Re: [PATCH 0/7] PMAG-AA frame buffer driver rework

From: Maciej W. Rozycki
Date: Sat Feb 27 2016 - 17:23:38 EST


Hi Tomi,

> > This is an update to the PMAG-AA frame buffer driver to adapt it to our
> > current APIs, which I promised a while ago (too much of a while, which I
> > apologise for).
>
> Interesting hardware. I like the names of these things. "Monolithic CMOS
> 64 x 64 Pixel Cursor Generator", "TURBOchannel Color Frame Buffer".

Indeed, it sounds so 1980s, doesn't it? It took me a while too to figure
out how to wire it to a monitor as the adapter has a TNC connector for its
output (the colour adapters from the line have a DA-3W3 connector each
instead). Fortunately at least video signalling is standard -- EIA-343A
-- so a modern VGA monitor can be used as long as it does sync-on-green.

> Also interesting to see the git history of a file going almost directly
> to the initial git commit.

FYI, for anything beyond you'd have to get at LMO's GIT history, e.g.:

$ git blame 66f0a432564b -- drivers/video/fbdev/pmag-aa-fb.c

on a checkout from <git://git.linux-mips.org/pub/scm/ralf/linux>.

Ralf had been running his MIPS tree in CVS since forever and consequently
had a slightly more detailed history of changes than Linus had with his
master copy. When the move from BitKeeper to GIT was made Ralf was kind
enough to go through the pain and converted his CVS repository to GIT
before merging with Linus.

There are older branches there as well, namely: linux-2.4, linux-2.2, and
linux-2.0 (anything beyond these is linear on master). Most operations on
the master branch are cut through at the initial Linus's commit though,
you need to name a commit preceding that point -- such as one immediately
before, which I referred to above -- to go past.

> Thanks, queued for 4.6.

Thanks.

I've since noticed blanking will need an update to interpret VESA modes
correctly, i.e. return errors for the unsupported ones. This is a minor
issue, so I'll group it with blanking support for the other two boards
which is missing for the Bt459 RAMDAC they use. The Bt459 can switch sync
generation off, so the full powerdown mode can be supported too in
addition to plain screen blanking (obviously you can't switch individual
sync signals in a composite signal).

Cursor support for the Bt459 would be good to have too; noted as a future
enhancement possibility.

Maciej