Critical bytecode bug with hash tables while dumping emacs.

[Vibhav Pant <vibhavp@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 727 bytes --]

While, working on `byte-switch`, I discovered a critical bug when a
hash table is present in the constant vector of a bytecode function.
If such a function is called when temacs dumps a binary during the
biuld process, all keys and their corresponding values change to that
of a different hash table. The following patch demonstrates this by
testing the value of 'foo` in a table mapping 'foo to 'bar, and
erroring out if the test fails. This causes temacs to error out,
failing the build.

Since the current elisp codebase, doesn't use printed representation
anywhere to represent hash tables, this bug doesn't come up. However,
this may fail builds in the future, if future code does that.

-- 
Vibhav Pant
vibhavp@gmail.com

[-- Attachment #2: wrong-hash-table.patch --]
[-- Type: text/x-patch, Size: 683 bytes --]

diff --git a/lisp/custom.el b/lisp/custom.el
index 70b6839db3..49330bb1a4 100644
--- a/lisp/custom.el
+++ b/lisp/custom.el
@@ -146,6 +146,10 @@ custom-declare-variable
 set to nil, as the value is no longer rogue."
   (put symbol 'standard-value (purecopy (list default)))
   ;; Maybe this option was rogue in an earlier version.  It no longer is.
+  (let ((ht #s(hash-table size 65 test eql rehash-size 1.5 rehash-threshold 0.8 data
+                          (foo bar))))
+    (unless (eq 'bar (gethash 'foo ht))
+      (error (format "This shouldn't be happening. Hash table: %s" ht))))
   (when (get symbol 'force-value)
     (put symbol 'force-value nil))
   (if (keywordp doc)