[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 22 Sep 2018 22:45:11 -0400
From: Rich Felker <dalias@...c.org>
To: musl@...ts.openwall.com
Subject: Re: un-UBify-strings
On Sat, Sep 22, 2018 at 10:32:34PM -0400, Rich Felker wrote:
> On Sun, Sep 23, 2018 at 02:11:42AM +0000, Pascal Cuoq wrote:
> > Hello Rich,
> >
> > On 23 Sep 2018, at 02:35, Rich Felker <dalias@...c.org<mailto:dalias@...c.org>> wrote:
> >
> > I've had this patch sitting around since 2016, and just updated it to
> > apply cleanly. Any objections?
> >
> > Your patch contains:
> >
> > ....
> > size_t __attribute__((__may_alias__)) *wd;
> > const size_t __attribute__((__may_alias__)) *ws;
> > ....
> >
> > In my experience, this use of __may_alias__ does not do anything.
> > See function f in the example below, which both GCC and Clang
> > optimize as if the programmer had not used __may_alias__ at all:
> > https://gcc.godbolt.org/z/Um4NU7
> >
> > You should use a typdef for the aliasing type, as shown for function
> > g (in with GCC and Clang do not apply the optimization).
> >
> > The example in GCC's documentation for __may_alias__ also uses a
> > typedef:
> > https://gcc.gnu.org/onlinedocs/gcc-4.0.4/gcc/Type-Attributes.html
>
> Thanks, this is very helpful. I'll prepare an updated version.
While I've got your attention, I'm also trying to fix the UB in
address range checks for implementing memmove as memcpy, etc. Is this
correct:
if ((uintptr_t)s-(uintptr_t)d-n <= -2*n) return memcpy(d, s, n);
?
Rich
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
