Re: kern/58225 - netbsd9/amd64 requires COMPAT_16 for 32bit support (fwd)

[Jason Thorpe <thorpej%me.com@localhost>]

> From what I can see (*) you've handled the (COMPAT_NETBSD32 & COMPAT_16)
> situation, but at the expense of mishandling (COMPAT_NETBSD32 & ! COMPAT_16) 
> situation.  You really should not be auto-loading a module
> that isn't desired.  As you can see, that brings in a whole lot of other
> cruft for COMPAT_* and COMPAT_NETBSD323_*.
> 
> What happens if we simply don't autoload compat_netbsd32_16 module?  If
> it is loaded through other means (modload(8) maybe) then we get the
> behaviour that you already implemented; if the module isn't loaded, we
> should reurn EINVAL.

Note the only reason I made the changes in the first place is because COMPAT_NETBSD32 signal delivery was already a complicated broken mess (and the changes addressed a reported bug).  Sure, more modules might get loaded now, but at least signal delivery works.

TBH, I don’t get what the big deal is here.  The NETBSD32 version is doing basically what the non-NETBSD32 version is doing: that is, if the process says is uses the old “sigcontext” type of trampoline, it auto-loads the compat_16 module.  And it’s been doing that ever since 2019 when *you* made that change.

-- thorpej