Re: [9front] fix race condition on writes to #v/vgactl

[Anthony Martin <ality@pbrane.org>]

qwx@sciops.net once said:
> Here's the final patch, mostly the same.  I'll commit this later
> unless there are any objections.

There's only two callers of resetscreenimage, two callers of
vgascreenwin, and four callers of deletescreenimage.

I would adjust the call sites to include them within the locked
sections, remove the locking inside them and get rid of the
repetitive underscore functions.

As an aside, all the wet floors involving locks in the vga code
could use some mopping up. Best keep the mold away.

Cheers,
  Anthony