From: Jan Beulich <jbeulich@suse.com>
To: Binutils <binutils@sourceware.org>
Cc: "H.J. Lu" <hjl.tools@gmail.com>
Subject: [PATCH 1/3] x86: use fixed-width type for codep and friends
Date: Fri, 19 May 2023 16:06:31 +0200 [thread overview]
Message-ID: <2274d914-c77f-37e7-5589-870a647e24a0@suse.com> (raw)
In-Reply-To: <cdaeb0ea-7ac0-c07b-6b7d-2478b8b5f437@suse.com>
This first of all removes a dependency on bfd_byte and unsigned char
being the same types. It further eliminates the need to mask by 0xff
when fetching values (which wasn't done fully consistently anyway),
improving code legibility.
While there, where possible add const.
--- a/opcodes/i386-dis.c
+++ b/opcodes/i386-dis.c
@@ -120,8 +120,8 @@ static void ATTRIBUTE_PRINTF_3 i386_dis_
struct dis_private {
/* Points to first byte not fetched. */
- bfd_byte *max_fetched;
- bfd_byte the_buffer[MAX_MNEM_SIZE];
+ uint8_t *max_fetched;
+ uint8_t the_buffer[MAX_MNEM_SIZE];
bfd_vma insn_start;
int orig_sizeflag;
};
@@ -133,7 +133,7 @@ enum address_mode
mode_64bit
};
-static const char *prefix_name (enum address_mode, int, int);
+static const char *prefix_name (enum address_mode, uint8_t, int);
enum x86_64_isa
{
@@ -149,9 +149,9 @@ struct instr_info
int prefixes;
/* REX prefix the current instruction. See below. */
- unsigned char rex;
+ uint8_t rex;
/* Bits of REX we've already used. */
- unsigned char rex_used;
+ uint8_t rex_used;
bool need_modrm;
bool need_vex;
@@ -168,10 +168,10 @@ struct instr_info
char obuf[MAX_OPERAND_BUFFER_SIZE];
char *obufp;
char *mnemonicendp;
- unsigned char *start_codep;
- unsigned char *insn_codep;
- unsigned char *codep;
- unsigned char *end_codep;
+ const uint8_t *start_codep;
+ uint8_t *insn_codep;
+ uint8_t *codep;
+ const uint8_t *end_codep;
signed char last_lock_prefix;
signed char last_repz_prefix;
signed char last_repnz_prefix;
@@ -186,7 +186,7 @@ struct instr_info
#define MAX_CODE_LENGTH 15
/* We can up to 14 ins->prefixes since the maximum instruction length is
15bytes. */
- unsigned char all_prefixes[MAX_CODE_LENGTH - 1];
+ uint8_t all_prefixes[MAX_CODE_LENGTH - 1];
disassemble_info *info;
struct
@@ -288,7 +288,7 @@ struct instr_info
to ADDR (exclusive) are valid. Returns true for success, false
on error. */
static bool
-fetch_code (struct disassemble_info *info, bfd_byte *until)
+fetch_code (struct disassemble_info *info, uint8_t *until)
{
int status = -1;
struct dis_private *priv = info->private_data;
@@ -8794,7 +8794,8 @@ static enum {
}
ckprefix (instr_info *ins)
{
- int newrex, i, length;
+ int i, length;
+ uint8_t newrex;
i = 0;
length = 0;
@@ -8804,7 +8805,7 @@ ckprefix (instr_info *ins)
if (!fetch_code (ins->info, ins->codep + 1))
return ckp_fetch_error;
newrex = 0;
- switch (*ins->codep & 0xff)
+ switch (*ins->codep)
{
/* REX prefixes family. */
case 0x40:
@@ -8824,7 +8825,7 @@ ckprefix (instr_info *ins)
case 0x4e:
case 0x4f:
if (ins->address_mode == mode_64bit)
- newrex = *ins->codep & 0xff;
+ newrex = *ins->codep;
else
return ckp_okay;
ins->last_rex_prefix = i;
@@ -8904,8 +8905,8 @@ ckprefix (instr_info *ins)
/* Rex is ignored when followed by another prefix. */
if (ins->rex)
return ckp_bogus;
- if ((*ins->codep & 0xff) != FWAIT_OPCODE)
- ins->all_prefixes[i++] = *ins->codep & 0xff;
+ if (*ins->codep != FWAIT_OPCODE)
+ ins->all_prefixes[i++] = *ins->codep;
ins->rex = newrex;
ins->codep++;
length++;
@@ -8917,7 +8918,7 @@ ckprefix (instr_info *ins)
prefix byte. */
static const char *
-prefix_name (enum address_mode mode, int pref, int sizeflag)
+prefix_name (enum address_mode mode, uint8_t pref, int sizeflag)
{
static const char *rexes [16] =
{
@@ -9136,7 +9137,7 @@ get_valid_dis386 (const struct dis386 *d
case USE_3BYTE_TABLE:
if (!fetch_code (ins->info, ins->codep + 2))
return &err_opcode;
- vindex = *ins->codep++ & 0xff;
+ vindex = *ins->codep++;
dp = &three_byte_table[dp->op[1].bytemode][vindex];
ins->end_codep = ins->codep;
if (!fetch_modrm (ins))
@@ -9243,7 +9244,7 @@ get_valid_dis386 (const struct dis386 *d
}
ins->need_vex = true;
ins->codep++;
- vindex = *ins->codep++ & 0xff;
+ vindex = *ins->codep++;
dp = &xop_table[vex_table_index][vindex];
ins->end_codep = ins->codep;
@@ -9308,7 +9309,7 @@ get_valid_dis386 (const struct dis386 *d
}
ins->need_vex = true;
ins->codep++;
- vindex = *ins->codep++ & 0xff;
+ vindex = *ins->codep++;
dp = &vex_table[vex_table_index][vindex];
ins->end_codep = ins->codep;
/* There is no MODRM byte for VEX0F 77. */
@@ -9343,7 +9344,7 @@ get_valid_dis386 (const struct dis386 *d
}
ins->need_vex = true;
ins->codep++;
- vindex = *ins->codep++ & 0xff;
+ vindex = *ins->codep++;
dp = &vex_table[dp->op[1].bytemode][vindex];
ins->end_codep = ins->codep;
/* There is no MODRM byte for VEX 77. */
@@ -9435,7 +9436,7 @@ get_valid_dis386 (const struct dis386 *d
ins->need_vex = true;
ins->codep++;
- vindex = *ins->codep++ & 0xff;
+ vindex = *ins->codep++;
dp = &evex_table[vex_table_index][vindex];
ins->end_codep = ins->codep;
if (!fetch_modrm (ins))
@@ -9782,11 +9783,10 @@ print_insn (bfd_vma pc, disassemble_info
goto out;
}
- ins.two_source_ops = ((*ins.codep & 0xff) == 0x62
- || (*ins.codep & 0xff) == 0xc8);
+ ins.two_source_ops = (*ins.codep == 0x62 || *ins.codep == 0xc8);
if ((ins.prefixes & PREFIX_FWAIT)
- && ((*ins.codep & 0xff) < 0xd8 || (*ins.codep & 0xff) > 0xdf))
+ && (*ins.codep < 0xd8 || *ins.codep > 0xdf))
{
/* Handle ins.prefixes before fwait. */
for (i = 0; i < ins.fwait_prefix && ins.all_prefixes[i];
@@ -9799,22 +9799,22 @@ print_insn (bfd_vma pc, disassemble_info
goto out;
}
- if ((*ins.codep & 0xff) == 0x0f)
+ if (*ins.codep == 0x0f)
{
unsigned char threebyte;
ins.codep++;
if (!fetch_code (info, ins.codep + 1))
goto fetch_error_out;
- threebyte = *ins.codep & 0xff;
+ threebyte = *ins.codep;
dp = &dis386_twobyte[threebyte];
ins.need_modrm = twobyte_has_modrm[threebyte];
ins.codep++;
}
else
{
- dp = &dis386[*ins.codep & 0xff];
- ins.need_modrm = onebyte_has_modrm[*ins.codep & 0xff];
+ dp = &dis386[*ins.codep];
+ ins.need_modrm = onebyte_has_modrm[*ins.codep];
ins.codep++;
}
@@ -10514,9 +10514,7 @@ static bool
dofloat (instr_info *ins, int sizeflag)
{
const struct dis386 *dp;
- unsigned char floatop;
-
- floatop = ins->codep[-1] & 0xff;
+ unsigned char floatop = ins->codep[-1];
if (ins->modrm.mod != 3)
{
@@ -10537,7 +10535,7 @@ dofloat (instr_info *ins, int sizeflag)
putop (ins, fgrps[dp->op[0].bytemode][ins->modrm.rm], sizeflag);
/* Instruction fnstsw is only one with strange arg. */
- if (floatop == 0xdf && (ins->codep[-1] & 0xff) == 0xe0)
+ if (floatop == 0xdf && ins->codep[-1] == 0xe0)
strcpy (ins->op_out[0], att_names16[0] + ins->intel_syntax);
}
else
@@ -11710,7 +11708,7 @@ get8s (instr_info *ins, bfd_vma *res)
{
if (!fetch_code (ins->info, ins->codep + 1))
return false;
- *res = (((bfd_vma) *ins->codep++ & 0xff) ^ 0x80) - 0x80;
+ *res = ((bfd_vma) *ins->codep++ ^ 0x80) - 0x80;
return true;
}
@@ -11719,8 +11717,8 @@ get16 (instr_info *ins, bfd_vma *res)
{
if (!fetch_code (ins->info, ins->codep + 2))
return false;
- *res = (bfd_vma) *ins->codep++ & 0xff;
- *res |= ((bfd_vma) *ins->codep++ & 0xff) << 8;
+ *res = *ins->codep++;
+ *res |= (bfd_vma) *ins->codep++ << 8;
return true;
}
@@ -11738,10 +11736,10 @@ get32 (instr_info *ins, bfd_vma *res)
{
if (!fetch_code (ins->info, ins->codep + 4))
return false;
- *res = *ins->codep++ & (bfd_vma) 0xff;
- *res |= (*ins->codep++ & (bfd_vma) 0xff) << 8;
- *res |= (*ins->codep++ & (bfd_vma) 0xff) << 16;
- *res |= (*ins->codep++ & (bfd_vma) 0xff) << 24;
+ *res = *ins->codep++;
+ *res |= (bfd_vma) *ins->codep++ << 8;
+ *res |= (bfd_vma) *ins->codep++ << 16;
+ *res |= (bfd_vma) *ins->codep++ << 24;
return true;
}
@@ -11764,14 +11762,14 @@ get64 (instr_info *ins, uint64_t *res)
if (!fetch_code (ins->info, ins->codep + 8))
return false;
- a = *ins->codep++ & 0xff;
- a |= (*ins->codep++ & 0xff) << 8;
- a |= (*ins->codep++ & 0xff) << 16;
- a |= (*ins->codep++ & 0xffu) << 24;
- b = *ins->codep++ & 0xff;
- b |= (*ins->codep++ & 0xff) << 8;
- b |= (*ins->codep++ & 0xff) << 16;
- b |= (*ins->codep++ & 0xffu) << 24;
+ a = *ins->codep++;
+ a |= (unsigned int) *ins->codep++ << 8;
+ a |= (unsigned int) *ins->codep++ << 16;
+ a |= (unsigned int) *ins->codep++ << 24;
+ b = *ins->codep++;
+ b |= (unsigned int) *ins->codep++ << 8;
+ b |= (unsigned int) *ins->codep++ << 16;
+ b |= (unsigned int) *ins->codep++ << 24;
*res = a + ((uint64_t) b << 32);
return true;
}
@@ -12450,7 +12448,7 @@ OP_I (instr_info *ins, int bytemode, int
case b_mode:
if (!fetch_code (ins->info, ins->codep + 1))
return false;
- op = *ins->codep++ & 0xff;
+ op = *ins->codep++;
break;
case v_mode:
USED_REX (REX_W);
@@ -12737,7 +12735,7 @@ OP_ESreg (instr_info *ins, int code, int
{
if (ins->intel_syntax)
{
- switch (ins->codep[-1] & 0xff)
+ switch (ins->codep[-1])
{
case 0x6d: /* insw/insl */
intel_operand_size (ins, z_mode, sizeflag);
@@ -12763,7 +12761,7 @@ OP_DSreg (instr_info *ins, int code, int
{
if (ins->intel_syntax)
{
- switch (ins->codep[-1] & 0xff)
+ switch (ins->codep[-1])
{
case 0x6f: /* outsw/outsl */
intel_operand_size (ins, z_mode, sizeflag);
@@ -13209,7 +13207,7 @@ OP_3DNowSuffix (instr_info *ins, int byt
place where an 8-bit immediate would normally go. ie. the last
byte of the instruction. */
ins->obufp = ins->mnemonicendp;
- mnemonic = Suffix3DNow[*ins->codep++ & 0xff];
+ mnemonic = Suffix3DNow[*ins->codep++];
if (mnemonic)
ins->obufp = stpcpy (ins->obufp, mnemonic);
else
@@ -13274,7 +13272,7 @@ CMP_Fixup (instr_info *ins, int bytemode
if (!fetch_code (ins->info, ins->codep + 1))
return false;
- cmp_type = *ins->codep++ & 0xff;
+ cmp_type = *ins->codep++;
if (cmp_type < ARRAY_SIZE (simd_cmp_op))
{
char suffix[3];
@@ -13724,7 +13722,7 @@ OP_REG_VexI4 (instr_info *ins, int bytem
if (!fetch_code (ins->info, ins->codep + 1))
return false;
- reg = *ins->codep++ & 0xff;
+ reg = *ins->codep++;
if (bytemode != x_mode && bytemode != scalar_mode)
abort ();
@@ -13768,7 +13766,7 @@ VPCMP_Fixup (instr_info *ins, int bytemo
if (!fetch_code (ins->info, ins->codep + 1))
return false;
- cmp_type = *ins->codep++ & 0xff;
+ cmp_type = *ins->codep++;
/* There are aliases for immediates 0, 1, 2, 4, 5, 6.
If it's the case, print suffix, otherwise - print the immediate. */
if (cmp_type < ARRAY_SIZE (simd_cmp_op)
@@ -13823,7 +13821,7 @@ VPCOM_Fixup (instr_info *ins, int bytemo
if (!fetch_code (ins->info, ins->codep + 1))
return false;
- cmp_type = *ins->codep++ & 0xff;
+ cmp_type = *ins->codep++;
if (cmp_type < ARRAY_SIZE (xop_cmp_op))
{
char suffix[3];
@@ -13870,7 +13868,7 @@ PCLMUL_Fixup (instr_info *ins, int bytem
if (!fetch_code (ins->info, ins->codep + 1))
return false;
- pclmul_type = *ins->codep++ & 0xff;
+ pclmul_type = *ins->codep++;
switch (pclmul_type)
{
case 0x10:
next prev parent reply other threads:[~2023-05-19 14:06 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-05-19 14:05 [PATCH 0/3] x86: a little bit of disassembler tidying Jan Beulich
2023-05-19 14:06 ` Jan Beulich [this message]
2023-05-19 14:06 ` [PATCH 2/3] x86: disassembling over-long insns Jan Beulich
2023-05-19 14:07 ` [PATCH 3/3] x86: convert two pointers to (indexing) integers Jan Beulich
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=2274d914-c77f-37e7-5589-870a647e24a0@suse.com \
--to=jbeulich@suse.com \
--cc=binutils@sourceware.org \
--cc=hjl.tools@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).