Re: Critical bytecode bug with hash tables while dumping emacs.

[Vibhav Pant <vibhavp@gmail.com>]

[-- Attachment #1.1: Type: text/plain, Size: 1259 bytes --]

Also, this one line patch crashes temacs with a SIGSEGV - the hash table
contents
are possibly getting corrupted because of a bad memory write.

On Fri, Jan 27, 2017 at 12:27 AM Vibhav Pant <vibhavp@gmail.com> wrote:

> On Thu, Jan 26, 2017 at 11:03 PM, Paul Eggert <eggert@cs.ucla.edu> wrote:
> > In that case the bug is not critical, right? One way to address the
> problem
> > is to say that code should not print hash tables before dumping.
>
> It's critical in the sense that any code loaded from loadup.el is
> effectively
> prohibited from using printed hash tables in any way. I've recently been
> working on adding a new 'switch` bytecode op (@ branch
> feature/byte-switch),
> which uses hash tables generated during compile time (so the constant
> vector
> stores printed hash tables) as jump tables. This bug breaks switch
> entirely.
>
> > we're planning to redo dumping anyway and can address this problem (if
> it still
> > occurs) then.
>
> I suspect this bug is related to purecopy, which I suppose isn't a part of
> the
> redo (or is it? I don't know much about the new dumping code). If anyone
> has any
> ideas about this issue, I'd appreciate some pointers on where to start.
>
> Thanks,
> Vibhav
> --
> Vibhav Pant
> vibhavp@gmail.com
>

[-- Attachment #1.2: Type: text/html, Size: 2160 bytes --]

[-- Attachment #2: wrong-hash-table.patch --]
[-- Type: text/x-patch, Size: 672 bytes --]

diff --git a/lisp/custom.el b/lisp/custom.el
index 70b6839db3..beccb2b545 100644
--- a/lisp/custom.el
+++ b/lisp/custom.el
@@ -145,6 +145,8 @@ custom-declare-variable
 `standard-value'.  At the same time, SYMBOL's property `force-value' is
 set to nil, as the value is no longer rogue."
   (put symbol 'standard-value (purecopy (list default)))
+  (message "%s" #s(hash-table size 65 test eql rehash-size 1.5 rehash-threshold 0.8 data
+                              (1 2 3 4 5 6 7 8 9 1 a a b b c c d d e e q q w w r r t t)))
   ;; Maybe this option was rogue in an earlier version.  It no longer is.
   (when (get symbol 'force-value)
     (put symbol 'force-value nil))