[Simh] NetBSD related issues

[Mark Pizzolato - Info Comm]

Hi Felix,

It seems there are two issues on your host:

1) the one you're called out (Large File Support).  This oversight was a longstanding bug for the NetBSD host platform.  The code at github now should work for you.  This issue was hard to find in older versions of simh.  The latest simh codebase has enriched the SHOW VERSION output to display various details about the host and simulator which relate to various features.

2) the 'ldconfig: not found'.  This will affect other things which you may not have noticed yet.  This issue is due to a recent change in NetBSD on x86 hosts which has converted from a.out to ELF binary/library formats.  I haven't tested a build on a NetBSD host recently.

Additionally, you mention:

> But disk images >=2 GiB still don't. (And this time, 1 MB really is
> 1000000 bytes and not 1 MiB as with 3.9-0 before :))
> 
> sim> set rq0 rauser=2146
> sim> set rq0 rauser=2147
> Invalid argument

Good observation.  However, your conclusion is a little off (as would be reasonable by what you are seeing).

Internally the limit computation is identical, however the limit in the latest code has been raised to reflect the true capacity of the 32bit value which is used to contain the disk size.  The old limit seems to have been chosen to align with the value of 2047/2048.

> The -L flag for RAUSER to set LBNs is also interesting:
> 
> sim> set -L rq0 rauser=4194304
> Invalid argument
> sim> set -L rq0 rauser=4194303
> Invalid argument
> sim> set -L rq0 rauser=4194302
> [ok]
> 
> Shouldn't 4194303 still work, as it is (just) under 2 GiB?

4194303 is the new upper LBN when large file support isn't available (the prior limit was 4000000), and you are right that 4194303 should be allowed.  The code now has a > instead of a >= in the limit test.

Thanks for your feedback.

The current github code will work better for you, but I'm still working on how to handle the ELF issue.

- Mark