Related to PR 29877.
Take the following PPC64-linux code (ignore the fact the inline-asm should be
improved a lot via use of clobbers).
static inline int sys_ppu_thread_set_priority(sys_ppu_thread_t thread_id,int
prio)
{
register uint64_t p1 __asm__ ("3") = thread_id;
register uint64_t p2 __asm__ ("4") = prio;
register uint64_t p3 __asm__ ("5"); register uint64_t p4 __asm__ ("6");
register uint64_t p5 __asm__ ("7"); register uint64_t p6 __asm__ ("8");
register uint64_t p7 __asm__ ("9");
register uint64_t p8 __asm__ ("10");
register uint64_t n __asm__ ("11") = 47;
__asm__ volatile ("sc" : "=r" (p1), "=r" (p2), "=r" (p3), "=r" (p4), "=r"
(p5), "=r" (p6), "=r" (p7), "=r" (p8), "=r" (n) : "r" (p1), "r" (p2), "r" (p3),
"r" (p4), "r" (p5), "r" (p6), "r" (p7), "r" (p8), "r" (n) : "0", "12", "lr",
"ctr", "xer", "cr0", "cr1", "cr5", "cr6", "cr7", "memory");
return (int)(p1);
}
int32_t P(int32_t priority){return 0;}
struct Thread {
sys_ppu_thread_t tid;
Thread();
void setPriority(int32_t p);
};
void Thread::setPriority(int32_t p)
{
sys_ppu_thread_t a = tid;
sys_ppu_thread_set_priority(a, P(p));
}
---------------- CUT --------------------
Right now we get in .optimized:
p1 = this->tid;
p2 = (uint64_t) P (p);
Which is wrong as the assignment of p1 happens before the call to P so we
clobber it during the setup of the call.
--
Summary: [4.1/4.2/4.3 Regression] TER breaks some inline-asm code
(again)
Product: gcc
Version: 4.3.0
Status: UNCONFIRMED
Keywords: wrong-code
Severity: blocker
Priority: P3
Component: tree-optimization
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: pinskia at gcc dot gnu dot org
GCC target triplet: powerpc64-linux-gnu
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33619
