Re: Make tree-ssa-strlen.c handle partial unterminated strings

[Martin Sebor <msebor at gmail dot com>]

On 05/05/2017 06:01 AM, Richard Sandiford wrote:
> tree-ssa-strlen.c looks for cases in which a string is built up using
> operations like:
>
>     memcpy (a, "foo", 4);
>     memcpy (a + 3, "bar", 4);
>     int x = strlen (a);
>
> As a side-effect, it optimises the non-final memcpys so that they don't
> include the nul terminator.
>
> However, after removing some "& ~0x1"s from tree-ssa-dse.c, the DSE pass
> does this optimisation itself (because it can tell that later memcpys
> overwrite the terminators).  The strlen pass wasn't able to handle these
> pre-optimised calls in the same way as the unoptimised ones.
>
> This patch adds support for tracking unterminated strings.

Oooh, very nice! :)  I spent some time on something like this
last summer (under bug 71304) but didn't finish it.  My patch
also handled NULs inserted by simple assignment.  I see your
patch handles them when they are inserted by a call to one of
the functions but not otherwise, as in the test case below.
Is that something that could be easily handled in your
approach?

  char a[30];

  int f1 (void)
  {
    __builtin_memcpy (a, "1234567", 7);
    __builtin_memcpy (a + 7, "", 1);
     return __builtin_strlen (a);   // is optimized to 7
  }

  int f2 (void)
  {
    __builtin_memcpy (a, "1234567", 7);
    a[7] = '\0';
     return __builtin_strlen (a);   // could be optimized to 7
  }

Martin