* make sure struct real_value's bitfields are packed
@ 2004-03-21 19:10 Alexandre Oliva
2004-03-24 10:51 ` Richard Henderson
0 siblings, 1 reply; 3+ messages in thread
From: Alexandre Oliva @ 2004-03-21 19:10 UTC (permalink / raw)
To: gcc
[-- Attachment #1: Type: text/plain, Size: 605 bytes --]
REAL_VALUE_TYPE_SIZE is computed assuming the bitfields in struct
real_value are packed into a single 32-bit word, but some ABIs specify
that bitfields with different underlying types don't share the same
underlying word.
This patch arranges for all bitfields in struct real_value to
explicitly have type unsigned int. exp was renamed to uexp and
accessor macros were introduced to turn it back into a signed value.
No performance differences are expected, since the macros effectively
do what the compiler would have done to sign-extend the values.
Bootstrapped on i686-pc-linux-gnu. Ok to install?
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: pack-real-bitfields.patch --]
[-- Type: text/x-patch, Size: 20771 bytes --]
Index: gcc/ChangeLog
from Alexandre Oliva <aoliva@redhat.com>
* real.h (struct real_value): Use the same type for all
bitfields. Rename exp to uexp.
(REAL_EXP, SET_REAL_EXP): New accessor macros for uexp.
Adjust all uses of exp...
* builtins.c: ... here, ...
* emit-rtl.c: ... here, and ...
* real.c: ... and here.
Index: gcc/builtins.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/builtins.c,v
retrieving revision 1.296
diff -u -p -r1.296 builtins.c
--- gcc/builtins.c 20 Mar 2004 04:52:49 -0000 1.296
+++ gcc/builtins.c 21 Mar 2004 05:19:54 -0000
@@ -6812,7 +6812,7 @@ fold_builtin (tree exp)
REAL_VALUE_TYPE cst;
real_convert (&cst, TYPE_MODE (type), &dconstpi);
- cst.exp -= 2;
+ SET_REAL_EXP (&cst, REAL_EXP (&cst) - 2);
return build_real (type, cst);
}
}
Index: gcc/emit-rtl.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/emit-rtl.c,v
retrieving revision 1.384
diff -u -p -r1.384 emit-rtl.c
--- gcc/emit-rtl.c 20 Mar 2004 23:08:56 -0000 1.384
+++ gcc/emit-rtl.c 21 Mar 2004 05:19:57 -0000
@@ -5325,7 +5325,7 @@ init_emit_once (int line_numbers)
REAL_VALUE_FROM_INT (dconstm2, -2, -1, double_mode);
dconsthalf = dconst1;
- dconsthalf.exp--;
+ SET_REAL_EXP (&dconsthalf, REAL_EXP (&dconsthalf) - 1);
real_arithmetic (&dconstthird, RDIV_EXPR, &dconst1, &dconst3);
Index: gcc/real.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/real.c,v
retrieving revision 1.140
diff -u -p -r1.140 real.c
--- gcc/real.c 11 Mar 2004 13:39:57 -0000 1.140
+++ gcc/real.c 21 Mar 2004 05:19:59 -0000
@@ -491,7 +491,7 @@ normalize (REAL_VALUE_TYPE *r)
if (i < 0)
{
r->class = rvc_zero;
- r->exp = 0;
+ SET_REAL_EXP (r, 0);
return;
}
@@ -503,14 +503,14 @@ normalize (REAL_VALUE_TYPE *r)
if (shift > 0)
{
- exp = r->exp - shift;
+ exp = REAL_EXP (r) - shift;
if (exp > MAX_EXP)
get_inf (r, r->sign);
else if (exp < -MAX_EXP)
get_zero (r, r->sign);
else
{
- r->exp = exp;
+ SET_REAL_EXP (r, exp);
lshift_significand (r, r, shift);
}
}
@@ -581,7 +581,7 @@ do_add (REAL_VALUE_TYPE *r, const REAL_V
}
/* Swap the arguments such that A has the larger exponent. */
- dexp = a->exp - b->exp;
+ dexp = REAL_EXP (a) - REAL_EXP (b);
if (dexp < 0)
{
const REAL_VALUE_TYPE *t;
@@ -589,7 +589,7 @@ do_add (REAL_VALUE_TYPE *r, const REAL_V
dexp = -dexp;
sign ^= subtract_p;
}
- exp = a->exp;
+ exp = REAL_EXP (a);
/* If the exponents are not identical, we need to shift the
significand of B down. */
@@ -639,7 +639,7 @@ do_add (REAL_VALUE_TYPE *r, const REAL_V
r->class = rvc_normal;
r->sign = sign;
- r->exp = exp;
+ SET_REAL_EXP (r, exp);
/* Re-normalize the result. */
normalize (r);
@@ -748,8 +748,8 @@ do_multiply (REAL_VALUE_TYPE *r, const R
for (j = 0; j < 2; ++j)
{
- int exp = (a->exp - (2*SIGSZ-1-i)*(HOST_BITS_PER_LONG/2)
- + (b->exp - (1-j)*(HOST_BITS_PER_LONG/2)));
+ int exp = (REAL_EXP (a) - (2*SIGSZ-1-i)*(HOST_BITS_PER_LONG/2)
+ + (REAL_EXP (b) - (1-j)*(HOST_BITS_PER_LONG/2)));
if (exp > MAX_EXP)
{
@@ -765,7 +765,7 @@ do_multiply (REAL_VALUE_TYPE *r, const R
memset (&u, 0, sizeof (u));
u.class = rvc_normal;
- u.exp = exp;
+ SET_REAL_EXP (&u, exp);
for (k = j; k < SIGSZ * 2; k += 2)
{
@@ -863,7 +863,7 @@ do_divide (REAL_VALUE_TYPE *r, const REA
rr->class = rvc_normal;
rr->sign = sign;
- exp = a->exp - b->exp + 1;
+ exp = REAL_EXP (a) - REAL_EXP (b) + 1;
if (exp > MAX_EXP)
{
get_inf (r, sign);
@@ -874,7 +874,7 @@ do_divide (REAL_VALUE_TYPE *r, const REA
get_zero (r, sign);
return true;
}
- rr->exp = exp;
+ SET_REAL_EXP (rr, exp);
inexact = div_significands (rr, a, b);
@@ -935,9 +935,9 @@ do_compare (const REAL_VALUE_TYPE *a, co
if (a->sign != b->sign)
return -a->sign - -b->sign;
- if (a->exp > b->exp)
+ if (REAL_EXP (a) > REAL_EXP (b))
ret = 1;
- else if (a->exp < b->exp)
+ else if (REAL_EXP (a) < REAL_EXP (b))
ret = -1;
else
ret = cmp_significands (a, b);
@@ -960,10 +960,10 @@ do_fix_trunc (REAL_VALUE_TYPE *r, const
break;
case rvc_normal:
- if (r->exp <= 0)
+ if (REAL_EXP (r) <= 0)
get_zero (r, r->sign);
- else if (r->exp < SIGNIFICAND_BITS)
- clear_significand_below (r, SIGNIFICAND_BITS - r->exp);
+ else if (REAL_EXP (r) < SIGNIFICAND_BITS)
+ clear_significand_below (r, SIGNIFICAND_BITS - REAL_EXP (r));
break;
default:
@@ -1099,7 +1099,7 @@ real_exponent (const REAL_VALUE_TYPE *r)
case rvc_nan:
return (unsigned int)-1 >> 1;
case rvc_normal:
- return r->exp;
+ return REAL_EXP (r);
default:
abort ();
}
@@ -1119,13 +1119,13 @@ real_ldexp (REAL_VALUE_TYPE *r, const RE
break;
case rvc_normal:
- exp += op0->exp;
+ exp += REAL_EXP (op0);
if (exp > MAX_EXP)
get_inf (r, r->sign);
else if (exp < -MAX_EXP)
get_zero (r, r->sign);
else
- r->exp = exp;
+ SET_REAL_EXP (r, exp);
break;
default:
@@ -1184,7 +1184,7 @@ real_identical (const REAL_VALUE_TYPE *a
return true;
case rvc_normal:
- if (a->exp != b->exp)
+ if (REAL_EXP (a) != REAL_EXP (b))
return false;
break;
@@ -1266,13 +1266,13 @@ real_to_integer (const REAL_VALUE_TYPE *
return i;
case rvc_normal:
- if (r->exp <= 0)
+ if (REAL_EXP (r) <= 0)
goto underflow;
/* Only force overflow for unsigned overflow. Signed overflow is
undefined, so it doesn't matter what we return, and some callers
expect to be able to use this routine for both signed and
unsigned conversions. */
- if (r->exp > HOST_BITS_PER_WIDE_INT)
+ if (REAL_EXP (r) > HOST_BITS_PER_WIDE_INT)
goto overflow;
if (HOST_BITS_PER_WIDE_INT == HOST_BITS_PER_LONG)
@@ -1286,7 +1286,7 @@ real_to_integer (const REAL_VALUE_TYPE *
else
abort ();
- i >>= HOST_BITS_PER_WIDE_INT - r->exp;
+ i >>= HOST_BITS_PER_WIDE_INT - REAL_EXP (r);
if (r->sign)
i = -i;
@@ -1328,7 +1328,7 @@ real_to_integer2 (HOST_WIDE_INT *plow, H
break;
case rvc_normal:
- exp = r->exp;
+ exp = REAL_EXP (r);
if (exp <= 0)
goto underflow;
/* Only force overflow for unsigned overflow. Signed overflow is
@@ -1383,7 +1383,7 @@ static unsigned long
rtd_divmod (REAL_VALUE_TYPE *num, REAL_VALUE_TYPE *den)
{
unsigned long q, msb;
- int expn = num->exp, expd = den->exp;
+ int expn = REAL_EXP (num), expd = REAL_EXP (den);
if (expn < expd)
return 0;
@@ -1404,7 +1404,7 @@ rtd_divmod (REAL_VALUE_TYPE *num, REAL_V
}
while (--expn >= expd);
- num->exp = expd;
+ SET_REAL_EXP (num, expd);
normalize (num);
return q;
@@ -1455,7 +1455,7 @@ real_to_decimal (char *str, const REAL_V
/* Estimate the decimal exponent, and compute the length of the string it
will print as. Be conservative and add one to account for possible
overflow or rounding error. */
- dec_exp = r.exp * M_LOG10_2;
+ dec_exp = REAL_EXP (&r) * M_LOG10_2;
for (max_digits = 1; dec_exp ; max_digits++)
dec_exp /= 10;
@@ -1484,7 +1484,7 @@ real_to_decimal (char *str, const REAL_V
and strip trailing decimal zeros. */
u = r;
- u.exp = SIGNIFICAND_BITS - 1;
+ SET_REAL_EXP (&u, SIGNIFICAND_BITS - 1);
/* Largest M, such that 10**2**M fits within SIGNIFICAND_BITS. */
m = floor_log2 (max_digits);
@@ -1508,15 +1508,16 @@ real_to_decimal (char *str, const REAL_V
while (--m >= 0);
/* Revert the scaling to integer that we performed earlier. */
- u.exp += r.exp - (SIGNIFICAND_BITS - 1);
+ SET_REAL_EXP (&u, REAL_EXP (&u) + REAL_EXP (&r)
+ - (SIGNIFICAND_BITS - 1));
r = u;
/* Find power of 10. Do this by dividing out 10**2**M when
this is larger than the current remainder. Fill PTEN with
the power of 10 that we compute. */
- if (r.exp > 0)
+ if (REAL_EXP (&r) > 0)
{
- m = floor_log2 ((int)(r.exp * M_LOG10_2)) + 1;
+ m = floor_log2 ((int)(REAL_EXP (&r) * M_LOG10_2)) + 1;
do
{
const REAL_VALUE_TYPE *ptentwo = ten_to_ptwo (m);
@@ -1552,7 +1553,7 @@ real_to_decimal (char *str, const REAL_V
do_multiply (&u, &v, ten);
/* Stop if we're now >= 1. */
- if (u.exp > 0)
+ if (REAL_EXP (&u) > 0)
break;
v = u;
@@ -1563,7 +1564,7 @@ real_to_decimal (char *str, const REAL_V
/* Find power of 10. Do this by multiplying in P=10**2**M when
the current remainder is smaller than 1/P. Fill PTEN with the
power of 10 that we compute. */
- m = floor_log2 ((int)(-r.exp * M_LOG10_2)) + 1;
+ m = floor_log2 ((int)(-REAL_EXP (&r) * M_LOG10_2)) + 1;
do
{
const REAL_VALUE_TYPE *ptentwo = ten_to_ptwo (m);
@@ -1690,7 +1691,7 @@ void
real_to_hexadecimal (char *str, const REAL_VALUE_TYPE *r, size_t buf_size,
size_t digits, int crop_trailing_zeros)
{
- int i, j, exp = r->exp;
+ int i, j, exp = REAL_EXP (r);
char *p, *first;
char exp_buf[16];
size_t max_digits;
@@ -1849,7 +1850,7 @@ real_from_string (REAL_VALUE_TYPE *r, co
}
r->class = rvc_normal;
- r->exp = exp;
+ SET_REAL_EXP (r, exp);
normalize (r);
}
@@ -1962,7 +1963,7 @@ real_from_integer (REAL_VALUE_TYPE *r, e
{
r->class = rvc_normal;
r->sign = high < 0 && !unsigned_p;
- r->exp = 2 * HOST_BITS_PER_WIDE_INT;
+ SET_REAL_EXP (r, 2 * HOST_BITS_PER_WIDE_INT);
if (r->sign)
{
@@ -2205,7 +2206,7 @@ real_maxval (REAL_VALUE_TYPE *r, int sig
r->sign = sign;
r->signalling = 0;
r->canonical = 0;
- r->exp = fmt->emax * fmt->log2_b;
+ SET_REAL_EXP (r, fmt->emax * fmt->log2_b);
np2 = SIGNIFICAND_BITS - fmt->p * fmt->log2_b;
memset (r->sig, -1, SIGSZ * sizeof (unsigned long));
@@ -2227,7 +2228,7 @@ real_2expN (REAL_VALUE_TYPE *r, int n)
else
{
r->class = rvc_normal;
- r->exp = n;
+ SET_REAL_EXP (r, n);
r->sig[SIGSZ-1] = SIG_MSB;
}
}
@@ -2275,38 +2276,38 @@ round_for_format (const struct real_form
the true base. */
if (fmt->log2_b != 1)
{
- int shift = r->exp & (fmt->log2_b - 1);
+ int shift = REAL_EXP (r) & (fmt->log2_b - 1);
if (shift)
{
shift = fmt->log2_b - shift;
r->sig[0] |= sticky_rshift_significand (r, r, shift);
- r->exp += shift;
+ SET_REAL_EXP (r, REAL_EXP (r) + shift);
}
}
/* Check the range of the exponent. If we're out of range,
either underflow or overflow. */
- if (r->exp > emax2)
+ if (REAL_EXP (r) > emax2)
goto overflow;
- else if (r->exp <= emin2m1)
+ else if (REAL_EXP (r) <= emin2m1)
{
int diff;
if (!fmt->has_denorm)
{
/* Don't underflow completely until we've had a chance to round. */
- if (r->exp < emin2m1)
+ if (REAL_EXP (r) < emin2m1)
goto underflow;
}
else
{
- diff = emin2m1 - r->exp + 1;
+ diff = emin2m1 - REAL_EXP (r) + 1;
if (diff > p2)
goto underflow;
/* De-normalize the significand. */
r->sig[0] |= sticky_rshift_significand (r, r, diff);
- r->exp += diff;
+ SET_REAL_EXP (r, REAL_EXP (r) + diff);
}
}
@@ -2335,19 +2336,20 @@ round_for_format (const struct real_form
/* Overflow. Means the significand had been all ones, and
is now all zeros. Need to increase the exponent, and
possibly re-normalize it. */
- if (++r->exp > emax2)
+ SET_REAL_EXP (r, REAL_EXP (r) + 1);
+ if (REAL_EXP (r) > emax2)
goto overflow;
r->sig[SIGSZ-1] = SIG_MSB;
if (fmt->log2_b != 1)
{
- int shift = r->exp & (fmt->log2_b - 1);
+ int shift = REAL_EXP (r) & (fmt->log2_b - 1);
if (shift)
{
shift = fmt->log2_b - shift;
rshift_significand (r, r, shift);
- r->exp += shift;
- if (r->exp > emax2)
+ SET_REAL_EXP (r, REAL_EXP (r) + shift);
+ if (REAL_EXP (r) > emax2)
goto overflow;
}
}
@@ -2355,7 +2357,7 @@ round_for_format (const struct real_form
}
/* Catch underflow that we deferred until after rounding. */
- if (r->exp <= emin2m1)
+ if (REAL_EXP (r) <= emin2m1)
goto underflow;
/* Clear out trailing garbage. */
@@ -2497,7 +2499,7 @@ real_hash (const REAL_VALUE_TYPE *r)
return h;
case rvc_normal:
- h |= r->exp << 3;
+ h |= REAL_EXP (r) << 3;
break;
case rvc_nan:
@@ -2586,7 +2588,7 @@ encode_ieee_single (const struct real_fo
if (denormal)
exp = 0;
else
- exp = r->exp + 127 - 1;
+ exp = REAL_EXP (r) + 127 - 1;
image |= exp << 23;
image |= sig;
break;
@@ -2616,7 +2618,7 @@ decode_ieee_single (const struct real_fo
{
r->class = rvc_normal;
r->sign = sign;
- r->exp = -126;
+ SET_REAL_EXP (r, -126);
r->sig[SIGSZ-1] = image << 1;
normalize (r);
}
@@ -2643,7 +2645,7 @@ decode_ieee_single (const struct real_fo
{
r->class = rvc_normal;
r->sign = sign;
- r->exp = exp - 127 + 1;
+ SET_REAL_EXP (r, exp - 127 + 1);
r->sig[SIGSZ-1] = image | SIG_MSB;
}
}
@@ -2770,7 +2772,7 @@ encode_ieee_double (const struct real_fo
if (denormal)
exp = 0;
else
- exp = r->exp + 1023 - 1;
+ exp = REAL_EXP (r) + 1023 - 1;
image_hi |= exp << 20;
image_hi |= sig_hi;
image_lo = sig_lo;
@@ -2817,7 +2819,7 @@ decode_ieee_double (const struct real_fo
{
r->class = rvc_normal;
r->sign = sign;
- r->exp = -1022;
+ SET_REAL_EXP (r, -1022);
if (HOST_BITS_PER_LONG == 32)
{
image_hi = (image_hi << 1) | (image_lo >> 31);
@@ -2860,7 +2862,7 @@ decode_ieee_double (const struct real_fo
{
r->class = rvc_normal;
r->sign = sign;
- r->exp = exp - 1023 + 1;
+ SET_REAL_EXP (r, exp - 1023 + 1);
if (HOST_BITS_PER_LONG == 32)
{
r->sig[SIGSZ-1] = image_hi | SIG_MSB;
@@ -2990,7 +2992,7 @@ encode_ieee_extended (const struct real_
case rvc_normal:
{
- int exp = r->exp;
+ int exp = REAL_EXP (r);
/* Recall that IEEE numbers are interpreted as 1.F x 2**exp,
whereas the intermediate representation is 0.F x 2**exp.
@@ -3076,7 +3078,7 @@ decode_ieee_extended (const struct real_
and decrease the exponent to match. In this case, Motorola
defines the explicit integer bit to be valid, so we don't
know whether the msb is set or not. */
- r->exp = fmt->emin;
+ SET_REAL_EXP (r, fmt->emin);
if (HOST_BITS_PER_LONG == 32)
{
r->sig[SIGSZ-1] = sig_hi;
@@ -3120,7 +3122,7 @@ decode_ieee_extended (const struct real_
{
r->class = rvc_normal;
r->sign = sign;
- r->exp = exp - 16383 + 1;
+ SET_REAL_EXP (r, exp - 16383 + 1);
if (HOST_BITS_PER_LONG == 32)
{
r->sig[SIGSZ-1] = sig_hi;
@@ -3414,7 +3416,7 @@ encode_ieee_quad (const struct real_form
if (denormal)
exp = 0;
else
- exp = r->exp + 16383 - 1;
+ exp = REAL_EXP (r) + 16383 - 1;
image3 |= exp << 16;
if (HOST_BITS_PER_LONG == 32)
@@ -3494,7 +3496,7 @@ decode_ieee_quad (const struct real_form
r->class = rvc_normal;
r->sign = sign;
- r->exp = -16382 + (SIGNIFICAND_BITS - 112);
+ SET_REAL_EXP (r, -16382 + (SIGNIFICAND_BITS - 112));
if (HOST_BITS_PER_LONG == 32)
{
r->sig[0] = image0;
@@ -3545,7 +3547,7 @@ decode_ieee_quad (const struct real_form
{
r->class = rvc_normal;
r->sign = sign;
- r->exp = exp - 16383 + 1;
+ SET_REAL_EXP (r, exp - 16383 + 1);
if (HOST_BITS_PER_LONG == 32)
{
@@ -3644,7 +3646,7 @@ encode_vax_f (const struct real_format *
case rvc_normal:
sig = (r->sig[SIGSZ-1] >> (HOST_BITS_PER_LONG - 24)) & 0x7fffff;
- exp = r->exp + 128;
+ exp = REAL_EXP (r) + 128;
image = (sig << 16) & 0xffff0000;
image |= sign;
@@ -3672,7 +3674,7 @@ decode_vax_f (const struct real_format *
{
r->class = rvc_normal;
r->sign = (image >> 15) & 1;
- r->exp = exp - 128;
+ SET_REAL_EXP (r, exp - 128);
image = ((image & 0x7f) << 16) | ((image >> 16) & 0xffff);
r->sig[SIGSZ-1] = (image << (HOST_BITS_PER_LONG - 24)) | SIG_MSB;
@@ -3720,7 +3722,7 @@ encode_vax_d (const struct real_format *
/* Add the sign and exponent. */
image0 |= sign;
- image0 |= (r->exp + 128) << 7;
+ image0 |= (REAL_EXP (r) + 128) << 7;
break;
default:
@@ -3755,7 +3757,7 @@ decode_vax_d (const struct real_format *
{
r->class = rvc_normal;
r->sign = (image0 >> 15) & 1;
- r->exp = exp - 128;
+ SET_REAL_EXP (r, exp - 128);
/* Rearrange the half-words of the external format into
proper ascending order. */
@@ -3820,7 +3822,7 @@ encode_vax_g (const struct real_format *
/* Add the sign and exponent. */
image0 |= sign;
- image0 |= (r->exp + 1024) << 4;
+ image0 |= (REAL_EXP (r) + 1024) << 4;
break;
default:
@@ -3855,7 +3857,7 @@ decode_vax_g (const struct real_format *
{
r->class = rvc_normal;
r->sign = (image0 >> 15) & 1;
- r->exp = exp - 1024;
+ SET_REAL_EXP (r, exp - 1024);
/* Rearrange the half-words of the external format into
proper ascending order. */
@@ -3970,7 +3972,7 @@ encode_i370_single (const struct real_fo
case rvc_normal:
sig = (r->sig[SIGSZ-1] >> (HOST_BITS_PER_LONG - 24)) & 0xffffff;
- exp = ((r->exp / 4) + 64) << 24;
+ exp = ((REAL_EXP (r) / 4) + 64) << 24;
image = sign | exp | sig;
break;
@@ -3998,7 +4000,7 @@ decode_i370_single (const struct real_fo
{
r->class = rvc_normal;
r->sign = sign;
- r->exp = (exp - 64) * 4;
+ SET_REAL_EXP (r, (exp - 64) * 4);
r->sig[SIGSZ-1] = sig << (HOST_BITS_PER_LONG - 24);
normalize (r);
}
@@ -4039,7 +4041,7 @@ encode_i370_double (const struct real_fo
image_hi >>= 8;
}
- exp = ((r->exp / 4) + 64) << 24;
+ exp = ((REAL_EXP (r) / 4) + 64) << 24;
image_hi |= sign | exp;
break;
@@ -4076,7 +4078,7 @@ decode_i370_double (const struct real_fo
{
r->class = rvc_normal;
r->sign = sign;
- r->exp = (exp - 64) * 4 + (SIGNIFICAND_BITS - 56);
+ SET_REAL_EXP (r, (exp - 64) * 4 + (SIGNIFICAND_BITS - 56));
if (HOST_BITS_PER_LONG == 32)
{
@@ -4174,7 +4176,7 @@ encode_c4x_single (const struct real_for
break;
case rvc_normal:
- exp = r->exp - 1;
+ exp = REAL_EXP (r) - 1;
sig = (r->sig[SIGSZ-1] >> (HOST_BITS_PER_LONG - 24)) & 0x7fffff;
if (r->sign)
{
@@ -4222,7 +4224,7 @@ decode_c4x_single (const struct real_for
}
sig = (sig << (HOST_BITS_PER_LONG - 24)) | SIG_MSB;
- r->exp = exp + 1;
+ SET_REAL_EXP (r, exp + 1);
r->sig[SIGSZ-1] = sig;
}
}
@@ -4247,7 +4249,7 @@ encode_c4x_extended (const struct real_f
break;
case rvc_normal:
- exp = r->exp - 1;
+ exp = REAL_EXP (r) - 1;
sig = r->sig[SIGSZ-1];
if (HOST_BITS_PER_LONG == 64)
@@ -4311,7 +4313,7 @@ decode_c4x_extended (const struct real_f
sig = sig << 1 << 31;
sig |= SIG_MSB;
- r->exp = exp + 1;
+ SET_REAL_EXP (r, exp + 1);
r->sig[SIGSZ-1] = sig;
}
}
Index: gcc/real.h
===================================================================
RCS file: /cvs/gcc/gcc/gcc/real.h,v
retrieving revision 1.76
diff -u -p -r1.76 real.h
--- gcc/real.h 23 Jan 2004 16:16:33 -0000 1.76
+++ gcc/real.h 21 Mar 2004 05:19:59 -0000
@@ -42,13 +42,22 @@ enum real_value_class {
struct real_value GTY(())
{
- ENUM_BITFIELD (real_value_class) class : 2;
+ /* Use the same underlying type for all bit-fields, so as to make
+ sure they're packed together, otherwise REAL_VALUE_TYPE_SIZE will
+ be miscomputed. */
+ unsigned int /* ENUM_BITFIELD (real_value_class) */ class : 2;
unsigned int sign : 1;
unsigned int signalling : 1;
unsigned int canonical : 1;
- signed int exp : EXP_BITS;
+ unsigned int uexp : EXP_BITS;
unsigned long sig[SIGSZ];
};
+
+#define REAL_EXP(REAL) \
+ ((int)((REAL)->uexp ^ (unsigned int)(1 << (EXP_BITS - 1))) \
+ - (1 << (EXP_BITS - 1)))
+#define SET_REAL_EXP(REAL, EXP) \
+ ((REAL)->uexp = ((unsigned int)(EXP) & (unsigned int)((1 << EXP_BITS) - 1)))
/* Various headers condition prototypes on #ifdef REAL_VALUE_TYPE, so it
needs to be a macro. We do need to continue to have a structure tag
[-- Attachment #3: Type: text/plain, Size: 188 bytes --]
--
Alexandre Oliva http://www.ic.unicamp.br/~oliva/
Red Hat Compiler Engineer aoliva@{redhat.com, gcc.gnu.org}
Free Software Evangelist oliva@{lsd.ic.unicamp.br, gnu.org}
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: make sure struct real_value's bitfields are packed
2004-03-21 19:10 make sure struct real_value's bitfields are packed Alexandre Oliva
@ 2004-03-24 10:51 ` Richard Henderson
2004-03-24 13:27 ` Alexandre Oliva
0 siblings, 1 reply; 3+ messages in thread
From: Richard Henderson @ 2004-03-24 10:51 UTC (permalink / raw)
To: Alexandre Oliva; +Cc: gcc
On Sun, Mar 21, 2004 at 02:00:43PM -0300, Alexandre Oliva wrote:
> * real.h (struct real_value): Use the same type for all
> bitfields. Rename exp to uexp.
> (REAL_EXP, SET_REAL_EXP): New accessor macros for uexp.
> Adjust all uses of exp...
> * builtins.c: ... here, ...
> * emit-rtl.c: ... here, and ...
> * real.c: ... and here.
Ug. Ok I guess. For the record, what ABI cares about this?
And does it *really* care about signed vs unsigned int? Not
just the enum thing?
r~
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: make sure struct real_value's bitfields are packed
2004-03-24 10:51 ` Richard Henderson
@ 2004-03-24 13:27 ` Alexandre Oliva
0 siblings, 0 replies; 3+ messages in thread
From: Alexandre Oliva @ 2004-03-24 13:27 UTC (permalink / raw)
To: Richard Henderson; +Cc: gcc
On Mar 23, 2004, Richard Henderson <rth@redhat.com> wrote:
> Ug. Ok I guess. For the record, what ABI cares about this?
frv-*linux*. IIRC, SH5 does as well.
> And does it *really* care about signed vs unsigned int? Not
> just the enum thing?
Hmm... I'm not sure, it's been a while since I wrote this patch
[tests] Yes, it does care about un/signed as well.
--
Alexandre Oliva http://www.ic.unicamp.br/~oliva/
Red Hat Compiler Engineer aoliva@{redhat.com, gcc.gnu.org}
Free Software Evangelist oliva@{lsd.ic.unicamp.br, gnu.org}
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2004-03-24 2:42 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2004-03-21 19:10 make sure struct real_value's bitfields are packed Alexandre Oliva
2004-03-24 10:51 ` Richard Henderson
2004-03-24 13:27 ` Alexandre Oliva
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).