Problem reported by Juanma Barranquero in:
https://lists.gnu.org/r/emacs-devel/2020-08/msg00279.html
and a similar glitch was reported by Eli Zaretskii (Bug#36597#67).
* src/pdumper.c: Remove -Wconversion pragma.
(ALLOW_IMPLICIT_CONVERSION, DISALLOW_IMPLICIT_CONVERSION):
Remove. All uses removed. Although -Wconversion may have
been helpful when writing pdumper.c it is now causing more
trouble than it’s worth here (just as in the rest of Emacs).
(dump_read_all): Avoid no-longer-necessary use of ‘size_t’
rather than ‘int’.
Commit 16a16645f5 has only changed a
comment in ‘struct Lisp_Hash_Table’, so the portable dumper doesn’t
need to be adapted.
* src/pdumper.c (dump_hash_table): Update hash code for ‘struct
Lisp_Hash_Table’.
* src/pdumper.c (hash_table_thaw): Pacify -Wconversion so
we can use make_nil_vector again.
* src/timefns.c (syms_of_timefns): Prefer make_nil_vector
to make_vector with Qnil.
* src/pdumper.c (hash_table_thaw): Use Fmake_vector. Suggested by
Pip Cet <pipcet@gmail.com>.
(dump_trace): Declare ATTRIBUTE_FORMAT_PRINTF, not
ATTRIBUTE_FORMAT((__printf__), so that we pick the right attribute
for MinGW.
* src/pdumper.c (dump_queue_empty_p): Avoid unnecessary call
to Fhash_table_count on a known hash table.
(dump_hash_table_list): !NILP, not CONSP.
(hash_table_freeze, hash_table_thaw): ASIZE, not Flength, on vectors.
Initialize in same order as struct.
(hash_table_thaw): make_nil_vector, not Fmake_vector with nil.
* src/pdumper.c (PRIdDUMP_OFF): New macro.
(EMACS_INT_XDIGITS): New constant.
(struct dump_context): Use dump_off for relocation counts.
All uses changed.
(dump_queue_enqueue, dump_queue_dequeue, Fdump_emacs_portable):
Don’t assume counts fit in ‘unsigned’ or ‘unsigned long’.
Use EMACS_INT_XDIGITS instead of assuming it’s 16.
This simplifies code, and helps performance in some cases (Bug#36597).
* src/lisp.h (hash_rehash_needed_p): Remove. All uses removed.
(hash_rehash_if_needed): Remove. All uses removed.
(struct Lisp_Hash_Table): Remove comment about rehashing hash tables.
* src/pdumper.c (thaw_hash_tables): New function.
(hash_table_thaw): New function.
(hash_table_freeze): New function.
(dump_hash_table): Simplify.
(dump_hash_table_list): New function.
(hash_table_contents): New function.
(Fdump_emacs_portable): Handle hash tables by eager rehashing.
(pdumper_load): Restore hash tables.
(init_pdumper_once): New function.
* src/image.c (image_set_transform [HAVE_XRENDER]): Use different filter
when scaling up vs scaling down.
* src/nsimage.m (ns_image_set_smoothing):
([EmacsImage setSmoothing:]): New functions.
* src/nsterm.h: Add definitions.
* src/nsterm.m (ns_dumpglyphs_image): Disable smoothing if requested.
GCC has removed the -fcheck-pointer bounds option, and the Linux
kernel has also removed support for Intel MPX, so there’s no point
to keeping this debugging option within Emacs.
* src/bytecode.c (BYTE_CODE_THREADED):
* src/lisp.h (DEFINE_LISP_SYMBOL, XSYMBOL, make_lisp_symbol):
Assume __CHKP__ is not defined.
* src/ptr-bounds.h: Remove. All uses of ptr_bounds_clip,
ptr_bounds_copy, ptr_bounds_init, ptr_bounds_set removed.
* src/xdisp.c (with_echo_area_buffer): Pass void * instead of
ptrdiff_t, since the values are typically pointers and this ports
better to (mostly-theoretical) hosts where ptrdiff_t is narrower
than intptr_t. All uses changed.
* configure.ac: Use AC_CHECK_FUNCS_ONCE for __lsan_ignore_object.
* src/buffer.c, src/data.c, src/emacs-module.c, src/regex-emacs.c:
* src/search.c: Use __lsan_ignore_object unconditionally, and don’t
include sanitizer/lsan_interface.h.
* src/lisp.h (__lsan_ignore_object): Provide a dummy in the
typical case where leak sanitization is not available.
* src/alloc.c (mark_maybe_object):
Use simpler way to avoid -fsanitize=undefined false alarms,
by converting the word tag to intptr_t first.
Omit now-unnecessary runtime overflow check.
(mark_memory): Work even if UINTPTR_MAX <= INT_MAX (!).
* src/process.c (wait_reading_process_output): Before the select,
check every interesting gnutls stream for available data in the
buffer. If some of them hit, and either there is no wait_proc or
the wait_proc is one of the gnutls streams with new data, set the
select timeout to 0 after the select, and merge the gnutls buffer
status into the select returns (bug#40665). This fixes a problem
where TLS connections would sometimes hang.
Emacs segfaults at the X11 initialization if XIM is set
and xim_styles is NULL. This patch avoids the crash.
* src/xfns.c: Check also if FRAME_X_XIM_STYLES(f) is NULL.
(Bug#42676) (Bug#42673) (Bug#42677)
Copyright-paperwork-exempt: yes
* src/xdisp.c (try_window): Fix logic of disabling margins when
cursor is close to BOB or EOB. Account for header-line, if any,
when computing the scroll margin in pixels. (Bug#42653)
We perform weird pointer arithmetic due to the layout of Lisp_Objects
holding symbols. ASan/UBSan warns about that (Bug#42530). Suppress
the warnings by performing the arithmetic on integer types and casting
back to pointers.
* src/alloc.c (mark_maybe_object, mark_memory): Temporarily cast
pointer to 'intptr_t'.
* src/emacs-module.c (initialize_environment): Call
__lsan_ignore_object only if HAVE_SANITIZER_LSAN_INTERFACE_H is
undefined. This fixes compilation on systems that don't have
__lsan_* functions.
We intentionally leak some objects. Prevent the ASan leak detector
from raising false alarms in these cases.
* configure.ac: Search for lsan_interface.h header.
* src/data.c (make_blv): Allow leaking of buffer-local values.
* src/buffer.c (enlarge_buffer_text): Allow leaking of buffer text.
* src/emacs-module.c (Fmodule_load, initialize_environment): Allow
intentional leak of runtime and environment objects if module
assertions are enabled.
We can't just use a hash lookup because a global and a local reference
might refer to the same Lisp object.
* src/emacs-module.c (module_free_global_ref): More precise check for
global liveness.
(cherry picked from commit 9f01ce6327)
We can't simply look up the Lisp object in the global reference table
because an invalid local and a valid global reference might refer to
the same object. Instead, we have to test the address of the global
reference against the stored references.
* src/emacs-module.c (module_global_reference_p): New helper function.
(value_to_lisp): Use it.
(cherry picked from commit 6355a3ec62)
Instead of storing the global values in a global 'emacs_value_storage'
object, store them as hash values alongside the reference counts.
That way the garbage collector takes care of cleaning them up.
* src/emacs-module.c (global_storage): Remove.
(struct module_global_reference): New pseudovector type.
(XMODULE_GLOBAL_REFERENCE): New helper function.
(module_make_global_ref, module_free_global_ref): Use
'module_global_reference' struct for global reference values.
(value_to_lisp, module_handle_nonlocal_exit): Adapt to deletion of
'global_storage'.
(cherry picked from commit 5c5eb97908)
This incorporates:
2020-07-30 work around some Oracle Studio attribute bugs
2020-07-29 fsusage, regex, stat-size: remove Cray support
2020-07-26 inttypes: remove support for AIX 4
2020-07-26 gettimeofday: remove workaround for Mac OS X 10.0
2020-07-26 don't require gl_LOCALTIME_BUFFER_DEFAULTS
2020-07-26 alloca: remove Cray-2 and Cray Y-MP support
2020-07-26 libgmp: remove dependency on havelib
2020-07-26 libgmp: remove HAVE_GMP, LIB_GMP
2020-07-25 multiarch: prepare for x86_64+arm64 universal in macOS 11
2020-07-25 sigprocmask: small autoconf macro improvement
2020-07-25 small autoconf macro improvements
2020-07-24 timespec: remove dependence on ‘verify’
2020-07-24 optimize a few more three-valued comparisons
2020-07-24 fix _GL_CMP parenthesization typo
2020-07-23 optimize three-valued comparison between integers
2020-07-24 doc: update for Mac OS X 10.13
2020-07-23 fchmodat, lchmod: use /proc on Cygwin
2020-07-21 inttypes: fix PRI*PTR and SCN*PTR on 64-bit native Windows
2020-07-12 libgmp: avoid warning when --without-libgmp is used
2020-07-12 libgmp: link to the correct shared library
* lib/mini-gmp-gnulib.c: Ignore -Wsuggest-attribute=malloc only for
* build-aux/config.guess, build-aux/config.sub:
* build-aux/install-sh, doc/misc/texinfo.tex, lib/c-strcasecmp.c:
* lib/c-strncasecmp.c, lib/fchmodat.c, lib/fsusage.c:
* lib/gettimeofday.c, lib/inttypes.in.h, lib/lchmod.c:
* lib/mini-gmp-gnulib.c, lib/nstrftime.c, lib/regex.h, lib/timespec.h:
* m4/alloca.m4, m4/getgroups.m4, m4/gettimeofday.m4:
* m4/gnulib-common.m4, m4/inttypes.m4, m4/libgmp.m4, m4/mktime.m4:
* m4/multiarch.m4:
Copy from Gnulib.
* lib/gnulib.mk.in, m4/gnulib-comp.m4: Regenerate.
* src/Makefile.in, test/Makefile.in (LIBGMP):
Rename from LIB_GMP for compatibility with Gnulib.
All uses changed.
* src/alloc.c (__builtin_unwind_init) [!HAVE___BUILTIN_UNWIND_INIT]:
Move from here ...
* src/lisp.h: ... to here, since flush_stack_call_func uses it.
* src/pdumper.c (dump_off_from_lisp): Avoid ‘return n;;’ to pacify
Oracle Studio.
* src/xdisp.c (face_at_pos): Pass ATTR_FILTER to
face_for_overlay_string.
* src/xfaces.c (face_for_overlay_string): Accept an additional
argument ATTR_INDEX and pass it to merge_face_ref for merging the
face at POS. This ensures a face from buffer text will not be
merged unless it specifies the :extend attribute. (Bug#42552)
* src/dispextern.h (face_for_overlay_string): Adjust prototype.
4b3085a7fe Fix last change
efdd4632c9 Fix Arabic shaping when column-number-mode is in effect
d5acc50941 Fix description of kmacro-* commands in the user manual
SELinux has used 'char *' in place of its typedef
'security_context_t' since 2014 (v2.3) because the latter was
inconvenient to use when paired with the 'const' qualifier. The
typedef has been kept around for compatibility with legacy callers,
but it is deprecated in SELinux v3.1. See the following URL for the
relevant announcement:
https://lore.kernel.org/selinux/20200710162034.GC1768200@localhost.localdomain/
* src/fileio.c (Fcopy_file, Ffile_selinux_context)
(Fset_file_selinux_context) [HAVE_LIBSELINUX]: Replace deprecated
'security_context_t' typedef with the equivalent 'char *'.
We can't just use a hash lookup because a global and a local reference
might refer to the same Lisp object.
* src/emacs-module.c (module_free_global_ref): More precise check for
global liveness.
* test/data/emacs-module/mod-test.c (Fmod_test_globref_invalid_free):
New test module function.
(emacs_module_init): Export it.
* test/src/emacs-module-tests.el
(module--test-assertions--globref-invalid-free): New unit test.
We can't simply look up the Lisp object in the global reference table
because an invalid local and a valid global reference might refer to
the same object. Instead, we have to test the address of the global
reference against the stored references.
* src/emacs-module.c (module_global_reference_p): New helper function.
(value_to_lisp): Use it.
* test/data/emacs-module/mod-test.c
(Fmod_test_invalid_store_copy): New test module function.
(emacs_module_init): Export it.
* test/src/emacs-module-tests.el
(module--test-assertions--load-non-live-object-with-global-copy):
New unit test.
* src/indent.c (scan_for_column, compute_motion): Pass -1,
instead of NEUTRAL_DIR, to 'composition_reseat_it'.
* src/composite.c (composition_reseat_it): Interpret negative
value of BIDI_LEVEL to mean the caller doesn't know what is the
bidi direction of the text. (Bug#41005)
