* libguile/jit.c (emit_entry_trampoline): Don't bother hackily trying to
save registers; the "jit_frame" call handles that.
(compile_return_values, compile_return_from_interrupt): Fix bug when
computing previous FP: no need to add frame_overhead_slots.
(emit_load_prev_fp_offset, emit_store_prev_fp_offset): Rename from
emit_load_prev_frame_size, emit_store_prev_frame_size.
(emit_push_frame): Adapt to emit_store_prev_frame_size. Don't
subtract off the frame_overhead_slots.
(scm_jit_enter_mcode): Comment out a printf for the time being.
* libguile/jit.c (emit_alloc_frame_for_sp, compile_current_thread): Fix
some ldxr/ldxi stxr/stxi confusions.
(compile_alloc_frame): Omit if the frame size is already correct.
(compile, compute_mcode, scm_sys_jit_compile, scm_jit_enter_mcode):
Add a bit more debugging.
* libguile/lightning/lib/lightning.c (_jit_emit): The default code
allocator will simply mmap a code buffer, try to emit into that
buffer, and if it fails, try again with a larger buffer. However the
buffer size starts at 0, for some reason. Why? I can't see the
reason. Change the default to 4096. In the future we will need to
implement our own allocator anyway so that we can pack multiple JIT
runs in one page.
* libguile/init.c (scm_i_init_guile): Call scm_init_jit ().
* libguile/jit.c (enter_mcode, exit_mcode): New static members; code
pointers for the JIT trampoline.
(emit_exit): New helper. The Lightning tramp/frame mechanism that we
use needs to exit via a jmp instead of a return. Adapt callers of
jit_ret.
(emit_entry_trampoline): Use the "frame" mechanism to enter the JIT.
(compile1): Add missing "break" after case statements. Oops!
(compile): Add prolog and "tramp" to compiled functions.
(initialize_jit): New local routine to init the JIT on demand.
(compute_mcode): New helper, to compile a function.
(scm_sys_jit_compile): New function, exported to Scheme as
%jit-compile.
(scm_jit_compute_mcode): Return the existing mcode if the function is
at the start.
(scm_jit_enter_mcode): Call the enter_mcode trampoline.
* libguile/jit.h (struct scm_jit_state): Declare, so we can make
pointers to it.
* libguile/threads.h (struct scm_thread): Add jit_state member.
* libguile/threads.c (on_thread_exit): Free the jit state.
* libguile/intrinsics.h: Add "intrinsic" for handle-interrupts code.
Unlike the other intrinsics, this one isn't a function.
* libguile/programs.c (try_parse_arity): Add cases for instructions used
in VM builtins.
(scm_primitive_call_ip): Return #f if call-ip not found.
* libguile/vm-engine.c (handle-interrupts): Get code from intrinsics.
* libguile/vm.c
* libguile/vm.c (instrumented_code, define_vm_builtins): Add
instrumentation to the builtins, so that they can be JIT-compiled.
(INIT_BUILTIN): Remove min-arity setting; the fallback min-arity
interpreter should figure it out.
(scm_bootstrap_vm): Call the new define_vm_builtins function.
* libguile/gsubr.c (primitive_call_ip): Return 0 if call IP not found.
(primitive_subr_idx): Interpret call ip == 0 as not-a-subr.
* module/system/vm/program.scm (program-arguments-alist): Allow a #f
call-ip.
* libguile/intrinsics.h:
* libguile/intrinsics.c (atomic_ref_scm, atomic_set_scm):
(atomic_swap_scm, atomic_compare_and_swap_scm): New intrinsics, given
that lightning doesn't know atomics.
(scm_bootstrap_intrinsics): Init new intrinsics.
* libguile/vm-engine.c (atomic-scm-ref/immediate)
(atomic-scm-set!/immediate, atomic-scm-swap!/immediate)
(atomic-scm-compare-and-swap!/immediate): Use intrinsics, to be like
the JIT.
* libguile/intrinsics.h (INDIRECT_INT64_INTRINSICS): New definition. If
true, int64 args and return values are passed by reference. Here to
make JIT easier.
* libguile/intrinsics.c (indirect_scm_to_int64, indirect_scm_to_uint64):
(indirect_scm_to_uint64_truncate, indirect_scm_from_int64):
(indirect_scm_from_uint64, indirect_lsh, indirect_rsh): New indirect
variants.
(scm_bootstrap_intrinsics): Use indirect variants as appropriate.
* libguile/vm-engine.c: Update to call indirect intrinsics if
appropriate.
* libguile/intrinsics.h:
* libguile/intrinsics.c (string_set_x): Change to take size_t and u32 as
args.
(allocate_words): Change to take size_t as arg.
* libguile/vm.c (expand_apply_argument): Rename from rest_arg_length,
and also handle the stack manipulation.
* libguile/vm-engine.c (expand-apply-argument): Update for intrinsic
change.
(call-scm-sz-u32): Rename from call-scm-u64-u64, as it matches its
uses and will compile better on 32-bit systems.
* module/system/vm/assembler.scm (define-scm-sz-u32-intrinsic):
(string-set!): Update for new instrinsic call inst.
* libguile/jit.c (compile_call_scm_sz_u32): Adapt.
* libguile/intrinsics.h (SCM_FOR_ALL_VM_INTRINSICS):
* libguile/intrinsics.c (error_wrong_num_args): Take the thread as an
arg, instead of the ostensible callee.
* libguile/vm-engine.c: Update callers of wrong-num-args intrinsic to
pass a thread instead.
* libguile/intrinsics.h (SCM_FOR_ALL_VM_INTRINSICS): Update prototype of
capture-continuation.
* libguile/jit.h:
* libguile/jit.c (scm_jit_enter_mcode): Return void, not the vra.
Instead, we expect the code to set vp->ip for the vra.
* libguile/vm-engine.c (instrument-entry, instrument-loop)
(return-values, abort): Adapt scm_jit_enter_mcode calling convention.
(capture-continuation): No need to pass an mra; the intrinsic will
read it from the stack.
* libguile/vm.c (capture_continuation): Remove mra arg, as we take mra
from the continuation.
(scm_call_n): Adapt to scm_jit_enter_mcode change.
* libguile/continuations.c (scm_i_continuation_to_frame): Adapt to vra
field renaming.
(scm_i_reinstate_continuation, grow_stack, copy_stack_and_call)
(scm_dynthrow): Take mra of continuation. Set on the vp before the
longjmp.
* libguile/continuations.h: Update scm_i_reinstate_continuation
prototype.
* libguile/dynstack.h:
* libguile/control.c (scm_suspendable_continuation_p):
* libguile/dynstack.c (PROMPT_WORDS, PROMPT_VRA, PROMPT_MRA):
(PROMPT_JMPBUF, scm_dynstack_push_prompt, scm_dynstack_find_prompt)
(scm_dynstack_wind_prompt): Store both virtual and machine return
addresses on the dynstack, for prompts.
* libguile/eval.c (eval): Pass NULL for mra.
* libguile/intrinsics.c (push_prompt): Add mra arg, and pass it to the
dynstack.
* libguile/intrinsics.h: Update prototypes so that continuation-related
intrinsics can save and restore the MRA.
* libguile/jit.h:
* libguile/jit.c: Return VRA when JIT code needs to tier down.
* libguile/stacks.c (find_prompt, scm_make_stack)
* libguile/throw.c (catch): Adapt find-prompt calls.
* libguile/vm-engine.c (instrument-entry, instrument-loop): Add logic to
continue with vcode after the mcode finishes.
(compose-continuation, capture-continuation, abort, prompt): Add logic
to pass NULL as captured MRA, but continue with mcode from new
continuations, if appropriate.
* libguile/vm.c (scm_i_vm_cont_to_frame, capture_stack)
(scm_i_capture_current_stack, reinstate_continuation_x)
(capture_continuation, compose_continuation_inner, compose_continuation)
(capture_delimited_continuation, abort_to_prompt): Adapt to plumb
around machine code continuations.
(scm_call_n): Check "mra_after_abort" field for machine code
continuation, if any.
* libguile/vm.h (struct scm_vm): Add "mra_after_abort" field.
(struct scm_vm_cont): Rename "ra" field to "vra" and add "mra" field.
* libguile/foreign.c (scm_i_foreign_call): Rename back from
foreign_call. Need a new trampoline that's easier to call from JIT,
until we actually rewrite the FFI in terms of the JIT.
(scm_register_foreign): Remove foreign_call intrinsic init.
* libguile/intrinsics.h (SCM_FOR_ALL_VM_INTRINSICS): Foreign-call
intrinsic sets return directly on stack.
* libguile/vm-engine.c (foreign-call): Adapt to new intrinsic behavior.
* libguile/vm.c (foreign_call, scm_bootstrap_vm): Add new intrinsic
wrapper.
* libguile/intrinsics.h (SCM_FOR_ALL_VM_INTRINSICS): Make intrinsics for
the VM hooks.
* libguile/vm-engine.c (RUN_HOOK): Run hooks through the intrinsics
table.
* libguile/vm.c (invoke_apply_hook, invoke_return_hook):
(invoke_next_hook, invoke_abort_hook): Rename, remove NOINLINE
attribute (as we call them through the intrinsics table).
(scm_bootstrap_vm): Init new intrinsics.
* libguile/intrinsics.h (SCM_FOR_ALL_VM_INTRINSICS): Add
unpack-values-object.
* libguile/vm-engine.c (subr-call): If the object is a values object,
call out to unpack-values-object. This is to avoid reifying
allocate-frame code in each jitted subr.
* libguile/vm.c (unpack_values_object, scm_bootstrap_vm): Define new
intrinsic.
* libguile/jit.c (compile_capture_continuation): Rename from call_cc now
that the call is elsewhere.
* libguile/vm-engine.c (call, tail-call): Remove needless SYNC_IP before
get-callee-vcode; the intrinsic can sync the ip if needed from the
frame.
(capture-continuation): Rename from call/cc, and leave the call itself
to tail-call.
* libguile/vm.c (vm_builtin_call_with_current_continuation_code): Update
to put the continuation in a local and then tail call.
(get_callee_vcode): Sync vp->ip if we error.
* libguile/intrinsics.h:
* libguile/vm.c (get_callee_vcode): Rename from apply_non_program, and
instead return the IP for the callee of a frame.
(scm_call_n, scm_bootstrap_vm): Adapt to get_callee_vcode change.
* libguile/vm-engine.c (call, tail-call, call/cc): Use
get_callee_vcode unconditionally. JIT will do this to avoid so much
code generation for calls.
Reported by Ricardo Wurmus <rekado@elephly.net>.
Fixes <https://bugs.gnu.org/32161>.
* libguile/r6rs-ports.c (custom_binary_port_seek): Use 'scm_to_off_t'
instead of 'scm_to_int'.
* test-suite/tests/r6rs-ports.test ("8.2.7 Input Ports")["custom binary
input port position, long offset"]: New test.
Fixes <https://bugs.gnu.org/28211>.
* libguile/vm-engine.c (call, call_label, handle_interrupts): Add
'new_fp' variable; set the dynamic link and return address of the frame
at NEW_FP before setting 'vp->fp'. This fixes a bug whereby, in a
multi-threaded context, the stack-marking code could run after vp->fp
has been set but before its dynamic link has been set, leading the
stack-walking code in 'scm_i_vm_mark_stack' to exit early on.
Previously 'vm-frame-older' would fail to traverse the chain of frames.
* libguile/libguile-2.2-gdb.scm (uint-type): New variable
(vm-frame): Fix "saved ip" and "saved fp" computation. The latter had
been broken roughly since commit
72353de77d.
(vm-frame-older): Return #f when IP is zero, not when FP is zero.
(vm-frame-function-name): Wrap 'vm-frame-program-debug-info' in
'false-if-exception'
Fixes <https://bugs.gnu.org/31879>.
* libguile/modules.c (import_obarray_mutex): New variable.
(resolve_duplicate_binding, module_imported_variable): Acquire it before
accessing an obarray.
* libguile/r6rs-ports.c (scm_get_bytevector_n, scm_get_bytevector_n_x):
Turn 'c_count' and related variables into a 'size_t', and use
'scm_to_size_t' instead of 'scm_to_uint'.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
Previously 'load-thunk-from-memory' would often throw to 'system-error'
based on a stale value in 'errno', leading to incorrect error messages.
* libguile/loader.c (load_thunk_from_memory): Set 'errno' to 0 before
jumping to cleanup in the ABORT preprocessor macro, and also in the case
when 'process_dynamic_segment' reports an error.
Previously 'load-thunk-from-memory' would often throw to 'system-error'
when passed an incorrect ELF file, leading to incorrect error messages.
* libguile/loader.c (load_thunk_from_memory): Reset 'errno' when
'check_elf_header' returns non-NULL.
* test-suite/tests/vm.test: New file.
* test-suite/Makefile.am (SCM_TESTS): Add it.
* libguile/posix.c (scm_crypt): Take 'scm_i_misc_mutex' right before
calling 'crypt'. Move 'SCM_SYSERROR' call after 'scm_dynwind_end'.
* test-suite/tests/posix.test ("crypt"): New test prefix.
* libguile/vm-engine.c (RUN_HOOK0, RUN_HOOK1): Remove.
(RUN_HOOK): Take hook name.
(APPLY_HOOK, RETURN_HOOK, NEXT_HOOK, ABORT_CONTINUATION_HOOK): Use
RUN_HOOK.
* libguile/vm.c (vm_dispatch_hook): Remove value count arg; hooks no
longer receive values (e.g. the return hook now uses
frame-return-values).
(vm_dispatch_abort_hook): Remove value count, which was bogus because
the active frame was the continuation which might contain other
locals, potentially unboxed, not the implicit return-values frame. In
the future we could push on an implicit return-values frame instead.
* module/system/vm/traps.scm (trap-in-procedure, trap-frame-finish):
(trap-in-dynamic-extent, trap-calls-to-procedure): Adapt abort hooks
to not take values. They weren't being used anyway!
* libguile/frames.c (scm_frame_return_values): New function, for use
when a frame is at "return-values".
(scm_init_frames_builtins): Register frame-return-values.
* libguile/vm-engine.c (RETURN_HOOK): Rename from POP_CONTINUATION_HOOK.
(call, call-label): Remove PUSH_CONTINUATION_HOOK; it's unneeded, as
you can always check the FP from an apply hook.
(return-values): Run return hook before popping frame.
* libguile/vm.c (vm_dispatch_return_hook): Rename from
vm_dispatch_pop_continuation_hook. Remove push continuation hook.
(scm_vm_return_hook):
* libguile/vm.h (SCM_VM_PUSH_CONTINUATION_HOOK): Remove.
(SCM_VM_RETURN_HOOK): Rename from SCM_VM_POP_CONTINUATION_HOOK.
* module/system/vm/frame.scm (frame-return-values): Export.
* module/system/vm/trace.scm (print-return, trace-calls-to-procedure)
(trace-calls-in-procedure): Adapt to not receiving values as
arguments.
* module/system/vm/traps.scm (trap-in-procedure, trap-frame-finish):
Adapt to return hook coming from returning frame.
(program-sources-by-line): Update to use match instead of pmatch.
* module/system/vm/traps.scm (trap-in-dynamic-extent)
(trap-calls-to-procedure): Adapt to return hook not receiving values.
* module/system/vm/vm.scm: Remove push continuation hook and rename
return hook.
* libguile/continuations.c (goto_continuation_code):
* libguile/control.c (compose_continuation_code): Add JIT
instrumentation. Unlike foreign or subr routines, we leave these as
closures; it doesn't make sense to JIT for any particular
continuation.
* libguile/foreign.c (get_foreign_stub_code): Use
scm_i_alloc_primitive_code_with_instrumentation. This will allow FFI
procedures to be JITted, and allow us to move the apply hook to
instrument-entry.
* lib/Makefile.am: Explicitly add flexmember. Already included though.
* libguile/gsubr.c (alloc_subr_idx, record_subr_name): Factor out an
"expected subr count" variable.
(alloc_chunk, alloc): Rework so that when the arena grows, old code is
still recognized as primitive.
(scm_i_alloc_primitive_code_with_instrumentation): Fix bug whereby the
JIT function data was initialized at a bogus address, because we were
adding to a uint32_t* instead of a char*. Make internally public.
(is_primitive_code): Recognize all allocated code as primitive.
(alloc_subr_code): Adapt to
scm_i_alloc_primitive_code_with_instrumentation rename.
(primitive_subr_idx): If the call IP isn't a subr-call, return a
sentinel value.
(scm_i_primitive_name, scm_subr_function): Allow for primitives that
aren't subrs.
* libguile/gsubr.h: Decalre
scm_i_alloc_primitive_code_with_instrumentation.
* libguile/gsubr.c: Reimplement to store subr names and procedures in a
side table, and to allocate fresh vcode for each subr. This allows
JIT of subrs, moves to a uniform all-code-starts-with-instrument-entry
regime, and also allows statprof to distinguish between subrs based on
IP.
* libguile/gsubr.h (SCM_SUBRF, SCM_SUBR_NAME): Call out to functions,
now that these are in a side table.
(scm_subr_function, scm_subr_name): New exports.
(scm_i_primitive_name): New internal function, for looking up a
primitive name based on IP.
(scm_apply_subr): Take the subr index.
* libguile/vm-engine.c (subr-call):
* libguile/jit.c (compile_subr_call): Adapt to take index as arg.
* module/statprof.scm (sample-stack-procs, count-call):
(stack-samples->procedure-data): Update to always record IP in stack
samples and call counts.
* module/system/vm/frame.scm (frame-procedure-name): Simplify.
(frame-instruction-pointer-or-primitive-procedure-name): Removed.
* libguile/programs.h:
* libguile/programs.c (scm_primitive_code_name): New function.
* module/system/vm/program.scm (primitive-code-name): New export.
