Re: Auto-generate module-loadable syscall table?

[christos%zoulas.com@localhost (Christos Zoulas)]

On May 8,  8:11pm, paul%vps1.whooppee.com@localhost (Paul Goyette) wrote:
-- Subject: Re: Auto-generate module-loadable syscall table?

| On Fri, 8 May 2015, Christos Zoulas wrote:
| 
| > In article <Pine.NEB.4.64.1505081756290.25140%vps1.whooppee.com@localhost>,
| > Paul Goyette  <paul%whooppee.com@localhost> wrote:
| >> I've just noticed that the table in sys/kern/kern_syscall.c is not
| >> generated from the kern/syscalls.master file.
| >>
| >> Is there some reason why kern/Makefile isn't used to create the table at
| >> the same time as creating the other various syscall-related sources?
| >
| > How would you do that? See if the name contains, _aio_, _lio_, compat etc?
| 
| Well, there's already a MODULAR keyword for kern/makesyscalls.sh so it 
| doesn't seem to be so difficult to possibly add one more parameter to 
| those lines (to identify the name of the containing module).  Then the 
| MODULAR syscalls could all be listed in a new generated source file 
| (perhaps kern/syscalls_autoload.h) and be included whereever neeeded 
| (for now, that means kern/kern_syscall.h AFAICT).
| 
| Or am I totally misunderstanding this comment from kern/Makefile?
| 
|  	MODULAR:attempt to autoload system call if not present

No, looks right to me. Actually the list is inconsistent right now (lfs for
example) so if you auto-gen it, it will be better.

christos