Inspired by a suggestion from Chris Gregory in:
http://lists.gnu.org/archive/html/emacs-devel/2016-12/msg00965.html
On my platform, this generates exactly the same machine insns.
* src/eval.c (prog_ignore): Rename from unwind_body, since
it’s more general than that. All callers changed.
(Fprog1): Simplify by using prog_ignore.
(Fwhile): Clarify by using prog_ignore.
This improves performance overall on my benchmark on x86-64,
since the interpreted program-counter resides in a machine
register rather than in RAM.
* etc/DEBUG, src/.gdbinit: Remove xbytecode GDB command, as there
is no longer a byte stack to decode.
* src/bytecode.c (struct byte_stack, byte_stack_list)
(relocate_byte_stack): Remove. All uses removed.
(FETCH): Simplify now that pc is now local (typically, in a
register) and no longer needs to be relocated.
(CHECK_RANGE): Remove. All uses now done inline, in a different way.
(BYTE_CODE_QUIT): Remove; now done by op_relative_branch.
(exec_byte_code): Allocate a copy of the function’s bytecode,
so that there is no problem if GC moves it.
* src/lisp.h (struct handler): Remove byte_stack member.
All uses removed.
* src/thread.c (unmark_threads): Remove. All uses removed.
* src/thread.h (struct thread_state): Remove m_byte_stack_list member.
All uses removed. m_stack_bottom is now the first non-Lisp field.
* src/bytecode.c (BYTE_MAINTAIN_TOP): Remove; no longer needed.
(struct byte_stack) [BYTE_MAINTAIN_TOP]:
Remove unused members ‘top’ and ‘bottom’.
(exec_byte_code): Nest inside { } to avoid GCC warning about
jumping over declaration when compiled with -DBYTE_CODE_SAFE.
* src/bytecode.c (exec_byte_code):
* src/image.c (parse_image_spec):
Prefer FUNCTIONP (x) to !NILP (Ffunctionp (x)).
* src/eval.c (FUNCTIONP): Move here ...
* src/lisp.h: ... from here. No longer inline, as that
bloats the text and does not help speed (at least on my platform).
(functionp): Remove this name, since callers use FUNCTIONP.
This does not let Emacs run, just build.
* lib-src/etags.c (main):
* lib-src/profile.c (main):
Use return, not exit.
* src/bytecode.c (BYTE_CODE_THREADED) [__CHKP__]:
Do not define, as -fcheck-pointer-bounds is incompatible with taking
addresses of labels.
* src/menu.c (Fx_popup_dialog): Use eassume, not eassert,
to pacify gcc -fcheck-pointer-bounds -Wnull-dereference.
* src/bytecode.c (exec_byte_code): Simplify, mostly by moving
initializers into decls, and by omitting some unnecessary
changes to ‘top’. This improves performance a bit on x86-64,
as it happens.
This improves performance overall on my benchmark on x86-64,
since the interpreted program-counter resides in a machine
register rather than in RAM.
* etc/DEBUG, src/.gdbinit: Remove xbytecode GDB command, as there
is no longer a byte stack to decode.
* src/bytecode.c (struct byte_stack, byte_stack_list)
(relocate_byte_stack): Remove. All uses removed.
(FETCH): Simplify now that pc is now local (typically, in a
register) and no longer needs to be relocated.
(CHECK_RANGE): Remove. All uses now done inline, in a different way.
(BYTE_CODE_QUIT): Remove; now done by op_relative_branch.
(exec_byte_code): Allocate a copy of the function’s bytecode,
so that there is no problem if GC moves it.
* src/lisp.h (struct handler): Remove byte_stack member.
All uses removed.
(SAFE_ALLOCA_LISP_EXTRA): New macro, a generalization of
SAFE_ALLOCA_LISP.
(SAFE_ALLOCA_LISP): Use it.
* src/bytecode.c (exec_byte_code): Remove MAX_ALLOCA-based limit
on bytecode maxdepth, by using SAFE_ALLOCA_LISP instead of alloca.
pipeline is fuller.
* src/bytecode.c (BYTE_CODE_QUIT): Check for GC, too. Do the
check only once every 256 times. This should be good enough, and
improves performance significantly on x86-64 as branch-prediction
typically assumes checking will not be done so the instruction
pipeline stays fuller.
(exec_byte_code): Set up the quit counter. Don’t call maybe_gc
directly, as BYTE_CODE_QUIT does that now.
* src/bytecode.c (BYTE_CODE_SAFE): Default to false, so that it
can be used outside #if. All uses of ‘defined BYTE_CODE_SAFE’
changed to ‘BYTE_CODE_SAFE’. Use BYTE_CODE_SAFE in plain
expressions instead of #if expressions when this is easy.
(struct byte_stack) [BYTE_CODE_SAFE]: Remove member ‘bottom’,
as it is no longer needed.
(exec_byte_code): Omit #if BYTE_CODE_SAFE when this is easy.
Simplify stack-overflow checking when BYTE_CODE_SAFE.
BYTE_CODE_METER hasn’t worked since 2013, and nobody seems to have
noticed. Perhaps we should remove it?
* src/bytecode.c (METER_2) [BYTE_CODE_METER]:
Use *aref_addr instead of AREF, since it needs to be an lvalue.
Although for decades ‘signal’ has been documented to not return,
a corner case in the Lisp debugger causes ‘signal’ to return.
Remove the corner case and adjust Emacs internals accordingly.
An alternative would be to document the corner case, but this
would complicate the Lisp API unnecessarily. (Bug#24047)
* src/eval.c (signal_or_quit): New function, with most of the
old contents of Fsignal.
(quit): New function, which uses signal_or_quit and which
might return. All keyboard-based callers of Fsignal (Qquit,
Qnil) changed to use this new function instead.
(Fsignal): Use signal_or_quit. Now _Noreturn. All callers
changed.
(xsignal): Move to lisp.h.
* src/lisp.h (xsignal): Now an inline function, as it's now
just an alias for Fsignal.
The new approach is less selective, but also less intrusive on C code.
* src/bytecode.c, src/image.c, src/keyboard.c: Ignore -Wclobbered.
* src/conf_post.h (NONVOLATILE): Remove. All uses removed.
It is no longer needed now that we assume GC_MARK_STACK ==
GC_MAKE_GCPROS_NOOPS.
* src/bytecode.c (BYTE_MAINTAIN_TOP): Remove.
All uses removed, and code simplified accordingly.
(BEFORE_POTENTIAL_GC, AFTER_POTENTIAL_GC): Remove, since they
are always no-ops now. All uses removed.
(MAYBE_GC): Remove. All uses replaced by maybe_gc,
since it is now equivalent.
* src/conf_post.h (GNUC_PREREQ): New macro.
Change uses of __GNUC_MINOR__ and __GNUC_PATCHLEVEL__ to use this
macro instead, for clarity and consistency.
(PRINTF_ARCHETYPE): New macro. Define it to __gnu_printf__ only
if glibc, since non-GNU platforms don’t necessarily support GNU
printf formats.
(ATTRIBUTE_FORMAT_PRINTF): Use it.
* src/bytecode.c (get_byte_code_arity): Omit unnecessary
runtime test for integer argument, unless debugging.
Use EMACS_INT for Emacs integers.
* src/eval.c (Ffunc_arity): Omit unused locals.
Avoid side effects in ‘if’ expr.
(lambda_arity): Use bool for boolean, and EMACS_INT for Emacs ints.
Work around Bug#22884 by rewording comments and strings to avoid ‘(’
at the start of a line unless it starts a function. This change
is a short-term hack; in the longer run we plan to fix cc-mode’s
performance for C files that have ‘(’ at the start of a line in a
comment or string.
* src/lisp.h (PUSH_HANDLER): Remove.
All callers changed to use push_handler directly.
* src/eval.c (internal_condition_case)
(internal_condition_case_1, internal_condition_case_2)
(internal_condition_case_n):
Use same pattern as for other invokers of push_handler.
(push_handler, push_handler_nosignal): Use call-by-value
instead of call-by-reference. All uses changed.
(push_handler): Simplify by rewriting in terms of
push_handler_nosignal.
(push_handler_nosignal): Profile any newly allocated memory.
Although \( is needed in docstrings in Elisp code, it is not needed in
docstrings in C code, since C function definitiions do not start with
a parenthesis. The backslashes made the docstrings a bit harder to
read and to format in columns. Also, some C docstrings had ( in
column 1 and this did not appear to be causing any problems. So,
simplify C docstrings by replacing \( with ( and \) with ).
This removes the need for GCPRO1 etc. Suggested by Stefan Monnier in:
http://lists.gnu.org/archive/html/emacs-devel/2015-08/msg00918.html
* doc/lispref/internals.texi (Writing Emacs Primitives):
* etc/NEWS:
Document the change.
* src/alloc.c (gcprolist, dump_zombies, MAX_ZOMBIES, zombies)
(nzombies, ngcs, avg_zombies, max_live, max_zombies, avg_live)
(Fgc_status, check_gcpros, relocatable_string_data_p, gc-precise):
* src/bytecode.c (mark_byte_stack) [BYTE_MARK_STACK]:
* src/eval.c (gcpro_level) [DEBUG_GCPRO]:
* src/lisp.h (struct handler.gcpro, struct gcpro, GC_MARK_STACK)
(GC_USE_GCPROS_AS_BEFORE, GC_MAKE_GCPROS_NOOPS)
(GC_MARK_STACK_CHECK_GCPROS, GC_USE_GCPROS_CHECK_ZOMBIES)
(BYTE_MARK_STACK, GCPRO1, GCPRO2, GCPRO3, GCPRO4, GCPRO5, GCPRO6)
(GCPRO7, UNGCPRO, RETURN_UNGCPRO):
Remove. All uses removed. The code now assumes
GC_MARK_STACK == GC_MAKE_GCPROS_NOOPS.
* src/bytecode.c (relocate_byte_stack):
Rename from unmark_byte_stack, since it now only relocates.
All callers changed.
* src/frame.c (make_frame): Add an IF_LINT to pacify GCC 5.2
with GCPROs removed.
* src/systime.h: Use EMACS_LISP_H as the canary instead of GCPRO1.
* test/automated/finalizer-tests.el (finalizer-basic)
(finalizer-circular-reference, finalizer-cross-reference)
(finalizer-error):
* test/automated/generator-tests.el (cps-test-iter-close-finalizer):
Remove tests, as they depend on gc-precise.
Fixes Bug#15880.
This patch also makes Q constants (e.g., Qnil) constant addresses
from the C point of view.
* make-docfile.c: Revamp to generate table of symbols, too.
Include <stdbool.h>.
(xstrdup): New function.
(main): Don't process the same file twice.
(SYMBOL): New constant in enum global_type.
(struct symbol): Turn 'value' member into a union, either v.value
for int or v.svalue for string. All uses changed.
(add_global): New arg svalue, which overrides value, so that globals
can have a string value.
(close_emacs_global): New arg num_symbols; all uses changed.
Output lispsym decl.
(write_globals): Output symbol globals too. Output more
ATTRIBUTE_CONST, now that Qnil etc. are C constants.
Output defsym_name table.
(scan_c_file): Move most of guts into ...
(scan_c_stream): ... new function. Scan for DEFSYMs and
record symbols found. Don't read past EOF if file doesn't
end in newline.
* alloc.c, bidi.c, buffer.c, bytecode.c, callint.c, casefiddle:
* casetab.c, category.c, ccl.c, charset.c, chartab.c, cmds.c, coding.c:
* composite.c, data.c, dbusbind.c, decompress.c, dired.c, dispnew.c:
* doc.c, editfns.c, emacs.c, eval.c, fileio.c, fns.c, font.c, fontset.c:
* frame.c, fringe.c, ftfont.c, ftxfont.c, gfilenotify.c, gnutls.c:
* image.c, inotify.c, insdel.c, keyboard.c, keymap.c, lread.c:
* macfont.m, macros.c, minibuf.c, nsfns.m, nsfont.m, nsimage.m:
* nsmenu.m, nsselect.m, nsterm.m, print.c, process.c, profiler.c:
* search.c, sound.c, syntax.c, term.c, terminal.c, textprop.c, undo.c:
* window.c, xdisp.c, xfaces.c, xfns.c, xftfont.c, xmenu.c, xml.c:
* xselect.c, xsettings.c, xterm.c:
Remove Q vars that represent symbols (e.g., Qnil, Qt, Qemacs).
These names are now defined automatically by make-docfile.
* alloc.c (init_symbol): New function.
(Fmake_symbol): Use it.
(c_symbol_p): New function.
(valid_lisp_object_p, purecopy): Use it.
* alloc.c (marked_pinned_symbols):
Use make_lisp_symbol instead of make_lisp_ptr.
(garbage_collect_1): Mark lispsym symbols.
(CHECK_ALLOCATED_AND_LIVE_SYMBOL): New macro.
(mark_object): Use it.
(sweep_symbols): Sweep lispsym symbols.
(symbol_uses_obj): New function.
(which_symbols): Use it. Work for lispsym symbols, too.
(init_alloc_once): Initialize Vpurify_flag here; no need to wait,
since Qt's address is already known now.
(syms_of_alloc): Add lispsym count to symbols_consed.
* buffer.c (init_buffer_once): Compare to Qnil, not to make_number (0),
when testing whether storage is all bits zero.
* dispextern (struct image_type):
* font.c (font_property_table):
* frame.c (struct frame_parm_table, frame_parms):
* keyboard.c (scroll_bar_parts, struct event_head):
* xdisp.c (struct props):
Use XSYMBOL_INIT (Qfoo) and struct Lisp_Symbol * rather than &Qfoo and
Lisp_Object *, since Qfoo is no longer an object whose address can be
taken. All uses changed.
* eval.c (run_hook): New function. Most uses of Frun_hooks changed to
use it, so that they no longer need to take the address of a Lisp sym.
(syms_of_eval): Don't use DEFSYM on Vrun_hooks, as it's a variable.
* frame.c (syms_of_frame): Add defsyms for the frame_parms table.
* keyboard.c (syms_of_keyboard): Don't DEFSYM Qmenu_bar here.
DEFSYM Qdeactivate_mark before the corresponding var.
* keymap.c (syms_of_keymap): Use DEFSYM for Qmenu_bar and Qmode_line
instead of interning their symbols; this avoids duplicates.
(LISP_INITIALLY, TAG_PTR)
(DEFINE_LISP_SYMBOL_BEGIN, DEFINE_LISP_SYMBOL_END, XSYMBOL_INIT):
New macros.
(LISP_INITIALLY_ZERO): Use it.
(enum symbol_interned, enum symbol_redirect, struct Lisp_Symbol)
(EXFUN, DEFUN_ARGS_MANY, DEFUN_ARGS_UNEVALLED, DEFUN_ARGS_*):
Move decls up, to avoid forward uses. Include globals.h earlier, too.
(make_lisp_symbol): New function.
(XSETSYMBOL): Use it.
(DEFSYM): Now just a placeholder for make-docfile.
* lread.c (DEFINE_SYMBOLS): Define, for globals.h.
(intern_sym): New function, with body taken from old intern_driver.
(intern_driver): Use it. Last arg is now Lisp integer, not ptrdiff_t.
All uses changed.
(define_symbol): New function.
(init_obarray): Define the C symbols taken from lispsym.
Use plain DEFSYM for Qt and Qnil.
* syntax.c (init_syntax_once): No need to worry about
Qchar_table_extra_slots.
* configure.ac (WERROR_CFLAGS): Don't disable -Wnested-externs.
While we're at it, don't disable -Wlogical-op either.
* src/bytecode.c: Include blockinput.h and keyboard.h rather
than rolling their APIs by hand.
* src/emacs.c: Include regex.h and rely on its and lisp.h's API
rather than rolling them by hand.
* src/lastfile.c: Include lisp.h, to check this file's API.
* src/lisp.h (lisp_eval_depth, my_edata, my_endbss, my_endbss_static):
New decls.
* src/regex.h (re_max_failures): New decl.
* src/unexcw.c, src/unexmacosx.c, src/unexw32.c:
Rely on lisp.h's API rather than rolling it by hand.
* src/vm-limit.c (__after_morecore_hook, __morecore, real_morecore):
Declare at top level, to pacify GCC -Wnested-externs.
Especially, C99 prohibits nesting a struct X inside struct Y if
struct X has a flexible array member.
Also, merge from gnulib, incorporating:
2013-11-03 intprops: port to Oracle Studio c99
* lib/intprops.h: Update from gnulib.
* src/alloc.c (struct sdata): New type.
(sdata): Implement in terms of struct sdata.
Remove u member; all uses replaced by next_vector, set_next_vector.
(SDATA_SELECTOR, SDATA_DATA, SDATA_DATA_OFFSET): Adjust to sdata change.
(SDATA_DATA_OFFSET): Now a constant, not a macro.
(struct sblock): Rename first_data member to data, which is now
a flexible array member. All uses changed.
(next_vector, set_next_vector, large_vector_vec): New functions.
(vector_alignment): New constant.
(roundup_size): Make it a multiple of ALIGNOF_STRUCT_LISP_VECTOR, too.
(struct large-vector): Now merely a NEXT member, since the old approach
ran afoul of stricter C99. All uses changed to use
large_vector_vec or large_vector_offset.
(large_vector_offset): New constant.
* src/dispnew.c: Include tparam.h, for tgetent.
Do not include term.h; no longer needed.
* src/gnutls.c (Fgnutls_boot): Don't continue after calling a _Noreturn.
* src/lisp.h (ENUM_BF) [__SUNPRO_C && __STDC__]: Use unsigned int.
(struct Lisp_Vector): Use a flexible array member for contents,
instead of a union with a member that is an array of size 1.
All uses changed.
(ALIGNOF_STRUCT_LISP_VECTOR): New constant, to make up for the
fact that the struct no longer contains a union.
(struct Lisp_Misc_Any, struct Lisp_Marker, struct Lisp_Overlay)
(struct Lisp_Save_Value, struct Lisp_Free):
Use unsigned, not int, for spacers, to avoid c99 warning.
(union specbinding): Use unsigned, not bool, for bitfield, as
bool is not portable to pre-C99 hosts.
* lisp/emacs-lisp/byte-opt.el (byte-optimize-form-code-walker):
Optimize under `condition-case' and `catch' if
byte-compile--use-old-handlers is nil.
(disassemble-offset): Handle new bytecodes.
* lisp/emacs-lisp/bytecomp.el (byte-pushcatch, byte-pushconditioncase)
(byte-pophandler): New byte codes.
(byte-goto-ops): Adjust accordingly.
(byte-compile--use-old-handlers): New var.
(byte-compile-catch): Use new byte codes depending on
byte-compile--use-old-handlers.
(byte-compile-condition-case--old): Rename from
byte-compile-condition-case.
(byte-compile-condition-case--new): New function.
(byte-compile-condition-case): New function that dispatches depending
on byte-compile--use-old-handlers.
(byte-compile-unwind-protect): Pass a function to byte-unwind-protect
when we can.
* lisp/emacs-lisp/cconv.el (cconv-convert, cconv-analyse-form): Adjust for
the new compilation scheme using the new byte-codes.
* src/alloc.c (Fgarbage_collect): Merge scans of handlerlist and catchlist,
and make them unconditional now that they're heap-allocated.
* src/bytecode.c (BYTE_CODES): Add Bpushcatch, Bpushconditioncase
and Bpophandler.
(bcall0): New function.
(exec_byte_code): Add corresponding cases. Improve error message when
encountering an invalid byte-code. Let Bunwind_protect accept
a function (rather than a list of expressions) as argument.
* src/eval.c (catchlist): Remove (merge with handlerlist).
(handlerlist, lisp_eval_depth): Not static any more.
(internal_catch, internal_condition_case, internal_condition_case_1)
(internal_condition_case_2, internal_condition_case_n):
Use PUSH_HANDLER.
(unwind_to_catch, Fthrow, Fsignal): Adjust to merged
handlerlist/catchlist.
(internal_lisp_condition_case): Use PUSH_HANDLER. Adjust to new
handlerlist which can only handle a single condition-case handler at
a time.
(find_handler_clause): Simplify since we only a single branch here
any more.
* src/lisp.h (struct handler): Merge struct handler and struct catchtag.
(PUSH_HANDLER): New macro.
(catchlist): Remove.
(handlerlist): Always declare.
This helps to avoid a few glitches dictated by C's aliasing rules.
* lisp.h (struct Lisp_Vector): Use union for next and
contents member. Adjust comment. Change related users.
* alloc.c (next_in_free_list, set_next_in_free_list): Remove.
Related users changed.
* buffer.c, bytecode.c, ccl.c, character.h, chartab.c, composite.c:
* composite.h, disptab.h, fns.c, fontset.c, indent.c, keyboard.c:
* lread.c, msdos.c, process.c, w32menu.c, window.c, xdisp.c:
* xfaces.c, xfont.c, xmenu.c: Related users changed.
* src/data.c: Change comparison functions' interface and
implementation
* src/lisp.h: Make arithcompare available for efficient two arg
comparisons
* src/bytecode.c: Use arithcompare
* src/fileio.c: Use new interface
* test/automated/data-tests.el: New tests for comparison functions
* etc/NEWS
