Problem reported by Daniel Colascione (Bug#41680).
* src/lread.c (read1): Check that AREF (tmp, COMPILED_BYTECODE)
is a string before subjecting it to STRING_MULTIBYTE.
Be more consistent about using AREF in the neighborhood,
to help prevent this sort of problem from recurring.
* src/lread.c (read_vector): If the vector is to short to be for
bytecodes don’t do bytecode processing for it, as the processing
might run past the end of the vector.
Check Lisp_Compiled objects better as they’re created,
so that the byte-code interpreter needn’t do the checks
each time it executes them. This improved performance
of ‘make compile-always’ by 1.5% on my platform. Also,
improve the quality of the (still-incomplete) checks, as
this is more practical now that they’re done less often.
* src/alloc.c (make_byte_code): Remove. All uses removed.
(Fmake_byte_code): Put a better (though still incomplete)
check here instead. Simplify by using Fvector instead
of make_uninit_vector followed by memcpy, and by using
XSETPVECTYPE instead of make_byte_code followed by XSETCOMPILED.
* src/bytecode.c (Fbyte_code): Do sanity check and conditional
translation to unibyte here instead of each time the function is
executed.
(exec_byte_code): Omit no-longer-necessary sanity and
unibyte checking. Use SCHARS instead of SBYTES where
either will do, as SCHARS is faster.
* src/eval.c (fetch_and_exec_byte_code): New function.
(funcall_lambda): Use it.
(funcall_lambda, lambda_arity, Ffetch_bytecode):
Omit no-longer-necessary sanity checks.
(Ffetch_bytecode): Add sanity check if actually fetching.
* src/lisp.h (XSETCOMPILED): Remove. All uses removed.
* src/lread.c (read1): Check byte-code objects more thoroughly,
albeit still incompletely, and do translation to unibyte here
instead of each time the function is executed.
(read1): Use XSETPVECYPE instead of make_byte_code.
(read_vector): Omit no-longer-necessary sanity check.
* src/buffer.h (fetch_char_advance, fetch_char_advance_no_check)
(buf_next_char_len, next_char_len, buf_prev_char_len)
(prev_char_len, inc_both, dec_both): New inline functions,
replacing the old character.h macros FETCH_CHAR_ADVANCE,
FETCH_CHAR_ADVANCE_NO_CHECK, BUF_INC_POS, INC_POS, BUF_DEC_POS,
DEC_POS, INC_BOTH, DEC_BOTH respectively. All callers changed.
These new functions all assume buffer primitives and so need
to be here rather than in character.h.
* src/casefiddle.c (make_char_unibyte): New static function,
replacing the old MAKE_CHAR_UNIBYTE macro. All callers changed.
(do_casify_unibyte_string): Use SINGLE_BYTE_CHAR_P instead
of open-coding it.
* src/ccl.c (GET_TRANSLATION_TABLE): New static function,
replacing the old macro of the same name.
* src/character.c (string_char): Omit 2nd arg. 3rd arg can no
longer be NULL. All callers changed.
* src/character.h (SINGLE_BYTE_CHAR_P): Move up.
(MAKE_CHAR_UNIBYTE, MAKE_CHAR_MULTIBYTE, PREV_CHAR_BOUNDARY)
(STRING_CHAR_AND_LENGTH, STRING_CHAR_ADVANCE)
(FETCH_STRING_CHAR_ADVANCE)
(FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
(FETCH_STRING_CHAR_ADVANCE_NO_CHECK, FETCH_CHAR_ADVANCE)
(FETCH_CHAR_ADVANCE_NO_CHECK, INC_POS, DEC_POS, INC_BOTH)
(DEC_BOTH, BUF_INC_POS, BUF_DEC_POS): Remove.
(make_char_multibyte): New static function, replacing
the old macro MAKE_CHAR_MULTIBYTE. All callers changed.
(CHAR_STRING_ADVANCE): Remove; all callers changed to use
CHAR_STRING.
(NEXT_CHAR_BOUNDARY): Remove; it was unused.
(raw_prev_char_len): New inline function, replacing the
old PREV_CHAR_BOUNDARY macro. All callers changed.
(string_char_and_length): New inline function, replacing the
old STRING_CHAR_AND_LENGTH macro. All callers changed.
(STRING_CHAR): Rewrite in terms of string_char_and_length.
(string_char_advance): New inline function, replacing the old
STRING_CHAR_ADVANCE macro. All callers changed.
(fetch_string_char_advance): New inline function, replacing the
old FETCH_STRING_CHAR_ADVANCE macro. All callers changed.
(fetch_string_char_as_multibyte_advance): New inline function,
replacing the old FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE macro.
All callers changed.
(fetch_string_char_advance_no_check): New inline function,
replacing the old FETCH_STRING_CHAR_ADVANCE_NO_CHECK macro. All
callers changed.
* src/regex-emacs.c (HEAD_ADDR_VSTRING): Remove; no longer used.
* src/syntax.c (scan_lists): Use dec_bytepos instead of
open-coding it.
* src/xdisp.c (string_char_and_length): Rename from
string_char_and_length to avoid name conflict with new function in
character.h. All callers changed.
When 'load-dangerous-libraries' was t, Emacs allowed loading .elc
files compiled by XEmacs. This patch removes the support for that use
case, and declares the variable obsolete.
* lisp/subr.el (load-dangerous-libraries): Declare obsolete.
* src/lread.c (Fload): Ignore its value, and thereby refuse to load
files byte compiled by XEmacs.
(syms_of_lread): Update doc string of 'bytecomp-version-regexp' to not
refer to it.
* doc/emacs/building.texi (Lisp Libraries): Remove its documentation.
* src/lread.c (read1): Guard against two 'struct Lisp_Vector *'
pointers differing only in their most significant bit. Problem
reported by Pip Cet (Bug#39529#22).
Problem reported by Federico Tedin (Bug#39529).
The problem was that dumping uses a hash table based on 'equal'
when purecopying compiled objects, but then modifies the compiled
objects while they are keys in the table. This no-no was uncovered
by the sxhash fixes in 2020-01-07T19:23:11Z!eggert@cs.ucla.edu.
Eli Zaretski pinpointed the patch that triggered the bug.
* src/lread.c (read1): When reading a compiled object, replace
its docstring with a unique negative integer instead of with 0,
so that purecopy doesn’t unify it with some other compiled object
that happens to have the same Lisp code.
(new_backquote_flag): Delete variable.
(load_error_old_style_backquotes): Delete function.
(force_new_style_backquotes): Delete variable.
(read_internal_start): Don't obey it any more.
Also, update description of debuginfo and sources to match
Debian and Red Hat more accurately, and move this sad tale
from INSTALL to etc/PROBLEMS which is a better home for it.
* Makefile.in (emacs_srcdir, install-c-src):
* configure.ac (emacs_srcdir, --with-install-srcdir):
* src/epaths.in (PATH_EMACS_SOURCE):
* src/lread.c (emacs-source-directory):
Remove. All uses removed.
* lisp/emacs-lisp/find-func.el (find-function-C-source):
Do not worry about compressed C sources.
* src/lread.c: Do not include <dosname.h>.
Without this change, on typical GNU/Linux distributions
like Debian, the first button of ‘C-h f car RET’ does not work
because the source code for ‘car’ is not installed (Bug#37527).
Fix this by installing the (compressed) C source code alongside
the (compressed) Lisp source code that is already installed.
This adds about 3 MB (about 2%) to the size of the installed files
on my platform.
* Makefile.in (emacs_srcdir): New macro.
(epaths-force): Substitute PATH_EMACS_SOURCE.
(install-c-src): New rule, that installs a copy of the C source
code if emacs_srcdir says to.
(install-arch-indep): Depend on it.
* configure.ac (emacs_srcdir): New var.
Add support for --disable-install-srcdir.
* lisp/emacs-lisp/find-func.el (find-function-C-source-directory):
Look in emacs-source-directory first.
(find-function-C-source): Also look for gzipped source files.
* lisp/startup.el (normal-top-level):
Also recode emacs-source-directory.
* src/epaths.in (PATH_EMACS_SOURCE): New macro.
* src/lread.c: Include dosname.h, for IS_ABSOLUTE_FILE_NAME.
(syms_of_lread): New var emacs-source-directory.
Quit or retry if fstat, lstat, stat or openat fail with EINTR.
This should fix some bugs on platforms where accessing files via
NFS can fail that way (Bug#9256).
* src/dired.c (file_attributes):
* src/fileio.c (file_directory_p) [O_PATH]:
Use emacs_openat instead of openat.
* src/dired.c (file_attributes): Use emacs_fstatat instead of fstatat.
* src/fileio.c (barf_or_query_if_file_exists, Frename_file):
* src/filelock.c (rename_lock_file):
Use emacs_fstatat instead of lstat.
* src/fileio.c (file_directory_p, Ffile_regular_p, Ffile_modes)
(Ffile_newer_than_file_p, Fverify_visited_file_modtime)
(Fset_visited_file_modtime, auto_save_1):
* src/lread.c (Fload):
* src/sysdep.c (get_current_dir_name_or_unreachable):
Use emacs_fstatat instead of stat.
* src/sysdep.c (emacs_fstatat, emacs_openat): New functions.
(emacs_open): Redo in terms of emacs_open.
186152ba40 Pacify gcc -Wunused-function on Ubuntu 18.04.3
4cd143aded Fix copyright years by hand
365e01cc9f Update copyright year to 2020
cd2c156163 ; * etc/NEWS: Make the description of XDG fallback more ac...
# Conflicts:
# etc/NEWS
# etc/refcards/ru-refcard.tex
On macOS, shared libraries typically have the suffix .dylib. This
commit switches the module suffix to .dylib on Darwin to account for
that. To also support the .so suffix, introduce the concept of a
secondary module suffix.
* configure.ac: Switch MODULES_SUFFIX to .dylib for Darwin, introduce
MODULES_SECONDARY_SUFFIX.
* src/lread.c (Fload, syms_of_lread): Also use
MODULES_SECONDARY_SUFFIX if defined.
* test/src/emacs-module-tests.el (module-darwin-secondary-suffix): New
unit test.
Problem reported by Andreas Schwab (Bug#37475).
* doc/lispref/files.texi (Writing to Files)
(Testing Accessibility, Kinds of Files):
Document that accessibility and file-type predicates return nil
if there is trouble determining accessibility or type.
* etc/NEWS: Adjust, and list the affected primitives.
* src/callproc.c (init_callproc): Go back to Ffile_exists_p.
* src/fileio.c (PICKY_EACCES, file_test_errno):
Remove. All uses removed.
(Ffile_name_case_insensitive_p, Ffile_exists_p, Ffile_symlink_p)
(Ffile_directory_p, Ffile_regular_p): Document that these
functions return nil if there is trouble.
(Ffile_name_case_insensitive_p, check_file_access)
(Ffile_writable_p, Ffile_symlink_p, Ffile_directory_p)
(Ffile_accessible_directory_p, Ffile_regular_p)
* src/lread.c (Fload):
Go back to treating trouble in determining the answer as if the
file were missing.
* src/fileio.c (Ffile_newer_than_file_p): Use file_attribute_errno
not file_test_errno, since returning nil is not appropriate when
there are two files to test; e.g., in the rare cases where both
file timestamps have overflowed then neither t nor nil is correct.
Problem reported by Tino Calancha (Bug#37445).
* src/emacs.c (init_cmdargs): Call file-truename only if
needed, i.e., if invocation-directory ends in "/i386/" on
WINDOWSNT.
* src/lread.c (readevalloop): If the sourcename is not
absolute, make it absolute. There is no need to convert
non-absolute files into truenames, since absolute files are
not converted into truenames.
(init_lread): Do not convert source-directory into a truename
at startup. There is no need to do so in a dumped Emacs since
an absolute file name suffices. The source directory might
not even exist any more, or might have been replaced by an
interloper who takes advantage of the truename calculation.
(syms_of_lread): Remove Qfile_truename; no longer needed.
* src/fileio.c (file_test_errno): Now extern.
* src/lread.c (Fload): Use file_test_errno instead,
since this is really just a file test (the attributes
are not given to the user).
Signal an error for file-oriented errors that are not tame
errors like ENOENT and ENOTDIR (Bug#37389).
Do this for primitives exposed to Lisp; the lower
level internal C API merely makes errno values available
to higher-level C code.
* doc/lispref/files.texi (Testing Accessibility)
(File Attributes, Extended Attributes): Do not say that the
functions return nil when the return value cannot be determined.
* etc/NEWS: Mention the change.
* src/dired.c (Ffile_attributes): Fix doc string confusion
about opening a file vs getting its attributes.
(file_attributes): Signal serious errors.
* src/fileio.c (check_existing, check_executable)
(check_writable): Remove. All callers changed to use
check_file_access or file_access_p.
(file_access_p, file_metadata_errno, file_attribute_errno)
(file_test_errno, check_file_access, check_emacs_readlinkat):
New functions.
* src/fileio.c (Ffile_executable_p, Ffile_readable_p)
(Ffile_name_case_insensitive_p, Frename_file, Ffile_exists_p):
(Ffile_symlink_p, Ffile_directory_p)
(Ffile_accessible_directory_p, Ffile_regular_p)
(Ffile_selinux_context, Ffile_acl, Ffile_modes)
(Ffile_newer_than_file_p, Fset_visited_file_modtime)
(Ffile_system_info):
* src/filelock.c (unlock_file, Ffile_locked_p):
* src/lread.c (Fload):
Signal serious errors.
* src/fileio.c (Ffile_writable_p): Remove unnecessary CHECK_STRING.
(emacs_readlinkat): Now static.
* src/filelock.c (current_lock_owner, lock_if_free): Return a
positive errno on error, and the negative of the old old value
on success. All callers changed.
* src/lread.c (openp): Propagate serious errno values to caller.
Problem for ‘read’ reported by Pip Cet in:
https://lists.gnu.org/r/emacs-devel/2019-08/msg00316.html
* src/fns.c (Frequire): Protect against circular current-load-list.
* src/lread.c (Fget_load_suffixes):
Protect against circular load-suffixes or load-file-rep-suffixes.
(Fload): Protect against circular loads-in-progress.
(openp): Protect against circular PATH and SUFFIXES.
(build_load_history): Protect against circular load-history or
current-load-list.
(readevalloop_eager_expand_eval): Protect against circular SUBFORMS.
(read1): Protect against circular data.
* test/src/lread-tests.el (lread-circular-hash): New test.
* doc/lispref/commands.texi (Reading One Event): Mention that ""
has the same effect as nil as a prompt.
* src/lread.c (Fread_event, Fread_char_exclusive, Fread_char):
Mention what happens when PROMPT is nil/"" in the doc string
(bug#15012).
I've seen segfaults where `infile` is nil when we get to
readbyte_from_file, presumably because Fload set it to NULL (via
close_infile_unwind) just before returning to its caller which was
probably itself within another read/load and for some reason
readevalloop didn't get to re-set `infile` like it used to do at every
iteration. I was not able to really track down the bug, but the way
`infile` was set/reset seemed fragile and managing it like a standard
dynamically-scoped var seems both safer (and more efficient since we
don't need readevalloop to constantly re-set it).
(readchar): Assert that `infile` is set if using a function the depends on it.
(readbyte_from_file): Assert that `infile` is set.
(close_infile_unwind): Reset `infile` to its previous value rather than
to NULL.
(Fload): Remember the previous value of `infile` before chaning it.
(readevalloop): Don't set `infile` any more.
* src/fns.c (cmpfn_eql, cmpfn_equal, cmpfn_user_defined)
(hashfn_eq, hashfn_equal, hashfn_eql, hashfn_user_defined):
* src/profiler.c (cmpfn_profiler, hashfn_profiler):
Use new calling convention where the return value is a fixnum
instead of EMACS_UINT. While we’re at it, put the hash table
at the end, since that’s a bit simpler and generates better
code (at least on the x86-64). All callers changed.
* src/fns.c (hash_lookup): Store fixnum rather than EMACS_UINT.
All callers changed.
(hash_put): Take a fixnum rather than an EMACS_UINT.
All callers changed. Remove unnecessary eassert (XUFIXNUM does it).
* src/lisp.h (struct hash_table_test):
Adjust signatures of cmpfn and hashfn.
Without this patch, (capitalize "x") can alter the match data,
which is not what users expect. Problem found by running
morse-tests-unnato-region in a stripped-down Emacs.
Perhaps ‘load’ should also save and restore the match data?
That would be a simpler fix, though arguably incompatible.
* src/lread.c (save_match_data_load): New function.
* src/chartab.c (uniprop_table):
* src/doc.c (reread_doc_file):
* src/eval.c (Fautoload_do_load):
* src/fns.c (Frequire): Use it.
This improves the performance of (read "%xFF") by about 25%
on my platform.
* src/lread.c: Include <vla.h>, so that we can better document
buffer sizes of arguments.
(invalid_radix_integer_format, stackbufsize): New constants.
(free_contents): Remove. All uses removed.
(invalid_radix_integer): New function.
(read_integer): New arg STACKBUF. Assume radix is in range.
All uses changed. Use STACKBUF to avoid calling malloc in the
usual case. Use grow_read_buffer to simplify.
(read1): Tune. Improve quality of diagnostic when
MOST_POSITIVE_FIXNUM < radix <= EMACS_INT_MAX.
This compiler is a bit pickier about checking conformance to
the C standard, ranging from syntax trivia (no extra ";" at
the top level) to portability trivia (warnings re conversion
between function and data pointers) to more-important stuff
like lack of support for some __attribute__ usages.
* src/dynlib.c (dynlib_addr): First argument is a function
pointer, not a data pointer. All callers changed.
* src/emacs-module.c (module_function_address):
Return module_funcptr, not void *. All uses changed.
* src/lisp.h (module_funcptr) [HAVE_MODULES]: New type.
* src/lread.c (union ieee754_double): Don’t assume the usual
semantics for converting signed to unsigned int when initializing
a bitfield, as the Oracle compiler complains and the C standard
is unclear.
* src/pdumper.c (ALLOW_IMPLICIT_CONVERSION): Make it clearer
that -Wsign-conversion is disabled everywhere in this file.
(dump_trace, dump_tailq_prepend, dump_tailq_append):
Don’t assume __attribute__.
(dump_object_self_representing_p): Don’t disable conversion
warnings; it’s not needed here.
(DEFINE_FROMLISP_FUNC): Avoid possible signal in integer
conversion from unsigned to signed.
(DEFINE_FROMLISP_FUNC, finish_dump_pvec): Avoid warning about
unreachable statements on platforms not supporting the
__attribute__.
(intmax_t_from_lisp, intmax_t_to_lisp, dump_off_from_lisp)
(dump_off_to_lisp, dump_emacs_reloc_immediate_lv)
(dump_emacs_reloc_immediate_ptrdiff_t)
(dump_emacs_reloc_immediate_intmax_t)
(dump_emacs_reloc_immediate_int, dump_emacs_reloc_immediate_bool):
Omit stray semicolon that violates C standard.
(dump_metadata_for_pdumper): Add cast to pacify compiler complaining
about conversion from function pointer to data pointer.
(Fdump_emacs_portable): Do not use CALLN to call a function
with zero arguments, as C99 prohibits empty initializers.
* src/xdisp.c (syms_of_xdisp): Do not nest calls to pure_list,
to work around a bug in Oracle Developer Studio 12.6.
See Bug#31676.
* lisp/emacs-lisp/byte-run.el
(byte-run--unescaped-character-literals-warning): New defun.
* src/lread.c (load_warn_unescaped_character_literals): Use new defun.
(syms_of_lread): Define symbol for new defun.
* lisp/emacs-lisp/bytecomp.el (byte-compile-from-buffer): Use new
defun.
* test/src/lread-tests.el (lread-tests--unescaped-char-literals):
test/lisp/emacs-lisp/bytecomp-tests.el
(bytecomp-tests--unescaped-char-literals): Adapt unit tests.
