In article <20041202203143.GA1725@piano.dberg.net>,
	spam@dberg.net (Dave B) writes:
>   Under NetBSD 1.5.3++, shmat() is failing with ENOMEM.  The port it
> fails on is Atari, but the same operation works fine under Alpha and
> i386 (also 1.5.3++).  Apologies, the test code below has to be run
> under ktrace for the symptom to be seen.
> 
>   I do have the options SYSV{MSG,SEM,SHM} in my kernel config (plus,
> the shmget() succeeding).  I've tried unlimiting datasize w/ulimit
> with no luck.  Is there something else I'm forgetting to configure?
> Any resource limits I'm not thinking of?  If not, and it's a bug, is
> there any use in submitting a PR, or is 1.5 retired?

There's no reason to think things would be any different in current...
MAXDSIZ on atari is still only 64Mb, while it's 1Gb on alpha and i386.
How much memory are you trying to mmap()? There might be a reason why
atari's limit is so low, but if you want to find out for yourself,
try building a kernel with "options MAXDSIZ (1*1024*1024*1024)".


Frederick