summaryrefslogtreecommitdiff
path: root/sys/arch/powerpc/fpu
AgeCommit message (Expand)Author
2022-09-20Typo in comment. No binary changes.rin
2022-09-20stfs{,x}{,u}: Switch to conversion algorithm specified by Power ISA.rin
2022-09-15fnm{add,sub}{,s}: Do not negate NaN.rin
2022-09-15fpu_execute(): Use FR() macro. No binary changes.rin
2022-09-14Fix logic for FPSCR[OX]; set FPSCR[FI] (and therefore FPSCR[XX]).rin
2022-09-14Fix logic for FPSCR[UX]:rin
2022-09-09G/C ``notyet''; OPC_load_st_62 == std{,u} are integer insns.rin
2022-09-08fpu_compare(): Assign r by using return value from fpu_sub(),rin
2022-09-08fpu_compare(): Fix inverted logic and comment it out for now.rin
2022-09-07Fix mnemonics for fm{add,sub}. NFC except for evcnt(9) names.rin
2022-09-07fpu_explode(): Fix sign for FTYPE_INT.rin
2022-09-07Turn fpu_sub() macro into static inline function for clarity.rin
2022-09-06Style sync with other parts of this file, as well as FreeBSD:rin
2022-09-06Fix errors in calculation of intermediate mantissa bits > 95.rin
2022-09-06fpu_sub(): Do not invert sign bit of NaN.rin
2022-09-06fpu_mul(): Do not multiply sign bits for NaN.rin
2022-09-06Fix priority for NaN propagation: frA > frB > frC.rin
2022-09-06FPU generates QNaN with all payload bits being cleared.rin
2022-09-06IEEE 754 and Power ISA require sqrt(+0) = +0 and sqrt(-0) = -0, not Inf.rin
2022-09-05Clean up fpu_extern.h; only a function public to external is fpu_emulate().rin
2022-09-05static-fy fpu_execute().rin
2022-09-05static-fy fpu_[ixsd]tof(). NFC.rin
2022-09-04Unconditionally include <sys/systm.h> for KASSERT{,MSG}().rin
2022-09-04Trailing white-space/TAB.rin
2022-09-04Simplify; no need to override ``type'' here. NFC.rin
2022-09-04Simplify logic to remove needless goto. NFC.rin
2022-09-04Update FR and FI (and undefined FPRF) for ``round to integer'' insns.rin
2022-09-04Fix logic for FI, FR, and FPRF fields of FPSCR.rin
2022-09-04Fix debug printf. Unbreak DEBUG build.rin
2022-09-04Remove extra break.rin
2022-09-03Localize addr and buf. NFC.rin
2022-09-03Fix weird logic. NFC.rin
2022-09-03Do not use ``cond'' for two purposes: flag and bit field.rin
2022-09-02fpu_implode(): Clear lower word for float just for sure.rin
2022-09-02Fix fres and frsqrte.rin
2022-09-02Make fpu_explode() and fpu_implode() take uint64_t and uint64_t *,rin
2022-09-02Use hi and lo for higher and lower words, instead of i and j.rin
2022-09-02fpu_subr.c: Convert #ifdef DIAGNOSTIC panic() --> KASSERT{,MSG}().rin
2022-09-02Fix off-by-one error in exponent when round a double to therin
2022-09-02Make fpu_ftod() return uint64_t value. NFC.rin
2022-09-01Do not raise SNaN exception for QNaN.rin
2022-09-01Set FPSCR[FPRF] for calculated result, for all instructionsrin
2022-09-01static-fy fpu_fto[ixsd](). NFC.rin
2022-09-01Further fix for fcti{w,d}{,z}.rin
2022-09-01Remove stale comment.rin
2022-08-30Fix fcti{w,d}{,z}.rin
2022-08-30Move FTYPE_* definitions from instr.h to fpu_emu.h.rin
2022-08-30Do not assert FPSCR[UX] when overflow.rin
2022-08-30Fix fsel; frB is chosen when frA is negative or NaN.rin
2022-08-30Fix mcrfs; Clear copied exception bits, and do not raise any exception.rin