Re: [Gcl-devel] SIGKILL on kbsd

[Camm Maguire]

Greetings, and thanks!  I've worked around this for now.  In general, I
was assuming that a successful brk call guaranteed the corresponding
allocation of memory, even though it is common for the pages to not
actually be added to the process until written.  On kfbsd, I limit the
break to PHYS_PAGES from sysconf.  Linux seems to give a brk failure
when the actual allocation would fail, though my testing here is
preliminary. 

Take care,

Robert Millan <address@hidden> writes:

> 2013/6/25 Camm Maguire <address@hidden>:
>> Greetings, and thanks for your reply!  This is on falla.  The code
>> probes for the maximum allocatable data segment at runtime. The ENOMEM
>> does not appear conservative enough, as I'm apparently allowed 34Gb, but
>> the system has much less virtual memory.
>
> Might be related to:
>
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=661283#53
>
>
>

-- 
Camm Maguire                                        address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah