[FBDEV BK] Updates and fixes.

[FBDEV BK] Updates and fixes.

[James Simmons]

As usually I have a standard diff at 
http://phoenix.infradead.org/~jsimmons/fbdev.diff.gz

It is against the latest BK tree. I haven'tt tried it yet for 2.5.67

Linus, please do a

	bk pull http://fbdev.bkbits.net/fbdev-2.5

This will update the following files:

 Documentation/devices.txt        |    7 
 arch/i386/boot/compressed/misc.c |    2 
 arch/i386/boot/video.S           |   34 +
 arch/i386/kernel/setup.c         |    4 
 drivers/video/aty/aty128fb.c     |   16 
 drivers/video/cfbimgblt.c        |    4 
 drivers/video/console/fbcon.c    |  585 +++++++++++-------------
 drivers/video/console/fbcon.h    |    1 
 drivers/video/controlfb.c        |   18 
 drivers/video/edid.h             |  138 +++++
 drivers/video/fbcmap.c           |   33 -
 drivers/video/fbmem.c            |  211 ++++----
 drivers/video/fbmon.c            |  922 +++++++++++++++++++++++++++++++--------
 drivers/video/i810/i810.h        |    7 
 drivers/video/i810/i810_accel.c  |  140 +++--
 drivers/video/i810/i810_dvt.c    |    3 
 drivers/video/i810/i810_gtf.c    |    7 
 drivers/video/i810/i810_main.c   |  207 +++-----
 drivers/video/i810/i810_main.h   |    4 
 drivers/video/imsttfb.c          |    4 
 drivers/video/logo/logo.c        |   69 +-
 drivers/video/modedb.c           |  105 ++++
 drivers/video/platinumfb.c       |   28 -
 drivers/video/pm2fb.c            |   22 
 drivers/video/radeonfb.c         |   54 +-
 drivers/video/retz3fb.c          |   16 
 drivers/video/riva/fbdev.c       |  380 +++++++---------
 drivers/video/sis/sis_main.h     |   62 +-
 drivers/video/softcursor.c       |  243 ++++------
 drivers/video/tdfxfb.c           |   18 
 drivers/video/tgafb.c            |    2 
 drivers/video/tridentfb.c        |    8 
 drivers/video/vesafb.c           |   23 
 drivers/video/vga16fb.c          |    6 
 include/asm-i386/setup.h         |    1 
 include/linux/fb.h               |   29 -
 include/linux/linux_logo.h       |    2 
 include/linux/pci_ids.h          |    1 
 include/video/edid.h             |   27 +
 39 files changed, 2140 insertions(+), 1303 deletions(-)

through these ChangeSets:

<jsimmons@kozmo.(none)> (03/04/12 1.1014)
   [FBDEV] Improved speed performance. We copy many bytes of data instead of just one at a time.
   
   [IMSTT FBDEV] Fixed a bug that caused the hardware to lock up when scrolling.

<jsimmons@maxwell.earthlink.net> (03/04/11 1.1011)
   [RIVA FBDEV] Cursor fixes. Almost done. At least it looks normal most of the time.

<jsimmons@kozmo.(none)> (03/04/10 1.1009)
   [RADEON FBDEV] Compile fixes.

<jsimmons@kozmo.(none)> (03/04/10 1.1007)
   [RADEON FBDEV] Detect 8 Megs of RAM not 8 Kilobytes.

<jsimmons@kozmo.(none)> (03/04/10 1.1006)
   [FBDEV] EDID support from OpenFirmware on PPC platoforms and from the BIOS on intel platforms.

<jsimmons@maxwell.earthlink.net> (03/04/10 1.1004)
   [FBDEV] Made the upper layer code always use the cursor mask of struct fb_cursor inside struct fb_info. This moved memory management of the mask and image data to the upper layers.

<jsimmons@maxwell.earthlink.net> (03/04/10 1.1003)
   [FBDEV] Made the upper layer code always use the cursor mask of struct fb_cursor inside struct fb_info. This moved memory management of the mask and image data to the upper layers.
   
   [RADEON FBDEV] Updates for the Radeon 9100.

<jsimmons@kozmo.(none)> (03/04/04 1.998)
   [FBDEV] Killed off shutting down IRQs. We need them for some types of hardware.
   
   [FBDEV] Cleanup with FB_CUR_SETCUR and the enable field. Fix to set the cursor shape when we change the size of the cursor.
   
   [I810 FBDEV] Updates to the new cursor code.
   
   [FBDEV GENERIC CURSOR] A memcpy optimization. Also only allocate a new mask field when the size of the cursor changes.

<jsimmons@maxwell.earthlink.net> (03/04/02 1.995)
   [FBDEV] Use C99 style.

<jsimmons@kozmo.(none)> (03/04/02 1.991.1.2)
   [FBDEV SOFT CURSOR] Test to see if kmalloc failed.
   
   [FBCON] Test to see if the user priovides there own work queue.

<jsimmons@kozmo.(none)> (03/04/02 1.991.1.1)
   [FBDEV] Final cursor code cleanups. Now the burden of handling the cursor code lies on the driver side. The reason for this is that a invalid cursor might come from userland.

<jsimmons@kozmo.(none)> (03/03/31 1.990)
   [FBDEV] Massive cleanups of the cursor api.

<jsimmons@maxwell.earthlink.net> (03/03/26 1.986)
   [FBDEV] Documentation on the device numbers of /dev/fb being mulitples of 32 is no longer true. Removed that info.
   
   [FBDEV] Logo fixes. Now we can display different color logos on screens of different color depths.
   
   [VGA16 FBDEV] Small compile error. Fixed it now.

<jsimmons@maxwell.earthlink.net> (03/03/26 1.984)
   [I810 FBDEV] Driver updates.
   
   [FBCON] Reversed some of my cursor changes.

<jsimmons@kozmo.(none)> (03/03/25 1.983)
   [FBDEV] The image color depth of zero hack has been killed.

<jsimmons@maxwell.earthlink.net> (03/03/25 1.982)
   [FBCON] Now we use workqueues so framebuffer code can always work in a process context.
   
   [GENERIC CURSOR] Safety check in case kmalloc failes

<jsimmons@maxwell.earthlink.net> (03/03/25 1.981)
   [FBCON] Could be called outside of a process context. This fixes that.

<jsimmons@maxwell.earthlink.net> (03/03/25 1.979)
   [RAGE 128/CONTROL/PLATNIUM FBDEV] PPC updates.
   
   [RADEON FBDEV] PLL fix for specific type of card.

Re: [FBDEV BK] Updates and fixes.

[Andrew Morton]

James Simmons <jsimmons@infradead.org> wrote:
>
> 
> As usually I have a standard diff at 
> http://phoenix.infradead.org/~jsimmons/fbdev.diff.gz
> 

There are a lot of compilation warnings.


drivers/char/agp/agp.h:50: warning: `global_cache_flush' defined but not used
drivers/char/agp/agp.h:50: warning: `global_cache_flush' defined but not used
drivers/char/agp/agp.h:50: warning: `global_cache_flush' defined but not used
drivers/video/aty/mach64_gx.c:194: warning: initialization from incompatible pointer type
drivers/video/aty/mach64_gx.c:486: warning: initialization from incompatible pointer type
drivers/video/aty/mach64_gx.c:602: warning: initialization from incompatible pointer type
drivers/video/aty/mach64_gx.c:726: warning: initialization from incompatible pointer type
drivers/video/aty/mach64_gx.c:873: warning: initialization from incompatible pointer type
drivers/video/console/fbcon.c: In function `fbcon_startup':
drivers/video/console/fbcon.c:530: warning: unused variable `irqres'
drivers/video/console/fbcon.c: At top level:
drivers/video/console/fbcon.c:206: warning: `fb_vbl_handler' defined but not used
drivers/video/riva/fbdev.c: In function `rivafb_cursor':
drivers/video/riva/fbdev.c:1472: warning: unused variable `s_idx'
drivers/video/riva/fbdev.c:1472: warning: unused variable `d_idx'
drivers/video/riva/fbdev.c:1472: warning: unused variable `j'



I tried your patch on a bunch of machines.

IBM A21P laptop:

Since 2.5.54, this machine has had a problem with the VGA console.  No fb
stuff configured at all.  During boot, as soon as the kernel output hits the
bottom of the screen (the point where it would first scroll), no more output
comes out, ever.  The machine boots up OK though.

#
# Console display driver support
#
CONFIG_VGA_CONSOLE=y
CONFIG_MDA_CONSOLE=y
CONFIG_DUMMY_CONSOLE=y

Disabling CONFIG_MDA_CONSOLE prevents that from happening.

When I enable framebuffer with your latest patch on this machine the display
screws up - it alternates between blackness and a strange flicker of
horizontal and vertical grey lines.  No text is visible.

This machine has:

ATI Technologies Inc Rage Mobility M3 AGP 2x (rev 02)





A second desktop machine has:

VGA compatible controller: nVidia Corporation NV17 [GeForce4 MX440] (rev a3)

When fbcon is enabled here the screen is full of random uninitialised gunk
and no text is readable.





Another machine has Cirrus hardware.  The Cirrus drivers do not compile.





A fourth machine has:

	nVidia Corporation NV11 [GeForce2 MX] (rev b2)

this seems to work OK.  The penguins initially have a white background, then
the background goes black, then they go away.  Perhaps that is intentional. 
But it gets to a readable login prompt.



A fifth machine has:

	ATI Technologies Inc Radeon VE QY

it seems to work OK as well.  The penguins started out on a black background.
I didn't test dual-head mode.

Re: [Linux-fbdev-devel] Re: [FBDEV BK] Updates and fixes.

[Benjamin Herrenschmidt]

> When I enable framebuffer with your latest patch on this machine the display
> screws up - it alternates between blackness and a strange flicker of
> horizontal and vertical grey lines.  No text is visible.
> 
> This machine has:
> 
> ATI Technologies Inc Rage Mobility M3 AGP 2x (rev 02)

This is a r128 based chip, which lacks some of the proper support for
LCD stuff (especially automatic retreival of the EDID from BIOS) that
we have in radeonfb. If you feed the driver with a suitable mode, it
should work

> A second desktop machine has:
> 
> VGA compatible controller: nVidia Corporation NV17 [GeForce4 MX440] (rev a3)
> 
> When fbcon is enabled here the screen is full of random uninitialised gunk
> and no text is readable.

I haven't yet checked which codebase is in there for rivafb, I have
started collecting various patches around & doing my own fixes, I have
a version here that works on PPC with GeForce2 & 4, though I still
have problems with accel on the GeForce4. It's a 2.4 code base right
now, I haven't had time to clean that up and release a patch though.

If 2.5 still has the old codebase, then GeForce4 isn't properly
supported yet.

Note that if you have a flat panel, then the problem of properly
retreiving the EDID to get a suitable default mode is also present
there, unless James added some support for it.

> 	ATI Technologies Inc Radeon VE QY
> 
> it seems to work OK as well.  The penguins started out on a black background.
> I didn't test dual-head mode.

Ok, I've seen that James added various fixes to 2.5 radeonfb, I still
have to port some of my 2.4 ones to 2.5, I except those chips to be
properly supported with fbdev.

Ben.

Re: [FBDEV BK] Updates and fixes.

[Linus Torvalds]

On Mon, 14 Apr 2003, James Simmons wrote:
> 
> Linus, please do a
> 
> 	bk pull http://fbdev.bkbits.net/fbdev-2.5

This seems to cause

	arch/i386/boot/setup.o(.text+0x9a4): In function `video':
	: undefined reference to `store_edid'

which is apparently because I don't configure VIDEO_SELECT.

I did this to make it compile, but you should check what the real issue 
is.

		Linus
----
# This is a BitKeeper generated patch for the following project:
# Project Name: Linux kernel tree
# This patch format is intended for GNU patch command version 2.5 or higher.
# This patch includes the following deltas:
#	           ChangeSet	1.1034  -> 1.1035 
#	arch/i386/boot/video.S	1.6     -> 1.7    
#
# The following is the BitKeeper ChangeSet Log
# --------------------------------------------
# 03/04/14	torvalds@home.transmeta.com	1.1035
# Store EDID only when CONFIG_VIDEO_SELECT is set and edid
# function actually exists.
# --------------------------------------------
#
diff -Nru a/arch/i386/boot/video.S b/arch/i386/boot/video.S
--- a/arch/i386/boot/video.S	Mon Apr 14 12:14:46 2003
+++ b/arch/i386/boot/video.S	Mon Apr 14 12:14:46 2003
@@ -133,9 +133,9 @@
 #ifdef CONFIG_VIDEO_RETAIN
 	call	restore_screen			# Restore screen contents
 #endif /* CONFIG_VIDEO_RETAIN */
+	call	store_edid
 #endif /* CONFIG_VIDEO_SELECT */
 	call	mode_params			# Store mode parameters
-	call	store_edid
 	popw	%ds				# Restore original DS
 	ret
 

Re: [FBDEV BK] Updates and fixes.

[James Simmons]

> There are a lot of compilation warnings. 
> 
> drivers/char/agp/agp.h:50: warning: `global_cache_flush' defined but not used

Not related.

> drivers/video/aty/mach64_gx.c:194: warning: initialization from incompatible pointer type

Haven't applied a few driver patches yet because I have been working on 
the upper layer stuff.

> drivers/video/console/fbcon.c: In function `fbcon_startup':
> drivers/video/console/fbcon.c:530: warning: unused variable `irqres'
> drivers/video/console/fbcon.c: At top level:
> drivers/video/console/fbcon.c:206: warning: `fb_vbl_handler' defined but not used

Which brings up the point. The driver specific code for handing the 
flashing (acorn, atarti, and mac generic) driver shoudl move that code to 
there respected driver. 

> drivers/video/riva/fbdev.c: In function `rivafb_cursor':

Cleaned up.

> #
> # Console display driver support
> #
> CONFIG_VGA_CONSOLE=y
> CONFIG_MDA_CONSOLE=y
> CONFIG_DUMMY_CONSOLE=y
> 
> Disabling CONFIG_MDA_CONSOLE prevents that from happening.

That is strange. I think it might have to do with the font height issue 
someone reported just recently. Before thee was a global variable 
video_font_height. This caused issues with framebuffer consoles since 
different VCs can have different font heights. Now it is private to each 
driver. This is probable related. I will test. I have a few more updates 
to vgacon coming later one.

> This machine has:
> 
> ATI Technologies Inc Rage Mobility M3 AGP 2x (rev 02)

The patch applied is the latest work from Monta Vista to get that chipset 
working on non ix86 platforms. As usual any changes to the Mach 64 driver 
affects a bunch of other machines. 

> VGA compatible controller: nVidia Corporation NV17 [GeForce4 MX440] (rev a3)
> 
> When fbcon is enabled here the screen is full of random uninitialised gunk
> and no text is readable.

Well you are using bleeding edge code here.  

> Another machine has Cirrus hardware.  The Cirrus drivers do not compile.

Haven't ported it over to the new api yet. I have been fixing the upper 
layer lately. As soon as I'm done I plan to do extensive house cleaning. 
Alot of the drivers where ported over as fast as possible and as such 
suffered.

> A fourth machine has:
> 
> 	nVidia Corporation NV11 [GeForce2 MX] (rev b2)
> 
> this seems to work OK.  The penguins initially have a white background, then
> the background goes black, then they go away.  Perhaps that is intentional. 
> But it gets to a readable login prompt.

That is the stable code. As for the white background that is strange.

> A fifth machine has:
> 
> 	ATI Technologies Inc Radeon VE QY
> 
> it seems to work OK as well.  The penguins started out on a black background.
> I didn't test dual-head mode.

Great. I don't think the driver supports dual head mode yet.

Re: [Linux-fbdev-devel] Re: [FBDEV BK] Updates and fixes.

[James Simmons]

> > ATI Technologies Inc Rage Mobility M3 AGP 2x (rev 02)
> 
> This is a r128 based chip, which lacks some of the proper support for
> LCD stuff (especially automatic retreival of the EDID from BIOS) that
> we have in radeonfb. If you feed the driver with a suitable mode, it
> should work

Oh. A Rage 128 chip with a LCD display. Then you are out of luck right 
now.

> I haven't yet checked which codebase is in there for rivafb, I have
> started collecting various patches around & doing my own fixes, I have
> a version here that works on PPC with GeForce2 & 4, though I still
> have problems with accel on the GeForce4. It's a 2.4 code base right
> now, I haven't had time to clean that up and release a patch though.
> 
> If 2.5 still has the old codebase, then GeForce4 isn't properly
> supported yet.

We it sort of is old but yet new. I did port alot of the code from 2.4.X. 
to the latest tree. I haven't ported "thee" latest patches tho.

> Note that if you have a flat panel, then the problem of properly
> retreiving the EDID to get a suitable default mode is also present
> there, unless James added some support for it.

It there but only for PPC platforms. Hopefully with the new fbmon.c code I 
will be able to have this work on ix86 very soon.

Re: [FBDEV BK] Updates and fixes.

[James Simmons]

> which is apparently because I don't configure VIDEO_SELECT.
> 
> I did this to make it compile, but you should check what the real issue 
> is.

Okay. I will work on it.

Re: [FBDEV BK] Updates and fixes.

[Andreas Henriksson]

On Tue, Apr 15, 2003 at 08:21:43PM +0100, James Simmons wrote:
> > 	nVidia Corporation NV11 [GeForce2 MX] (rev b2)
> > 
> > this seems to work OK.  The penguins initially have a white background, then
> > the background goes black, then they go away.  Perhaps that is intentional. 
> > But it gets to a readable login prompt.
> 
> That is the stable code. As for the white background that is strange.
> 

This happens for me too (white background that goes away when there
eventually is text that hits that area and overwrites it). 
2.5.67 + extra stuff from bk + pnp + fbdev from april 11 with an Intel 8xx
integrated card.

$ lspci | grep VGA
00:02.0 VGA compatible controller: Intel Corp. 82815 CGC [Chipset Graphics Controller] (rev 02)

I also have an additional problem, the whole picture jumps slightly up
and down all the time, plus I have a thin corruptionline from side to
side that walks up and down on the screen. Like an old tv.

My lilo.conf append line looks like this:
append="video=i810fb:xres:1024,yres:768,bpp:16,hsync1:30,hsync2:82, \
vsync1:50,vsync2:75,accel"

I've tried several different modes but they all appear to give about the
same effect.

... and I'm using a TFT screen (ViewSonic ViewPanel VG181), if that
matters (not using DVI though).

At the same time I installed the 2.5 kernel I also upgraded to XFree86
4.2.1 (from 4.1, the default in debian woody) and the exact same
behaviour happened in X too (jumping slightly up and down, thin line of
corruption and I couldn't get the xvideo extension to show anything but
a blue screen in programs that used xv. [mplayer]). It all went away
when I upgraded to XFree86 4.3. Don't know if that'll help you any but I
thought it's better to mention it all.


I'll try a new kernel (with a newer fbdev patch) as soon as the dust
settles. If you want me to try out any special patch or kernel version
tell me and I'll do it as soon as I can.


Regards,
Andreas Henriksson

Re: [Linux-fbdev-devel] Re: [FBDEV BK] Updates and fixes.

[Geert Uytterhoeven]

On Tue, 15 Apr 2003, James Simmons wrote:
> > drivers/video/aty/mach64_gx.c:194: warning: initialization from incompatible pointer type
> 
> Haven't applied a few driver patches yet because I have been working on 
> the upper layer stuff.

Try this one. I thought I had sent it to Linus with my latest batch...

To: linus
Subject: [PATCH] ATI Mach64 GX compile fix

Atyfb: Add missing parts of reversal of Mobility changes, allowing ATI Mach64
GX support to compile again.

--- linux-2.5.x/drivers/video/aty/mach64_gx.c	Tue Mar 25 10:06:59 2003
+++ linux-m68k-2.5.x/drivers/video/aty/mach64_gx.c	Wed Mar 26 11:18:58 2003
@@ -119,7 +119,7 @@
 }
 
 static int aty_var_to_pll_514(const struct fb_info *info, u32 vclk_per,
-			      u32 bpp, u32 width, union aty_pll *pll)
+			      u8 bpp, union aty_pll *pll)
 {
 	/*
 	 *  FIXME: use real calculations instead of using fixed values from the old
@@ -338,7 +338,7 @@
      */
 
 static int aty_var_to_pll_18818(const struct fb_info *info, u32 vclk_per,
-				u32 bpp, u32 width, union aty_pll *pll)
+				u8 bpp, union aty_pll *pll)
 {
 	u32 MHz100;		/* in 0.01 MHz */
 	u32 program_bits;
@@ -494,7 +494,7 @@
      */
 
 static int aty_var_to_pll_1703(const struct fb_info *info, u32 vclk_per,
-			       u32 bpp, u32 width, union aty_pll *pll)
+			       u32 vclk_per, u8 bpp, union aty_pll *pll)
 {
 	u32 mhz100;		/* in 0.01 MHz */
 	u32 program_bits;
@@ -610,7 +610,7 @@
      */
 
 static int aty_var_to_pll_8398(const struct fb_info *info, u32 vclk_per,
-			       u32 bpp, u32 width, union aty_pll *pll)
+			       u32 vclk_per, u8 bpp, union aty_pll *pll)
 {
 	u32 tempA, tempB, fOut, longMHz100, diff, preDiff;
 
@@ -734,7 +734,7 @@
      */
 
 static int aty_var_to_pll_408(const struct fb_info *info, u32 vclk_per,
-			      u32 bpp, u32 width, union aty_pll *pll)
+			      u8 bpp, union aty_pll *pll)
 {
 	u32 mhz100;		/* in 0.01 MHz */
 	u32 program_bits;

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds

Re: [FBDEV BK] Updates and fixes.

[Pavel Machek]

Hi!

>    [RIVA FBDEV] Cursor fixes. Almost done. At least it looks normal most of the time.

I still see cursor problems on 2.5.68
(vesafb): softcursor is not being hidden
properly (backspace in bash creates
solid line), and "small"/"big" cursor
setting seems to somehow propagate
around virtual consoles...

At least cursor no longer hides itself...

				Pavel
-- 
				Pavel
Written on sharp zaurus, because my Velo1 broke. If you have Velo you don't need...