bug#21666: Bug is indeed undo-list and GC related

[phillip.lord@russet.org.uk (Phillip Lord)]

You might try looking at commits where the undo-list handling has been
changed since 19efb9.

To a first approximation, git log --author="Phillip Lord" should show
most of these.

But these might be good ones to try.

ae18cc20f899335cede183f18221b6aae539fd51
957b05c615ee749b569d9fa2b214b2a2d8fa9bda
20aa42e8204f8f0139ba3880cb32ddf88acc9bf4

Phil


Jorgen Schäfer <jorgen.schaefer@gmail.com> writes:

> No, I can not. As mentioned, the problem occurs after a day or two of
> usage. If you understand the problem better now that you know what prevents
> it, I am of course happy to run test cases.
>
> Eli Zaretskii <eliz@gnu.org> schrieb am Mi., 9. März 2016 16:15:
>
>> > From: Jorgen Schäfer <jorgen.schaefer@gmail.com>
>> > Date: Wed, 09 Mar 2016 10:14:35 +0000
>> >
>> > I do not know how GC works in Emacs Lisp, but it seems to me that an
>> Emacs Lisp program should not be
>> > able to segfault Emacs in this manner. It is interesting that this
>> problem was introduced sometime after
>> > 19efb9 – an Emacs built from that commit did not exhibit the problem.
>>
>> Can you bisect it more narrowly?  "After 19efb9" leaves a lot of
>> alternatives...
>>
>> Thanks.
>>