Re: [PATCH] New radeonfb fork

Re: [PATCH] New radeonfb fork

[torn]

Hi Ben, I'm trying your patch on a toshiba notebook (satellite 1900-303)
with an ati radeon mobility m6. It's a bit early to say that (almost)
everything works good, but at the moment it seems so.

During the kernel compilation  I encountered some problems due tu missing
#defines. In detail, the probles was in the replacement of
PCI_DEVICE_ID_PDC_1841 in .._PDC_ADMA100 and of .._CMD_680 in .._SII_680.
Adding the old lines solved the problem. Maybe you are using a kernel
patch that modifies that code, and you did the diff against that sources.
Or (more probably) there is another logic reason I don't see.

Another thing I must tell you is that with your driver, but also with the
old radeonfb and with the xfree86 driver, some strange screen flicker
occours sometimes. When this happens I see several undefined horizontal
lines all over the LCD screen. Changing video mode a couple of times makes
the screen return to normality. I guessed that this behavoir is linked to
pixclock changes (changing the pixclock value via fbset 2 or 3 times makes
the screen flicker, another couple of pixclock changes make everyting
return to normality). Many people with notebooks with mobility M6s graphic
cards reported this problem. With windows this never happens. This is VERY
annoying.
I hope I can help the driver developement, if you want me to try something
in particular, just tell me.

Paride
torn@autistici.org





-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: Re: [PATCH] New radeonfb fork

[Benjamin Herrenschmidt]

On Wed, 2003-04-09 at 14:38, torn@autistici.org wrote:

> During the kernel compilation  I encountered some problems due tu missing
> #defines. In detail, the probles was in the replacement of
> PCI_DEVICE_ID_PDC_1841 in .._PDC_ADMA100 and of .._CMD_680 in .._SII_680.
> Adding the old lines solved the problem. Maybe you are using a kernel
> patch that modifies that code, and you did the diff against that sources.
> Or (more probably) there is another logic reason I don't see.

Yah, my patch brings pci_ids.h to 2.4.21 version, I'll fix that in
the next patch release

> Another thing I must tell you is that with your driver, but also with the
> old radeonfb and with the xfree86 driver, some strange screen flicker
> occours sometimes. When this happens I see several undefined horizontal
> lines all over the LCD screen. Changing video mode a couple of times makes
> the screen return to normality. I guessed that this behavoir is linked to
> pixclock changes (changing the pixclock value via fbset 2 or 3 times makes
> the screen flicker, another couple of pixclock changes make everyting
> return to normality). Many people with notebooks with mobility M6s graphic
> cards reported this problem. With windows this never happens. This is VERY
> annoying.
> I hope I can help the driver developement, if you want me to try something
> in particular, just tell me.

Looks like the PLL isn't locking properly. The latest driver I'll
release (today hopefully) includes a workaround that _might_ help

> Paride
> torn@autistici.org
> 
> 
> 
> 
> 
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> Linux-fbdev-devel mailing list
> Linux-fbdev-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-fbdev-devel
-- 
Benjamin Herrenschmidt <benh@kernel.crashing.org>


-------------------------------------------------------
This SF.net email is sponsored by: Etnus, makers of TotalView, The debugger 
for complex code. Debugging C/C++ programs can leave you feeling lost and 
disoriented. TotalView can help you find your way. Available on major UNIX 
and Linux platforms. Try it free. www.etnus.com

Re: Re: [PATCH] New radeonfb fork

[paride]

Benjamin Herrenschmidt wrote:
> Looks like the PLL isn't locking properly. The latest driver I'll
> release (today hopefully) includes a workaround that _might_ help

I hope so, fixing this issue will make many toshiba and hp users very 
happy. There are two more thing I think it's better to tell you:

1) This happened also with the standard radeonfb. When the driver is 
loaded, the whole screen image is moded up of about 3 or 4 mm. So the 
upper part of the screen is cut, and at the bottom there is a black 
strip. "fbset -vsync high" fixes this perfectly, so this isn't a real 
problem, maybe is just a characteristic of my hardware, telling you 
shouldn't hurt.

2) when switching from x to console, happens that when the text scrolls 
outside the screen, it reappears in the bottom part of the screen, very 
bigger, and everything gets messed up. I don't know if this could be 
related to the flicker problem, but it never happened with the old 
radeonfb. Playing a bit with fbset solves the problem. I didn't 
understand exactly what causes this, I'm investigating.

paride



-------------------------------------------------------
This SF.net email is sponsored by: Etnus, makers of TotalView, The debugger 
for complex code. Debugging C/C++ programs can leave you feeling lost and 
disoriented. TotalView can help you find your way. Available on major UNIX 
and Linux platforms. Try it free. www.etnus.com

Re: Re: [PATCH] New radeonfb fork

[Benjamin Herrenschmidt]

On Fri, 2003-04-11 at 14:47, paride wrote:
> Benjamin Herrenschmidt wrote:
> > Looks like the PLL isn't locking properly. The latest driver I'll
> > release (today hopefully) includes a workaround that _might_ help
> 
> I hope so, fixing this issue will make many toshiba and hp users very 
> happy. There are two more thing I think it's better to tell you:
> 
> 1) This happened also with the standard radeonfb. When the driver is 
> loaded, the whole screen image is moded up of about 3 or 4 mm. So the 
> upper part of the screen is cut, and at the bottom there is a black 
> strip. "fbset -vsync high" fixes this perfectly, so this isn't a real 
> problem, maybe is just a characteristic of my hardware, telling you 
> shouldn't hurt.

Just to make sure I understood you, this also happens with my
driver or not ? It's possible that your monitor EDID information
as provided by the BIOS is slightly incorrect, I'll look if there
is no parsing error on my side. Can you send me a dmesg output ?

> 2) when switching from x to console, happens that when the text scrolls 
> outside the screen, it reappears in the bottom part of the screen, very 
> bigger, and everything gets messed up. I don't know if this could be 
> related to the flicker problem, but it never happened with the old 
> radeonfb. Playing a bit with fbset solves the problem. I didn't 
> understand exactly what causes this, I'm investigating.

Isn't fixed by my latest version (URL below) ? 
http://www.penguinppc.org/~benh/radeonfb-041103-2.4.20.diff
http://www.penguinppc.org/~benh/radeonfb-041103-2.4.21-pre7.diff


Ben.



-------------------------------------------------------
This SF.net email is sponsored by: Etnus, makers of TotalView, The debugger 
for complex code. Debugging C/C++ programs can leave you feeling lost and 
disoriented. TotalView can help you find your way. Available on major UNIX 
and Linux platforms. Try it free. www.etnus.com