Re: [PATCH] Compress kernel modules on installation.

[Willy Tarreau]

On Tue, Feb 26, 2008 at 02:30:16PM +0200, Adrian Bunk wrote:
> On Tue, Feb 26, 2008 at 11:22:03AM +0100, Willy Tarreau wrote:
> > On Tue, Feb 26, 2008 at 11:14:55AM +0200, Adrian Bunk wrote:
> > > On Mon, Feb 25, 2008 at 11:21:38PM +0100, Willy Tarreau wrote:
> >...
> > > > Have you tried keeping the module names intact (.ko, not .ko.gz) ?
> > > > It's what I was doing with modutils in 2.4 and what I'm still doing
> > > > with module-init-tools in 2.6. While I don't particularly use mkinitrd,
> > > > I think that keeping the name intact is preferable and should help.
> > > 
> > > How would you see if, and if yes with what program, a module was 
> > > compressed if the name is kept intact?
> > 
> > depmod/modinfo/insmod/modprobe already know it. And quite honnestly,
> > I don't know about any other program which really needs to process
> > those files once installed. Well, maybe ksymoops, but I'd have to
> > check, as I don't recall having ever been annoyed with this.
> >...
> 
> depmod/modinfo/insmod/modprobe know only if you compile 
> module-init-tools with zlib support.
> 
> And what about the busybox versions?
> 
> A different name would e.g.:
> - easily allow proper error handling if the userspace modules program 
>   doesn't support the compression used
> - better scale to support additional compressions
> - give the user a hint what is happening and what might be the problem
>   when anythig goes wrong

I agree, but right now module.dep references existing files with their
real names. Maybe something should define exactly what it should contain
(eg: module_name.ko even if .ko.gz is used) so that all tools relying on
it do not stop after noticing that the file referenced there does not
exist.

regards,
Willy

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/