)

h8300's arch_ptrace() does

	if ((addr & 3) || addr < 0 || addr >= sizeof(struct user)) {

ok, the third comparison saved us.  Lucky...

In fact most architectures seem to have copy-n-pasted that test.  I guess
that going from signed to unsigned is safer than vice-versa.



Now what about frv?

	case PTRACE_PEEKUSR: {
		tmp = 0;
		ret = -EIO;
		if ((addr & 3) || addr < 0)
			break;

we broke that.  Maybe it doesn't matter, don't know.



So yeah, it's _probably_ OK, but I'm not sure that it would be smart to
merge it and hope.  Nor would it be a good use of time to check all that
code just to fix a sparse warning.