This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
[COMMITTED PATCH] fix x86_64-nacl PLT nops
- From: Roland McGrath <mcgrathr at google dot com>
- To: binutils at sourceware dot org
- Date: Tue, 17 Apr 2012 12:53:20 -0700
- Subject: [COMMITTED PATCH] fix x86_64-nacl PLT nops
I used the wrong nop sequence in the x86_64's first PLT entry for NaCl
targets. I'd forgotten the issue that 32-byte aligned boundaries must
be valid instruction boundaries.
I've fixed it with this change, committed as obvious (obvious enough for a
simple change in code that I wrote recently).
Thanks,
Roland
bfd/
2012-04-17 Roland McGrath <mcgrathr@google.com>
* elf64-x86-64.c (elf_x86_64_nacl_plt0_entry): Fix nop padding
so that 32-byte boundary is a proper instruction boundary.
diff --git a/bfd/elf64-x86-64.c b/bfd/elf64-x86-64.c
index 8ca811c..3b1c0cc 100644
--- a/bfd/elf64-x86-64.c
+++ b/bfd/elf64-x86-64.c
@@ -5134,13 +5134,16 @@ static const bfd_byte elf_x86_64_nacl_plt0_entry[NACL_PLT_ENTRY_SIZE] =
0x4d, 0x01, 0xfb, /* add %r15, %r11 */
0x41, 0xff, 0xe3, /* jmpq *%r11 */
- /* 41 bytes of nop to pad out to the standard size. */
+ /* 9-byte nop sequence to pad out to the next 32-byte boundary. */
+ 0x2e, 0x0f, 0x1f, 0x84, 0, 0, 0, 0, 0, /* nopl %cs:0x0(%rax,%rax,1) */
+
+ /* 32 bytes of nop to pad out to the standard size. */
0x66, 0x66, 0x66, 0x66, 0x66, 0x66, /* excess data32 prefixes */
0x2e, 0x0f, 0x1f, 0x84, 0, 0, 0, 0, 0, /* nopw %cs:0x0(%rax,%rax,1) */
0x66, 0x66, 0x66, 0x66, 0x66, 0x66, /* excess data32 prefixes */
0x2e, 0x0f, 0x1f, 0x84, 0, 0, 0, 0, 0, /* nopw %cs:0x0(%rax,%rax,1) */
- 0x66, 0x66, /* excess data32 prefixes */
- 0x2e, 0x0f, 0x1f, 0x84, 0, 0, 0, 0, 0, /* nopw %cs:0x0(%rax,%rax,1) */
+ 0x66, /* excess data32 prefix */
+ 0x90 /* nop */
};
static const bfd_byte elf_x86_64_nacl_plt_entry[NACL_PLT_ENTRY_SIZE] =