[PATCH] FB driver for Cirrus GD5446 graphic hardware.

[Pavel Pisa]

Hello Joel,

thanks for review.

I would check if Alex is available now.
But to move things forward I would push the work if
time allows me.

On Thursday 13 September 2012 18:45:19 Joel Sherrill wrote:
> Thanks for posting this so quickly. Here are some quick comments:
>
> + No Makefile.am changes?

Not for now. I think, that it should be separate patch.
Discussion on IRC. For simple test next

To test it, next change is enough

--- a/c/src/lib/libbsp/i386/pc386/Makefile.am
+++ b/c/src/lib/libbsp/i386/pc386/Makefile.am
@@ -83,3 +83,4 @@ include_HEADERS += ../../i386/shared/comm/i386_io.h
 libbsp_a_SOURCES += console/defkeymap.c
-libbsp_a_SOURCES += console/fb_vga.c
+#libbsp_a_SOURCES += console/fb_vga.c
+libbsp_a_SOURCES += console/fb_cirrus.c
 libbsp_a_SOURCES += console/keyboard.c

I incline to the next as interim solution

--- a/c/src/lib/libbsp/i386/pc386/Makefile.am
+++ b/c/src/lib/libbsp/i386/pc386/Makefile.am
@@ -83,3 +83,4 @@ include_HEADERS += ../../i386/shared/comm/i386_io.h
 libbsp_a_SOURCES += console/defkeymap.c
+if ENABLE_VGA_CIRRUS
 libbsp_a_SOURCES += console/fb_vga.c
+else !ENABLE_VGA_CIRRUS
+libbsp_a_SOURCES += console/fb_cirrus.c
+endif !ENABLE_VGA_CIRRUS
 libbsp_a_SOURCES += console/keyboard.c

It should be replaced by infrastructure for runtime driver registration.

> + C++ style comments need to be C
> + Some one-line C comments do not have a space before the */
> + Use single blank lines not multiple ones in a row

OK

> + No license terms at top of file

What is the best template in the RTEMS sources?

> I assume it works as as we get more into the graphics-addon patches
> I will be able to duplicate your results.

It should work with any/old version of the graphic code
or other independent projects.

The rtems-graphics-toolkit is huge amount of the code and
subprojects and number of copyend and changed code is huge too.

Alex's version can be cloned from

  git clone https://github.com/alex-sever-h/rtems-graphics-toolkit.git

I can help with some filterring and squashing changes
to help maintain readable history but I do not think
that changes load should flow through mailinglist.

They should get in through pull after testing.

Best wishes,

             Pavel