Hi all, gcc-4.2.2 seems to generating wrong/misaligned code for movapd. I have used the same test case mentione here (for almost the similar bug) http://gcc.gnu.org/bugzilla/attachment.cgi?id=6012
The relavent information about the version and the files are as follows: The version of gcc: gcc -v Using built-in specs. Target: i386-redhat-linux Configured with: ../../src/gcc-4.2.2/configure --prefix=/depot/gcc-4.2.2-static --disable-shared --enable-threads=posix --disable-checking --with-system-zlib --enable-__cxa_atexit --disable-libunwind-exceptions --enable-languages=c,c++,objc,fortran --with-cpu=generic --host=i386-redhat-linux Thread model: posi $ gcc -g -O2 -funsigned-bitfields -fsigned-char -ffloat-store -Wformat -msse2 -mfpmath=sse -c sse.c $ objdump -Sd sse.o > sse_asm.txt $ grep movapd sse_asm.txt a3: 66 0f 28 c2 movapd %xmm2,%xmm0 cb: 66 0f 28 c1 movapd %xmm1,%xmm0 f6: 66 0f 28 c4 movapd %xmm4,%xmm0 165: 66 0f 29 9d 38 fe ff movapd %xmm3,0xfffffe38(%ebp) #<============ 185: 66 0f 28 9d 38 fe ff movapd 0xfffffe38(%ebp),%xmm3 Is this a known issue? If so, is there are there any suggested workarounds (other than upgrading to later versions :-) ? Regards, Gowri Kumar
double sin(double x);
double cos(double x);
typedef struct {
double x, y, z;
} DVECTOR;
typedef struct {
DVECTOR m[4];
} DMATRIX;
void
create_geo_to_topo(DMATRIX *xfp, double lat, double lon)
{
double cos_lat;
{
double sin_lon, cos_lon, sin_lat;
sin_lon = sin(lon);
cos_lon = cos(lon);
sin_lat = sin(lat);
cos_lat = cos(lat);
xfp->m[0].x = -sin_lon;
xfp->m[1].x = cos_lon;
xfp->m[2].x = 0.0;
xfp->m[0].y = -cos_lon * sin_lat;
xfp->m[1].y = -sin_lon * sin_lat;
xfp->m[2].y = cos_lat;
xfp->m[0].z = cos_lon * cos_lat;
xfp->m[1].z = sin_lon * cos_lat;
xfp->m[2].z = sin_lat;
{
DVECTOR geo;
DVECTOR lla;
lla.x = lat;
lla.y = lon;
lla.z = 0.0;
geo_lla_xyz( &lla, &geo);
xfp->m[3].x = -geo.x * xfp->m[0].x -
geo.y * xfp->m[1].x -
geo.z * xfp->m[2].x;
xfp->m[3].y = -geo.x * xfp->m[0].y -
geo.y * xfp->m[1].y -
geo.z * xfp->m[2].y;
xfp->m[3].z = -geo.x * xfp->m[0].z -
geo.y * xfp->m[1].z -
geo.z * xfp->m[2].z;
}
}
}
sse.i
Description: Binary data
sse.s
Description: Binary data
sse.o: file format elf32-i386
Disassembly of section .text:
00000000 <create_geo_to_topo>:
void
create_geo_to_topo(DMATRIX *xfp, double lat, double lon)
{
0: 55 push %ebp
1: 89 e5 mov %esp,%ebp
3: 53 push %ebx
4: 81 ec d4 01 00 00 sub $0x1d4,%esp
a: 8b 45 0c mov 0xc(%ebp),%eax
d: 8b 5d 08 mov 0x8(%ebp),%ebx
10: 89 85 50 fe ff ff mov %eax,0xfffffe50(%ebp)
16: 8b 45 10 mov 0x10(%ebp),%eax
19: 89 85 54 fe ff ff mov %eax,0xfffffe54(%ebp)
1f: 8b 45 14 mov 0x14(%ebp),%eax
22: 89 85 48 fe ff ff mov %eax,0xfffffe48(%ebp)
28: 8b 45 18 mov 0x18(%ebp),%eax
2b: 89 85 4c fe ff ff mov %eax,0xfffffe4c(%ebp)
double cos_lat;
{
double sin_lon, cos_lon, sin_lat;
sin_lon = sin(lon);
31: f2 0f 10 85 48 fe ff movsd 0xfffffe48(%ebp),%xmm0
38: ff
39: f2 0f 11 04 24 movsd %xmm0,(%esp)
3e: e8 fc ff ff ff call 3f <create_geo_to_topo+0x3f>
cos_lon = cos(lon);
43: f2 0f 10 85 48 fe ff movsd 0xfffffe48(%ebp),%xmm0
4a: ff
4b: f2 0f 11 04 24 movsd %xmm0,(%esp)
50: dd 5d e8 fstpl 0xffffffe8(%ebp)
53: e8 fc ff ff ff call 54 <create_geo_to_topo+0x54>
sin_lat = sin(lat);
58: f2 0f 10 85 50 fe ff movsd 0xfffffe50(%ebp),%xmm0
5f: ff
60: f2 0f 11 04 24 movsd %xmm0,(%esp)
65: dd 5d e0 fstpl 0xffffffe0(%ebp)
68: e8 fc ff ff ff call 69 <create_geo_to_topo+0x69>
cos_lat = cos(lat);
6d: f2 0f 10 85 50 fe ff movsd 0xfffffe50(%ebp),%xmm0
74: ff
75: f2 0f 11 04 24 movsd %xmm0,(%esp)
7a: dd 5d d8 fstpl 0xffffffd8(%ebp)
7d: e8 fc ff ff ff call 7e <create_geo_to_topo+0x7e>
xfp->m[0].x = -sin_lon;
xfp->m[1].x = cos_lon;
xfp->m[2].x = 0.0;
xfp->m[0].y = -cos_lon * sin_lat;
xfp->m[1].y = -sin_lon * sin_lat;
xfp->m[2].y = cos_lat;
xfp->m[0].z = cos_lon * cos_lat;
xfp->m[1].z = sin_lon * cos_lat;
xfp->m[2].z = sin_lat;
{
DVECTOR geo;
DVECTOR lla;
lla.x = lat;
lla.y = lon;
lla.z = 0.0;
geo_lla_xyz( &lla, &geo);
82: 8d 85 70 fe ff ff lea 0xfffffe70(%ebp),%eax
88: f2 0f 10 55 e8 movsd 0xffffffe8(%ebp),%xmm2
8d: f2 0f 10 4d e0 movsd 0xffffffe0(%ebp),%xmm1
92: f2 0f 10 65 d8 movsd 0xffffffd8(%ebp),%xmm4
97: f2 0f 10 1d 00 00 00 movsd 0x0,%xmm3
9e: 00
9f: 66 0f 57 ed xorpd %xmm5,%xmm5
a3: 66 0f 28 c2 movapd %xmm2,%xmm0
a7: 66 0f 57 c3 xorpd %xmm3,%xmm0
ab: f2 0f 11 45 d0 movsd %xmm0,0xffffffd0(%ebp)
b0: f2 0f 10 45 d0 movsd 0xffffffd0(%ebp),%xmm0
b5: dd 5d f0 fstpl 0xfffffff0(%ebp)
b8: f2 0f 11 03 movsd %xmm0,(%ebx)
bc: f2 0f 11 4b 18 movsd %xmm1,0x18(%ebx)
c1: f2 0f 11 6b 30 movsd %xmm5,0x30(%ebx)
c6: f2 0f 59 55 f0 mulsd 0xfffffff0(%ebp),%xmm2
cb: 66 0f 28 c1 movapd %xmm1,%xmm0
cf: f2 0f 59 4d f0 mulsd 0xfffffff0(%ebp),%xmm1
d4: 66 0f 57 c3 xorpd %xmm3,%xmm0
d8: f2 0f 11 45 c8 movsd %xmm0,0xffffffc8(%ebp)
dd: f2 0f 10 45 c8 movsd 0xffffffc8(%ebp),%xmm0
e2: f2 0f 59 45 d8 mulsd 0xffffffd8(%ebp),%xmm0
e7: f2 0f 11 45 c0 movsd %xmm0,0xffffffc0(%ebp)
ec: f2 0f 10 45 c0 movsd 0xffffffc0(%ebp),%xmm0
f1: f2 0f 11 43 08 movsd %xmm0,0x8(%ebx)
f6: 66 0f 28 c4 movapd %xmm4,%xmm0
fa: f2 0f 59 45 d0 mulsd 0xffffffd0(%ebp),%xmm0
ff: f2 0f 11 45 b8 movsd %xmm0,0xffffffb8(%ebp)
104: f2 0f 10 45 b8 movsd 0xffffffb8(%ebp),%xmm0
109: f2 0f 11 43 20 movsd %xmm0,0x20(%ebx)
10e: f2 0f 10 45 f0 movsd 0xfffffff0(%ebp),%xmm0
113: f2 0f 11 43 38 movsd %xmm0,0x38(%ebx)
118: f2 0f 11 4d b0 movsd %xmm1,0xffffffb0(%ebp)
11d: f2 0f 10 45 b0 movsd 0xffffffb0(%ebp),%xmm0
122: f2 0f 11 43 10 movsd %xmm0,0x10(%ebx)
127: f2 0f 11 55 a8 movsd %xmm2,0xffffffa8(%ebp)
12c: f2 0f 10 45 a8 movsd 0xffffffa8(%ebp),%xmm0
131: f2 0f 11 63 40 movsd %xmm4,0x40(%ebx)
136: f2 0f 11 43 28 movsd %xmm0,0x28(%ebx)
13b: f2 0f 10 85 50 fe ff movsd 0xfffffe50(%ebp),%xmm0
142: ff
143: f2 0f 11 85 58 fe ff movsd %xmm0,0xfffffe58(%ebp)
14a: ff
14b: f2 0f 10 85 48 fe ff movsd 0xfffffe48(%ebp),%xmm0
152: ff
153: f2 0f 11 85 60 fe ff movsd %xmm0,0xfffffe60(%ebp)
15a: ff
15b: 89 44 24 04 mov %eax,0x4(%esp)
15f: 8d 85 58 fe ff ff lea 0xfffffe58(%ebp),%eax
165: 66 0f 29 9d 38 fe ff movapd %xmm3,0xfffffe38(%ebp)
16c: ff
16d: f2 0f 11 ad 68 fe ff movsd %xmm5,0xfffffe68(%ebp)
174: ff
175: 89 04 24 mov %eax,(%esp)
178: e8 fc ff ff ff call 179 <create_geo_to_topo+0x179>
xfp->m[3].x = -geo.x * xfp->m[0].x -
17d: f2 0f 10 a5 70 fe ff movsd 0xfffffe70(%ebp),%xmm4
184: ff
185: 66 0f 28 9d 38 fe ff movapd 0xfffffe38(%ebp),%xmm3
18c: ff
18d: f2 0f 10 95 78 fe ff movsd 0xfffffe78(%ebp),%xmm2
194: ff
195: f2 0f 10 8d 80 fe ff movsd 0xfffffe80(%ebp),%xmm1
19c: ff
19d: f2 0f 11 65 a0 movsd %xmm4,0xffffffa0(%ebp)
1a2: f2 0f 10 45 a0 movsd 0xffffffa0(%ebp),%xmm0
1a7: 66 0f 57 c3 xorpd %xmm3,%xmm0
1ab: f2 0f 11 45 98 movsd %xmm0,0xffffff98(%ebp)
1b0: f2 0f 10 03 movsd (%ebx),%xmm0
1b4: f2 0f 11 55 80 movsd %xmm2,0xffffff80(%ebp)
1b9: f2 0f 11 45 90 movsd %xmm0,0xffffff90(%ebp)
1be: f2 0f 10 45 98 movsd 0xffffff98(%ebp),%xmm0
1c3: f2 0f 59 45 90 mulsd 0xffffff90(%ebp),%xmm0
1c8: f2 0f 11 45 88 movsd %xmm0,0xffffff88(%ebp)
1cd: f2 0f 10 43 18 movsd 0x18(%ebx),%xmm0
1d2: f2 0f 11 8d 60 ff ff movsd %xmm1,0xffffff60(%ebp)
1d9: ff
1da: f2 0f 11 85 78 ff ff movsd %xmm0,0xffffff78(%ebp)
1e1: ff
1e2: f2 0f 10 45 80 movsd 0xffffff80(%ebp),%xmm0
1e7: f2 0f 59 85 78 ff ff mulsd 0xffffff78(%ebp),%xmm0
1ee: ff
1ef: f2 0f 11 85 70 ff ff movsd %xmm0,0xffffff70(%ebp)
1f6: ff
1f7: f2 0f 10 45 88 movsd 0xffffff88(%ebp),%xmm0
1fc: f2 0f 5c 85 70 ff ff subsd 0xffffff70(%ebp),%xmm0
203: ff
204: f2 0f 11 85 68 ff ff movsd %xmm0,0xffffff68(%ebp)
20b: ff
20c: f2 0f 10 43 30 movsd 0x30(%ebx),%xmm0
211: f2 0f 11 85 58 ff ff movsd %xmm0,0xffffff58(%ebp)
218: ff
219: f2 0f 10 85 60 ff ff movsd 0xffffff60(%ebp),%xmm0
220: ff
221: f2 0f 59 85 58 ff ff mulsd 0xffffff58(%ebp),%xmm0
228: ff
229: f2 0f 11 85 50 ff ff movsd %xmm0,0xffffff50(%ebp)
230: ff
231: f2 0f 10 85 68 ff ff movsd 0xffffff68(%ebp),%xmm0
238: ff
239: f2 0f 5c 85 50 ff ff subsd 0xffffff50(%ebp),%xmm0
240: ff
241: f2 0f 11 85 48 ff ff movsd %xmm0,0xffffff48(%ebp)
248: ff
249: f2 0f 10 85 48 ff ff movsd 0xffffff48(%ebp),%xmm0
250: ff
251: f2 0f 11 43 48 movsd %xmm0,0x48(%ebx)
geo.y * xfp->m[1].x -
geo.z * xfp->m[2].x;
xfp->m[3].y = -geo.x * xfp->m[0].y -
256: f2 0f 11 a5 90 fe ff movsd %xmm4,0xfffffe90(%ebp)
25d: ff
25e: f2 0f 10 85 90 fe ff movsd 0xfffffe90(%ebp),%xmm0
265: ff
266: 66 0f 57 c3 xorpd %xmm3,%xmm0
26a: f2 0f 11 85 a0 fe ff movsd %xmm0,0xfffffea0(%ebp)
271: ff
272: f2 0f 10 43 08 movsd 0x8(%ebx),%xmm0
277: f2 0f 11 95 b0 fe ff movsd %xmm2,0xfffffeb0(%ebp)
27e: ff
27f: f2 0f 11 85 40 ff ff movsd %xmm0,0xffffff40(%ebp)
286: ff
287: f2 0f 10 85 a0 fe ff movsd 0xfffffea0(%ebp),%xmm0
28e: ff
28f: f2 0f 59 85 40 ff ff mulsd 0xffffff40(%ebp),%xmm0
296: ff
297: f2 0f 11 85 38 ff ff movsd %xmm0,0xffffff38(%ebp)
29e: ff
29f: f2 0f 10 43 20 movsd 0x20(%ebx),%xmm0
2a4: f2 0f 11 8d c0 fe ff movsd %xmm1,0xfffffec0(%ebp)
2ab: ff
2ac: f2 0f 11 85 30 ff ff movsd %xmm0,0xffffff30(%ebp)
2b3: ff
2b4: f2 0f 10 85 b0 fe ff movsd 0xfffffeb0(%ebp),%xmm0
2bb: ff
2bc: f2 0f 59 85 30 ff ff mulsd 0xffffff30(%ebp),%xmm0
2c3: ff
2c4: f2 0f 11 85 28 ff ff movsd %xmm0,0xffffff28(%ebp)
2cb: ff
2cc: f2 0f 10 85 38 ff ff movsd 0xffffff38(%ebp),%xmm0
2d3: ff
2d4: f2 0f 5c 85 28 ff ff subsd 0xffffff28(%ebp),%xmm0
2db: ff
2dc: f2 0f 11 85 20 ff ff movsd %xmm0,0xffffff20(%ebp)
2e3: ff
2e4: f2 0f 10 43 38 movsd 0x38(%ebx),%xmm0
2e9: f2 0f 11 85 18 ff ff movsd %xmm0,0xffffff18(%ebp)
2f0: ff
2f1: f2 0f 10 85 c0 fe ff movsd 0xfffffec0(%ebp),%xmm0
2f8: ff
2f9: f2 0f 59 85 18 ff ff mulsd 0xffffff18(%ebp),%xmm0
300: ff
301: f2 0f 11 85 10 ff ff movsd %xmm0,0xffffff10(%ebp)
308: ff
309: f2 0f 10 85 20 ff ff movsd 0xffffff20(%ebp),%xmm0
310: ff
311: f2 0f 5c 85 10 ff ff subsd 0xffffff10(%ebp),%xmm0
318: ff
319: f2 0f 11 85 08 ff ff movsd %xmm0,0xffffff08(%ebp)
320: ff
321: f2 0f 10 85 08 ff ff movsd 0xffffff08(%ebp),%xmm0
328: ff
329: f2 0f 11 43 50 movsd %xmm0,0x50(%ebx)
geo.y * xfp->m[1].y -
geo.z * xfp->m[2].y;
xfp->m[3].z = -geo.x * xfp->m[0].z -
32e: f2 0f 11 a5 88 fe ff movsd %xmm4,0xfffffe88(%ebp)
335: ff
336: f2 0f 10 85 88 fe ff movsd 0xfffffe88(%ebp),%xmm0
33d: ff
33e: 66 0f 57 c3 xorpd %xmm3,%xmm0
342: f2 0f 11 85 98 fe ff movsd %xmm0,0xfffffe98(%ebp)
349: ff
34a: f2 0f 10 43 10 movsd 0x10(%ebx),%xmm0
34f: f2 0f 11 85 00 ff ff movsd %xmm0,0xffffff00(%ebp)
356: ff
357: f2 0f 10 85 98 fe ff movsd 0xfffffe98(%ebp),%xmm0
35e: ff
35f: f2 0f 59 85 00 ff ff mulsd 0xffffff00(%ebp),%xmm0
366: ff
367: f2 0f 11 85 f8 fe ff movsd %xmm0,0xfffffef8(%ebp)
36e: ff
36f: f2 0f 11 95 a8 fe ff movsd %xmm2,0xfffffea8(%ebp)
376: ff
377: f2 0f 10 43 28 movsd 0x28(%ebx),%xmm0
37c: f2 0f 11 8d b8 fe ff movsd %xmm1,0xfffffeb8(%ebp)
383: ff
384: f2 0f 11 85 f0 fe ff movsd %xmm0,0xfffffef0(%ebp)
38b: ff
38c: f2 0f 10 85 a8 fe ff movsd 0xfffffea8(%ebp),%xmm0
393: ff
394: f2 0f 59 85 f0 fe ff mulsd 0xfffffef0(%ebp),%xmm0
39b: ff
39c: f2 0f 11 85 e8 fe ff movsd %xmm0,0xfffffee8(%ebp)
3a3: ff
3a4: f2 0f 10 85 f8 fe ff movsd 0xfffffef8(%ebp),%xmm0
3ab: ff
3ac: f2 0f 5c 85 e8 fe ff subsd 0xfffffee8(%ebp),%xmm0
3b3: ff
3b4: f2 0f 11 85 e0 fe ff movsd %xmm0,0xfffffee0(%ebp)
3bb: ff
3bc: f2 0f 10 43 40 movsd 0x40(%ebx),%xmm0
3c1: f2 0f 11 85 d8 fe ff movsd %xmm0,0xfffffed8(%ebp)
3c8: ff
3c9: f2 0f 10 85 b8 fe ff movsd 0xfffffeb8(%ebp),%xmm0
3d0: ff
3d1: f2 0f 59 85 d8 fe ff mulsd 0xfffffed8(%ebp),%xmm0
3d8: ff
3d9: f2 0f 11 85 d0 fe ff movsd %xmm0,0xfffffed0(%ebp)
3e0: ff
3e1: f2 0f 10 85 e0 fe ff movsd 0xfffffee0(%ebp),%xmm0
3e8: ff
3e9: f2 0f 5c 85 d0 fe ff subsd 0xfffffed0(%ebp),%xmm0
3f0: ff
3f1: f2 0f 11 85 c8 fe ff movsd %xmm0,0xfffffec8(%ebp)
3f8: ff
3f9: f2 0f 10 85 c8 fe ff movsd 0xfffffec8(%ebp),%xmm0
400: ff
401: f2 0f 11 43 58 movsd %xmm0,0x58(%ebx)
geo.y * xfp->m[1].z -
geo.z * xfp->m[2].z;
}
}
}
406: 81 c4 d4 01 00 00 add $0x1d4,%esp
40c: 5b pop %ebx
40d: 5d pop %ebp
40e: c3 ret
