__builtin___clear_cache gives “illegal hardware instruction”

[fxcoudert]

Reduced from gcc.dg/builtins-64.c:

$ cat builtins-64.c 
/* { dg-do run } */
/* { dg-require-effective-target alloca } */

int main ()
{
  char *mem = __builtin_alloca (40);
  __builtin___clear_cache (mem, mem + 40);
  return 0;
}
$ ./a.out
zsh: illegal hardware instruction  ./a.out

The backtrace is:

* thread #1, queue = 'com.apple.main-thread', stop reason = EXC_BAD_INSTRUCTION (code=1, subcode=0xd53b0023)
    frame #0: 0x0000000100003f30 a.out`__aarch64_sync_cache_range(base=0x000000016fdffa40, end=0x000000016fdffa68) at sync-cache.c:42:5
   39  	  if (! cache_info)
   40  	    /* CTR_EL0 [3:0] contains log2 of icache line size in words.
   41  	       CTR_EL0 [19:16] contains log2 of dcache line size in words.  */
-> 42  	    asm volatile ("mrs\t%0, ctr_el0":"=r" (cache_info));
   43  	
   44  	  icache_lsize = 4 << (cache_info & 0xF);
   45  	  dcache_lsize = 4 << ((cache_info >> 16) & 0xF);
Target 0: (a.out) stopped.
(lldb) bt
* thread #1, queue = 'com.apple.main-thread', stop reason = EXC_BAD_INSTRUCTION (code=1, subcode=0xd53b0023)
  * frame #0: 0x0000000100003f30 a.out`__aarch64_sync_cache_range(base=0x000000016fdffa40, end=0x000000016fdffa68) at sync-cache.c:42:5
    frame #1: 0x0000000100003ef4 a.out`main + 52
    frame #2: 0x00000001d29e8a04 libdyld.dylib`start + 4

The disassembly from gcc is:

	.arch armv8-a
	.text
	.align	2
	.globl _main
_main:
LFB0:
	.cfi_startproc
	stp	x29, x30, [sp, -32]!
	.cfi_def_cfa_offset 32
	.cfi_offset 29, -32
	.cfi_offset 30, -24
	mov	x29, sp
	.cfi_def_cfa_register 29
	sub	sp, sp, #48
	mov	x0, sp
	add	x0, x0, 15
	lsr	x0, x0, 4
	lsl	x0, x0, 4
	str	x0, [x29, 24]
	ldr	x0, [x29, 24]
	add	x0, x0, 40
	mov	x1, x0
	ldr	x0, [x29, 24]
	bl	___clear_cache
	mov	w0, 0
	mov	sp, x29
	ldp	x29, x30, [sp], 32
	.cfi_restore 30
	.cfi_restore 29
	.cfi_def_cfa 31, 0
	ret
	.cfi_endproc

while clang does:

	.section	__TEXT,__text,regular,pure_instructions
	.build_version macos, 11, 0	sdk_version 11, 0
	.globl	_main                   ; -- Begin function main
	.p2align	2
_main:                                  ; @main
	.cfi_startproc
; %bb.0:
	sub	sp, sp, #80             ; =80
	stp	x29, x30, [sp, #64]     ; 16-byte Folded Spill
	add	x29, sp, #64            ; =64
	.cfi_def_cfa w29, 16
	.cfi_offset w30, -8
	.cfi_offset w29, -16
	adrp	x8, ___stack_chk_guard@GOTPAGE
	ldr	x8, [x8, ___stack_chk_guard@GOTPAGEOFF]
	ldr	x8, [x8]
	stur	x8, [x29, #-8]
	mov	w9, #0
	str	w9, [sp, #12]
	add	x8, sp, #16             ; =16
	str	x8, [sp]
	ldr	x8, [sp]
	add	x1, x8, #40             ; =40
	mov	x0, x8
	bl	___clear_cache
	adrp	x8, ___stack_chk_guard@GOTPAGE
	ldr	x8, [x8, ___stack_chk_guard@GOTPAGEOFF]
	ldr	x8, [x8]
	ldur	x10, [x29, #-8]
	subs	x8, x8, x10
	b.ne	LBB0_2
	b	LBB0_1
LBB0_1:
	mov	w0, #0
	ldp	x29, x30, [sp, #64]     ; 16-byte Folded Reload
	add	sp, sp, #80             ; =80
	ret
LBB0_2:
	bl	___stack_chk_fail
	brk	#0x1
	.cfi_endproc

[iains]

should be fixed.