Subject: Re: Failed mmap regression test
To: Leo Weppelman <leo@wau.mis.ah.nl>
From: Ignatios Souvatzis <is@jocelyn.rhein.de>
List: current-users
Date: 04/04/2000 20:34:38
On Tue, Apr 04, 2000 at 01:25:52PM +0200, Leo Weppelman wrote:
> On the Atari 1.4W/32MB:
> 
> regress ===> mmap
> >>> MAPPING FILE <<<
>     CHECKING RESIDENCY
>     LOCKING RANGE
>     CHECKING RESIDENCY
>     UNLOCKING RANGE
> >>> MAPPING 128 PAGE ANONYMOUS REGION <<<
> mmap: mmap anon #1: Segmentation fault
> FAILED
> 
> I haven't seen any panics (yet ;-)

Did you try unlimit first?

I've seen that Segmentation fault, too, (without unlimit) and looked a bit
into it;
it happens inside something called by err()... I guess this should not happen,
either.

errno was 9, for me. 9 from mmap is invalid file descriptor, according to
the docs... which can't happen for a anonymous mmap(), right?

Which is why I thought the system is messed up at this point already, and
so that the malloc needed by err() fails, and the system panics later.

	-is