PKE tests run identically on SPARC/Solaris and x86/Linux.
* sky-pke.c (pke_io_{read,write}_buffer): Endianness fixes aka
"E-fixes" in register and FIFO read/writes.
(pke_code_{pkemscalf,pkemscal}): E-fixes in VU CIA setting.
(pke_code_{mpg,unpack}): E-fixes in VU memory & tracking updates.
(pke_code_direct): E-fixes in GPUIF FIFO stuffing.
* sky-pke.h (PKE_MEM_WRITE): E-fixes in trace file writing.
* sky-vu0.c (vu0_attach): Allocate micro/data memory with zalloc
to guarantee sufficient (16-byte) alignment.
* sky-vu1.c (vu1_attach): Ditto.
(vu1_io_read_register_window): *PARTIAL* E-fixes in register accesses.
* sky-libvpe.c (gif_write): E-fixes in GPUIF FIFO stuffing.
* sky-gpuif.c (gif_io_{read,write}_buffer): E-fixes in
register and FIFO read/writes.
* sky-dma.c (do_dma_transfer_tag): E-fixes in tag reading.
mechanism is starting to subside.
* sky-pke.h (PKE_FLAG_INT_NOLOOP): Added device flag to indicate
presence of stalled & interrupted PKEcode.
* sky-pke.c (pke_issue): Added PKEcode interrupt bit handling.
(pke_flip_dbf): Changed double-buffering logic to match SCEI
clarification.
(pke_code_*): Added interrupt bit stalling clause.
(pke_code_pkems*): Added ITOP/ITOPS transmission code.
(pke_code_unpack): Added more careful logic for processing
overflows of VU data memory addresses.
* sky-pke.h (PKE_MEM_READ): Removed "read" entry from FIFO trace.
* sky-pke.c (pke_attach): Set trace file to line buffering iff
open.
(pke_io_read_buffer, pke_io_write_buffer): Handle erroneous
reads/writes by zero-padding.
(pke_io_write_buffer): Switch to more bit-field definition macros.
(pke_issue): Remove "stalled" entry from FIFO trace.
(pke_pc_advance): Correct logic for DMA-tag-skipping, PKEcode
classification.
(pke_code_mskpath3): Sketch of possible PATH3 masking method.
(pke_code_mpg): Keep order of lower/upper VU words as supplied.
(pke_code_unpack): Logic change for wl/cl/num unpacking. Weird.
little struct.
interp.c: Update. Also add floating point Max/Min functions.
mips.igen: Remove r5900 tag from any floating point instructions.
r5900.igen: Rewrite. Implement *all* floating point insns (except ld/st).
r5400.igen: Tag mdmx functions as being mdmx specific.
were hammered in today's runs. Work is beginning in endian-proofing
the code.
* sky-pke.c (pke1_issue): Issue on correct PKE device.
(pke_io_write_buffer, pke_code_mpg, pke_code_unpack): Perform more
endian conversions.
(pke_code_mpg, pke_code_direct): Add operand alignment assertions.
(pke_code_mpg): Correct VU stall checks. Correct VU opcode
transfer ordering.
(pke_code_direct): Correct typos in DIRECT operand accessing.
(pke_code_unpack): Correct conditional sign-extension handling.
* sky-gpuif.c (gif_io_read_buffer, gif_io_write_buffer): Correct
assertion polarity.
(gif_read_tag): Disable faulty DMA-tag testing code.
the SCEI PKE simulator's output on its own test sample (tsv432.in).
* sky-pke.h (PKE_MEM_READ, PKE_MEM_WRITE, PKE_REG_MASK_SET): Add
trace file records.
* sky-pke.c: (pke_track_write): Removed function. Replaced with
in-line modifications to VU tracking tables.
(pke_attach): Attach VU tracking tables. Use line buffering on
trace files.
(pke_issue): Spit out additional trace records.
(pke_pc_operand_bits): Correct bitfield masking error.
(*): Replace sim_read/write with kludge PKE_MEM_READ/WRITE
throughout.
(pke_code_unpack): Correct numerous small bugs in operand decoding
etc.
A few PKE instructions even run correctly! Next missing function of
interest: FIFO pruning.
* sky-pke.c (pke_issue): Take extra SIM_DESC argument.
(pke_attach): Attach correct PKE0/PKE1 device. Open trace file if
VIF{0,1}_TRACE_FILE env. var. is defined.
(pke_io_write_buffer): Classify words in FIFO quadword. Use
kludgey sim_core routines to access DMA registers.
(pke_pc_advance): Add PKEcode classification. Correct DMA tag
skipping. Emit trace records.
(pke_pc_fifo): Add PKEcode operand classification.
(pke_check_stall): Perform stall checks against updated register
scheme.
(pke_code_unpack): Correct operand-count calculation.
(pke_code_stmask): Correct instruction skipping.
* sky-pke.h (PKE_MEM_WRITE, PKE_MEM_READ): New kludge macros.
(BIT_MASK_BTW): Corrected off-by-one error.
(enum wordclass): Classify words in a FIFO quadword.
* sky-dma.c (dma_io_read_buffer): Correct address checking assertions.
* sky-engine.c (engine_run): Pass along SIM_DESC to PKE
instruction issue code.
* handling of super duper packed UNPACK arguments
* skipping of in-progress instruction on break/stop
* interrupt generation to 5900
* PATH2/PATH3 status checking & masking
* ability to write to FIFO one word (instead of quadword) at a time
update v850, tic80 and mips simulators.
IGEN - Prepend prefix to more generated symbols and macros
(idecode_issue, instruction_word).
IGEN - Add -Wnowith option to supress warnings about word size
inflicts in input files.
MIPS - Clean up Makefile.in, m16.igen, m16.dc (new), m16run.c (new) so
that a mips16 simulator built using IGEN can be compiled.
Use the bfd-processor name in the sim-engine switch.
Add nr_cpus argument to sim_engine_run.
tic80, v850, d30v, mips, common:
Update
mips: Fill in bfd-processor field of model records so that
they match ../bfd/archures.
For sim/mips, enable multi-sim support when mips64vr5400-elf is target.
For sim/igen, allow specification of a default machine (will need
more work later).
(ANDI): Was missing mipsIV model, fix assembler syntax.
(do_c_cond_fmt): New function.
(C.cond.fmt): Handle mips I-III which do not support CC field
separatly.
(bc1): Handle mips IV which do not have a delaed FCC separatly.
(SDR): Mask paddr when BigEndianMem, not the converse as specified
in IV3.2 spec.
(DMULT, DMULTU): Force use of hosts 64bit multiplication. Handle
vr5000 which saves LO in a GPR separatly.
* configure.in (enable-sim-igen): For vr5000, select vr5000
specific instructions.
* configure: Re-generate.
MIPS simulator so that correctly writes the value of CIA back int PC
(the global previously used) when the simulation halts.
Fix implementation of DELAY_SLOT and NULLIFY_NEXT_INSTRUCTION macros.
calls) and sim_open so that they uses the virtual memory data transfer
functions sim_read & sim_write. This eliminates all code (other than
in load_memory & store_memory) that makes assumptions about the
implementation of the underlying memory model.
When address translation of insn fetch fails raise exception immediatly.
Use address_word as type of all address variables (instead of unsigned64),
the former is configured as either 32 or 64 bit type.
Always compile fpu code (no #if has fpu)
* gencode.c (SDBBP,DERET): Added (3900) insns.
(RFE): Turn on for 3900.
* interp.c (DebugBreakPoint,DEPC,Debug,Debug_*): Added.
(dsstate): Made global.
(SUBTARGET_R3900): Added.
(CANCELDELAYSLOT): New.
(SignalException): Ignore SystemCall rather than ignore and
terminate. Add DebugBreakPoint handling.
(decode_coproc): New insns RFE, DERET; and new registers Debug
and DEPC protected by SUBTARGET_R3900.
(sim_engine_run): Use CANCELDELAYSLOT rather than clearing
bits explicitly.
* Makefile.in,configure.in: Add mips subtarget option.
* configure: Update.
Add file sim-hload.c - generic load for hardware only simulators.
Review each simulators sim_open, sim_load, sim_create_inferior so that
they more closely match required behavour.
that image properties such as endianness can be checked.
More strongly document the expected behavour of each of the sim_*
interfaces.
Add default endian argument to simulator config macro
SIM_AC_OPTION_ENDIAN. Use in sim_config.
* interp.c: #include bfd.h.
(target_byte_order): Delete.
(sim_kind, myname, big_endian_p): New static locals.
(sim_open): Set sim_kind, myname. Move call to set_endianness to
after argument parsing. Recognize -E arg, set endianness accordingly.
(sim_load): Return SIM_RC. New arg abfd. Call sim_load_file to
load file into simulator. Set PC from bfd.
(sim_create_inferior): Return SIM_RC. Delete arg start_address.
(set_endianness): Use big_endian_p instead of target_byte_order.
* Make-common.in (CSEARCH): Do not include the gdb directory in
the search path.
* Make-common.in (SIM_ENDIAN, SIM_HOSTENDIAN, SIM_INLINE,
SIM_WARNING): Drop, requiring the simulator specific Makefile.in
to explicitly incorporate these.
* aclocal.m4 (--enable-sim-alignment); New option. Strongly
specify the alignment restrictions of the target architecture -
without this option all alignment restrictions are accomodated.
(--enable-sim-assert): New option. Conditionally compile in
assertion statements.
(--enable-sim-float): New option. Strongly specify the target's
floating point support.
(--enable-sim-hardware): New option. Specify the hardware devices
included in the simulation.
(--enable-sim-packages): New option. Specify the hardware
packages included in the simulation.
(--enable-sim-regparm): New option. Specify that parameters be
passed in registers instead of on the stack.
(--enable-sim-reserved-bits): New option. Specify that reserved
bits within an instruction are are correctly set.
(--enable-sim-smp): New option. Specify the level of SMP support
to be included in the simulator.
(--enable-sim-stdcall): New option. Specify an alternative
function call convention.
(--enable-sim-xor-endian): New option. Configure xor-endian
support used by some targets to implement bi-endian support.
(SignalException): Pass floating point cases to mips16_entry.
(ValueFPR): Don't restrict fmt_single and fmt_word to even
registers.
(StoreFPR): Likewise. Also, don't clobber fpr + 1 for fmt_single
or fmt_word.
(COP_LW): Pass fmt_word rather than fmt_uninterpreted to StoreFPR,
and then set the state to fmt_uninterpreted.
(COP_SW): Temporarily set the state to fmt_word while calling
ValueFPR.
scheme which is more compatible with WinGDB builds.
* configure.in: Improve comment on how to run autoconf.
* configure: Re-run autoconf to get new ../common/aclocal.m4.
* Makefile.in: Use autoconf substitution to install common
makefile fragment.
(mips16_entry): New static function.
(SignalException): Look for mips16 entry and exit instructions.
(simulate): Use the correct index when setting fpr_state after
doing a pending move.
force a 64 bit multiplication.
(build_instruction) [OR]: In mips16 mode, don't do anything if the
destination register is 0, since that is the default mips16 nop
instruction.
(build_endian_shift): Don't check proc64.
(build_instruction): Always set memval to uword64. Cast op2 to
uword64 when shifting it left in memory instructions. Always use
the same code for stores--don't special case proc64.
relative operands.
(build_instruction): Call JALDELAYSLOT rather than DELAYSLOT for a
jal instruction.
* interp.c (simJALDELAYSLOT): Define.
(JALDELAYSLOT): Define.
(INDELAYSLOT, INJALDELAYSLOT): Define.
(simulate): Clear simJALDELAYSLOT when simDELAYSLOT is cleared.
* gencode.c (build_instruction, case JUMP): Truncate PC to 32 bits.
* interp.c (CHECKHILO): Define away.
(simSIGINT): New macro.
(membank_size): Increase from 1MB to 2MB.
(control_c): New function.
(sim_resume): Rename parameter signal to signal_number. Add local
variable prev. Call signal before and after simulate.
(sim_stop_reason): Add simSIGINT support.
(sim_warning, sim_error, dotrace, SignalException): Define as stdarg
functions always.
(sim_warning): Delete call to SignalException. Do call printf_filtered
if logfh is NULL.
(AddressTranslation): Add #ifdef DEBUG around debugging message and
a call to sim_warning.
* gencode.c (inst_type): Add mips16 instruction encoding types.
(GETDATASIZEINSN): Define.
(MIPS_DECODE): Add REG flag to dsllv, dsrav, and dsrlv. Add
jalx. Add LEFT flag to mfhi and mflo. Add RIGHT flag to mthi and
mtlo.
(MIPS16_DECODE): New table, for mips16 instructions.
(bitmap_val): New static function.
(struct mips16_op): Define.
(mips16_op_table): New table, for mips16 operands.
(build_mips16_operands): New static function.
(process_instructions): If PC is odd, decode a mips16
instruction. Break out instruction handling into new
build_instruction function.
(build_instruction): New static function, broken out of
process_instructions. Check modifiers rather than flags for SHIFT
bit count and m[ft]{hi,lo} direction.
(usage): Pass program name to fprintf.
(main): Remove unused variable this_option_optind. Change
``*loptarg++'' to ``loptarg++''.
(my_strtoul): Parenthesize && within ||.
* interp.c (sim_trace): If tracefh is NULL, set it to stderr.
(LoadMemory): Accept a halfword pAddr if vAddr is odd.
(simulate): If PC is odd, fetch a 16 bit instruction, and
increment PC by 2 rather than 4.
* configure.in: Add case for mips16*-*-*.
* configure: Rebuild.
* interp.c (SignalException): Check for explicit terminating
breakpoint value.
* gencode.c: Pass instruction value through SignalException()
calls for Trap, Breakpoint and Syscall.
* interp.c (SquareRoot): Add HAVE_SQRT check to ensure sqrt() is
only used on those hosts that provide it.
* configure.in: Add sqrt() to list of functions to be checked for.
* config.in: Re-generated.
* configure: Re-generated.
expanding STORE RIGHT, to fix swr.
* support.h (SIGNEXTEND): If the sign bit is not set, explicitly
clear the high bits.
* interp.c (Convert): Fix fmt_single to fmt_long to not truncate.
Fix float to int conversions to produce signed values.
* interp.c (sim_monitor): Improved monitor printf
simulation. Tidied up simulator warnings, and added "--log" option
for directing warning message output.
* gencode.c: Use sim_warning() rather than WARNING macro.
getopt1.o, rather than on gencode.c. Link objects together.
Don't link against -liberty.
(gencode.o, getopt.o, getopt1.o): New targets.
* gencode.c: Include <ctype.h> and "ansidecl.h".
(AND): Undefine after including "ansidecl.h".
(ULONG_MAX): Define if not defined.
(OP_*): Don't define macros; now defined in opcode/mips.h.
(main): Call my_strtoul rather than strtoul.
(my_strtoul): New static function.
instead of `long long' and `unsigned long long' data types.
* interp.c: #include sysdep.h to get signals, and define default
for SIGBUS.
* (Convert): Work around for Visual-C++ compiler bug with type
conversion.
* support.h: Make things compile under Visual-C++ by using
__int64 instead of `long long'. Change many refs to long long
into word64/uword64 typedefs.
INSTALL_PROGRAM, INSTALL_DATA): Use autoconf-set values.
(docdir): Removed.
* configure.in (AC_PREREQ): autoconf 2.5 or higher.
(AC_PROG_INSTALL): Added.
(AC_PROG_CC): Moved to before configure.host call.
* configure: Rebuilt.
* configure.in: Define @SIMCONF@ depending on mips target.
* configure: Rebuild.
* Makefile.in (run): Add @SIMCONF@ to control simulator
construction.
* gencode.c: Change LOADDRMASK to 64bit memory model only.
* interp.c: Remove some debugging, provide more detailed error
messages, update memory accesses to use LOADDRMASK.
AC_CHECK_LIB, and AC_CHECK_FUNCS. Change AC_OUTPUT to set
stamp-h.
* configure: Rebuild.
* config.in: New file, generated by autoheader.
* interp.c: Include "config.h". Include <stdlib.h>, <string.h>,
and <strings.h> if they exist. Replace #ifdef sun with #ifdef
HAVE_ANINT and HAVE_AINT, as appropriate.
* Makefile.in (run): Use @LIBS@ rather than -lm.
(interp.o): Depend upon config.h.
(Makefile): Just rebuild Makefile.
(clean): Remove stamp-h.
(mostlyclean): Make the same as clean, not as distclean.
(config.h, stamp-h): New targets.
* interp.c (xfer_direct_word, xfer_direct_long,
swap_direct_word, swap_direct_long, xfer_big_word,
xfer_big_long, xfer_little_word, xfer_little_long,
swap_word,swap_long): Added.
* interp.c (ColdReset): Provide function indirection to
host<->simulated_target transfer routines.
* interp.c (sim_store_register, sim_fetch_register): Updated to
make use of indirected transfer routines.
* interp.c (Convert): Provide round-to-nearest and round-to-zero
support for Sun hosts.
* Makefile.in (gencode): Ensure the host compiler and libraries
used for cross-hosted build.
Allow a DOS hosted version of the simulator to be built. NOTE: The FP
is still not complete, since round-to-nearest and round-to-zero have
not been implemented generically.
* gencode.c, interp.c: Replaced explicit long long references with
WORD64HI, WORD64LO, SET64HI and SET64LO macro calls.
* support.h (SET64LO, SET64HI): Macros added.
This is an intermediate checkin. The work of removing "long long"
usage is not yet finished. These changes are clean, and have been
sitting on my machine for a while (whilst doing other work), and it is
safer for them to be checked in.
support.
* interp.c: Added dineroIII, and BSD profiling support. Also
run-time FP handling.
At the moment the options are still mostly build-time controlled,
rather than run-time. Also work still needs to be done to remove (long
long) usage (However this is trivial, just time-consuming).
The out-standing instruction work to be done is in supporting round
and trunc for FP operations, and providing better exception support.
the run-time support code (interp.c) to provide better tracing, and
also to add profiling and architecture specific support. At the moment
the simulator has a fixed size, fixed address memory area, and
simulates a subset of the IDT monitor calls (enough to execute test
programs).
The other major feature (could even be a bug) is that the simulator
makes use of the GCC "long long" extension. Work has been started to
make this a build configuration option... but there is still a lot of
this to be done.
