GNU bug report logs - #75593
31.0.50; Faulty macro kills Emacs

Previous Next

To reply to this bug, email your comments to 75593 AT debbugs.gnu.org.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Alexander Prähauser <ahprae <at> protonmail.com>
To: bug-gnu-emacs <at> gnu.org
Subject: 31.0.50; Faulty macro kills Emacs
Date: Wed, 15 Jan 2025 19:00:01 +0000

[Message part 1 (text/plain, inline)]

A macro I've defined in causes Emacs to crash. The macro is faulty but I
still don't think it should do that. I've attached a file with a
functioning version of the macro, the version that crashes with comments
where it is changed and where it crashes and a configuration that causes
it to crash. Careful, evaluating that buffer will kill Emacs. I've run
`emacs -Q' in gdb and added a logging file. The output of `bt-full' is 


#0  0x00005555557537d9 in ??? ()
#1  0x0000000000000560 in ??? ()
#2  0x0000000000000002 in ??? ()
#3  0x0000000000000002 in ??? ()
#4  0x00007fffefb5b66b in ??? ()
#5  0x00007ffff0b592c3 in ??? ()
#6  0x00007ffff0b48b73 in ??? ()
#7  0x00007ffff0b5e593 in ??? ()
#8  0x0000555555727b78 in ??? ()
#9  0x0000000000000000 in ??? ()




In GNU Emacs 31.0.50 (build 1, x86_64-pc-linux-gnu, GTK+ Version
 3.24.41, cairo version 1.18.2) of 2025-01-15 built on gentoolaptop
Repository revision: 16c89c5ae5ec0c002c327793e783f0a943bacb0d
Repository branch: master
Windowing system distributor 'The X.Org Foundation', version 11.0.12101014
System Description: Gentoo Linux

Configured features:
ACL CAIRO DBUS FREETYPE GIF GLIB GMP GNUTLS GPM GSETTINGS HARFBUZZ JPEG
LCMS2 LIBSYSTEMD LIBXML2 MODULES NATIVE_COMP NOTIFY INOTIFY PDUMPER PNG
RSVG SECCOMP SOUND SQLITE3 THREADS TIFF TOOLKIT_SCROLL_BARS TREE_SITTER
WEBP X11 XDBE XIM XINPUT2 XPM GTK3 ZLIB

Important settings:
  value of $LANG: en_US.utf8
  locale-coding-system: utf-8-unix

Major mode: Text

Minor modes in effect:
  icicle-mode: t
  minibuffer-depth-indicate-mode: t
  gdb-many-windows: t
  mixed-pitch-mode: t
  Info-breadcrumbs-in-mode-line-mode: t
  recentf-mode: t
  global-edit-server-edit-mode: t
  global-goto-address-mode: t
  goto-address-mode: t
  auto-insert-mode: t
  dynamic-completion-mode: t
  electric-pair-mode: t
  corfu-history-mode: t
  delete-selection-mode: t
  global-undo-tree-mode: t
  undo-tree-mode: t
  desktop-save-mode: t
  d-emacs-mode: t
  emms-mode-line-mode: t
  emms-playing-time-display-mode: t
  emms-playing-time-mode: t
  rainbow-delimiters-mode: t
  global-org-modern-mode: t
  TeX-PDF-mode: t
  fancy-compilation-mode: t
  eshell-vterm-mode: t
  yas-global-mode: t
  yas-minor-mode: t
  all-the-icons-completion-mode: t
  global-aggressive-indent-mode: t
  org-roam-db-autosync-mode: t
  corfu-prescient-mode: t
  global-corfu-mode: t
  corfu-mode: t
  marginalia-mode: t
  global-git-commit-mode: t
  magit-auto-revert-mode: t
  savehist-mode: t
  smart-mark-mode: t
  windmove-mode: t
  straight-use-package-mode: t
  straight-package-neutering-mode: t
  override-global-mode: t
  tooltip-mode: t
  global-eldoc-mode: t
  eldoc-mode: t
  show-paren-mode: t
  electric-indent-mode: t
  mouse-wheel-mode: t
  global-prettify-symbols-mode: t
  tab-bar-history-mode: t
  tab-bar-mode: t
  file-name-shadow-mode: t
  global-font-lock-mode: t
  font-lock-mode: t
  blink-cursor-mode: t
  undelete-frame-mode: t
  minibuffer-regexp-mode: t
  line-number-mode: t
  global-visual-line-mode: t
  visual-line-mode: t
  transient-mark-mode: t
  auto-composition-mode: t
  auto-encryption-mode: t
  auto-compression-mode: t

Load-path shadows:
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/bookmark+ hides /home/alex/.emacs.d/elpa/bookmark+-20230325.160624/bookmark+
/home/alex/.emacs.d/straight/build/corfu-terminal/corfu-terminal hides /home/alex/.emacs.d/elpa/corfu-terminal-0.7/corfu-terminal
/home/alex/.emacs.d/straight/build/corfu-terminal/corfu-terminal-autoloads hides /home/alex/.emacs.d/elpa/corfu-terminal-0.7/corfu-terminal-autoloads
/home/alex/.emacs.d/straight/build/corfu/corfu hides /home/alex/.emacs.d/elpa/corfu-20250103.2029/corfu
/home/alex/.emacs.d/straight/build/corfu/corfu-quick hides /home/alex/.emacs.d/elpa/corfu-20250103.2029/corfu-quick
/home/alex/.emacs.d/straight/build/corfu/corfu-popupinfo hides /home/alex/.emacs.d/elpa/corfu-20250103.2029/corfu-popupinfo
/home/alex/.emacs.d/straight/build/corfu/corfu-info hides /home/alex/.emacs.d/elpa/corfu-20250103.2029/corfu-info
/home/alex/.emacs.d/straight/build/corfu/corfu-indexed hides /home/alex/.emacs.d/elpa/corfu-20250103.2029/corfu-indexed
/home/alex/.emacs.d/straight/build/corfu/corfu-history hides /home/alex/.emacs.d/elpa/corfu-20250103.2029/corfu-history
/home/alex/.emacs.d/straight/build/corfu/corfu-echo hides /home/alex/.emacs.d/elpa/corfu-20250103.2029/corfu-echo
/home/alex/.emacs.d/straight/build/corfu/corfu-autoloads hides /home/alex/.emacs.d/elpa/corfu-20250103.2029/corfu-autoloads
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/doremi hides /home/alex/.emacs.d/elpa/doremi-20220406.185820/doremi
/home/alex/.emacs.d/straight/build/consult/consult hides /home/alex/.emacs.d/elpa/consult-20250107.927/consult
/home/alex/.emacs.d/straight/build/consult/consult-xref hides /home/alex/.emacs.d/elpa/consult-20250107.927/consult-xref
/home/alex/.emacs.d/straight/build/consult/consult-register hides /home/alex/.emacs.d/elpa/consult-20250107.927/consult-register
/home/alex/.emacs.d/straight/build/consult/consult-org hides /home/alex/.emacs.d/elpa/consult-20250107.927/consult-org
/home/alex/.emacs.d/straight/build/consult/consult-kmacro hides /home/alex/.emacs.d/elpa/consult-20250107.927/consult-kmacro
/home/alex/.emacs.d/straight/build/consult/consult-info hides /home/alex/.emacs.d/elpa/consult-20250107.927/consult-info
/home/alex/.emacs.d/straight/build/consult/consult-imenu hides /home/alex/.emacs.d/elpa/consult-20250107.927/consult-imenu
/home/alex/.emacs.d/straight/build/consult/consult-flymake hides /home/alex/.emacs.d/elpa/consult-20250107.927/consult-flymake
/home/alex/.emacs.d/straight/build/consult/consult-compile hides /home/alex/.emacs.d/elpa/consult-20250107.927/consult-compile
/home/alex/.emacs.d/straight/build/consult/consult-autoloads hides /home/alex/.emacs.d/elpa/consult-20250107.927/consult-autoloads
/home/alex/.emacs.d/elpa/org-projectile-helm-20230817.801/org-projectile-helm hides /home/alex/.emacs.d/elpa/org-projectile-20230817.851/org-projectile-helm
/home/alex/.emacs.d/straight/build/popon/popon hides /home/alex/.emacs.d/elpa/popon-0.13/popon
/home/alex/.emacs.d/straight/build/popon/popon-autoloads hides /home/alex/.emacs.d/elpa/popon-0.13/popon-autoloads
/home/alex/.emacs.d/elpa/transient-20250108.1351/transient hides /usr/local/share/emacs/31.0.50/lisp/transient
/home/alex/.emacs.d/elpa/bind-key-20230203.2004/bind-key hides /usr/local/share/emacs/31.0.50/lisp/bind-key
/home/alex/.emacs.d/elpa/modus-themes-20250107.1034/theme-loaddefs hides /usr/local/share/emacs/31.0.50/lisp/theme-loaddefs
/home/alex/.emacs.d/elpa/use-package-20230426.2324/use-package-lint hides /usr/local/share/emacs/31.0.50/lisp/use-package/use-package-lint
/home/alex/.emacs.d/elpa/use-package-20230426.2324/use-package-jump hides /usr/local/share/emacs/31.0.50/lisp/use-package/use-package-jump
/home/alex/.emacs.d/elpa/use-package-ensure-system-package-20221209.2013/use-package-ensure-system-package hides /usr/local/share/emacs/31.0.50/lisp/use-package/use-package-ensure-system-package
/home/alex/.emacs.d/elpa/use-package-20230426.2324/use-package-ensure hides /usr/local/share/emacs/31.0.50/lisp/use-package/use-package-ensure
/home/alex/.emacs.d/elpa/use-package-20230426.2324/use-package hides /usr/local/share/emacs/31.0.50/lisp/use-package/use-package
/home/alex/.emacs.d/elpa/use-package-20230426.2324/use-package-diminish hides /usr/local/share/emacs/31.0.50/lisp/use-package/use-package-diminish
/home/alex/.emacs.d/elpa/use-package-20230426.2324/use-package-delight hides /usr/local/share/emacs/31.0.50/lisp/use-package/use-package-delight
/home/alex/.emacs.d/elpa/use-package-20230426.2324/use-package-core hides /usr/local/share/emacs/31.0.50/lisp/use-package/use-package-core
/home/alex/.emacs.d/elpa/use-package-20230426.2324/use-package-bind-key hides /usr/local/share/emacs/31.0.50/lisp/use-package/use-package-bind-key
/home/alex/.emacs.d/straight/build/compat/compat hides /usr/local/share/emacs/31.0.50/lisp/emacs-lisp/compat
/home/alex/.emacs.d/straight/build/seq/seq hides /usr/local/share/emacs/31.0.50/lisp/emacs-lisp/seq
/home/alex/.emacs.d/elpa/avy-act-20241226.2351/avy-act hides /home/alex/.emacs.d/site-lisp/avy-act/avy-act
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/hexrgb hides /home/alex/.emacs.d/site-lisp/doremi/hexrgb
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/doremi hides /home/alex/.emacs.d/site-lisp/doremi/doremi
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles-fn hides /home/alex/.emacs.d/site-lisp/icicles/icicles-fn
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/vline hides /home/alex/.emacs.d/site-lisp/icicles/vline
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/thingatpt+ hides /home/alex/.emacs.d/site-lisp/icicles/thingatpt+
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/synonyms hides /home/alex/.emacs.d/site-lisp/icicles/synonyms
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/pp+ hides /home/alex/.emacs.d/site-lisp/icicles/pp+
/home/alex/.emacs.d/site-lisp/palette hides /home/alex/.emacs.d/site-lisp/icicles/palette
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/lacarte hides /home/alex/.emacs.d/site-lisp/icicles/lacarte
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/info+ hides /home/alex/.emacs.d/site-lisp/icicles/info+
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icomplete+ hides /home/alex/.emacs.d/site-lisp/icicles/icomplete+
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles hides /home/alex/.emacs.d/site-lisp/icicles/icicles
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles-var hides /home/alex/.emacs.d/site-lisp/icicles/icicles-var
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles-opt hides /home/alex/.emacs.d/site-lisp/icicles/icicles-opt
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles-mode hides /home/alex/.emacs.d/site-lisp/icicles/icicles-mode
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles-mcmd hides /home/alex/.emacs.d/site-lisp/icicles/icicles-mcmd
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles-mac hides /home/alex/.emacs.d/site-lisp/icicles/icicles-mac
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles-face hides /home/alex/.emacs.d/site-lisp/icicles/icicles-face
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles-doc2 hides /home/alex/.emacs.d/site-lisp/icicles/icicles-doc2
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles-doc1 hides /home/alex/.emacs.d/site-lisp/icicles/icicles-doc1
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles-cmd2 hides /home/alex/.emacs.d/site-lisp/icicles/icicles-cmd2
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles-cmd1 hides /home/alex/.emacs.d/site-lisp/icicles/icicles-cmd1
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/icicles-chg hides /home/alex/.emacs.d/site-lisp/icicles/icicles-chg
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/hl-line+ hides /home/alex/.emacs.d/site-lisp/icicles/hl-line+
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/hexrgb hides /home/alex/.emacs.d/site-lisp/icicles/hexrgb
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/fuzzy-match hides /home/alex/.emacs.d/site-lisp/icicles/fuzzy-match
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/doremi hides /home/alex/.emacs.d/site-lisp/icicles/doremi
/home/alex/.emacs.d/site-lisp/doremi/doremi-frm hides /home/alex/.emacs.d/site-lisp/icicles/doremi-frm
/home/alex/.emacs.d/site-lisp/dired+ hides /home/alex/.emacs.d/site-lisp/icicles/dired+
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/crosshairs hides /home/alex/.emacs.d/site-lisp/icicles/crosshairs
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/col-highlight hides /home/alex/.emacs.d/site-lisp/icicles/col-highlight
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/bookmark+ hides /home/alex/.emacs.d/site-lisp/icicles/bookmark+
/home/alex/.emacs.d/elpa/Icicles-20241217.3924/apropos-fn+var hides /home/alex/.emacs.d/site-lisp/icicles/apropos-fn+var
/home/alex/.emacs.d/site-lisp/d-emacs/d-emacs hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/d-emacs
/home/alex/.emacs.d/site-lisp/d-emacs/d-emacs-pkg hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/d-emacs-pkg
/home/alex/.emacs.d/site-lisp/d-emacs/killermacro hides /home/alex/.emacs.d/site-lisp/d-emacs/d-emacs-dirs/killermacro
/home/alex/.emacs.d/site-lisp/add-tex-envs hides /home/alex/.emacs.d/site-lisp/DEmacs.old/DEmacs/DEmacs-basic/add-tex-envs
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-macroexpansions hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-macroexpansions
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-commands hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-commands
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-functions hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-functions
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-modes hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-modes
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-tutorial hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-tutorial
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-constants hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-constants
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-customs hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-customs
/home/alex/.emacs.d/site-lisp/Daselt/d-lua/d-lua hides /home/alex/.emacs.d/site-lisp/Daselt/d-lua/auto/d-lua
/home/alex/.emacs.d/site-lisp/Daselt/auto/DLua hides /home/alex/.emacs.d/site-lisp/Daselt/d-lua/auto/DLua
/home/alex/.emacs.d/site-lisp/d-emacs/d-emacs-stump/d-emacs-stump hides /home/alex/.emacs.d/site-lisp/d-emacs/d-emacs-stump/test/d-emacs-stump
/home/alex/.emacs.d/site-lisp/d-emacs/d-dfk/d-dfk-customs hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-dfk/d-dfk-customs
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-emacs-special-example-user-defined-bindlists hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-emacs/d-emacs-special-example-user-defined-bindlists
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-emacs-eval-maps hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-emacs/d-emacs-eval-maps
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-emacs-commands hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-emacs/d-emacs-commands
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-emacs-functions hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-emacs/d-emacs-functions
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-emacs-customs hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-emacs/d-emacs-customs
/home/alex/.emacs.d/site-lisp/d-emacs/old/d-emacs-constants hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-emacs/d-emacs-constants
/home/alex/.emacs.d/site-lisp/d-emacs/d-lua/d-lua-eval-commands hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-lua/d-lua-eval-commands
/home/alex/.emacs.d/site-lisp/d-emacs/d-lua/d-lua-eval hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-lua/d-lua-eval
/home/alex/.emacs.d/site-lisp/d-emacs/d-lua/d-lua-functions hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-lua/d-lua-functions
/home/alex/.emacs.d/site-lisp/d-emacs/d-lua/d-lua-customs hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-lua/d-lua-customs
/home/alex/.emacs.d/site-lisp/d-emacs/d-lua/d-lua-constants hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-lua/d-lua-constants
/home/alex/.emacs.d/site-lisp/d-emacs/d-emacs-stump/d-stump-functions hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-stump/d-stump-functions
/home/alex/.emacs.d/site-lisp/d-emacs/d-emacs-stump/d-stump-customs hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-stump/d-stump-customs
/home/alex/.emacs.d/site-lisp/d-emacs/d-emacs-stump/d-stump-constants hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-stump/d-stump-constants
/home/alex/.emacs.d/site-lisp/d-emacs/d-tri/d-tri-functions hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-tri/d-tri-functions
/home/alex/.emacs.d/site-lisp/d-emacs/d-tri/d-tri-customs hides /home/alex/.emacs.d/site-lisp/Daselt/d-emacs/pkg-configs/d-tri/d-tri-customs

Features:
(shadow mail-extr gnus-msg emacsbug d-emacs-icicles icicles two-column
icicles-mode dired+ mb-depth+ mb-depth icicles-cmd2 icicles-cmd1
icicles-mcmd icicles-fn icicles-var apropos-fn+var apropos icicles-opt
ffap- fuzzy-match icicles-face cus-start display-line-numbers gdb-mi gud
misearch multi-isearch image-file image-converter image-dired-dired
image-dired image-dired-tags image-dired-external image-dired-util
mixed-pitch corfu-quick dabbrev cape-char vertico-directory isearch+
isearch-prop menu-bar+ bookmark+ bookmark+-key bookmark+-bmu info+
fit-frame help-fns+ wid-edit+ bookmark+-lit pp+ dired-x bookmark+-1
smtpmail gnus-registry registry gnus-art mm-uu mml2015 mm-view mml-smime
smime gnutls dig gnus-sum gnus-group gnus-undo gnus-start gnus-dbus
gnus-cloud nnimap nnmail mail-source nnoo gnus-spec gnus-win gnus-int
gnus-range imap rfc2104 utf7 d-emacs-org d-emacs-eww d-emacs-gnus
d-emacs-vterm d-emacs-emacs d-emacs-corfu d-emacs-doremi d-emacs-cdlatex
d-emacs-pdf-tools vc-git vc-dispatcher recentf tree-widget edit-server
epa-file goto-addr autoinsert completion elec-pair corfu-history delsel
undo-tree desktop frameset d-emacs d-emacs-mode d-emacs-stump
d-emacs-dirs d-emacs-bind d-emacs-dfk d-emacs-xkb smartparens-c
cc-mode-expansions cc-mode cc-fonts cc-guess cc-menus cc-cmds cc-styles
cc-align cc-engine cc-vars cc-defs d-emacs-coords d-emacs-base avy-act
eww vtable mule-util shr pixel-fill kinsoku mm-url gnus nnheader range
emms-idapi-browser emms-idapi emms-idapi-musicbrainz emms-mpris
emms-librefm-stream emms-librefm-scrobbler emms-playlist-limit emms-i18n
emms-history emms-score emms-stream-info emms-metaplaylist-mode
emms-bookmarks emms-cue emms-mode-line-icon emms-browser sort
emms-volume emms-volume-sndioctl emms-volume-mixerctl emms-volume-pulse
emms-volume-amixer emms-playlist-sort emms-last-played emms-player-xine
emms-lyrics emms-streams emms-show-all emms-tag-editor emms-tag-tracktag
emms-mark emms-mode-line emms-info-native emms-info-native-spc
emms-info-native-mp3 emms-info-native-ogg emms-info-native-opus
emms-info-native-flac emms-info-native-vorbis bindat emms-info-exiftool
emms-info-tinytag emms-info-metaflac emms-info-opusinfo
emms-info-ogginfo emms-info-mp3info emms-player-vlc emms-player-mpv
emms-playing-time emms-player-mplayer emms-setup emms-player-mpd
emms-url emms-cache emms-info emms-later-do emms-playlist-mode
emms-source-playlist emms-source-file locate emms-player-simple emms
emms-compat empv back-button nav-flash smartrep rainbow-blocks
org-rainbow-tags dired-rainbow dired-hacks-utils dired-aux
rainbow-delimiters rainbow-mode modus-themes abyss-theme
yaml-mode-expansions yaml-mode julia-snail spinner pulse popup eldoc-box
flycheck ox-md org-modern math-delimiters add-tex-envs cdlatex reftex
reftex-loaddefs reftex-vars latex-extra preview smartparens-latex
latex-mode-expansions latex latex-flymake flymake tex-ispell tex-style
tex dbus texmathp auctex mentor url-scgi mentor-rpc mentor-files
mentor-data xml-rpc fancy-compilation eshell-vterm em-term esh-ext
esh-proc esh-opt esh-io esh-arg esh-module esh-module-loaddefs esh-util
vterm tramp trampver tramp-integration files-x tramp-message
tramp-compat tramp-loaddefs face-remap term disp-table ehelp
vterm-module term/xterm xterm unicode-fonts font-utils dynamic-fonts
ztree ztree-diff ztree-diff-model ztree-dir ztree-view ztree-protocol
ztree-util yasnippet elgrep async grep dedicated auto-dictionary
flyspell ispell all-the-icons-completion all-the-icons-dired
all-the-icons all-the-icons-faces data-material data-weathericons
data-octicons data-fileicons data-faicons data-alltheicons go-translate
gt-text-utility gt-engine-echo gt-engine-libre gt-engine-chatgpt
gt-engine-youdao gt-engine-stardict gt-engine-deepl gt-engine-google-rpc
gt-engine-google gt-engine-bing gt-extension gt-faces gt-core plz
gt-httpx gptel-extensions gptel gptel-org gptel-openai tabgo tab-line
dired-quick-sort ls-lisp aggressive-indent visible-mark
org-noter-org-roam org-noter org-noter-djvu org-noter-nov org-noter-pdf
pdf-tools pdf-view jka-compr pdf-cache pdf-info tq pdf-util pdf-macs
image-mode exif org-noter-core ekg-import-subdirectories triples
org-roam-export use-package-ensure-system-package system-packages
consult-org-roam consult-org-roam-buffer org-ref org-ref-core openalex
request org-ref-glossary org-ref-bibtex smartparens-html
html-mode-expansions sgml-mode facemenu avy doi-utils url-http url-auth
url-gw nsm org-ref-utils org-ref-export citeproc citeproc-itemgetters
citeproc-cite citeproc-biblatex citeproc-bibtex ol-bibtex parse-time
citeproc-subbibs citeproc-sort citeproc-name citeproc-formatters
citeproc-proc citeproc-disamb citeproc-itemdata
citeproc-generic-elements citeproc-macro citeproc-choose citeproc-date
citeproc-number smartparens-rst smartparens-markdown rst compile
citeproc-context citeproc-prange citeproc-style citeproc-locale
citeproc-term citeproc-rt citeproc-lib citeproc-s queue ox-pandoc ht
ox-org ox-odt rng-loc rng-uri rng-parse rng-match rng-dt rng-util
rng-pttrn nxml-parse nxml-ns nxml-enc xmltok nxml-util ox-latex
ox-icalendar ox-html table ox-ascii ox-publish ox org-ref-misc-links
org-ref-label-link org-ref-ref-links org-ref-citation-links xref project
org-ref-bibliography-links hydra lv bibtex-completion parsebib
org-roam-migrate org-roam-log org-roam-mode org-roam-capture org-roam-id
org-roam-node org-roam-db emacsql-sqlite-builtin sqlite org-roam-utils
org-roam-compat org-roam org-capture emacsql-sqlite emacsql
emacsql-compiler org-brain picture org-agenda org-attach embark-org
smartparens-org the-org-mode-expansions org-element org-persist org-id
org-refile org-element-ast inline org ob ob-tangle ob-ref ob-lob
ob-table ob-exp org-macro org-src sh-script smie treesit executable
ob-comint org-pcomplete org-list org-footnote org-faces org-entities
foldout noutline outline ob-emacs-lisp ob-core ob-eval org-cycle
org-table org-keys oc org-loaddefs cal-menu calendar cal-loaddefs ol
org-fold org-fold-core org-compat avl-tree generator org-version
org-macs biblio biblio-download biblio-dissemin biblio-ieee biblio-hal
biblio-dblp biblio-crossref biblio-arxiv timezone biblio-doi biblio-core
let-alist url-queue url-file ido bibtex iso8601 cape pass f s
password-store auth-source-pass kind-icon svg-lib svg dom xml
corfu-prescient vertico-prescient prescient char-fold windsize
corfu-terminal popon corfu consult-gh consult-gh-autoloads
compat-autoloads seq-autoloads embark-consult consult magit-bookmark
bookmark embark marginalia orderless magit-submodule magit-blame
magit-stash magit-reflog magit-bisect magit-push magit-pull magit-fetch
magit-clone magit-remote magit-commit magit-sequence magit-notes
magit-worktree magit-tag magit-merge magit-branch magit-reset
magit-files magit-refs magit-status magit magit-repos magit-apply
magit-wip magit-log which-func imenu magit-diff smerge-mode diff
diff-mode track-changes git-commit log-edit message sendmail yank-media
puny rfc822 mml mml-sec epa epg rfc6068 epg-config gnus-util
text-property-search time-date mailabbrev mail-utils gmm-utils
mailheader pcvs-util add-log magit-core magit-autorevert autorevert
filenotify magit-margin magit-transient magit-process with-editor shell
pcomplete comint ansi-osc server ansi-color magit-mode transient
benchmark magit-git magit-base magit-section format-spec cursor-sensor
crm persistent-soft list-utils pcache eieio-base savehist vertico-buffer
vertico-mouse vertico compat smartparens-config smartparens-text
smartparens loadhist dash expand-region text-mode-expansions
er-basic-expansions expand-region-core expand-region-custom
easy-kill-extras easy-kill-line-edge easy-kill-buffer easy-kill-to-char
easy-kill smart-mark buffer-move windmove anchored-transpose second-sel
doremi-frm frame-cmds palette eyedropper crosshairs col-highlight vline
hl-line+ hl-line advice misc-cmds rect faces+ thingatpt+ frame-fns avoid
hexrgb doremi-cmd cus-theme cus-edit pp wid-edit doremi info-colors
winner ring curry-compose rebinder gnus-dired finder-inf dired-du
find-dired edmacro auto-install cl ffap thingatpt dired dired-loaddefs
straight-autoloads straight quelpa-use-package quelpa mm-decode
mm-bodies mm-encode mail-parse rfc2231 rfc2047 rfc2045 mm-util
ietf-drums mail-prsvr lisp-mnt help-fns radix-tree comp comp-cstr
cl-extra warnings comp-run comp-common kmacro functions rx use-package
use-package-ensure use-package-delight use-package-diminish
use-package-bind-key bind-key easy-mmode use-package-core edebug debug
backtrace help-mode find-func sanityinc-tomorrow-bright-theme
color-theme-sanityinc-tomorrow color allout cus-load Icicles-autoloads
abyss-theme-autoloads aggressive-indent-autoloads
all-the-icons-completion-autoloads all-the-icons-dired-autoloads
auctex-cluttex-autoloads auto-dictionary-autoloads avy-act-autoloads
back-button-autoloads backup-walker-autoloads buffer-move-autoloads
cape-autoloads cdlatex-autoloads citar-embark-autoloads citar-autoloads
color-theme-sanityinc-tomorrow-autoloads consult-flycheck-autoloads
consult-lsp-autoloads consult-org-roam-autoloads
corfu-prescient-autoloads corfu-terminal-autoloads corfu-autoloads
dedicated-autoloads default-text-scale-autoloads diminish-autoloads
dired-quick-sort-autoloads dired-rainbow-autoloads
dired-hacks-utils-autoloads doremi-autoloads dynamic-fonts-autoloads
easy-kill-extras-autoloads eldoc-box-autoloads elgrep-autoloads
embark-consult-autoloads consult-autoloads embark-autoloads
emms-autoloads empv-autoloads eshell-vterm-autoloads
expand-region-autoloads fancy-compilation-autoloads
flycheck-package-autoloads flycheck-autoloads forge-autoloads
closql-autoloads ghub-autoloads gnuplot-autoloads go-translate-autoloads
gptel-autoloads julia-snail-autoloads kind-icon-autoloads
latex-extra-autoloads auctex-autoloads tex-site
latex-pretty-symbols-autoloads lsp-julia-autoloads julia-mode-autoloads
lsp-ui-autoloads lsp-mode-autoloads marginalia-autoloads
mentor-autoloads mixed-pitch-autoloads modus-themes-autoloads
mood-line-autoloads nav-flash-autoloads orderless-autoloads
org-brain-autoloads org-modern-autoloads org-noter-autoloads
helm-autoloads helm-core-autoloads async-autoloads
org-rainbow-tags-autoloads org-ref-prettify-autoloads org-ref-autoloads
citeproc-autoloads htmlize-autoloads org-roam-bibtex-autoloads
bibtex-completion-autoloads biblio-autoloads biblio-core-autoloads
org-roam-autoloads emacsql-autoloads ox-pandoc-autoloads
package-build-autoloads derived package-lint-autoloads paredit-autoloads
parsebib-autoloads pass-autoloads f-autoloads
password-store-otp-autoloads password-store-autoloads
perspective-autoloads plz-autoloads popon-autoloads popup-autoloads
quelpa-autoloads rainbow-blocks-autoloads rainbow-mode-autoloads
smart-mark-autoloads smartparens-autoloads smartrep-autoloads
smudge-autoloads oauth2-autoloads speed-type-autoloads spinner-autoloads
sqlite3-autoloads string-inflection-autoloads svg-lib-autoloads
swiper-autoloads ivy-autoloads sx-autoloads markdown-mode-autoloads
tabgo-autoloads tablist-autoloads telega-autoloads
treemacs-all-the-icons-autoloads all-the-icons-autoloads
treemacs-icons-dired-autoloads treemacs-magit-autoloads magit-autoloads
pcase transient-autoloads magit-section-autoloads
treemacs-persp-autoloads persp-mode-autoloads
treemacs-projectile-autoloads projectile-autoloads
treemacs-tab-bar-autoloads treemacs-autoloads posframe-autoloads
avy-autoloads treepy-autoloads triples-autoloads try-autoloads
unicode-fonts-autoloads ucs-utils-autoloads font-utils-autoloads
persistent-soft-autoloads list-utils-autoloads pcache-autoloads
url-scgi-autoloads use-package-ensure-system-package-autoloads
system-packages-autoloads vertico-prescient-autoloads vertico-autoloads
prescient-autoloads visible-mark-autoloads visual-fill-column-autoloads
vterm-autoloads wfnames-autoloads windresize-autoloads
windsize-autoloads with-editor-autoloads xml-rpc-autoloads
yaml-autoloads yaml-mode-autoloads yasnippet-autoloads zotero-autoloads
oauth-autoloads info dash-autoloads ztree-autoloads package browse-url
xdg url url-proxy url-privacy url-expand url-methods url-history
url-cookie generate-lisp-file url-domsuf url-util mailcap url-handlers
url-parse auth-source cl-seq eieio eieio-core cl-macs icons
password-cache json subr-x map byte-opt gv bytecomp byte-compile
url-vars cl-loaddefs cl-lib rmc iso-transl tooltip cconv eldoc paren
electric uniquify ediff-hook vc-hooks lisp-float-type elisp-mode mwheel
term/x-win x-win term/common-win x-dnd touch-screen tool-bar dnd fontset
image regexp-opt fringe tabulated-list replace newcomment text-mode
lisp-mode prog-mode register page tab-bar menu-bar rfn-eshadow isearch
easymenu timer select scroll-bar mouse jit-lock font-lock syntax
font-core term/tty-colors frame minibuffer nadvice seq simple cl-generic
indonesian philippine cham georgian utf-8-lang misc-lang vietnamese
tibetan thai tai-viet lao korean japanese eucjp-ms cp51932 hebrew greek
romanian slovak czech european ethiopic indian cyrillic chinese
composite emoji-zwj charscript charprop case-table epa-hook
jka-cmpr-hook help abbrev obarray oclosure cl-preloaded button loaddefs
theme-loaddefs faces cus-face macroexp files window text-properties
overlay sha1 md5 base64 format env code-pages mule custom widget keymap
hashtable-print-readable backquote threads dbusbind inotify lcms2
dynamic-setting system-font-setting font-render-setting cairo gtk
x-toolkit xinput2 x multi-tty move-toolbar make-network-process
tty-child-frames native-compile emacs)

Memory information:
((conses 16 1934237 2454408) (symbols 48 115220 13) (strings 32 483653 150979)
 (string-bytes 1 14288706) (vectors 16 154921) (vector-slots 8 2655804 546103)
 (floats 8 2409 9168) (intervals 56 34847 21741) (buffers 992 42))

[killermacro.el (text/plain, attachment)]

[gdb.txt (text/plain, attachment)]

Message #8 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Alexander Prähauser <ahprae <at> protonmail.com>
Cc: 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Wed, 15 Jan 2025 22:21:40 +0200

> Date: Wed, 15 Jan 2025 19:00:01 +0000
> From:  Alexander Prähauser via "Bug reports for GNU Emacs,
>  the Swiss army knife of text editors" <bug-gnu-emacs <at> gnu.org>
> 
> A macro I've defined in causes Emacs to crash. The macro is faulty but I
> still don't think it should do that. I've attached a file with a
> functioning version of the macro, the version that crashes with comments
> where it is changed and where it crashes and a configuration that causes
> it to crash. Careful, evaluating that buffer will kill Emacs. I've run
> `emacs -Q' in gdb and added a logging file. The output of `bt-full' is 

Could you please show a full recipe for reproducing this?  After
starting "emacs -Q", what should I type, step by step, to trigger the
crash?

Thanks.

Message #11 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Alexander Prähauser <ahprae <at> protonmail.com>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Wed, 15 Jan 2025 21:38:18 +0000

"Eli Zaretskii" <eliz <at> gnu.org> writes:

emacs -Q killermacro.el (or wherever you saved the file)
M-x eval-buffer

That should do it.

>> Date: Wed, 15 Jan 2025 19:00:01 +0000
>> From:  Alexander Prähauser via "Bug reports for GNU Emacs,
>>  the Swiss army knife of text editors" <bug-gnu-emacs <at> gnu.org>
>>
>> A macro I've defined in causes Emacs to crash. The macro is faulty but I
>> still don't think it should do that. I've attached a file with a
>> functioning version of the macro, the version that crashes with comments
>> where it is changed and where it crashes and a configuration that causes
>> it to crash. Careful, evaluating that buffer will kill Emacs. I've run
>> `emacs -Q' in gdb and added a logging file. The output of `bt-full' is
>
> Could you please show a full recipe for reproducing this?  After
> starting "emacs -Q", what should I type, step by step, to trigger the
> crash?
>
> Thanks.

Message #14 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Pip Cet <pipcet <at> protonmail.com>
To: Alexander Prähauser via Bug reports for GNU Emacs, the Swiss army knife of text editors
 <bug-gnu-emacs <at> gnu.org>, Stefan Monnier <monnier <at> iro.umontreal.ca>,
 Mattias Engdegård <mattiase <at> acm.org>
Cc: Alexander Prähauser <ahprae <at> protonmail.com>,
 Eli Zaretskii <eliz <at> gnu.org>, 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Wed, 15 Jan 2025 22:20:42 +0000

Alexander Prähauser via "Bug reports for GNU Emacs, the Swiss army knife of text editors" <bug-gnu-emacs <at> gnu.org> writes:

> "Eli Zaretskii" <eliz <at> gnu.org> writes:
>
> emacs -Q killermacro.el (or wherever you saved the file)
> M-x eval-buffer
>
> That should do it.

Reproducible here: the problem is that Feval does not verify its
LEXICAL argument is a valid lexical environment.  In your case, it's a
non-list cons cell.

Two ways to fix this:

1. Call Flength on the lexical environment.  This will throw an error if
the environment is invalid, including when it is circular, which might
otherwise result in unquittable infloops (this is currently subject of
bug#75520).

However, this may slow down a very critical piece of code (OTOH, it can
also conceivably speed it up, by ensuring that the lexical environment
is in a high-priority cache).


2. Revise all code which assumes Vinternal_interpreter_environment isn't
too bad.  In particular, this code:

	  for (Lisp_Object p = Vinternal_interpreter_environment;
	       !NILP (p); p = XCDR(p))
	    {
	      Lisp_Object e = XCAR (p);
	      if (SYMBOLP (e))
		dynvars = Fcons(e, dynvars);
	    }

should be

          Lisp_Object p = Vinternal_interpreter_environment;
          FOR_EACH_TAIL (p)
	    {
	      Lisp_Object e = XCAR (p);
	      if (SYMBOLP (e))
		dynvars = Fcons(e, dynvars);
	    }

(As virtually all places which used !NILP instead of CONSP were fixed a
while ago, my cocci script did not catch this case.)

We'll probably have to do both.

Here's (1):

commit 651362816bd5237b0f5aaea70d72a29aa9b8852e (HEAD)
Author: Pip Cet <pipcet <at> protonmail.com>
Date:   Wed Jan 15 22:17:43 2025 +0000

    Ensure Vinternal_interpreter_environment isn't malformed (bug#75593)
    
    * src/eval.c (Feval): Call 'Flength' on the lexical environment,
    ensuring it is a list.

diff --git a/src/eval.c b/src/eval.c
index 941d121c2fb..2619bd57415 100644
--- a/src/eval.c
+++ b/src/eval.c
@@ -2446,8 +2446,11 @@ DEFUN ("eval", Feval, Seval, 1, 2, 0,
   (Lisp_Object form, Lisp_Object lexical)
 {
   specpdl_ref count = SPECPDL_INDEX ();
-  specbind (Qinternal_interpreter_environment,
-	    CONSP (lexical) || NILP (lexical) ? lexical : list_of_t);
+  if (CONSP (lexical) || NILP (lexical))
+    Flength (lexical);
+  else
+    lexical = list_of_t;
+  specbind (Qinternal_interpreter_environment, lexical);
   return unbind_to (count, eval_sub (form));
 }
 

Message #20 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Pip Cet <pipcet <at> protonmail.com>
Cc: ahprae <at> protonmail.com, mattiase <at> acm.org, monnier <at> iro.umontreal.ca,
 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Thu, 16 Jan 2025 08:50:37 +0200

> Date: Wed, 15 Jan 2025 22:20:42 +0000
> From: Pip Cet <pipcet <at> protonmail.com>
> Cc: Eli Zaretskii <eliz <at> gnu.org>, Alexander Prähauser <ahprae <at> protonmail.com>, 75593 <at> debbugs.gnu.org
> 
> --- a/src/eval.c
> +++ b/src/eval.c
> @@ -2446,8 +2446,11 @@ DEFUN ("eval", Feval, Seval, 1, 2, 0,
>    (Lisp_Object form, Lisp_Object lexical)
>  {
>    specpdl_ref count = SPECPDL_INDEX ();
> -  specbind (Qinternal_interpreter_environment,
> -	    CONSP (lexical) || NILP (lexical) ? lexical : list_of_t);
> +  if (CONSP (lexical) || NILP (lexical))
> +    Flength (lexical);
> +  else
> +    lexical = list_of_t;
> +  specbind (Qinternal_interpreter_environment, lexical);
>    return unbind_to (count, eval_sub (form));
>  }

Isn't there a cheaper way of verifying LEXICAL is a list, not a cons
cell?

How about calling Flength only "#ifdef ENABLE_CHECKING", and otherwise
just verifying that LEXICAL is a list?  After all, passing a circular
list to this function means some Lisp program shoots itself in the
foot, so we could tell "don't do that", like in other cases when Emacs
gives Lisp programmers enough rope to hang themselves.

IOW, can we prevent crashes in a way that is cheap in production
builds, since 'eval' is quite a hotspot?

Message #23 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Pip Cet <pipcet <at> protonmail.com>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: ahprae <at> protonmail.com, mattiase <at> acm.org, monnier <at> iro.umontreal.ca,
 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Thu, 16 Jan 2025 10:38:21 +0000

"Eli Zaretskii" <eliz <at> gnu.org> writes:

>> Date: Wed, 15 Jan 2025 22:20:42 +0000
>> From: Pip Cet <pipcet <at> protonmail.com>
>> Cc: Eli Zaretskii <eliz <at> gnu.org>, Alexander Prähauser <ahprae <at> protonmail.com>, 75593 <at> debbugs.gnu.org
>>
>> --- a/src/eval.c
>> +++ b/src/eval.c
>> @@ -2446,8 +2446,11 @@ DEFUN ("eval", Feval, Seval, 1, 2, 0,
>>    (Lisp_Object form, Lisp_Object lexical)
>>  {
>>    specpdl_ref count = SPECPDL_INDEX ();
>> -  specbind (Qinternal_interpreter_environment,
>> -	    CONSP (lexical) || NILP (lexical) ? lexical : list_of_t);
>> +  if (CONSP (lexical) || NILP (lexical))
>> +    Flength (lexical);
>> +  else
>> +    lexical = list_of_t;
>> +  specbind (Qinternal_interpreter_environment, lexical);
>>    return unbind_to (count, eval_sub (form));
>>  }
>
> Isn't there a cheaper way of verifying LEXICAL is a list, not a cons
> cell?

I don't think there is, no.  As I explained, it's unlikely to be a
significant expense: lexical environments are not very large, usually,
and all other users of Vinternal_interpreter_environment are optimized
for small lists which are scanned until we find the right entry.  If we
wanted to optimize for large lexical environments, we might be able to,
but we've decided not to do that.

> How about calling Flength only "#ifdef ENABLE_CHECKING", and otherwise
> just verifying that LEXICAL is a list?

ENABLE_CHECKING shouldn't affect code behavior in such ways, no.  I
don't think a crash is acceptable here, but if people end up disabling
ENABLE_CHECKING and using XCAR on a non-cons, a crash is the *best*
possible outcome: more likely, we'd just be creating nonsensical
Lisp_Objects which may cause memory corruption.

Making behavior depend on ENABLE_CHECKING in such ways will cause bugs
in !ENABLE_CHECKING builds that cannot be debugged because the
corresponding ENABLE_CHECKING build will never reach the problematic
code.  IOW, it'd defeat the purpose of ENABLE_CHECKING.

> After all, passing a circular
> list to this function means some Lisp program shoots itself in the
> foot, so we could tell "don't do that", like in other cases when Emacs
> gives Lisp programmers enough rope to hang themselves.

Flength does precisely that.  If you think a crash, unquittable infloop,
or memory corruption is the right way to tell people "don't do that", I
disagree completely.  It's not what the rest of Emacs does, it's not the
right thing to do, and a performance concern that is entirely
hypothetical at this point is not a good reason to start doing it.

(Also, this is not primarily about circular lists, it's about non-lists
in general).

> IOW, can we prevent crashes in a way that is cheap in production
> builds, since 'eval' is quite a hotspot?

If there's any evidence that lexenvs are large enough that scanning them
linearly in this one place is a significant expense, we might want to
rewrite the entire lexenv code: all of it is O(N) in the size of the
environment: it assumes lexenvs are small enough to scan linearly, and
so does the fixed Feval.

(I think it would be possible to allow a hash table as the non-list
final CDR of Feval's lexenv argument, but we'd still end up overriding
bindings with an alist, so it'd be a bit of a Frankenstein affair).

Pip

Message #26 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Pip Cet <pipcet <at> protonmail.com>
Cc: ahprae <at> protonmail.com, mattiase <at> acm.org, monnier <at> iro.umontreal.ca,
 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Thu, 16 Jan 2025 13:36:08 +0200

> Date: Thu, 16 Jan 2025 10:38:21 +0000
> From: Pip Cet <pipcet <at> protonmail.com>
> Cc: monnier <at> iro.umontreal.ca, mattiase <at> acm.org, ahprae <at> protonmail.com, 75593 <at> debbugs.gnu.org
> 
> "Eli Zaretskii" <eliz <at> gnu.org> writes:
> 
> >> Date: Wed, 15 Jan 2025 22:20:42 +0000
> >> From: Pip Cet <pipcet <at> protonmail.com>
> >> Cc: Eli Zaretskii <eliz <at> gnu.org>, Alexander Prähauser <ahprae <at> protonmail.com>, 75593 <at> debbugs.gnu.org
> >>
> >> --- a/src/eval.c
> >> +++ b/src/eval.c
> >> @@ -2446,8 +2446,11 @@ DEFUN ("eval", Feval, Seval, 1, 2, 0,
> >>    (Lisp_Object form, Lisp_Object lexical)
> >>  {
> >>    specpdl_ref count = SPECPDL_INDEX ();
> >> -  specbind (Qinternal_interpreter_environment,
> >> -	    CONSP (lexical) || NILP (lexical) ? lexical : list_of_t);
> >> +  if (CONSP (lexical) || NILP (lexical))
> >> +    Flength (lexical);
> >> +  else
> >> +    lexical = list_of_t;
> >> +  specbind (Qinternal_interpreter_environment, lexical);
> >>    return unbind_to (count, eval_sub (form));
> >>  }
> >
> > Isn't there a cheaper way of verifying LEXICAL is a list, not a cons
> > cell?
> 
> I don't think there is, no.  As I explained, it's unlikely to be a
> significant expense: lexical environments are not very large, usually,
> and all other users of Vinternal_interpreter_environment are optimized
> for small lists which are scanned until we find the right entry.  If we
> wanted to optimize for large lexical environments, we might be able to,
> but we've decided not to do that.

OK, so can you then benchmark this?  E.g., by byte-compiling all the
*.el files in the tree with and without the change?  Or maybe there's
something appropriate in the benchmark suite?  Or both?

> > How about calling Flength only "#ifdef ENABLE_CHECKING", and otherwise
> > just verifying that LEXICAL is a list?
> 
> ENABLE_CHECKING shouldn't affect code behavior in such ways, no.

?? It already does.  We discover problems and bugs earlier all the
time when running a build with --enable-checking, because it aborts
before it runs into a potentially fatal problem.

> I
> don't think a crash is acceptable here, but if people end up disabling
> ENABLE_CHECKING and using XCAR on a non-cons, a crash is the *best*
> possible outcome: more likely, we'd just be creating nonsensical
> Lisp_Objects which may cause memory corruption.

I think the best outcome is to signal an error, not to crash, if that
is feasible.

> Making behavior depend on ENABLE_CHECKING in such ways will cause bugs
> in !ENABLE_CHECKING builds that cannot be debugged because the
> corresponding ENABLE_CHECKING build will never reach the problematic
> code.

I wrote that under the assumption that we could test for a proper list
without calling Flength.  If that is not possible/feasible, that point
is moot, and IMO we should only see if the run-time price is
acceptable.

> (Also, this is not primarily about circular lists, it's about non-lists
> in general).

That should be easy to detect up front, no?  I thought the issue was
about a cons cell vs a proper list.

> > IOW, can we prevent crashes in a way that is cheap in production
> > builds, since 'eval' is quite a hotspot?
> 
> If there's any evidence that lexenvs are large enough that scanning them
> linearly in this one place is a significant expense

I didn't say that; it remains to be established.  Thus my suggestion
above to benchmark.  But if there is a less expensive way, we could
take it regardless: one doesn't have to prove that a simple test is
less expensive than scanning the entire list structure.

Message #29 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Pip Cet <pipcet <at> protonmail.com>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: ahprae <at> protonmail.com, mattiase <at> acm.org, monnier <at> iro.umontreal.ca,
 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Thu, 16 Jan 2025 12:08:37 +0000

"Eli Zaretskii" <eliz <at> gnu.org> writes:

>> Date: Thu, 16 Jan 2025 10:38:21 +0000
>> From: Pip Cet <pipcet <at> protonmail.com>
>> Cc: monnier <at> iro.umontreal.ca, mattiase <at> acm.org, ahprae <at> protonmail.com, 75593 <at> debbugs.gnu.org
>>
>> "Eli Zaretskii" <eliz <at> gnu.org> writes:
>>
>> >> Date: Wed, 15 Jan 2025 22:20:42 +0000
>> >> From: Pip Cet <pipcet <at> protonmail.com>
>> >> Cc: Eli Zaretskii <eliz <at> gnu.org>, Alexander Prähauser <ahprae <at> protonmail.com>, 75593 <at> debbugs.gnu.org
>> >>
>> >> --- a/src/eval.c
>> >> +++ b/src/eval.c
>> >> @@ -2446,8 +2446,11 @@ DEFUN ("eval", Feval, Seval, 1, 2, 0,
>> >>    (Lisp_Object form, Lisp_Object lexical)
>> >>  {
>> >>    specpdl_ref count = SPECPDL_INDEX ();
>> >> -  specbind (Qinternal_interpreter_environment,
>> >> -	    CONSP (lexical) || NILP (lexical) ? lexical : list_of_t);
>> >> +  if (CONSP (lexical) || NILP (lexical))
>> >> +    Flength (lexical);
>> >> +  else
>> >> +    lexical = list_of_t;
>> >> +  specbind (Qinternal_interpreter_environment, lexical);
>> >>    return unbind_to (count, eval_sub (form));
>> >>  }
>> >
>> > Isn't there a cheaper way of verifying LEXICAL is a list, not a cons
>> > cell?
>>
>> I don't think there is, no.  As I explained, it's unlikely to be a
>> significant expense: lexical environments are not very large, usually,
>> and all other users of Vinternal_interpreter_environment are optimized
>> for small lists which are scanned until we find the right entry.  If we
>> wanted to optimize for large lexical environments, we might be able to,
>> but we've decided not to do that.
>
> OK, so can you then benchmark this?  E.g., by byte-compiling all the
> *.el files in the tree with and without the change?  Or maybe there's
> something appropriate in the benchmark suite?  Or both?

I'll see what I can do.  The patch certainly can be improved by checking
NILP (lexical) first, and only calling Flength if it's a cons cell, and
that appears to be the common case: building Emacs only calls Feval with
a cons cell only once or twice, and in those cases, it's a single-item
list.

>> > How about calling Flength only "#ifdef ENABLE_CHECKING", and otherwise
>> > just verifying that LEXICAL is a list?
>>
>> ENABLE_CHECKING shouldn't affect code behavior in such ways, no.
>
> ?? It already does.  We discover problems and bugs earlier all the
> time when running a build with --enable-checking, because it aborts
> before it runs into a potentially fatal problem.

Calling Flength is very different from aborting, IMHO.

>> I
>> don't think a crash is acceptable here, but if people end up disabling
>> ENABLE_CHECKING and using XCAR on a non-cons, a crash is the *best*
>> possible outcome: more likely, we'd just be creating nonsensical
>> Lisp_Objects which may cause memory corruption.
>
> I think the best outcome is to signal an error, not to crash, if that
> is feasible.

Indeed.  That's what Flength does.

>> Making behavior depend on ENABLE_CHECKING in such ways will cause bugs
>> in !ENABLE_CHECKING builds that cannot be debugged because the
>> corresponding ENABLE_CHECKING build will never reach the problematic
>> code.
>
> I wrote that under the assumption that we could test for a proper list
> without calling Flength.  If that is not possible/feasible, that point
> is moot, and IMO we should only see if the run-time price is
> acceptable.

(We can use list_length, not Flength, but the difference won't be
significant.  Or Fproper_list_p, of course, which might make it clearer
that we're not interested in the length at all.)

>> (Also, this is not primarily about circular lists, it's about non-lists
>> in general).
>
> That should be easy to detect up front, no?  I thought the issue was
> about a cons cell vs a proper list.

Sorry, but I don't see a way of checking that we're looking at a proper
list that's cheaper that list_length: we need to find the last cons
cell, so we need to walk the list, and counting cells while we're doing
so should not affect performance.

>> > IOW, can we prevent crashes in a way that is cheap in production
>> > builds, since 'eval' is quite a hotspot?
>>
>> If there's any evidence that lexenvs are large enough that scanning them
>> linearly in this one place is a significant expense
>
> I didn't say that; it remains to be established.  Thus my suggestion
> above to benchmark.  But if there is a less expensive way, we could
> take it regardless: one doesn't have to prove that a simple test is
> less expensive than scanning the entire list structure.

I don't think there is a simple test.  I may be wrong, of course.

Pip

Message #32 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Pip Cet <pipcet <at> protonmail.com>
Cc: ahprae <at> protonmail.com, mattiase <at> acm.org, monnier <at> iro.umontreal.ca,
 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Thu, 16 Jan 2025 14:47:54 +0200

> Date: Thu, 16 Jan 2025 12:08:37 +0000
> From: Pip Cet <pipcet <at> protonmail.com>
> Cc: monnier <at> iro.umontreal.ca, mattiase <at> acm.org, ahprae <at> protonmail.com, 75593 <at> debbugs.gnu.org
> 
> >> (Also, this is not primarily about circular lists, it's about non-lists
> >> in general).
> >
> > That should be easy to detect up front, no?  I thought the issue was
> > about a cons cell vs a proper list.
> 
> Sorry, but I don't see a way of checking that we're looking at a proper
> list that's cheaper that list_length: we need to find the last cons
> cell, so we need to walk the list, and counting cells while we're doing
> so should not affect performance.

No, I meant to detect objects that are not CONSP, but something else:
a number, a symbol, a string, etc.  I thought this was what you meant
by "non-lists".

Message #35 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Pip Cet <pipcet <at> protonmail.com>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: ahprae <at> protonmail.com, mattiase <at> acm.org, monnier <at> iro.umontreal.ca,
 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Thu, 16 Jan 2025 13:06:12 +0000

"Eli Zaretskii" <eliz <at> gnu.org> writes:

>> Date: Thu, 16 Jan 2025 12:08:37 +0000
>> From: Pip Cet <pipcet <at> protonmail.com>
>> Cc: monnier <at> iro.umontreal.ca, mattiase <at> acm.org, ahprae <at> protonmail.com, 75593 <at> debbugs.gnu.org
>>
>> >> (Also, this is not primarily about circular lists, it's about non-lists
>> >> in general).
>> >
>> > That should be easy to detect up front, no?  I thought the issue was
>> > about a cons cell vs a proper list.
>>
>> Sorry, but I don't see a way of checking that we're looking at a proper
>> list that's cheaper that list_length: we need to find the last cons
>> cell, so we need to walk the list, and counting cells while we're doing
>> so should not affect performance.
>
> No, I meant to detect objects that are not CONSP, but something else:
> a number, a symbol, a string, etc.  I thought this was what you meant
> by "non-lists".

Oh, my apologies.  I said "non-list" when what I meant was an "improper
list", in Emacs jargon.  Dotted "lists" caused the crash for the OP;
circular "lists" cause infloops, but no crashes.  The code already
treats !CONSP !NILP objects specially (only Qt as a value is documented,
the rest are silently treated like Qt is).

Pip

Message #38 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Stefan Monnier <monnier <at> iro.umontreal.ca>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: ahprae <at> protonmail.com, mattiase <at> acm.org, Pip Cet <pipcet <at> protonmail.com>,
 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Thu, 16 Jan 2025 13:17:17 -0500

> IOW, can we prevent crashes in a way that is cheap in production
> builds, since 'eval' is quite a hotspot?

I would actually be surprised if `Feval` is a hotspot.

AFAIK `eval_sub` is sometimes a hotspot (tho only when running
non-compiled code, which *should* be somewhat infrequent), but I don't
think `Feval` is.


        Stefan

Message #41 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Stefan Monnier <monnier <at> iro.umontreal.ca>
To: Pip Cet <pipcet <at> protonmail.com>
Cc: ahprae <at> protonmail.com, mattiase <at> acm.org, Eli Zaretskii <eliz <at> gnu.org>,
 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Thu, 16 Jan 2025 13:41:35 -0500

> If we wanted to optimize for large lexical environments, we might be
> able to, but we've decided not to do that.

Indeed.  The lack of optimization is for two reasons:

- As you say, we expect lexical envs to be small.
- We also expect that performance of interpreted code is not
  very significant.

> If there's any evidence that lexenvs are large enough that scanning them
> linearly in this one place is a significant expense, we might want to
> rewrite the entire lexenv code: all of it is O(N) in the size of the
> environment: it assumes lexenvs are small enough to scan linearly, and
> so does the fixed Feval.

I suggest to simply compare the time to bootstrap with and without the
`Flength` call.  This way we can base our decision on actual data rather
than just our intuition.

> (I think it would be possible to allow a hash table as the non-list
> final CDR of Feval's lexenv argument, but we'd still end up overriding
> bindings with an alist, so it'd be a bit of a Frankenstein affair).

Also, since lexenvs are mostly modified by extending them (without
affecting the base env since it might have been captured by a closure)
and then later reverting to the unextended base, hash-tables tend to
work rather poorly.

More common choices to handle large lexical environments are balanced
binary trees.  Other options generally require first replacing the var
names with something else, like small integers (e.g. DeBruijn indices),
but that requires a kind of preprocessing which we'd usually call
"compilation" (tho it can be simpler than what `bytecomp.el` does).


        Stefan "who likes «Myers stacks»: like singly-linked lists
                but with O(log N) time for `nthcdr`"

Message #44 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Stefan Monnier <monnier <at> iro.umontreal.ca>
Cc: ahprae <at> protonmail.com, mattiase <at> acm.org, pipcet <at> protonmail.com,
 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Thu, 16 Jan 2025 21:23:19 +0200

> From: Stefan Monnier <monnier <at> iro.umontreal.ca>
> Cc: Pip Cet <pipcet <at> protonmail.com>,  mattiase <at> acm.org,
>   ahprae <at> protonmail.com,  75593 <at> debbugs.gnu.org
> Date: Thu, 16 Jan 2025 13:17:17 -0500
> 
> > IOW, can we prevent crashes in a way that is cheap in production
> > builds, since 'eval' is quite a hotspot?
> 
> I would actually be surprised if `Feval` is a hotspot.
> 
> AFAIK `eval_sub` is sometimes a hotspot (tho only when running
> non-compiled code, which *should* be somewhat infrequent), but I don't
> think `Feval` is.

I hope you are right.  The benchmark should tell us.

Message #47 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Stefan Monnier <monnier <at> iro.umontreal.ca>
Cc: ahprae <at> protonmail.com, mattiase <at> acm.org, pipcet <at> protonmail.com,
 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Thu, 16 Jan 2025 21:25:36 +0200

> From: Stefan Monnier <monnier <at> iro.umontreal.ca>
> Cc: Eli Zaretskii <eliz <at> gnu.org>,  mattiase <at> acm.org,  ahprae <at> protonmail.com,
>   75593 <at> debbugs.gnu.org
> Date: Thu, 16 Jan 2025 13:41:35 -0500
> 
> I suggest to simply compare the time to bootstrap with and without the
> `Flength` call.  This way we can base our decision on actual data rather
> than just our intuition.

I suggested to byte-compile all the Lisp files for that very reason.
It should be a better measurement, since "make bootstrap" does some
stuff that is utterly unrelated.

Message #50 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Pip Cet <pipcet <at> protonmail.com>
To: Stefan Monnier <monnier <at> iro.umontreal.ca>
Cc: ahprae <at> protonmail.com, mattiase <at> acm.org, Eli Zaretskii <eliz <at> gnu.org>,
 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Fri, 17 Jan 2025 11:34:58 +0000

"Stefan Monnier" <monnier <at> iro.umontreal.ca> writes:

>> IOW, can we prevent crashes in a way that is cheap in production
>> builds, since 'eval' is quite a hotspot?
>
> I would actually be surprised if `Feval` is a hotspot.

Do we need to fix the case where the lexenv variable is modified by Lisp
code while it is the lexical environment?

(defmacro f () x)
(let ((l (list (cons 'l nil))))
  (setcdr (car l) l)
  (eval '(progn
	   (setcdr l 3)
	   (f))
	l))

crashes (this is the dotted-list case; the circular case causes an
unquittable infloop).

My original plan was to avoid looking at what Fmake_interpreted_closure
does with its lexenv.  It currently seems to me that it is simply
resurrected as Vinternal_interpreter_environment upon execution.

If that's true, maybe we should simply fix the Qmacro code in eval_sub
and hope for the best.


diff --git a/src/eval.c b/src/eval.c
index 941d121c2fb..43a7cb78cb6 100644
--- a/src/eval.c
+++ b/src/eval.c
@@ -2658,12 +2658,12 @@ eval_sub (Lisp_Object form)
 
 	  /* Make the macro aware of any defvar declarations in scope. */
 	  Lisp_Object dynvars = Vmacroexp__dynvars;
-	  for (Lisp_Object p = Vinternal_interpreter_environment;
-	       !NILP (p); p = XCDR(p))
+	  Lisp_Object p = Vinternal_interpreter_environment;
+	  FOR_EACH_TAIL (p)
 	    {
 	      Lisp_Object e = XCAR (p);
 	      if (SYMBOLP (e))
-		dynvars = Fcons(e, dynvars);
+		dynvars = Fcons (e, dynvars);
 	    }
 	  if (!EQ (dynvars, Vmacroexp__dynvars))
 	    specbind (Qmacroexp__dynvars, dynvars);

Message #53 received at 75593 <at> debbugs.gnu.org (full text, mbox):

From: Stefan Monnier <monnier <at> iro.umontreal.ca>
To: Pip Cet <pipcet <at> protonmail.com>
Cc: ahprae <at> protonmail.com, mattiase <at> acm.org, Eli Zaretskii <eliz <at> gnu.org>,
 75593 <at> debbugs.gnu.org
Subject: Re: bug#75593: 31.0.50; Faulty macro kills Emacs
Date: Fri, 17 Jan 2025 08:49:27 -0500

> Do we need to fix the case where the lexenv variable is modified by Lisp
> code while it is the lexical environment?
>
> (defmacro f () x)
> (let ((l (list (cons 'l nil))))
>   (setcdr (car l) l)
>   (eval '(progn
> 	   (setcdr l 3)
> 	   (f))
> 	l))
>
> crashes (this is the dotted-list case; the circular case causes an
> unquittable infloop).

😈

> If that's true, maybe we should simply fix the Qmacro code in eval_sub
> and hope for the best.

Sounds good to me,


        Stefan

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.