aarch64: use PTR_ARG and SIZE_ARG instead of DELOUSE

DELOUSE was added to asm code to make them compatible with non-LP64 ABIs, but it is an unfortunate name and the code was not compatible with ABIs where pointer and size_t are different. Glibc currently only supports the LP64 ABI so these macros are not really needed or tested, but for now the name is changed to be more meaningful instead of removing them completely. Some DELOUSE macros were dropped: clone, strlen and strnlen used it unnecessarily. The out of tree ILP32 patches are currently not maintained and will likely need a rework to rebase them on top of the time64 changes.
2025-10-12 19:04:54 +03:00 · 2020-12-17 10:03:05 +00:00
parent f9de8bfe1a
commit 45b1e17e91
28 changed files with 74 additions and 75 deletions
--- a/sysdeps/aarch64/__longjmp.S
+++ b/sysdeps/aarch64/__longjmp.S
@@ -46,7 +46,7 @@ ENTRY (__longjmp)
 	cfi_offset(d14, JB_D14<<3)
 	cfi_offset(d15, JB_D15<<3)

-	DELOUSE (0)
+	PTR_ARG (0)

 	ldp	x19, x20, [x0, #JB_X19<<3]
 	ldp	x21, x22, [x0, #JB_X21<<3]
--- a/sysdeps/aarch64/dl-tlsdesc.S
+++ b/sysdeps/aarch64/dl-tlsdesc.S
@@ -75,7 +75,7 @@
 	.align 2
 _dl_tlsdesc_return:
 	BTI_C
-	DELOUSE (0)
+	PTR_ARG (0)
 	ldr	PTR_REG (0), [x0, #PTR_SIZE]
 	RET
 	cfi_endproc
@@ -99,7 +99,7 @@ _dl_tlsdesc_undefweak:
 	BTI_C
 	str	x1, [sp, #-16]!
 	cfi_adjust_cfa_offset (16)
-	DELOUSE (0)
+	PTR_ARG (0)
 	ldr	PTR_REG (0), [x0, #PTR_SIZE]
 	mrs	x1, tpidr_el0
 	sub	PTR_REG (0), PTR_REG (0), PTR_REG (1)
@@ -145,7 +145,7 @@ _dl_tlsdesc_undefweak:
 	.align 2
 _dl_tlsdesc_dynamic:
 	BTI_C
-	DELOUSE (0)
+	PTR_ARG (0)

 	/* Save just enough registers to support fast path, if we fall
 	   into slow path we will save additional registers.  */
--- a/sysdeps/aarch64/memchr.S
+++ b/sysdeps/aarch64/memchr.S
@@ -61,8 +61,8 @@
   string, counting trailing zeros identifies exactly which byte matched.  */

 ENTRY (MEMCHR)
-	DELOUSE (0)
-	DELOUSE (2)
+	PTR_ARG (0)
+	SIZE_ARG (2)
 	bic	src, srcin, 15
 	cbz	cntin, L(nomatch)
 	ld1	{vdata.16b}, [src]
--- a/sysdeps/aarch64/memcmp.S
+++ b/sysdeps/aarch64/memcmp.S
@@ -42,9 +42,9 @@
 #define tmp2		x8

 ENTRY_ALIGN (memcmp, 6)
-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
+	PTR_ARG (0)
+	PTR_ARG (1)
+	SIZE_ARG (2)

 	subs	limit, limit, 16
 	b.lo	L(less16)
--- a/sysdeps/aarch64/memcpy.S
+++ b/sysdeps/aarch64/memcpy.S
@@ -73,9 +73,9 @@
 */

 ENTRY_ALIGN (MEMCPY, 6)
-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
+	PTR_ARG (0)
+	PTR_ARG (1)
+	SIZE_ARG (2)

 	add	srcend, src, count
 	add	dstend, dstin, count
@@ -209,9 +209,9 @@ END (MEMCPY)
 libc_hidden_builtin_def (MEMCPY)

 ENTRY_ALIGN (MEMMOVE, 4)
-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
+	PTR_ARG (0)
+	PTR_ARG (1)
+	SIZE_ARG (2)

 	add	srcend, src, count
 	add	dstend, dstin, count
--- a/sysdeps/aarch64/memrchr.S
+++ b/sysdeps/aarch64/memrchr.S
@@ -59,8 +59,8 @@
   string, counting trailing zeros identifies exactly which byte matched.  */

 ENTRY (__memrchr)
-	DELOUSE (0)
-	DELOUSE (2)
+	PTR_ARG (0)
+	SIZE_ARG (2)
 	add	end, srcin, cntin
 	sub	endm1, end, 1
 	bic	src, endm1, 15
--- a/sysdeps/aarch64/memset.S
+++ b/sysdeps/aarch64/memset.S
@@ -31,8 +31,8 @@

 ENTRY_ALIGN (MEMSET, 6)

-	DELOUSE (0)
-	DELOUSE (2)
+	PTR_ARG (0)
+	SIZE_ARG (2)

 	dup	v0.16B, valw
 	add	dstend, dstin, count
--- a/sysdeps/aarch64/multiarch/memchr_nosimd.S
+++ b/sysdeps/aarch64/multiarch/memchr_nosimd.S
@@ -64,8 +64,8 @@

 ENTRY_ALIGN (MEMCHR, 6)

-	DELOUSE (0)
-	DELOUSE (2)
+	PTR_ARG (0)
+	SIZE_ARG (2)

 	/* Do not dereference srcin if no bytes to compare. */
 	cbz	cntin, L(none_chr)
--- a/sysdeps/aarch64/multiarch/memcpy_advsimd.S
+++ b/sysdeps/aarch64/multiarch/memcpy_advsimd.S
@@ -64,9 +64,9 @@
   from the end.  */

 ENTRY (__memcpy_simd)
-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
+	PTR_ARG (0)
+	PTR_ARG (1)
+	SIZE_ARG (2)

 	add	srcend, src, count
 	add	dstend, dstin, count
@@ -181,9 +181,9 @@ libc_hidden_builtin_def (__memcpy_simd)


 ENTRY (__memmove_simd)
-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
+	PTR_ARG (0)
+	PTR_ARG (1)
+	SIZE_ARG (2)

 	add	srcend, src, count
 	add	dstend, dstin, count
--- a/sysdeps/aarch64/multiarch/memcpy_falkor.S
+++ b/sysdeps/aarch64/multiarch/memcpy_falkor.S
@@ -73,9 +73,9 @@
 #if IS_IN (libc)
 ENTRY_ALIGN (__memcpy_falkor, 6)

-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
+	PTR_ARG (0)
+	PTR_ARG (1)
+	SIZE_ARG (2)

 	cmp	count, 32
 	add	srcend, src, count
@@ -218,9 +218,9 @@ libc_hidden_builtin_def (__memcpy_falkor)

 ENTRY_ALIGN (__memmove_falkor, 6)

-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
+	PTR_ARG (0)
+	PTR_ARG (1)
+	SIZE_ARG (2)

 	cmp	count, 32
 	add	srcend, src, count
--- a/sysdeps/aarch64/multiarch/memcpy_thunderx.S
+++ b/sysdeps/aarch64/multiarch/memcpy_thunderx.S
@@ -81,9 +81,9 @@

 ENTRY_ALIGN (MEMMOVE, 6)

-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
+	PTR_ARG (0)
+	PTR_ARG (1)
+	SIZE_ARG (2)

 	sub	tmp1, dstin, src
 	cmp	count, 96
@@ -95,9 +95,9 @@ END (MEMMOVE)
 libc_hidden_builtin_def (MEMMOVE)
 ENTRY (MEMCPY)

-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
+	PTR_ARG (0)
+	PTR_ARG (1)
+	SIZE_ARG (2)

 	prfm	PLDL1KEEP, [src]
 	add	srcend, src, count
--- a/sysdeps/aarch64/multiarch/memcpy_thunderx2.S
+++ b/sysdeps/aarch64/multiarch/memcpy_thunderx2.S
@@ -97,9 +97,9 @@

 ENTRY_ALIGN (MEMMOVE, 6)

-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
+	PTR_ARG (0)
+	PTR_ARG (1)
+	SIZE_ARG (2)

 	add	srcend, src, count
 	cmp	count, 16
@@ -127,9 +127,9 @@ libc_hidden_builtin_def (MEMMOVE)
 	.p2align 4
 ENTRY (MEMCPY)

-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
+	PTR_ARG (0)
+	PTR_ARG (1)
+	SIZE_ARG (2)

 	add	srcend, src, count
 	cmp	count, 16
--- a/sysdeps/aarch64/multiarch/memset_base64.S
+++ b/sysdeps/aarch64/multiarch/memset_base64.S
@@ -36,8 +36,8 @@

 ENTRY_ALIGN (MEMSET, 6)

-	DELOUSE (0)
-	DELOUSE (2)
+	PTR_ARG (0)
+	SIZE_ARG (2)

 	bfi	valw, valw, 8, 8
 	bfi	valw, valw, 16, 16
--- a/sysdeps/aarch64/multiarch/memset_kunpeng.S
+++ b/sysdeps/aarch64/multiarch/memset_kunpeng.S
@@ -31,8 +31,8 @@

 ENTRY_ALIGN (MEMSET, 6)

-	DELOUSE (0)
-	DELOUSE (2)
+	PTR_ARG (0)
+	SIZE_ARG (2)

 	dup	v0.16B, valw
 	add	dstend, dstin, count
--- a/sysdeps/aarch64/multiarch/strlen_asimd.S
+++ b/sysdeps/aarch64/multiarch/strlen_asimd.S
@@ -86,7 +86,7 @@
   character, return the length, if not, continue in the main loop.  */

 ENTRY (__strlen_asimd)
-        DELOUSE (0)
+	PTR_ARG (0)

 	and	tmp1, srcin, MIN_PAGE_SIZE - 1
 	cmp	tmp1, MIN_PAGE_SIZE - 32
--- a/sysdeps/aarch64/setjmp.S
+++ b/sysdeps/aarch64/setjmp.S
@@ -33,7 +33,7 @@ END (_setjmp)
 libc_hidden_def (_setjmp)

 ENTRY (__sigsetjmp)
-	DELOUSE (0)
+	PTR_ARG (0)

 1:
 	stp	x19, x20, [x0, #JB_X19<<3]
--- a/sysdeps/aarch64/strchr.S
+++ b/sysdeps/aarch64/strchr.S
@@ -56,7 +56,7 @@
   string, counting trailing zeros identifies exactly which byte matched.  */

 ENTRY (strchr)
-	DELOUSE (0)
+	PTR_ARG (0)
 	bic	src, srcin, 15
 	dup	vrepchr.16b, chrin
 	ld1	{vdata.16b}, [src]
--- a/sysdeps/aarch64/strchrnul.S
+++ b/sysdeps/aarch64/strchrnul.S
@@ -54,7 +54,7 @@
   string, counting trailing zeros identifies exactly which byte matched.  */

 ENTRY (__strchrnul)
-	DELOUSE (0)
+	PTR_ARG (0)
 	bic	src, srcin, 15
 	dup	vrepchr.16b, chrin
 	ld1	{vdata.16b}, [src]
--- a/sysdeps/aarch64/strcmp.S
+++ b/sysdeps/aarch64/strcmp.S
@@ -62,8 +62,8 @@
   NUL too in big-endian, byte-reverse the data before the NUL check.  */

 ENTRY(strcmp)
-	DELOUSE (0)
-	DELOUSE (1)
+	PTR_ARG (0)
+	PTR_ARG (1)
 	sub	off2, src2, src1
 	mov	zeroones, REP8_01
 	and	tmp, src1, 7
--- a/sysdeps/aarch64/strcpy.S
+++ b/sysdeps/aarch64/strcpy.S
@@ -73,8 +73,8 @@
   string, counting trailing zeros identifies exactly which byte matched.  */

 ENTRY (STRCPY)
-	DELOUSE (0)
-	DELOUSE (1)
+	PTR_ARG (0)
+	PTR_ARG (1)
 	bic	src, srcin, 15
 	mov	wtmp, 0xf00f
 	ld1	{vdata.16b}, [src]
--- a/sysdeps/aarch64/strlen.S
+++ b/sysdeps/aarch64/strlen.S
@@ -54,8 +54,7 @@
   string, counting trailing zeros identifies exactly which byte matched.  */

 ENTRY (STRLEN)
-	DELOUSE (0)
-	DELOUSE (1)
+	PTR_ARG (0)
 	bic	src, srcin, 15
 	mov	wtmp, 0xf00f
 	ld1	{vdata.16b}, [src]
--- a/sysdeps/aarch64/strnlen.S
+++ b/sysdeps/aarch64/strnlen.S
@@ -55,9 +55,8 @@
 #define REP8_80 0x8080808080808080

 ENTRY_ALIGN_AND_PAD (__strnlen, 6, 9)
-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
+	PTR_ARG (0)
+	SIZE_ARG (1)
 	cbz	limit, L(hit_limit)
 	mov	zeroones, #REP8_01
 	bic	src, srcin, #15
--- a/sysdeps/aarch64/strrchr.S
+++ b/sysdeps/aarch64/strrchr.S
@@ -59,7 +59,7 @@
   if the relevant byte matched the NUL end of string.  */

 ENTRY(strrchr)
-	DELOUSE (0)
+	PTR_ARG (0)
 	bic	src, srcin, 15
 	dup	vrepchr.16b, chrin
 	mov	wtmp, 0x3003
--- a/sysdeps/aarch64/sysdep.h
+++ b/sysdeps/aarch64/sysdep.h
@@ -25,12 +25,14 @@
 # define AARCH64_R(NAME)	R_AARCH64_ ## NAME
 # define PTR_REG(n)		x##n
 # define PTR_LOG_SIZE		3
-# define DELOUSE(n)
+# define PTR_ARG(n)
+# define SIZE_ARG(n)
 #else
 # define AARCH64_R(NAME)	R_AARCH64_P32_ ## NAME
 # define PTR_REG(n)		w##n
 # define PTR_LOG_SIZE		2
-# define DELOUSE(n)		mov     w##n, w##n
+# define PTR_ARG(n)		mov     w##n, w##n
+# define SIZE_ARG(n)		mov     w##n, w##n
 #endif

 #define PTR_SIZE	(1<<PTR_LOG_SIZE)
--- a/sysdeps/unix/sysv/linux/aarch64/clone.S
+++ b/sysdeps/unix/sysv/linux/aarch64/clone.S
@@ -33,13 +33,12 @@
 */
        .text
 ENTRY(__clone)
-	DELOUSE (0)
-	DELOUSE (1)
-	DELOUSE (2)
-	DELOUSE (3)
-	DELOUSE (4)
-	DELOUSE (5)
-	DELOUSE (6)
+	PTR_ARG (0)
+	PTR_ARG (1)
+	PTR_ARG (3)
+	PTR_ARG (4)
+	PTR_ARG (5)
+	PTR_ARG (6)
 	/* Save args for the child.  */
 	mov	x10, x0
 	mov	x11, x2
--- a/sysdeps/unix/sysv/linux/aarch64/getcontext.S
+++ b/sysdeps/unix/sysv/linux/aarch64/getcontext.S
@@ -30,7 +30,7 @@
 	.text

 ENTRY(__getcontext)
-	DELOUSE (0)
+	PTR_ARG (0)
 	/* The saved context will return to the getcontext() call point
 	   with a return value of 0 */
 	str	xzr,	  [x0, oX0 +  0 * SZREG]
--- a/sysdeps/unix/sysv/linux/aarch64/setcontext.S
+++ b/sysdeps/unix/sysv/linux/aarch64/setcontext.S
@@ -34,7 +34,7 @@
 	.text

 ENTRY (__setcontext)
-	DELOUSE (0)
+	PTR_ARG (0)
 	/* Save a copy of UCP.  */
 	mov	x9, x0

--- a/sysdeps/unix/sysv/linux/aarch64/swapcontext.S
+++ b/sysdeps/unix/sysv/linux/aarch64/swapcontext.S
@@ -27,7 +27,7 @@

 	.text
 ENTRY(__swapcontext)
-	DELOUSE (0)
+	PTR_ARG (0)
 	/* Set the value returned when swapcontext() returns in this context.
 	   And set up x1 to become the return address of the caller, so we
 	   can return there with a normal RET instead of an indirect jump.  */