in SIMD assmebler, changed .cleanup to a local symbol

git-svn-id: http://gnuradio.org/svn/gnuradio/trunk@8684 221aa14e-8319-0410-a670-987f0aec2ac5

diff --git a/gnuradio-core/src/lib/filter/complex_dotprod_sse.S b/gnuradio-core/src/lib/filter/complex_dotprod_sse.S
index 0e1cd337d5c8f789453aca66ad0922dfbdd10f91..f7448486414577d1ebdc66d42f8b590a846b2080 100644 (file)
--- a/gnuradio-core/src/lib/filter/complex_dotprod_sse.S
+++ b/gnuradio-core/src/lib/filter/complex_dotprod_sse.S
@@ -105,7 +105,7 @@ GLOB_SYMB(complex_dotprod_sse):
        movaps  %xmm5, %xmm7 
 
        shrl    $2, %ecx                # n_2_complex_blocks / 4
-       je      .cleanup                # if zero, take short path
+       je      .Lcleanup               # if zero, take short path
 
        # finish setup and loop priming
 
@@ -188,7 +188,7 @@ GLOB_SYMB(complex_dotprod_sse):
        # At this point, xmm4 contains 2x2 partial sums.  We need
        # to compute a "horizontal complex add" across xmm4.  
        
-.cleanup:                              # xmm4 = r1 i2 r3 i4
+.Lcleanup:                             # xmm4 = r1 i2 r3 i4
        movl    20(%ebp), %eax          # @result
        movhlps %xmm4, %xmm0            # xmm0 = ?? ?? r1 r2
        addps   %xmm4, %xmm0            # xmm0 = ?? ?? r1+r3 i2+i4

diff --git a/gnuradio-core/src/lib/filter/complex_dotprod_sse64.S b/gnuradio-core/src/lib/filter/complex_dotprod_sse64.S
index de258bc7ad22463aefdd3c36ca98df513736e19a..3096da20b4dba3275d4731c304df996cbd1ff119 100644 (file)
--- a/gnuradio-core/src/lib/filter/complex_dotprod_sse64.S
+++ b/gnuradio-core/src/lib/filter/complex_dotprod_sse64.S
@@ -103,7 +103,7 @@ GLOB_SYMB(complex_dotprod_sse):
        shr     $2, %rdx                # n_2_complex_blocks / 4
        movaps  %xmm5, %xmm7 
 
-       je      .cleanup                # if zero, take short path
+       je      .Lcleanup               # if zero, take short path
 
        # finish setup and loop priming
 
@@ -186,7 +186,7 @@ GLOB_SYMB(complex_dotprod_sse):
        # At this point, xmm4 contains 2x2 partial sums.  We need
        # to compute a "horizontal complex add" across xmm4.  
        
-.cleanup:                              # xmm4 = r1 i2 r3 i4
+.Lcleanup:                             # xmm4 = r1 i2 r3 i4
        movhlps %xmm4, %xmm0            # xmm0 = ?? ?? r1 r2
        addps   %xmm4, %xmm0            # xmm0 = ?? ?? r1+r3 i2+i4
        movlps  %xmm0, (%rcx)           # store low 2x32 bits (complex) to memory

diff --git a/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse.S b/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse.S
index 54b815abd8fb1d41c64aa591925f8c6571a28af9..98563f9054e33baac1f14a831eb41aa81009e35d 100644 (file)
--- a/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse.S
+++ b/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse.S
@@ -102,7 +102,7 @@ GLOB_SYMB(fcomplex_dotprod_sse):
        movaps  %xmm5, %xmm7 
 
        shrl    $2, %ecx                # n_2_complex_blocks / 4
-       je      .cleanup                # if zero, take short path
+       je      .Lcleanup               # if zero, take short path
 
        # finish setup and loop priming
 
@@ -171,7 +171,7 @@ GLOB_SYMB(fcomplex_dotprod_sse):
        # At this point, xmm4 contains 2x2 partial sums.  We need
        # to compute a "horizontal complex add" across xmm4.  
        
-.cleanup:                              # xmm4 = r1 i2 r3 i4
+.Lcleanup:                             # xmm4 = r1 i2 r3 i4
        movl    20(%ebp), %eax          # @result
        movhlps %xmm4, %xmm0            # xmm0 = ?? ?? r1 r2
        addps   %xmm4, %xmm0            # xmm0 = ?? ?? r1+r3 i2+i4

diff --git a/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse64.S b/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse64.S
index 31027c6e263092a42b40e7877278cbeeed02b6b2..b260bffcdc78339a692467b3726fd3879d32402f 100644 (file)
--- a/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse64.S
+++ b/gnuradio-core/src/lib/filter/fcomplex_dotprod_sse64.S
@@ -99,7 +99,7 @@ GLOB_SYMB(fcomplex_dotprod_sse):
        movaps  %xmm5, %xmm7 
 
        shr     $2, %rdx                # n_2_complex_blocks / 4
-       je      .cleanup                # if zero, take short path
+       je      .Lcleanup               # if zero, take short path
 
        # finish setup and loop priming
 
@@ -168,7 +168,7 @@ GLOB_SYMB(fcomplex_dotprod_sse):
        # At this point, xmm4 contains 2x2 partial sums.  We need
        # to compute a "horizontal complex add" across xmm4.  
        
-.cleanup:                              # xmm4 = r1 i2 r3 i4
+.Lcleanup:                             # xmm4 = r1 i2 r3 i4
        movhlps %xmm4, %xmm0            # xmm0 = ?? ?? r1 r2
        addps   %xmm4, %xmm0            # xmm0 = ?? ?? r1+r3 i2+i4
        movlps  %xmm0, (%rcx)           # store low 2x32 bits (complex) to memory

diff --git a/gnuradio-core/src/lib/filter/float_dotprod_sse.S b/gnuradio-core/src/lib/filter/float_dotprod_sse.S
index 5d5359fe7616e12f6b9564ce6ae92ea5160a7825..a8f03b83f864de64588526149f2e91084b678059 100644 (file)
--- a/gnuradio-core/src/lib/filter/float_dotprod_sse.S
+++ b/gnuradio-core/src/lib/filter/float_dotprod_sse.S
@@ -97,7 +97,7 @@ GLOB_SYMB(float_dotprod_sse):
        movaps  %xmm5, %xmm7 
 
        shrl    $2, %ecx                # n_4_float_blocks / 4
-       je      .cleanup                # if zero, take short path
+       je      .Lcleanup               # if zero, take short path
 
        # finish setup and loop priming
 
@@ -150,7 +150,7 @@ GLOB_SYMB(float_dotprod_sse):
        # to compute a "horizontal add" across xmm4.  
        # This is a fairly nasty operation...
        
-.cleanup:                              # xmm4 = d1 d2 d3 d4
+.Lcleanup:                             # xmm4 = d1 d2 d3 d4
        xorps   %xmm0, %xmm0            # xmm0 = 0 0 0 0  (may be unnecessary)
        movhlps %xmm4, %xmm0            # xmm0 = 0 0 d1 d2
        addps   %xmm4, %xmm0            # xmm0 = d1 d2 d1+d3 d2+d4

diff --git a/gnuradio-core/src/lib/filter/float_dotprod_sse64.S b/gnuradio-core/src/lib/filter/float_dotprod_sse64.S
index c725b69fb70a3e8140470be7b2cf4e7efd6d8938..696f2cc0d100ef8fd81c1179eeaae940b2070ddf 100644 (file)
--- a/gnuradio-core/src/lib/filter/float_dotprod_sse64.S
+++ b/gnuradio-core/src/lib/filter/float_dotprod_sse64.S
@@ -94,7 +94,7 @@ GLOB_SYMB(float_dotprod_sse):
        movaps  %xmm5, %xmm7 
 
        shr     $2, %rdx                # n_4_float_blocks / 4
-       je      .cleanup                # if zero, take short path
+       je      .Lcleanup               # if zero, take short path
 
        # finish setup and loop priming
 
@@ -147,7 +147,7 @@ GLOB_SYMB(float_dotprod_sse):
        # to compute a "horizontal add" across xmm4.  
        # This is a fairly nasty operation...
        
-.cleanup:                              # xmm4 = d1 d2 d3 d4
+.Lcleanup:                             # xmm4 = d1 d2 d3 d4
        xorps   %xmm0, %xmm0            # xmm0 = 0 0 0 0  (may be unnecessary)
        movhlps %xmm4, %xmm0            # xmm0 = 0 0 d1 d2
        addps   %xmm4, %xmm0            # xmm0 = d1 d2 d1+d3 d2+d4