optimize repeating loading of aux shadow base

_From [bruen...@google.com](https://code.google.com/u/109494838902877177630/) on March 22, 2014 11:08:09_

We can optimize to share these loads of the aux shadow base:

+307 m4 @0x4f9c9a9c 64 8b 15 58 00 00 00 mov %fs:0x00000058 -> %edx
+314 m4 @0x4fa568a8 8b 52 04 mov 0x04(%edx) -> %edx
+317 m4 @0x505d47d4 64 8b 0d 58 00 00 00 mov %fs:0x00000058 -> %ecx
+324 m4 @0x505cd950 0b 51 04 or 0x04(%ecx) %edx -> %edx
+327 m4 @0x4fa52cf0 64 8b 0d 58 00 00 00 mov %fs:0x00000058 -> %ecx
+334 m4 @0x4f9c927c 89 51 04 mov %edx -> 0x04(%ecx)
+337 m4 @0x4fa530d0
+337 m4 @0x505cbe9c
+337 m4 @0x4f9c9a44
+337 m4 @0x4f9c5d90
+337 L3 66 0f ef c9 pxor %xmm1 %xmm1 -> %xmm1

_Original issue: http://code.google.com/p/drmemory/issues/detail?id=1486_