Use JUMPTARGET in x86-64 mathvec

When PLT may be used, JUMPTARGET should be used instead calling the
function directly.

	* sysdeps/x86_64/fpu/multiarch/svml_d_cos2_core_sse4.S
	(_ZGVbN2v_cos_sse4): Use JUMPTARGET to call cos.
	* sysdeps/x86_64/fpu/multiarch/svml_d_cos4_core_avx2.S
	(_ZGVdN4v_cos_avx2): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_cos8_core_avx512.S
	(_ZGVdN4v_cos): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_exp2_core_sse4.S
	(_ZGVbN2v_exp_sse4): Use JUMPTARGET to call exp.
	* sysdeps/x86_64/fpu/multiarch/svml_d_exp4_core_avx2.S
	(_ZGVdN4v_exp_avx2): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_exp8_core_avx512.S
	(_ZGVdN4v_exp): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_log2_core_sse4.S
	(_ZGVbN2v_log_sse4): Use JUMPTARGET to call log.
	* sysdeps/x86_64/fpu/multiarch/svml_d_log4_core_avx2.S
	(_ZGVdN4v_log_avx2): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_log8_core_avx512.S
	(_ZGVdN4v_log): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_pow2_core_sse4.S
	(_ZGVbN2vv_pow_sse4): Use JUMPTARGET to call pow.
	* sysdeps/x86_64/fpu/multiarch/svml_d_pow4_core_avx2.S
	(_ZGVdN4vv_pow_avx2): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_pow8_core_avx512.S
	(_ZGVdN4vv_pow): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_sin2_core_sse4.S
	(_ZGVbN2v_sin_sse4): Use JUMPTARGET to call sin.
	* sysdeps/x86_64/fpu/multiarch/svml_d_sin4_core_avx2.S
	(_ZGVdN4v_sin_avx2): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_sin8_core_avx512.S
	(_ZGVdN4v_sin): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_sincos2_core_sse4.S
	(_ZGVbN2vvv_sincos_sse4): Use JUMPTARGET to call sin and cos.
	* sysdeps/x86_64/fpu/multiarch/svml_d_sincos4_core_avx2.S
	(_ZGVdN4vvv_sincos_avx2): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_d_sincos8_core_avx512.S
	(_ZGVdN4vvv_sincos): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core_avx512.S
	(_ZGVdN8v_cosf): Use JUMPTARGET to call cosf.
	* sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core_sse4.S
	(_ZGVbN4v_cosf_sse4): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core_avx2.S
	(_ZGVdN8v_cosf_avx2): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_expf16_core_avx512.S
	(_ZGVdN8v_expf): Use JUMPTARGET to call expf.
	* sysdeps/x86_64/fpu/multiarch/svml_s_expf4_core_sse4.S
	(_ZGVbN4v_expf_sse4): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_expf8_core_avx2.S
	(_ZGVdN8v_expf_avx2): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_logf16_core_avx512.S
	(_ZGVdN8v_logf): Use JUMPTARGET to call logf.
	* sysdeps/x86_64/fpu/multiarch/svml_s_logf4_core_sse4.S
	(_ZGVbN4v_logf_sse4): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_logf8_core_avx2.S
	(_ZGVdN8v_logf_avx2): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_powf16_core_avx512.S
	(_ZGVdN8vv_powf): Use JUMPTARGET to call powf.
	* sysdeps/x86_64/fpu/multiarch/svml_s_powf4_core_sse4.S
	(_ZGVbN4vv_powf_sse4): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_powf8_core_avx2.S
	(_ZGVdN8vv_powf_avx2): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_sincosf16_core_avx512.S
	(_ZGVdN8vv_powf): Use JUMPTARGET to call sinf and cosf.
	* sysdeps/x86_64/fpu/multiarch/svml_s_sincosf4_core_sse4.S
	(_ZGVbN4vvv_sincosf_sse4): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_sincosf8_core_avx2.S
	(_ZGVdN8vvv_sincosf_avx2): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_sinf16_core_avx512.S
	(_ZGVdN8v_sinf): Use JUMPTARGET to call sinf.
	* sysdeps/x86_64/fpu/multiarch/svml_s_sinf4_core_sse4.S
	(_ZGVbN4v_sinf_sse4): Likewise.
	* sysdeps/x86_64/fpu/multiarch/svml_s_sinf8_core_avx2.S
	(_ZGVdN8v_sinf_avx2): Likewise.
	* sysdeps/x86_64/fpu/svml_d_wrapper_impl.h (WRAPPER_IMPL_SSE2):
	Use JUMPTARGET to call callee.
	(WRAPPER_IMPL_SSE2_ff): Likewise.
	(WRAPPER_IMPL_SSE2_fFF): Likewise.
	(WRAPPER_IMPL_AVX): Likewise.
	(WRAPPER_IMPL_AVX_ff): Likewise.
	(WRAPPER_IMPL_AVX_fFF): Likewise.
	(WRAPPER_IMPL_AVX512): Likewise.
	(WRAPPER_IMPL_AVX512_ff): Likewise.
	* sysdeps/x86_64/fpu/svml_s_wrapper_impl.h (WRAPPER_IMPL_SSE2):
	Likewise.
	(WRAPPER_IMPL_SSE2_ff): Likewise.
	(WRAPPER_IMPL_SSE2_fFF): Likewise.
	(WRAPPER_IMPL_AVX): Likewise.
	(WRAPPER_IMPL_AVX_ff): Likewise.
	(WRAPPER_IMPL_AVX_fFF): Likewise.
	(WRAPPER_IMPL_AVX512): Likewise.
	(WRAPPER_IMPL_AVX512_ff): Likewise.
	(WRAPPER_IMPL_AVX512_fFF): Likewise.

39 files changed, 224 insertions, 130 deletions

diff --git a/ChangeLog b/ChangeLog
index adf1752bf7..1eb8bd868a 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,97 @@
+2016-03-16  H.J. Lu  <hongjiu.lu@intel.com>
+
+	* sysdeps/x86_64/fpu/multiarch/svml_d_cos2_core_sse4.S
+	(_ZGVbN2v_cos_sse4): Use JUMPTARGET to call cos.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_cos4_core_avx2.S
+	(_ZGVdN4v_cos_avx2): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_cos8_core_avx512.S
+	(_ZGVdN4v_cos): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_exp2_core_sse4.S
+	(_ZGVbN2v_exp_sse4): Use JUMPTARGET to call exp.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_exp4_core_avx2.S
+	(_ZGVdN4v_exp_avx2): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_exp8_core_avx512.S
+	(_ZGVdN4v_exp): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_log2_core_sse4.S
+	(_ZGVbN2v_log_sse4): Use JUMPTARGET to call log.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_log4_core_avx2.S
+	(_ZGVdN4v_log_avx2): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_log8_core_avx512.S
+	(_ZGVdN4v_log): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_pow2_core_sse4.S
+	(_ZGVbN2vv_pow_sse4): Use JUMPTARGET to call pow.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_pow4_core_avx2.S
+	(_ZGVdN4vv_pow_avx2): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_pow8_core_avx512.S
+	(_ZGVdN4vv_pow): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_sin2_core_sse4.S
+	(_ZGVbN2v_sin_sse4): Use JUMPTARGET to call sin.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_sin4_core_avx2.S
+	(_ZGVdN4v_sin_avx2): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_sin8_core_avx512.S
+	(_ZGVdN4v_sin): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_sincos2_core_sse4.S
+	(_ZGVbN2vvv_sincos_sse4): Use JUMPTARGET to call sin and cos.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_sincos4_core_avx2.S
+	(_ZGVdN4vvv_sincos_avx2): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_d_sincos8_core_avx512.S
+	(_ZGVdN4vvv_sincos): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_cosf16_core_avx512.S
+	(_ZGVdN8v_cosf): Use JUMPTARGET to call cosf.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_cosf4_core_sse4.S
+	(_ZGVbN4v_cosf_sse4): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_cosf8_core_avx2.S
+	(_ZGVdN8v_cosf_avx2): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_expf16_core_avx512.S
+	(_ZGVdN8v_expf): Use JUMPTARGET to call expf.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_expf4_core_sse4.S
+	(_ZGVbN4v_expf_sse4): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_expf8_core_avx2.S
+	(_ZGVdN8v_expf_avx2): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_logf16_core_avx512.S
+	(_ZGVdN8v_logf): Use JUMPTARGET to call logf.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_logf4_core_sse4.S
+	(_ZGVbN4v_logf_sse4): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_logf8_core_avx2.S
+	(_ZGVdN8v_logf_avx2): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_powf16_core_avx512.S
+	(_ZGVdN8vv_powf): Use JUMPTARGET to call powf.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_powf4_core_sse4.S
+	(_ZGVbN4vv_powf_sse4): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_powf8_core_avx2.S
+	(_ZGVdN8vv_powf_avx2): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_sincosf16_core_avx512.S
+	(_ZGVdN8vv_powf): Use JUMPTARGET to call sinf and cosf.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_sincosf4_core_sse4.S
+	(_ZGVbN4vvv_sincosf_sse4): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_sincosf8_core_avx2.S
+	(_ZGVdN8vvv_sincosf_avx2): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_sinf16_core_avx512.S
+	(_ZGVdN8v_sinf): Use JUMPTARGET to call sinf.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_sinf4_core_sse4.S
+	(_ZGVbN4v_sinf_sse4): Likewise.
+	* sysdeps/x86_64/fpu/multiarch/svml_s_sinf8_core_avx2.S
+	(_ZGVdN8v_sinf_avx2): Likewise.
+	* sysdeps/x86_64/fpu/svml_d_wrapper_impl.h (WRAPPER_IMPL_SSE2):
+	Use JUMPTARGET to call callee.
+	(WRAPPER_IMPL_SSE2_ff): Likewise.
+	(WRAPPER_IMPL_SSE2_fFF): Likewise.
+	(WRAPPER_IMPL_AVX): Likewise.
+	(WRAPPER_IMPL_AVX_ff): Likewise.
+	(WRAPPER_IMPL_AVX_fFF): Likewise.
+	(WRAPPER_IMPL_AVX512): Likewise.
+	(WRAPPER_IMPL_AVX512_ff): Likewise.
+	* sysdeps/x86_64/fpu/svml_s_wrapper_impl.h (WRAPPER_IMPL_SSE2):
+	Likewise.
+	(WRAPPER_IMPL_SSE2_ff): Likewise.
+	(WRAPPER_IMPL_SSE2_fFF): Likewise.
+	(WRAPPER_IMPL_AVX): Likewise.
+	(WRAPPER_IMPL_AVX_ff): Likewise.
+	(WRAPPER_IMPL_AVX_fFF): Likewise.
+	(WRAPPER_IMPL_AVX512): Likewise.
+	(WRAPPER_IMPL_AVX512_ff): Likewise.
+	(WRAPPER_IMPL_AVX512_fFF): Likewise.
+
 2016-03-16  Samuel Thibault  <samuel.thibault@ens-lyon.org>
 
 	* sysdeps/mach/hurd/openat.c (__openat): Add missing ellipsis.
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cos2_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cos2_core_sse4.S
index 088fcae067..4d2ebf7aaf 100644
--- a/sysdeps/x86_64/fpu/multiarch/svml_d_cos2_core_sse4.S
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cos2_core_sse4.S
@@ -205,7 +205,7 @@ ENTRY (_ZGVbN2v_cos_sse4)
         shlq      $4, %r15
         movsd     200(%rsp,%r15), %xmm0
 
-        call      cos@PLT
+        call      JUMPTARGET(cos)
 
         movsd     %xmm0, 264(%rsp,%r15)
         jmp       .LBL_1_8
@@ -215,7 +215,7 @@ ENTRY (_ZGVbN2v_cos_sse4)
         shlq      $4, %r15
         movsd     192(%rsp,%r15), %xmm0
 
-        call      cos@PLT
+        call      JUMPTARGET(cos)
 
         movsd     %xmm0, 256(%rsp,%r15)
         jmp       .LBL_1_7
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cos4_core_avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cos4_core_avx2.S
index 4e653216d9..54f7e5e89e 100644
--- a/sysdeps/x86_64/fpu/multiarch/svml_d_cos4_core_avx2.S
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cos4_core_avx2.S
@@ -188,7 +188,7 @@ ENTRY (_ZGVdN4v_cos_avx2)
         vmovsd    328(%rsp,%r15), %xmm0
         vzeroupper
 
-        call      cos@PLT
+        call      JUMPTARGET(cos)
 
         vmovsd    %xmm0, 392(%rsp,%r15)
         jmp       .LBL_1_8
@@ -199,7 +199,7 @@ ENTRY (_ZGVdN4v_cos_avx2)
         vmovsd    320(%rsp,%r15), %xmm0
         vzeroupper
 
-        call      cos@PLT
+        call      JUMPTARGET(cos)
 
         vmovsd    %xmm0, 384(%rsp,%r15)
         jmp       .LBL_1_7
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_cos8_core_avx512.S b/sysdeps/x86_64/fpu/multiarch/svml_d_cos8_core_avx512.S
index 1cac1d827a..874bd80d23 100644
--- a/sysdeps/x86_64/fpu/multiarch/svml_d_cos8_core_avx512.S
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_cos8_core_avx512.S
@@ -221,7 +221,7 @@ WRAPPER_IMPL_AVX512 _ZGVdN4v_cos
         movzbl    %r12b, %r15d
         shlq      $4, %r15
         vmovsd    1160(%rsp,%r15), %xmm0
-        call      cos@PLT
+        call      JUMPTARGET(cos)
         vmovsd    %xmm0, 1224(%rsp,%r15)
         jmp       .LBL_1_8
 
@@ -229,7 +229,7 @@ WRAPPER_IMPL_AVX512 _ZGVdN4v_cos
         movzbl    %r12b, %r15d
         shlq      $4, %r15
         vmovsd    1152(%rsp,%r15), %xmm0
-        call      cos@PLT
+        call      JUMPTARGET(cos)
         vmovsd    %xmm0, 1216(%rsp,%r15)
         jmp       .LBL_1_7
 #endif
@@ -438,7 +438,7 @@ WRAPPER_IMPL_AVX512 _ZGVdN4v_cos
         vzeroupper
         vmovsd    1160(%rsp,%r15), %xmm0
 
-        call      cos@PLT
+        call      JUMPTARGET(cos)
 
         vmovsd    %xmm0, 1224(%rsp,%r15)
         jmp       .LBL_2_8
@@ -450,7 +450,7 @@ WRAPPER_IMPL_AVX512 _ZGVdN4v_cos
         vzeroupper
         vmovsd    1152(%rsp,%r15), %xmm0
 
-        call      cos@PLT
+        call      JUMPTARGET(cos)
 
         vmovsd    %xmm0, 1216(%rsp,%r15)
         jmp       .LBL_2_7
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_exp2_core_sse4.S b/sysdeps/x86_64/fpu/multiarch/svml_d_exp2_core_sse4.S
index 445b230152..9a779593cd 100644
--- a/sysdeps/x86_64/fpu/multiarch/svml_d_exp2_core_sse4.S
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_exp2_core_sse4.S
@@ -207,7 +207,7 @@ ENTRY (_ZGVbN2v_exp_sse4)
         shlq      $4, %r15
         movsd     200(%rsp,%r15), %xmm0
 
-        call      exp@PLT
+        call      JUMPTARGET(exp)
 
         movsd     %xmm0, 264(%rsp,%r15)
         jmp       .LBL_1_8
@@ -217,7 +217,7 @@ ENTRY (_ZGVbN2v_exp_sse4)
         shlq      $4, %r15
         movsd     192(%rsp,%r15), %xmm0
 
-        call      exp@PLT
+        call      JUMPTARGET(exp)
 
         movsd     %xmm0, 256(%rsp,%r15)
         jmp       .LBL_1_7
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_exp4_core_avx2.S b/sysdeps/x86_64/fpu/multiarch/svml_d_exp4_core_avx2.S
index 25f9e28941..2a35fe3846 100644
--- a/sysdeps/x86_64/fpu/multiarch/svml_d_exp4_core_avx2.S
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_exp4_core_avx2.S
@@ -193,7 +193,7 @@ ENTRY (_ZGVdN4v_exp_avx2)
         vmovsd    328(%rsp,%r15), %xmm0
         vzeroupper
 
-        call      exp@PLT
+        call      JUMPTARGET(exp)
 
         vmovsd    %xmm0, 392(%rsp,%r15)
         jmp       .LBL_1_8
@@ -204,7 +204,7 @@ ENTRY (_ZGVdN4v_exp_avx2)
         vmovsd    320(%rsp,%r15), %xmm0
         vzeroupper
 
-        call      exp@PLT
+        call      JUMPTARGET(exp)
 
         vmovsd    %xmm0, 384(%rsp,%r15)
         jmp       .LBL_1_7
diff --git a/sysdeps/x86_64/fpu/multiarch/svml_d_exp8_core_avx512.S b/sysdeps/x86_64/fpu/multiarch/svml_d_exp8_core_avx512.S
index 74f1d2ce7b..456792dec6 100644
--- a/sysdeps/x86_64/fpu/multiarch/svml_d_exp8_core_avx512.S
+++ b/sysdeps/x86_64/fpu/multiarch/svml_d_exp8_core_avx512.S
@@ -223,7 +223,7 @@ WRAPPER_IMPL_AVX512 _ZGVdN4v_exp
         movzbl    %r12b, %r15d
         shlq      $4, %r15
         vmovsd    1160(%rsp,%r15), %xmm0
-        call      exp@PLT
+        call      JUMPTARGET(exp)
         vmovsd    %xmm0, 1224(%rsp,%r15)
         jmp       .LBL_1_8
 
@@ -231,7 +231,7 @@ WRAPPER_IMPL_AVX512 _ZGVdN4v_exp
         movzbl    %r12b, %r15d
         shlq      $4, %r15
         vmovsd    1152(%rsp,%r15), %xmm0
-        call      exp@PLT
+        call      JUMPTARGET(exp)
         vmovsd    %xmm0, 1216(%rsp,%r15)
         jmp       .LBL_1_7
 #endif
@@ -438,7 +438,7 @@ WRAPPER_IMPL_AVX512 _ZGVdN4v_exp
         vmovsd    1160(%rsp,%r15), %xmm0
         vzeroupper
         vmovsd    1160(%rsp,%r15), %xmm0
-        call      exp@PLT
+        call      JUMPTARGET(exp)
         vmovsd    %xmm0, 1224(%rsp,%r15)
         jmp       .LBL_2_8
 
@@ -448,7 +448,7 @@ WRAPPER_IMPL_AVX512 _ZGVdN4v_exp
         vmovsd    1152(%rsp,%r15), %xmm0
         vzeroupper
         vmovsd    1152(%rsp,%r15), %xmm0
-        call      exp@PLT
+        call      JUMPTARGET(exp)
         vmovsd    %xmm0, 12