ipa-sra: Testcase that removing a "returns_nonnull" retval works

Since we can now remove return values of functions with return_nonnull
type attribute, I'll feel a bit safer if we can test this does not ICE
when someone attempts to access a non-existent call LHS.  Eventually
we should probably drop the attribute when this happens.

gcc/testsuite/ChangeLog:

2021-11-15  Martin Jambor  <mjambor@suse.cz>

	* gcc.dg/ipa/ipa-sra-ret-nonull.c: New test.
This commit is contained in:
Martin Jambor 2021-11-16 10:44:53 +01:00
parent 9ceaf0fee3
commit 9f7fc82014
No known key found for this signature in database
GPG Key ID: BF63C1BC3FA43540

View File

@ -0,0 +1,40 @@
/* { dg-do compile } */
/* { dg-options "-O2 -fdump-ipa-sra-details" } */
volatile void *gp;
volatile void *gq;
char buf[16];
__attribute__((returns_nonnull, noinline))
static char *
foo (char *p, char *q)
{
gq = q;
gp = p;
return q;
}
__attribute__((returns_nonnull, noinline))
static char *
bar (char *p, char *q)
{
return foo (p, q) + 8;
}
__attribute__((noipa))
static char *
get_charp (void)
{
return &buf[0];
}
int
main ()
{
char *r;
asm volatile ("" : : : "memory");
r = bar (get_charp (), get_charp ());
return 0;
}
/* { dg-final { scan-ipa-dump-times "Will SKIP return." 2 "sra" } } */