* src/xdisp.c (get_large_narrowing_begv, get_large_narrowing_zv)
(get_medium_narrowing_begv, get_medium_narrowing_zv): Use
'ptrdiff_t' instead of 'int', to prevent integer overflow in
large buffers. (Bug#72497)
This fixes bugs that crashed Emacs when the Lisp interpreter was fed
bad code.
* src/eval.c (FletX, Flet, internal_lisp_condition_case)
(funcall_lambda): Hoist symbol-with-pos elimination and type checks to a
dominating position for efficiency. This also plugs at least two typing
holes. (Mea culpa.)
* test/src/eval-tests.el (eval-bad-specbind): New regression test.
This reverts commit daefd6771a.
* src/sfnt.c (sfnt_read_OS_2_table): Restore realignment after
s_family_class, as it occupies byte 32, not 34. Reported by Pip
Cet <pipcet@protonmail.com>.
* src/image.c (struct image_type): Minor grammatical
corrections.
(image_destroy_x_image): [HAVE_NS]: Do not release
Emacs_Pix_Containers, which are identical to Emacs_Pixmaps and
consequently always released with the `struct image'.
(bug#72255)
* src/treesit.c (Ftreesit_parser_included_ranges)
(Ftreesit_query_capture, treesit_traverse_sibling_helper)
(treesit_traverse_match_predicate): Fix punctiation and documentation
style in comments and docstrings.
* src/pgtkfns.c (parse_resource_key): Avoid overflow by making array
larger, if a key is RESOURCE_KEY_MAX_LEN long. Do not merge to master,
since it's fixed in a different way there.
* doc/emacs/android.texi (Android Environment): Adjust
documentation to match.
* java/org/gnu/emacs/EmacsNoninteractive.java (main1): New
function. Remove initialization of EmacsNative hither.
(main): Acquire an ApplicationInfo or LoadedApk, as the case may
be on the host system, derive a ClassLoader from the result, and
load and call `main1' from within this class loader.
* src/android-emacs.c (main):
* src/android.c (setEmacsParams): Do not override
LD_LIBRARY_PATH or set EMACS_LD_LIBRARY_PATH. This enables
Emacs to execute subprocesses in certain "fortified" Android
systems, amongst other things.
* src/androidfns.c (Fx_display_mm_width, Fx_display_mm_height)
(Fandroid_display_monitor_attributes_list)
(Fandroid_external_storage_available_p)
(Fandroid_request_storage_access): Verify that a display
connection or service object is available.
* src/androidselect.c (Fandroid_get_clipboard)
(Fandroid_browse_url_internal, Fandroid_get_clipboard_targets)
(Fandroid_get_clipboard_data, Fandroid_notifications_notify):
Moderate tone of error messages.
* src/image.c (free_image_cache): Unconditionally release image
cache, as this function is only called with its existence
already established.
* src/xfaces.c (free_frame_faces): Clear FRAME_IMAGE_CACHE (f).
(bug#71929)
* src/android-emacs.c (main): Do not attempt to load the
bootstrap class path, which is redundant on all Android releases.
* src/android.c (initEmacs, android_browse_url): Do not assume
exceptions will be raised if GetStringUTFChars fails. Decode
Android JNI strings as Qandroid_jni.
* src/androidvfs.c (android_saf_check_nonnull): New function.
(android_saf_new_mkdir): Likewise.
Backport of commit 1b595b4598e7 since it turns out this "minor typo"
had disastrous consequences in pathological cases and explains
some of the problems seen in bug#71644 and bug#63040.
* src/marker.c (buf_bytepos_to_charpos): Don't compare byte-positions
with char-positions.
* src/w32.c (globals_of_w32): Move re-initialization of
Vlibrary_cache from here...
* src/emacs.c (main): ...to here, as it must be after load_pdump.
(Bug#71916)
* configure.ac (W32_LIBS): Don't link with -lusp10 on non-Cygwin
systems.
* src/emacs.c (main): Call globals_of_w32 before the startup
directory is initialized.
* src/w32.c (maybe_load_unicows_dll): Call
load_unicows_dll_for_w32fns.
* src/w32.h: Update prototypes.
* src/w32fns.c (Fx_create_frame, w32_create_tip_frame): Do not
register the Uniscribe font driver when unavailable.
(pfnSHFileOperationW): New function pointer.
(Fsystem_move_file_to_trash): Load UNICOWS.DLL if not yet
loaded. Call SHFileOperationW through said function pointer.
(pfnShellExecuteExW): New function pointer.
(Fw32_shell_execute) [!CYGWIN]: Load UNICOWS.DLL if not yet
loaded. Call ShellExecuteExW through said function pointer.
(pfnShell_NotifyIconW): New function pointer.
(add_tray_notification, delete_tray_notification): Call
Shell_NotifyIconW through said function pointer.
(Fw32_notification_notify): Load UNICOWS.DLL.
(Fw32_notification_close): Return if Shell_NotifyIconW is
unavailable, as when UNICOWS.DLL has yet to be loaded.
(load_unicows_dll_for_w32fns): New function.
* src/w32notify.c (pfnReadDirectoryChangesW): New function
pointer.
(watch_completion, remove_watch, Fw32notify_add_watch)
(Fw32notify_rm_watch, Fw32notify_valid_p): Call
ReadDirectoryChangesW through said function pointer, and assert
its presence.
(globals_of_w32notify): Load ReadDirectoryChangesW from
KERNEL32.DLL.
* src/w32uniscribe.c (pfnScriptItemize, pfnScriptShape)
(pfnScriptPlace, pfnScriptGetGlyphABCWidth, pfnScriptFreeCache)
(pfnScriptGetCMap): New function pointers.
(uniscribe_close, uniscribe_shape, uniscribe_encode_char)
(uniscribe_check_otf_1): Call Uniscribe functions through the
same.
(syms_of_w32uniscribe_for_pdumper): Load Uniscribe library and
required functions from the same, and if unavailable, return
while leaving uniscribe_available intact. On Cygwin, simply
assign USP10.DLL functions to the said new function pointers.
To reproduce the problem:
0. emacs -Q
1. eval: (add-to-list 'major-mode-remap-alist '(c-mode . c-ts-mode))
2. C-x v L
3. in the *vc-change-log* buffer move point to the commit 20af58d3a1
4. type D
5. crash caused by diff-font-lock-syntax fontification that uses treesit
Emacs: 6f2036243f (2024-06-23, latest master)
Tree-sitter: 3da7deed (2024-06-08, version 0.22.6)
The immediate cause of the crash is that tree-sitter accessed a node's
tree, but the tree is already deleted. Commenting out the
ts_tree_delete line in treesit_ensure_parsed can "fix" the crash.
What happended, I think, is this:
1. Buffer modified, parser->need_reparse set to true,
parser->timestamp incremented.
2. A node is created from the parser, this node has the old tree but
the new timestamp (bad!).
3. Parser re-parses (treesit_ensure_parsed), new tree created, old
tree deleted.
4. Ftreesit_query_capture accessed the old node, and the old tree,
crash.
We shouldn't bump the parser timestamp when we set
parser->need_reparse to true; instead, we should bump the timestamp
when we actually reparsed and created a new tree.
* src/treesit.c (treesit_record_change): Don't bump parser timestamp.
(treesit_sync_visible_region): Don't bump parser timestamp.
(Ftreesit_parser_set_included_ranges): Don't bump parser timestamp.
(treesit_ensure_parsed): Bump parser timestamp.
(Ftreesit_query_capture): Add node check.
* src/xfaces.c (free_realized_face): Always free realized face
structures, and avoid sending X requests when the display is not
available, not when it is.
