This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFC][PATCH][AArch64] Cleanup frame pointer usage


Jiong Wang wrote:

test.c
===
struct K {
  int a;
  int b;
  int c;
  int d;
  char e;
  short f;
  long g;
  float h;
  double i;
};

void foo (int, struct K *);

void test (int i)
{
  struct K k = {
   .a = 5,
   .b = 0,
   .c = i,
  };

  foo (5, &k);
}

There are 2 separate latent bugs here, both unrelated to this patch. If I build with -fomit-frame pointer I get:

	str	x30, [sp, -64]!
	mov	w2, 5
	mov	x1, sp
	str	w2, [sp, 16]
	add	x1, x1, 20
	stp	xzr, xzr, [x1]
	add	x1, sp, 16
	str	w0, [sp, 24]
	mov	x0, sp
	add	x0, x0, 36
	stp	xzr, xzr, [x0]
	mov	w0, w2
	str	xzr, [sp, 52]
	str	wzr, [sp, 60]
	bl	foo
	ldr	x30, [sp], 64
	ret

If I use a compiler from early January (which also includes my patch) I get:

	str	x30, [sp, -64]!
	mov	w2, 5
	stp	xzr, xzr, [sp, 16]
	add	x1, sp, 16
	str	w0, [sp, 24]
	mov	w0, w2
	stp	xzr, xzr, [sp, 32]
	stp	xzr, xzr, [sp, 48]
	str	w2, [sp, 16]
	bl	foo
	ldr	x30, [sp], 64
	ret

Clearly a nasty regression was introduced in the structure initialization code...

Wilco

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]