Re: [patch/rfc/testsuite] Test GDB on not-so-little core files

[Andrew Cagney <ac131313@redhat.com>]

> First something trivial:
> 
>    likely that the variable will occure early in the core file (an
>                                  ^^^^^^ typo
> 
> I checked Single Unix Spec v3 on rlimit:
> 
>   http://www.opengroup.org/onlinepubs/007904975/functions/getrlimit.html
> 
> There is an RLIMIT_AS which applies as well, so you might as well
> maximize that.

M'kay.

> 180	  *(char*)0 = 0;
> (gdb) PASS: gdb.base/bigcore.exp: next
> print heap
> $1 = (struct list *) 0x77e4fff0

[...]

> (gdb) PASS: gdb.base/bigcore.exp: load corefile
> print heap
> $16 = (struct list *) 0x77fbffe0
> (gdb) FAIL: gdb.base/bigcore.exp: check heap (address 0x77e4fff0)

At first glance it looks like a bug in the HP/UX code, but I wonder. 
Could the startup code be allocating an extra bit of memory leading to 
that address being moved?  Might need to tweak things so that the 
program under GDB is the one that is made to dump core.

> testcase /house/chastain/gdb/s1/gdb/testsuite/gdb.base/bigcore.exp completed in 183 seconds

Yea, contrast that to my GNU/Linux box which manages:

 > testcase src/gdb/testsuite/gdb.base/bigcore.exp completed in 4 seconds

Andrew