Clearly.
>The only way to do the (3) just before starting a getty whould be add a
>setfont and a reset of the console in an rc file because the init is
>execve'd, so we have not any chance to do it after init is running.
No - we can put anything we like in the rc scripts that init runs, so
it's no problem to signal the kernel at the appropriate point by whatever
means. The kernel can then clean up itself.
>Whould you really like to force an user program to restore the screen
>after the kernel have done the penguin show ?
Why not - a lot of the boot sequence *is* done in user mode. Setting
up swap, mounting filesystems, fscking, configuring the network... this
is just one teeny weeny extra.
>I don't like too much this idea because it breaks _all_ the
installations:
>this way we force _all_ the linux sysadmins to edit an rc file and add a
>setfont and a reset ...
>... and I think this is a no no.
If the admin makes a conscious decision to install the bootlogo, they
can make sure they do it properly, and include any other system
configuration that entails too. Here's the point - the bootlogo is not
enabled by default. It only 'breaks' installations where the admin
has chosen to install it, but made a botch job of it.
<PLUG PLUG PLUG>
I posted a bootlogo implementation last year (and updated it last month),
which does the above, shows a pretty logo for a while *and* cleans up
after itself. And the only command you need to turn it off is 'echo'!
You can find it on http://www.linuxhq.com/patch/20-p0555.html
</PLUG>
Cheers,
John
-- remember i was always true; remember that i always tried remember i loved only you; remember me and smile... for it's better to forget than to remember me and cry <a href="file:///dev/null">:-p</a>