Re: [FBDEV UPDATE] Newer patch.

Re: [FBDEV UPDATE] Newer patch.

[Meelis Roos]

> I have fixed the problems you have reported. I have a newer patch. Note
> this is updated with the LCD support. I like to see if the patch works on
> sparc. I has updates from the latest 2.4.X kernels. Please give it a try.

sparc64 (Ultra 5 with onboard mach64), debian unstable.

With all Mach64 suboptions set to Yes:

drivers/video/aty/atyfb_base.c: In function `atyfb_init':
drivers/video/aty/atyfb_base.c:2476: error: `XL_CHIP_ID' undeclared (first use in this function)
drivers/video/aty/atyfb_base.c:2476: error: (Each undeclared identifier is reported only once
drivers/video/aty/atyfb_base.c:2476: error: for each function it appears in.)
drivers/video/aty/atyfb_base.c:2552: warning: implicit declaration of function `aty_ld_pll_ct'
drivers/video/aty/atyfb_base.c:2341: warning: unused variable `lcd_ofs'
drivers/video/aty/atyfb_base.c:2342: warning: unused variable `driv_inf_tab'
drivers/video/aty/atyfb_base.c:2342: warning: unused variable `sig'
drivers/video/aty/atyfb_base.c:2342: warning: unused variable `rom_addr'

There are actually 2 lines with XL_CHIP_ID and this is not seen in any
other place under drivers/video/aty nor in include directory. Maybe
IS_XL(pdev->device) is the right thing?

-- 
Meelis Roos (mroos@linux.ee)

Re: [FBDEV UPDATE] Newer patch.

[Meelis Roos]

> drivers/video/aty/atyfb_base.c: In function `atyfb_init':
> drivers/video/aty/atyfb_base.c:2476: error: `XL_CHIP_ID' undeclared (first use in this function)

These went away with IS_XL(...), I don't know if this is actually
correct change.

> drivers/video/aty/atyfb_base.c:2552: warning: implicit declaration of function `aty_ld_pll_ct'
> drivers/video/aty/atyfb_base.c:2341: warning: unused variable `lcd_ofs'
> drivers/video/aty/atyfb_base.c:2342: warning: unused variable `driv_inf_tab'
> drivers/video/aty/atyfb_base.c:2342: warning: unused variable `sig'
> drivers/video/aty/atyfb_base.c:2342: warning: unused variable `rom_addr'

These of course remain.

Since the error is now go I get to a next bug:

make -f scripts/Makefile.build obj=drivers/video/console
/bin/sh: -c: line 1: unexpected EOF while looking for matching `''
/bin/sh: -c: line 2: syntax error: unexpected end of file
make[3]: *** [drivers/video/console/promcon_tbl.c] Error 2


-- 
Meelis Roos (mroos@linux.ee)

Re: [FBDEV UPDATE] Newer patch.

[Ben Collins]

On Thu, Oct 30, 2003 at 06:38:46PM +0000, James Simmons wrote:
> 
> > I noticed one thing, and that is that the mach64 used to use software
> > cursor it seems (I remember wondering why atyfb_cursor was never used
> > anywhere). It's now using the hw cursor.
> 
> Yeap. I'm in the process of getting several driver to use there hardware 
> cursors.
>  
> > Also, I notice with this new code that the random vertical shifting of
> > the console doesn't occur anymore like it does with current 2.6.0-test8
> > code. For as long as I can remember 2.6.0-test, and way back into
> > 2.5.5x, this has been a problem with highly active console programs
> > (mutt, vim, etc...). Good to see it's going away :)
> 
> :-)
> 
> I have fixed the problems you have reported. I have a newer patch. Note 
> this is updated with the LCD support. I like to see if the patch works on 
> sparc. I has updates from the latest 2.4.X kernels. Please give it a try.
> 
> http://phoenix.infradead.org/~jsimmons/fbdev.diff.gz
> 
> Let me know the results.

FYI, this new code is working for me on my Blade100. The cursor is much
better now.

-- 
Debian     - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/
WatchGuard - http://www.watchguard.com/

Re: [FBDEV UPDATE] Newer patch.

[James Simmons]

> I noticed one thing, and that is that the mach64 used to use software
> cursor it seems (I remember wondering why atyfb_cursor was never used
> anywhere). It's now using the hw cursor.

Yeap. I'm in the process of getting several driver to use there hardware 
cursors.
 
> Also, I notice with this new code that the random vertical shifting of
> the console doesn't occur anymore like it does with current 2.6.0-test8
> code. For as long as I can remember 2.6.0-test, and way back into
> 2.5.5x, this has been a problem with highly active console programs
> (mutt, vim, etc...). Good to see it's going away :)

:-)

I have fixed the problems you have reported. I have a newer patch. Note 
this is updated with the LCD support. I like to see if the patch works on 
sparc. I has updates from the latest 2.4.X kernels. Please give it a try.

http://phoenix.infradead.org/~jsimmons/fbdev.diff.gz

Let me know the results.

Re: [FBDEV UPDATE] Newer patch.

[Ben Collins]

On Thu, Oct 23, 2003 at 11:50:48PM +0100, James Simmons wrote:
> 
> > The cursor has changed from a nice underline to a solid white block. 
> 
> I seen the problem. Its the wrong color for the background color for the 
> cursor. I haven't been able to figure out why it went wrong. The specs are 
> not to clear on this.
> 
> > Not only that,
> > but the block is bigger than the font it is over (if I am on top of
> > adjacent letters, it covers the entire letter I am on, plus a couple of
> > pixels of the letter to the right).
> 
> Ug. That code is straight from the old driver. Will fix.
> 
> > In additition, the cursor now disappears while typing, and navigating
> > around (on the command line left and right, or even in an editor when
> > moving the cursor up and down). This disappearing while typing or
> > navigating is _really_ annoying. If I go left or right a lot, I have to
> > keep stopping to see where the cursor actually is.
> 
> I seen this problem last night with the NVIDIA fbdev driver. I think I 
> know what the problem is. I will try a fix tonight. 

I noticed one thing, and that is that the mach64 used to use software
cursor it seems (I remember wondering why atyfb_cursor was never used
anywhere). It's now using the hw cursor.

Also, I notice with this new code that the random vertical shifting of
the console doesn't occur anymore like it does with current 2.6.0-test8
code. For as long as I can remember 2.6.0-test, and way back into
2.5.5x, this has been a problem with highly active console programs
(mutt, vim, etc...). Good to see it's going away :)

-- 
Debian     - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/
WatchGuard - http://www.watchguard.com/

Re: [FBDEV UPDATE] Newer patch.

[James Simmons]

> The cursor has changed from a nice underline to a solid white block. 

I seen the problem. Its the wrong color for the background color for the 
cursor. I haven't been able to figure out why it went wrong. The specs are 
not to clear on this.

> Not only that,
> but the block is bigger than the font it is over (if I am on top of
> adjacent letters, it covers the entire letter I am on, plus a couple of
> pixels of the letter to the right).

Ug. That code is straight from the old driver. Will fix.

> In additition, the cursor now disappears while typing, and navigating
> around (on the command line left and right, or even in an editor when
> moving the cursor up and down). This disappearing while typing or
> navigating is _really_ annoying. If I go left or right a lot, I have to
> keep stopping to see where the cursor actually is.

I seen this problem last night with the NVIDIA fbdev driver. I think I 
know what the problem is. I will try a fix tonight. 

Re: [FBDEV UPDATE] Newer patch.

[Ben Collins]

On Wed, Oct 22, 2003 at 06:31:22PM +0100, James Simmons wrote:
> 
> Hi folks. 
> 
>   I have a new patch against 2.6.0-test8. This patch is a few fixes and I 
> added back in functionality for switching the video mode for fbcon via 
> fbset again. Give it a try and let me know the results.
> 
> http://phoenix.infradead.org/~jsimmons/fbdev.diff.gz

The changes to mach64_cursor.c really bork things up somehow. The cursor
has changed from a nice underline to a solid white block. Not only that,
but the block is bigger than the font it is over (if I am on top of
adjacent letters, it covers the entire letter I am on, plus a couple of
pixels of the letter to the right).

In additition, the cursor now disappears while typing, and navigating
around (on the command line left and right, or even in an editor when
moving the cursor up and down). This disappearing while typing or
navigating is _really_ annoying. If I go left or right a lot, I have to
keep stopping to see where the cursor actually is.

FYI, this is on an UltraSPARC Blade 100, Mach64. Atleast things didn't
break completely :) Definitely need this fixed though.

-- 
Debian     - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/
WatchGuard - http://www.watchguard.com/

Re: [FBDEV UPDATE] Newer patch.

[Geert Uytterhoeven]

On Wed, 22 Oct 2003, James Simmons wrote:
>   I have a new patch against 2.6.0-test8. This patch is a few fixes and I 
> added back in functionality for switching the video mode for fbcon via 
> fbset again. Give it a try and let me know the results.
> 
> http://phoenix.infradead.org/~jsimmons/fbdev.diff.gz

This patch accidentally includes drivers/video/logo/logo_linux_clut224.c.

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 UPDATE] Newer patch.

[James Simmons]

>     The attached patch is needed to make tdfxfb compile after
> applying this patch and also in test8-mm1 (so presumably in your older
> patch as well) (tdfxfb_imageblt calls cfb_imageblt).

Thanks.
 
>     tdfx is still badly broken in -mm1 both before and after replacing
> the older fbdev patch in -mm1 with your new one.  The behavior is much
> the same as reported with other drivers -- out of range frequencies
> and the same backtraces.  With fbset working I can set a new
> resolution which gets me a barely usable console -- lots of
> artifacts.

The out frequencies range problems existed before the api change for many 
drivers. Now that i2c support is being added to drivers this will go away.
The backtraces is from having debug info turned on. I haven't traced that 
problem yet. Are the artifacts from shrinking the screen or enlarging it?

Re: [FBDEV UPDATE] Newer patch.

[jhf]

[-- Attachment #1.1: Type: text/plain, Size: 1081 bytes --]

On Wed, Oct 22, 2003 at 06:31:22PM +0100, James Simmons wrote:
> 
> Hi folks. 
> 
>   I have a new patch against 2.6.0-test8. This patch is a few fixes and I 
> added back in functionality for switching the video mode for fbcon via 
> fbset again. Give it a try and let me know the results.
> 
> http://phoenix.infradead.org/~jsimmons/fbdev.diff.gz

    The attached patch is needed to make tdfxfb compile after
applying this patch and also in test8-mm1 (so presumably in your older
patch as well) (tdfxfb_imageblt calls cfb_imageblt).

    tdfx is still badly broken in -mm1 both before and after replacing
the older fbdev patch in -mm1 with your new one.  The behavior is much
the same as reported with other drivers -- out of range frequencies
and the same backtraces.  With fbset working I can set a new
resolution which gets me a barely usable console -- lots of
artifacts.

    I don't have time to test against vanilla -test8; maybe later.
Thanks!


-- 
Joseph Fannin
jhf@rivenstone.net

"That's all I have to say about that." -- Forrest Gump.

[-- Attachment #1.2: out --]
[-- Type: text/plain, Size: 924 bytes --]

diff -aur linux-2.6.0-test8-mm1_orig/drivers/video/Makefile /usr/src/linux-2.6.0-test8-mm1_duo/drivers/video/Makefile
--- linux-2.6.0-test8-mm1_orig/drivers/video/Makefile	2003-10-22 15:30:33.000000000 -0400
+++ /usr/src/linux-2.6.0-test8-mm1_duo/drivers/video/Makefile	2003-10-22 16:38:27.000000000 -0400
@@ -32,7 +32,7 @@
 obj-$(CONFIG_FB_CYBER)            += cyberfb.o
 obj-$(CONFIG_FB_CYBER2000)        += cyber2000fb.o cfbfillrect.o cfbcopyarea.o cfbimgblt.o
 obj-$(CONFIG_FB_SGIVW)            += sgivwfb.o cfbfillrect.o cfbcopyarea.o cfbimgblt.o
-obj-$(CONFIG_FB_3DFX)             += tdfxfb.o
+obj-$(CONFIG_FB_3DFX)             += tdfxfb.o cfbimgblt.o
 obj-$(CONFIG_FB_MAC)              += macfb.o macmodes.o cfbfillrect.o cfbcopyarea.o cfbimgblt.o 
 obj-$(CONFIG_FB_HP300)            += hpfb.o cfbfillrect.o cfbimgblt.o
 obj-$(CONFIG_FB_OF)               += offb.o cfbfillrect.o cfbimgblt.o cfbcopyarea.o

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

[FBDEV UPDATE] Newer patch.

[James Simmons]

Hi folks. 

  I have a new patch against 2.6.0-test8. This patch is a few fixes and I 
added back in functionality for switching the video mode for fbcon via 
fbset again. Give it a try and let me know the results.

http://phoenix.infradead.org/~jsimmons/fbdev.diff.gz