[Bug tree-optimization/80576] dead strcpy and strncpy followed by memset not eliminated
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80576 --- Comment #6 from Jeffrey A. Law --- *** Bug 81810 has been marked as a duplicate of this bug. ***
[Bug tree-optimization/80576] dead strcpy and strncpy followed by memset not eliminated
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80576 --- Comment #5 from Jeffrey A. Law --- Interestingly enough, it never comes up in gcc AFAICT, at least not the str* followed by memcpy/memset variant that shows up in that BZ. I don't think it'd be terribly hard to support, but if it's not going to help real code I'm not too inclined to add the complexity. The non-constant memcpy followed by a full object memset isn't optimized either. But that's a function of the non-constant sized memcpy. It'd likely fall into the same bucket too. Let me do some instrumentation on this variant.
[Bug tree-optimization/80576] dead strcpy and strncpy followed by memset not eliminated
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80576
--- Comment #4 from Martin Sebor ---
I've seen code similar to it but I don't remember how close they were. I
imagine it comes up quite a bit as GCC itself transforms code in various ways.
DSE already eliminates stores with constant sizes but it doesn't handle the
non-constant ones:
$ cat pr80576.c && gcc -O2 -S -Wall -Wextra -Wpedantic
-fdump-tree-optimized=/dev/stdout pr80576.c
extern char a[32];
void f (const char *s)
{
__builtin_memcpy (a, s, 7);// eliminated
__builtin_memset (a, 0, sizeof a);
}
void g (const char *s, unsigned n)
{
__builtin_memcpy (a, s, n);// not eliminated
__builtin_memset (a, 0, sizeof a);
}
;; Function f (f, funcdef_no=0, decl_uid=1909, cgraph_uid=1, symbol_order=0)
f (const char * s)
{
[local count: 1073741824]:
__builtin_memset (, 0, 32); [tail call]
return;
}
;; Function g (g, funcdef_no=1, decl_uid=1913, cgraph_uid=2, symbol_order=1)
g (const char * s, unsigned int n)
{
long unsigned int _1;
[local count: 1073741824]:
_1 = (long unsigned int) n_2(D);
__builtin_memcpy (, s_4(D), _1);
__builtin_memset (, 0, 32); [tail call]
return;
}
[Bug tree-optimization/80576] dead strcpy and strncpy followed by memset not eliminated
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80576 Jeffrey A. Law changed: What|Removed |Added CC||law at redhat dot com --- Comment #3 from Jeffrey A. Law --- I happened to stumble across this while looking for something else in BZ. I don't think it'd be hard to do this in DSE. The question I have is how common this is in practice. Have you seen this kind of code in the wild?
[Bug tree-optimization/80576] dead strcpy and strncpy followed by memset not eliminated
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80576 Eric Gallager changed: What|Removed |Added Status|UNCONFIRMED |NEW Last reconfirmed||2017-08-29 CC||egallager at gcc dot gnu.org Ever confirmed|0 |1 --- Comment #2 from Eric Gallager --- Confirmed.
[Bug tree-optimization/80576] dead strcpy and strncpy followed by memset not eliminated
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80576 Martin Sebor changed: What|Removed |Added See Also||https://gcc.gnu.org/bugzill ||a/show_bug.cgi?id=79715 --- Comment #1 from Martin Sebor --- See also bug 79715.
