mirror of
https://github.com/AuxXxilium/linux_dsm_epyc7002.git
synced 2024-12-11 21:46:46 +07:00
c896862105
This patch fixes rounding bug in emulation for double float operating on PowerPC platform. When pack double float operand, it need to truncate the tail due to the limited precision. If the truncated part is not zero, the last bit of work bit (totally 3 bits) need to '|' 1. This patch is completed in _FP_FRAC_SRS_2(X,N,sz) (arch/powerpc/math-emu/op-2.h). Originally the code leftwards rotates the operand to just keep the truncated part, then check whether it is zero. However, the number it rotates is not correct when N is not smaller than _FP_W_TYPE_SIZE, and it will cause the work bit '|' 1 in the improper case. This patch fixes this issue. Signed-off-by: Liu Yu <b13201@freescale.com> Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
435 lines
13 KiB
C
435 lines
13 KiB
C
/*
|
|
* Basic two-word fraction declaration and manipulation.
|
|
*/
|
|
|
|
#define _FP_FRAC_DECL_2(X) _FP_W_TYPE X##_f0, X##_f1
|
|
#define _FP_FRAC_COPY_2(D,S) (D##_f0 = S##_f0, D##_f1 = S##_f1)
|
|
#define _FP_FRAC_SET_2(X,I) __FP_FRAC_SET_2(X, I)
|
|
#define _FP_FRAC_HIGH_2(X) (X##_f1)
|
|
#define _FP_FRAC_LOW_2(X) (X##_f0)
|
|
#define _FP_FRAC_WORD_2(X,w) (X##_f##w)
|
|
|
|
#define _FP_FRAC_SLL_2(X,N) \
|
|
do { \
|
|
if ((N) < _FP_W_TYPE_SIZE) \
|
|
{ \
|
|
if (__builtin_constant_p(N) && (N) == 1) \
|
|
{ \
|
|
X##_f1 = X##_f1 + X##_f1 + (((_FP_WS_TYPE)(X##_f0)) < 0); \
|
|
X##_f0 += X##_f0; \
|
|
} \
|
|
else \
|
|
{ \
|
|
X##_f1 = X##_f1 << (N) | X##_f0 >> (_FP_W_TYPE_SIZE - (N)); \
|
|
X##_f0 <<= (N); \
|
|
} \
|
|
} \
|
|
else \
|
|
{ \
|
|
X##_f1 = X##_f0 << ((N) - _FP_W_TYPE_SIZE); \
|
|
X##_f0 = 0; \
|
|
} \
|
|
} while (0)
|
|
|
|
#define _FP_FRAC_SRL_2(X,N) \
|
|
do { \
|
|
if ((N) < _FP_W_TYPE_SIZE) \
|
|
{ \
|
|
X##_f0 = X##_f0 >> (N) | X##_f1 << (_FP_W_TYPE_SIZE - (N)); \
|
|
X##_f1 >>= (N); \
|
|
} \
|
|
else \
|
|
{ \
|
|
X##_f0 = X##_f1 >> ((N) - _FP_W_TYPE_SIZE); \
|
|
X##_f1 = 0; \
|
|
} \
|
|
} while (0)
|
|
|
|
/* Right shift with sticky-lsb. */
|
|
#define _FP_FRAC_SRS_2(X,N,sz) \
|
|
do { \
|
|
if ((N) < _FP_W_TYPE_SIZE) \
|
|
{ \
|
|
X##_f0 = (X##_f1 << (_FP_W_TYPE_SIZE - (N)) | X##_f0 >> (N) | \
|
|
(__builtin_constant_p(N) && (N) == 1 \
|
|
? X##_f0 & 1 \
|
|
: (X##_f0 << (_FP_W_TYPE_SIZE - (N))) != 0)); \
|
|
X##_f1 >>= (N); \
|
|
} \
|
|
else \
|
|
{ \
|
|
X##_f0 = (X##_f1 >> ((N) - _FP_W_TYPE_SIZE) | \
|
|
(((X##_f1 << (2 * _FP_W_TYPE_SIZE - (N))) | \
|
|
X##_f0) != 0)); \
|
|
X##_f1 = 0; \
|
|
} \
|
|
} while (0)
|
|
|
|
#define _FP_FRAC_ADDI_2(X,I) \
|
|
__FP_FRAC_ADDI_2(X##_f1, X##_f0, I)
|
|
|
|
#define _FP_FRAC_ADD_2(R,X,Y) \
|
|
__FP_FRAC_ADD_2(R##_f1, R##_f0, X##_f1, X##_f0, Y##_f1, Y##_f0)
|
|
|
|
#define _FP_FRAC_SUB_2(R,X,Y) \
|
|
__FP_FRAC_SUB_2(R##_f1, R##_f0, X##_f1, X##_f0, Y##_f1, Y##_f0)
|
|
|
|
#define _FP_FRAC_CLZ_2(R,X) \
|
|
do { \
|
|
if (X##_f1) \
|
|
__FP_CLZ(R,X##_f1); \
|
|
else \
|
|
{ \
|
|
__FP_CLZ(R,X##_f0); \
|
|
R += _FP_W_TYPE_SIZE; \
|
|
} \
|
|
} while(0)
|
|
|
|
/* Predicates */
|
|
#define _FP_FRAC_NEGP_2(X) ((_FP_WS_TYPE)X##_f1 < 0)
|
|
#define _FP_FRAC_ZEROP_2(X) ((X##_f1 | X##_f0) == 0)
|
|
#define _FP_FRAC_OVERP_2(fs,X) (X##_f1 & _FP_OVERFLOW_##fs)
|
|
#define _FP_FRAC_EQ_2(X, Y) (X##_f1 == Y##_f1 && X##_f0 == Y##_f0)
|
|
#define _FP_FRAC_GT_2(X, Y) \
|
|
((X##_f1 > Y##_f1) || (X##_f1 == Y##_f1 && X##_f0 > Y##_f0))
|
|
#define _FP_FRAC_GE_2(X, Y) \
|
|
((X##_f1 > Y##_f1) || (X##_f1 == Y##_f1 && X##_f0 >= Y##_f0))
|
|
|
|
#define _FP_ZEROFRAC_2 0, 0
|
|
#define _FP_MINFRAC_2 0, 1
|
|
|
|
/*
|
|
* Internals
|
|
*/
|
|
|
|
#define __FP_FRAC_SET_2(X,I1,I0) (X##_f0 = I0, X##_f1 = I1)
|
|
|
|
#define __FP_CLZ_2(R, xh, xl) \
|
|
do { \
|
|
if (xh) \
|
|
__FP_CLZ(R,xl); \
|
|
else \
|
|
{ \
|
|
__FP_CLZ(R,xl); \
|
|
R += _FP_W_TYPE_SIZE; \
|
|
} \
|
|
} while(0)
|
|
|
|
#if 0
|
|
|
|
#ifndef __FP_FRAC_ADDI_2
|
|
#define __FP_FRAC_ADDI_2(xh, xl, i) \
|
|
(xh += ((xl += i) < i))
|
|
#endif
|
|
#ifndef __FP_FRAC_ADD_2
|
|
#define __FP_FRAC_ADD_2(rh, rl, xh, xl, yh, yl) \
|
|
(rh = xh + yh + ((rl = xl + yl) < xl))
|
|
#endif
|
|
#ifndef __FP_FRAC_SUB_2
|
|
#define __FP_FRAC_SUB_2(rh, rl, xh, xl, yh, yl) \
|
|
(rh = xh - yh - ((rl = xl - yl) > xl))
|
|
#endif
|
|
|
|
#else
|
|
|
|
#undef __FP_FRAC_ADDI_2
|
|
#define __FP_FRAC_ADDI_2(xh, xl, i) add_ssaaaa(xh, xl, xh, xl, 0, i)
|
|
#undef __FP_FRAC_ADD_2
|
|
#define __FP_FRAC_ADD_2 add_ssaaaa
|
|
#undef __FP_FRAC_SUB_2
|
|
#define __FP_FRAC_SUB_2 sub_ddmmss
|
|
|
|
#endif
|
|
|
|
/*
|
|
* Unpack the raw bits of a native fp value. Do not classify or
|
|
* normalize the data.
|
|
*/
|
|
|
|
#define _FP_UNPACK_RAW_2(fs, X, val) \
|
|
do { \
|
|
union _FP_UNION_##fs _flo; _flo.flt = (val); \
|
|
\
|
|
X##_f0 = _flo.bits.frac0; \
|
|
X##_f1 = _flo.bits.frac1; \
|
|
X##_e = _flo.bits.exp; \
|
|
X##_s = _flo.bits.sign; \
|
|
} while (0)
|
|
|
|
|
|
/*
|
|
* Repack the raw bits of a native fp value.
|
|
*/
|
|
|
|
#define _FP_PACK_RAW_2(fs, val, X) \
|
|
do { \
|
|
union _FP_UNION_##fs _flo; \
|
|
\
|
|
_flo.bits.frac0 = X##_f0; \
|
|
_flo.bits.frac1 = X##_f1; \
|
|
_flo.bits.exp = X##_e; \
|
|
_flo.bits.sign = X##_s; \
|
|
\
|
|
(val) = _flo.flt; \
|
|
} while (0)
|
|
|
|
|
|
/*
|
|
* Multiplication algorithms:
|
|
*/
|
|
|
|
/* Given a 1W * 1W => 2W primitive, do the extended multiplication. */
|
|
|
|
#define _FP_MUL_MEAT_2_wide(fs, R, X, Y, doit) \
|
|
do { \
|
|
_FP_FRAC_DECL_4(_z); _FP_FRAC_DECL_2(_b); _FP_FRAC_DECL_2(_c); \
|
|
\
|
|
doit(_FP_FRAC_WORD_4(_z,1), _FP_FRAC_WORD_4(_z,0), X##_f0, Y##_f0); \
|
|
doit(_b_f1, _b_f0, X##_f0, Y##_f1); \
|
|
doit(_c_f1, _c_f0, X##_f1, Y##_f0); \
|
|
doit(_FP_FRAC_WORD_4(_z,3), _FP_FRAC_WORD_4(_z,2), X##_f1, Y##_f1); \
|
|
\
|
|
__FP_FRAC_ADD_4(_FP_FRAC_WORD_4(_z,3),_FP_FRAC_WORD_4(_z,2), \
|
|
_FP_FRAC_WORD_4(_z,1),_FP_FRAC_WORD_4(_z,0), \
|
|
0, _b_f1, _b_f0, 0, \
|
|
_FP_FRAC_WORD_4(_z,3),_FP_FRAC_WORD_4(_z,2), \
|
|
_FP_FRAC_WORD_4(_z,1),_FP_FRAC_WORD_4(_z,0)); \
|
|
__FP_FRAC_ADD_4(_FP_FRAC_WORD_4(_z,3),_FP_FRAC_WORD_4(_z,2), \
|
|
_FP_FRAC_WORD_4(_z,1),_FP_FRAC_WORD_4(_z,0), \
|
|
0, _c_f1, _c_f0, 0, \
|
|
_FP_FRAC_WORD_4(_z,3),_FP_FRAC_WORD_4(_z,2), \
|
|
_FP_FRAC_WORD_4(_z,1),_FP_FRAC_WORD_4(_z,0)); \
|
|
\
|
|
/* Normalize since we know where the msb of the multiplicands \
|
|
were (bit B), we know that the msb of the of the product is \
|
|
at either 2B or 2B-1. */ \
|
|
_FP_FRAC_SRS_4(_z, _FP_WFRACBITS_##fs-1, 2*_FP_WFRACBITS_##fs); \
|
|
R##_f0 = _FP_FRAC_WORD_4(_z,0); \
|
|
R##_f1 = _FP_FRAC_WORD_4(_z,1); \
|
|
} while (0)
|
|
|
|
/* This next macro appears to be totally broken. Fortunately nowhere
|
|
* seems to use it :-> The problem is that we define _z[4] but
|
|
* then use it in _FP_FRAC_SRS_4, which will attempt to access
|
|
* _z_f[n] which will cause an error. The fix probably involves
|
|
* declaring it with _FP_FRAC_DECL_4, see previous macro. -- PMM 02/1998
|
|
*/
|
|
#define _FP_MUL_MEAT_2_gmp(fs, R, X, Y) \
|
|
do { \
|
|
_FP_W_TYPE _x[2], _y[2], _z[4]; \
|
|
_x[0] = X##_f0; _x[1] = X##_f1; \
|
|
_y[0] = Y##_f0; _y[1] = Y##_f1; \
|
|
\
|
|
mpn_mul_n(_z, _x, _y, 2); \
|
|
\
|
|
/* Normalize since we know where the msb of the multiplicands \
|
|
were (bit B), we know that the msb of the of the product is \
|
|
at either 2B or 2B-1. */ \
|
|
_FP_FRAC_SRS_4(_z, _FP_WFRACBITS##_fs-1, 2*_FP_WFRACBITS_##fs); \
|
|
R##_f0 = _z[0]; \
|
|
R##_f1 = _z[1]; \
|
|
} while (0)
|
|
|
|
|
|
/*
|
|
* Division algorithms:
|
|
* This seems to be giving me difficulties -- PMM
|
|
* Look, NetBSD seems to be able to comment algorithms. Can't you?
|
|
* I've thrown printks at the problem.
|
|
* This now appears to work, but I still don't really know why.
|
|
* Also, I don't think the result is properly normalised...
|
|
*/
|
|
|
|
#define _FP_DIV_MEAT_2_udiv_64(fs, R, X, Y) \
|
|
do { \
|
|
extern void _fp_udivmodti4(_FP_W_TYPE q[2], _FP_W_TYPE r[2], \
|
|
_FP_W_TYPE n1, _FP_W_TYPE n0, \
|
|
_FP_W_TYPE d1, _FP_W_TYPE d0); \
|
|
_FP_W_TYPE _n_f3, _n_f2, _n_f1, _n_f0, _r_f1, _r_f0; \
|
|
_FP_W_TYPE _q_f1, _q_f0, _m_f1, _m_f0; \
|
|
_FP_W_TYPE _rmem[2], _qmem[2]; \
|
|
/* I think this check is to ensure that the result is normalised. \
|
|
* Assuming X,Y normalised (ie in [1.0,2.0)) X/Y will be in \
|
|
* [0.5,2.0). Furthermore, it will be less than 1.0 iff X < Y. \
|
|
* In this case we tweak things. (this is based on comments in \
|
|
* the NetBSD FPU emulation code. ) \
|
|
* We know X,Y are normalised because we ensure this as part of \
|
|
* the unpacking process. -- PMM \
|
|
*/ \
|
|
if (_FP_FRAC_GT_2(X, Y)) \
|
|
{ \
|
|
/* R##_e++; */ \
|
|
_n_f3 = X##_f1 >> 1; \
|
|
_n_f2 = X##_f1 << (_FP_W_TYPE_SIZE - 1) | X##_f0 >> 1; \
|
|
_n_f1 = X##_f0 << (_FP_W_TYPE_SIZE - 1); \
|
|
_n_f0 = 0; \
|
|
} \
|
|
else \
|
|
{ \
|
|
R##_e--; \
|
|
_n_f3 = X##_f1; \
|
|
_n_f2 = X##_f0; \
|
|
_n_f1 = _n_f0 = 0; \
|
|
} \
|
|
\
|
|
/* Normalize, i.e. make the most significant bit of the \
|
|
denominator set. CHANGED: - 1 to nothing -- PMM */ \
|
|
_FP_FRAC_SLL_2(Y, _FP_WFRACXBITS_##fs /* -1 */); \
|
|
\
|
|
/* Do the 256/128 bit division given the 128-bit _fp_udivmodtf4 \
|
|
primitive snagged from libgcc2.c. */ \
|
|
\
|
|
_fp_udivmodti4(_qmem, _rmem, _n_f3, _n_f2, 0, Y##_f1); \
|
|
_q_f1 = _qmem[0]; \
|
|
umul_ppmm(_m_f1, _m_f0, _q_f1, Y##_f0); \
|
|
_r_f1 = _rmem[0]; \
|
|
_r_f0 = _n_f1; \
|
|
if (_FP_FRAC_GT_2(_m, _r)) \
|
|
{ \
|
|
_q_f1--; \
|
|
_FP_FRAC_ADD_2(_r, _r, Y); \
|
|
if (_FP_FRAC_GE_2(_r, Y) && _FP_FRAC_GT_2(_m, _r)) \
|
|
{ \
|
|
_q_f1--; \
|
|
_FP_FRAC_ADD_2(_r, _r, Y); \
|
|
} \
|
|
} \
|
|
_FP_FRAC_SUB_2(_r, _r, _m); \
|
|
\
|
|
_fp_udivmodti4(_qmem, _rmem, _r_f1, _r_f0, 0, Y##_f1); \
|
|
_q_f0 = _qmem[0]; \
|
|
umul_ppmm(_m_f1, _m_f0, _q_f0, Y##_f0); \
|
|
_r_f1 = _rmem[0]; \
|
|
_r_f0 = _n_f0; \
|
|
if (_FP_FRAC_GT_2(_m, _r)) \
|
|
{ \
|
|
_q_f0--; \
|
|
_FP_FRAC_ADD_2(_r, _r, Y); \
|
|
if (_FP_FRAC_GE_2(_r, Y) && _FP_FRAC_GT_2(_m, _r)) \
|
|
{ \
|
|
_q_f0--; \
|
|
_FP_FRAC_ADD_2(_r, _r, Y); \
|
|
} \
|
|
} \
|
|
_FP_FRAC_SUB_2(_r, _r, _m); \
|
|
\
|
|
R##_f1 = _q_f1; \
|
|
R##_f0 = _q_f0 | ((_r_f1 | _r_f0) != 0); \
|
|
/* adjust so answer is normalized again. I'm not sure what the \
|
|
* final sz param should be. In practice it's never used since \
|
|
* N is 1 which is always going to be < _FP_W_TYPE_SIZE... \
|
|
*/ \
|
|
/* _FP_FRAC_SRS_2(R,1,_FP_WFRACBITS_##fs); */ \
|
|
} while (0)
|
|
|
|
|
|
#define _FP_DIV_MEAT_2_gmp(fs, R, X, Y) \
|
|
do { \
|
|
_FP_W_TYPE _x[4], _y[2], _z[4]; \
|
|
_y[0] = Y##_f0; _y[1] = Y##_f1; \
|
|
_x[0] = _x[3] = 0; \
|
|
if (_FP_FRAC_GT_2(X, Y)) \
|
|
{ \
|
|
R##_e++; \
|
|
_x[1] = (X##_f0 << (_FP_WFRACBITS-1 - _FP_W_TYPE_SIZE) | \
|
|
X##_f1 >> (_FP_W_TYPE_SIZE - \
|
|
(_FP_WFRACBITS-1 - _FP_W_TYPE_SIZE))); \
|
|
_x[2] = X##_f1 << (_FP_WFRACBITS-1 - _FP_W_TYPE_SIZE); \
|
|
} \
|
|
else \
|
|
{ \
|
|
_x[1] = (X##_f0 << (_FP_WFRACBITS - _FP_W_TYPE_SIZE) | \
|
|
X##_f1 >> (_FP_W_TYPE_SIZE - \
|
|
(_FP_WFRACBITS - _FP_W_TYPE_SIZE))); \
|
|
_x[2] = X##_f1 << (_FP_WFRACBITS - _FP_W_TYPE_SIZE); \
|
|
} \
|
|
\
|
|
(void) mpn_divrem (_z, 0, _x, 4, _y, 2); \
|
|
R##_f1 = _z[1]; \
|
|
R##_f0 = _z[0] | ((_x[0] | _x[1]) != 0); \
|
|
} while (0)
|
|
|
|
|
|
/*
|
|
* Square root algorithms:
|
|
* We have just one right now, maybe Newton approximation
|
|
* should be added for those machines where division is fast.
|
|
*/
|
|
|
|
#define _FP_SQRT_MEAT_2(R, S, T, X, q) \
|
|
do { \
|
|
while (q) \
|
|
{ \
|
|
T##_f1 = S##_f1 + q; \
|
|
if (T##_f1 <= X##_f1) \
|
|
{ \
|
|
S##_f1 = T##_f1 + q; \
|
|
X##_f1 -= T##_f1; \
|
|
R##_f1 += q; \
|
|
} \
|
|
_FP_FRAC_SLL_2(X, 1); \
|
|
q >>= 1; \
|
|
} \
|
|
q = (_FP_W_TYPE)1 << (_FP_W_TYPE_SIZE - 1); \
|
|
while (q) \
|
|
{ \
|
|
T##_f0 = S##_f0 + q; \
|
|
T##_f1 = S##_f1; \
|
|
if (T##_f1 < X##_f1 || \
|
|
(T##_f1 == X##_f1 && T##_f0 < X##_f0)) \
|
|
{ \
|
|
S##_f0 = T##_f0 + q; \
|
|
if (((_FP_WS_TYPE)T##_f0) < 0 && \
|
|
((_FP_WS_TYPE)S##_f0) >= 0) \
|
|
S##_f1++; \
|
|
_FP_FRAC_SUB_2(X, X, T); \
|
|
R##_f0 += q; \
|
|
} \
|
|
_FP_FRAC_SLL_2(X, 1); \
|
|
q >>= 1; \
|
|
} \
|
|
} while (0)
|
|
|
|
|
|
/*
|
|
* Assembly/disassembly for converting to/from integral types.
|
|
* No shifting or overflow handled here.
|
|
*/
|
|
|
|
#define _FP_FRAC_ASSEMBLE_2(r, X, rsize) \
|
|
do { \
|
|
if (rsize <= _FP_W_TYPE_SIZE) \
|
|
r = X##_f0; \
|
|
else \
|
|
{ \
|
|
r = X##_f1; \
|
|
r <<= _FP_W_TYPE_SIZE; \
|
|
r += X##_f0; \
|
|
} \
|
|
} while (0)
|
|
|
|
#define _FP_FRAC_DISASSEMBLE_2(X, r, rsize) \
|
|
do { \
|
|
X##_f0 = r; \
|
|
X##_f1 = (rsize <= _FP_W_TYPE_SIZE ? 0 : r >> _FP_W_TYPE_SIZE); \
|
|
} while (0)
|
|
|
|
/*
|
|
* Convert FP values between word sizes
|
|
*/
|
|
|
|
#define _FP_FRAC_CONV_1_2(dfs, sfs, D, S) \
|
|
do { \
|
|
_FP_FRAC_SRS_2(S, (_FP_WFRACBITS_##sfs - _FP_WFRACBITS_##dfs), \
|
|
_FP_WFRACBITS_##sfs); \
|
|
D##_f = S##_f0; \
|
|
} while (0)
|
|
|
|
#define _FP_FRAC_CONV_2_1(dfs, sfs, D, S) \
|
|
do { \
|
|
D##_f0 = S##_f; \
|
|
D##_f1 = 0; \
|
|
_FP_FRAC_SLL_2(D, (_FP_WFRACBITS_##dfs - _FP_WFRACBITS_##sfs)); \
|
|
} while (0)
|
|
|