>From binutils-2.19/gold/x86_64.cc:
// Nop sequences of various lengths.
const char nop1[1] = { 0x90 }; // nop
const char nop2[2] = { 0x66, 0x90 }; // xchg %ax %ax
const char nop3[3] = { 0x8d, 0x76, 0x00 }; // leal 0(%esi),%esi
const char nop4[4] = { 0x8d, 0x74, 0x26, 0x00}; // leal 0(%esi,1),%esi
This table was copied from i386.cc. However, most of these sequences are not
actually NOPs on x86-64! For example, nop3, which actually disassembles to lea
0(%rsi),%esi, has the effect of clearing the high 32 bits of %rsi. This can be
verified with a quick test program:
unsigned long x = 0x12345678ABCDEF00;
printf("x=%lx\n", x);
asm volatile (".byte 0x8d, 0x76, 0x00" : "=S" (x) : "S" (x));
printf("x=%lx\n", x);
Output:
x=12345678abcdef00
x=abcdef00
--
Summary: nops[] in binutils/gold/x86_64.cc are not actually x86-
64 NOPs
Product: binutils
Version: 2.19
Status: NEW
Severity: normal
Priority: P2
Component: gold
AssignedTo: ian at airs dot com
ReportedBy: anders at kaseorg dot com
CC: bug-binutils at gnu dot org
http://sourceware.org/bugzilla/show_bug.cgi?id=10419
------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.
_______________________________________________
bug-binutils mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/bug-binutils
