[racket-dev] Apparently random segfaults in DrDr.

From: Neil Van Dyke (neil at neilvandyke.org)
Date: Tue Oct 9 17:21:31 EDT 2012

> could be a regression from 5.2.1 to 5.2.

5.2.1 to 5.3

Neil Van Dyke wrote at 10/09/2012 05:20 PM:
> The other day, I was reproducibly crashing fresh DrRacket 5.3 
> processes by invoking the Macro Stepper on a particular file.  (Not a 
> seg fault; Linux out-of-memory killer would kill it after it got to 
> close to 3GB RAM usage, since I don't use swap space.)
> By commenting-out macro uses, I was able to get DrRacket to not crash, 
> but its RAM usage would fluctuate between 1.5GB and 800MB.  However, 
> after the Macro Stepper rendered this time that it didn't crash, when 
> I returned to DrRacket main window, the Definitions pane was blank 
> white (scrollbar still there, and indicating some text about the size 
> of my program, but no actual text visible in it, nor selectable).
> What was confusing was that the program was not very big, and made 
> only light and simple use of macros.  But my attempts to trim it down 
> to a smaller test case were unsuccessful.
> I don't recall seeing problems like this with the Macro Stepper 
> before.  Even really complicated and questionable uses of macros, it 
> has handled like a champ.  I suppose this could be a regression from 
> 5.2.1 to 5.2.
> If someone is really interested, I will try to find the problematic 
> file later tonight/tomorrow.
> Neil V.
> John Clements wrote at 10/09/2012 05:00 PM:
>> The message below suggests that evaluating "stepper-tool.rkt" caused… a seg fault. I'm chalking this up to "oh, DrRacket sometimes crashes", but I admit that it makes me a bit uncomfortable.
>> Advice or suggestions?
>> John
>> Begin forwarded message:
>>> From:drdr at racket-lang.org
>>> Subject: [DrDr] R25433 (timeout 2) (unclean 2) (stderr 6) (changes 67)
>>> Date: October 9, 2012 8:04:40 AM PDT
>>> To:clements at racket-lang.org
>>> DrDr has finished building push #25433 after 57.98m.
>>> http://drdr.racket-lang.org/25433/
>>> A file you are responsible for has a condition that may need inspecting.
>>>   unclean:
>>>     http://drdr.racket-lang.org/25433/collects/stepper/stepper-tool.rkt
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.racket-lang.org/dev/archive/attachments/20121009/eada8d6d/attachment.html>

Posted on the dev mailing list.