S390: Implement mempcpy with help of memcpy. [BZ #19765]

There exist optimized memcpy functions on s390, but no optimized mempcpy.
This patch adds mempcpy entry points in memcpy.S files, which
use the memcpy implementation. Now mempcpy itself is also an IFUNC function
as memcpy is and the variants are listed in ifunc-impl-list.c.

The s390 string.h does not define _HAVE_STRING_ARCH_mempcpy.
Instead mempcpy string/string.h inlines memcpy() + n.
If n is constant and small enough, GCC emits instructions like mvi or mvc
and avoids the function call to memcpy.
If n is not constant, then memcpy is called and n is added afterwards.
If _HAVE_STRING_ARCH_mempcpy would be defined, mempcpy would be called in
every case.

According to PR70140 "Inefficient expansion of __builtin_mempcpy"
(https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70140) GCC should handle a
call to mempcpy in the same way as memcpy. Then either the mempcpy macro
in string/string.h has to be removed or _HAVE_STRING_ARCH_mempcpy has to
be defined for S390.

ChangeLog:

	[BZ #19765]
	* sysdeps/s390/mempcpy.S: New File.
	* sysdeps/s390/multiarch/mempcpy.c: Likewise.
	* sysdeps/s390/multiarch/Makefile (sysdep_routines): Add mempcpy.
	* sysdeps/s390/multiarch/ifunc-impl-list.c (__libc_ifunc_impl_list):
	Add mempcpy variants.
	* sysdeps/s390/s390-32/memcpy.S: Add mempcpy entry point.
	(memcpy): Adjust to be usable from mempcpy entry point.
	(__memcpy_mvcle): Likewise.
	* sysdeps/s390/s390-64/memcpy.S: Likewise.
	* sysdeps/s390/s390-32/multiarch/memcpy-s390.S: Add entry points
	____mempcpy_z196, ____mempcpy_z10 and add __GI_ symbols for mempcpy.
	(__memcpy_z196): Adjust to be usable from mempcpy entry point.
	(__memcpy_z10): Likewise.
	* sysdeps/s390/s390-64/multiarch/memcpy-s390x.S: Likewise.

2 files changed, 54 insertions, 22 deletions

diff --git a/sysdeps/s390/s390-64/memcpy.S b/sysdeps/s390/s390-64/memcpy.S
index 9d60a14ccc..e3ace0c87b 100644
--- a/sysdeps/s390/s390-64/memcpy.S
+++ b/sysdeps/s390/s390-64/memcpy.S
@@ -27,19 +27,27 @@
 
 
        .text
+ENTRY(__mempcpy)
+	.machine "z900"
+	lgr     %r1,%r2             # Use as dest
+	la      %r2,0(%r4,%r2)      # Return dest + n
+	j	.L_Z900_start
+END(__mempcpy)
+#ifndef USE_MULTIARCH
+libc_hidden_def (__mempcpy)
+weak_alias (__mempcpy, mempcpy)
+libc_hidden_builtin_def (mempcpy)
+#endif
 
-#ifdef USE_MULTIARCH
-ENTRY(__memcpy_default)
-#else
 ENTRY(memcpy)
-#endif
 	.machine "z900"
+	lgr     %r1,%r2             # r1: Use as dest ; r2: Return dest
+.L_Z900_start:
 	ltgr    %r4,%r4
 	je      .L_Z900_4
 	aghi    %r4,-1
 	srlg    %r5,%r4,8
 	ltgr    %r5,%r5
-	lgr     %r1,%r2
 	jne     .L_Z900_13
 .L_Z900_3:
 	larl    %r5,.L_Z900_15
@@ -57,25 +65,24 @@ ENTRY(memcpy)
 	j       .L_Z900_3
 .L_Z900_15:
 	mvc     0(1,%r1),0(%r3)
-
-#ifdef USE_MULTIARCH
-END(__memcpy_default)
-#else
 END(memcpy)
+#ifndef USE_MULTIARCH
 libc_hidden_builtin_def (memcpy)
 #endif
 
 ENTRY(__memcpy_mvcle)
-       # Using as standalone function will result in unexpected
-       # results since the length field is incremented by 1 in order to
-       # compensate the changes already done in the functions above.
-       aghi    %r4,1               # length + 1
-       lgr     %r5,%r4             # source length
-       lgr     %r4,%r3             # source address
-       lgr     %r3,%r5             # destination length = source length
+	# Using as standalone function will result in unexpected
+	# results since the length field is incremented by 1 in order to
+	# compensate the changes already done in the functions above.
+	lgr     %r0,%r2             # backup return dest [ + n ]
+	aghi    %r4,1               # length + 1
+	lgr     %r5,%r4             # source length
+	lgr     %r4,%r3             # source address
+	lgr     %r2,%r1             # destination address
+	lgr     %r3,%r5             # destination length = source length
 .L_MVCLE_1:
-       mvcle   %r2,%r4,0           # thats it, MVCLE is your friend
-       jo      .L_MVCLE_1
-       lgr     %r2,%r1             # return destination address
-       br      %r14
+	mvcle   %r2,%r4,0           # thats it, MVCLE is your friend
+	jo      .L_MVCLE_1
+	lgr     %r2,%r0             # return destination address
+	br      %r14
 END(__memcpy_mvcle)
diff --git a/sysdeps/s390/s390-64/multiarch/memcpy-s390x.S b/sysdeps/s390/s390-64/multiarch/memcpy-s390x.S
index 8f54526b76..0f5a36edd1 100644
--- a/sysdeps/s390/s390-64/multiarch/memcpy-s390x.S
+++ b/sysdeps/s390/s390-64/multiarch/memcpy-s390x.S
@@ -29,12 +29,20 @@
 
 #if defined SHARED && IS_IN (libc)
 
+ENTRY(____mempcpy_z196)
+	.machine "z196"
+	lgr     %r1,%r2         # Use as dest
+	la      %r2,0(%r4,%r2)  # Return dest + n
+	j	.L_Z196_start
+END(____mempcpy_z196)
+
 ENTRY(__memcpy_z196)
 	.machine "z196"
+	lgr     %r1,%r2         # r1: Use as dest ; r2: Return dest
+.L_Z196_start:
 	ltgr    %r4,%r4
 	je      .L_Z196_4
 	aghi    %r4,-1
-	lgr     %r1,%r2
 	srlg    %r5,%r4,8
 	ltgr    %r5,%r5
 	jne     .L_Z196_5
@@ -58,11 +66,19 @@ ENTRY(__memcpy_z196)
 	mvc     0(1,%r1),0(%r3)
 END(__memcpy_z196)
 
+ENTRY(____mempcpy_z10)
+	.machine "z10"
+	lgr     %r1,%r2         # Use as dest
+	la      %r2,0(%r4,%r2)  # Return dest + n
+	j	.L_Z10_start
+END(____mempcpy_z10)
+
 ENTRY(__memcpy_z10)
 	.machine "z10"
+	lgr     %r1,%r2         # r1: Use as dest ; r2: Return dest
+.L_Z10_start:
 	cgije   %r4,0,.L_Z10_4
 	aghi    %r4,-1
-	lgr     %r1,%r2
 	srlg    %r5,%r4,8
 	cgijlh  %r5,0,.L_Z10_13
 .L_Z10_3:
@@ -84,14 +100,23 @@ ENTRY(__memcpy_z10)
 	mvc     0(1,%r1),0(%r3)
 END(__memcpy_z10)
 
+# define __mempcpy ____mempcpy_default
 #endif /* SHARED && IS_IN (libc) */
 
+#define memcpy __memcpy_default
 #include "../memcpy.S"
+#undef memcpy
 
 #if defined SHARED && IS_IN (libc)
 .globl   __GI_memcpy
 .set     __GI_memcpy,__memcpy_default
+.globl   __GI_mempcpy
+.set     __GI_mempcpy,____mempcpy_default
+.globl   __GI___mempcpy
+.set     __GI___mempcpy,____mempcpy_default
 #else
 .globl   memcpy
 .set     memcpy,__memcpy_default
+.weak    mempcpy
+.set     mempcpy,__mempcpy
 #endif