[Openmcl-devel] mystery SEGV starting 64bit ccl on linux

[Bit Twiddler]

The reason why I remember it being there is that I was
trying to figure out whether mmap() returned a pointer
to the beginning or the end of the memory block in
that case. I was getting a bus error where you guys
set the size of the block.

But, since you're not putting it in the call any more,
it probably doesn't matter.

On Thu, Nov 18, 2010 at 4:04 PM, Gary Byers <gb at clozure.com> wrote:

>
>
> On Thu, 18 Nov 2010, Bit Twiddler wrote:
>
>  I'm still not sure what the problem is, since I've been dealing
>> with so many different versions of ccl over the last few days.
>>
>> I distinctly remember seeing MAP_GROWSDOWN in the
>> call to mmap() from MapMemoryForStack(), but I just looked
>> in the 1.5 release, and it isn't there. So, I must have been
>> running something else.
>>
>
> It was removed from the SVN sources after 1.5 was released
> (and after the Linux kernel change's effects started getting
> reported.)
>
> If you've built the lisp kernel from recent sources, the
> MAP_GROWSDOWN should be gone.  If you're unsure, it wouldn't
> hurt to build it again, just to be sure.  (That's especially
> true if the problem is intermittent.)
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clozure.com/pipermail/openmcl-devel/attachments/20101118/7beaa430/attachment.htm>