[PATCH 2/4] radeonfb: add new module parameter to force PLL calculation

[PATCH 2/4] radeonfb: add new module parameter to force PLL calculation

[aherrman]

On my Acer Aspire 5100 (with 0x5975) I need to force PLL caclulation.
Most probably the BIOS dividers are incorrect/insufficient. Without pll
calculation console goes blank.

Signed-off-by: Andreas Herrmann <aherrman@arcor.de>
---
 drivers/video/aty/radeon_base.c |    7 ++++++-
 1 files changed, 6 insertions(+), 1 deletions(-)

diff --git a/drivers/video/aty/radeon_base.c b/drivers/video/aty/radeon_base.c
index 90a3957..f672c0c 100644
--- a/drivers/video/aty/radeon_base.c
+++ b/drivers/video/aty/radeon_base.c
@@ -268,6 +268,7 @@ static int mirror = 0;
 static int panel_yres = 0;
 static int force_dfp = 0;
 static int force_measure_pll = 0;
+static int force_pll_calc = 0;
 #ifdef CONFIG_MTRR
 static int nomtrr = 0;
 #endif
@@ -1590,7 +1591,7 @@ static int radeonfb_set_par(struct fb_info *info)
 
 		pixClock = 100000000 / rinfo->panel_info.clock;
 
-		if (rinfo->panel_info.use_bios_dividers) {
+		if (rinfo->panel_info.use_bios_dividers && !force_pll_calc) {
 			nopllcalc = 1;
 			newmode->ppll_div_3 = rinfo->panel_info.fbk_divider |
 				(rinfo->panel_info.post_divider << 16);
@@ -2495,6 +2496,8 @@ static int __init radeonfb_setup (char *options)
 			nomodeset = 1;
 		} else if (!strncmp(this_opt, "force_measure_pll", 17)) {
 			force_measure_pll = 1;
+		} else if (!strncmp(this_opt, "force_pll_calc", 17)) {
+			force_pll_calc = 1;
 		} else if (!strncmp(this_opt, "ignore_edid", 11)) {
 			ignore_edid = 1;
 #if defined(CONFIG_PM) && defined(CONFIG_X86)
@@ -2550,6 +2553,8 @@ module_param(monitor_layout, charp, 0);
 MODULE_PARM_DESC(monitor_layout, "Specify monitor mapping (like XFree86)");
 module_param(force_measure_pll, bool, 0);
 MODULE_PARM_DESC(force_measure_pll, "Force measurement of PLL (debug)");
+module_param(force_pll_calc, bool, 0400);
+MODULE_PARM_DESC(force_pll_calc, "Force calculation of PLL");
 #ifdef CONFIG_MTRR
 module_param(nomtrr, bool, 0);
 MODULE_PARM_DESC(nomtrr, "bool: disable use of MTRR registers");
-- 
1.5.3



-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/

Re: [PATCH 2/4] radeonfb: add new module parameter to force PLL calculation

[Benjamin Herrenschmidt]

On Tue, 2007-09-04 at 12:59 +0200, aherrman@arcor.de wrote:
> On my Acer Aspire 5100 (with 0x5975) I need to force PLL caclulation.
> Most probably the BIOS dividers are incorrect/insufficient. Without pll
> calculation console goes blank.
> 
> Signed-off-by: Andreas Herrmann <aherrman@arcor.de>

I don't like those tunables. First we should get a look at what values
we obtain from the BIOS. Could be something with the parsing of ATOM
BIOS. In any case, we might be able to detect we got wrong values or use
subsystem vendor/device ID to blacklist.

Ben.

> ---
>  drivers/video/aty/radeon_base.c |    7 ++++++-
>  1 files changed, 6 insertions(+), 1 deletions(-)
> 
> diff --git a/drivers/video/aty/radeon_base.c b/drivers/video/aty/radeon_base.c
> index 90a3957..f672c0c 100644
> --- a/drivers/video/aty/radeon_base.c
> +++ b/drivers/video/aty/radeon_base.c
> @@ -268,6 +268,7 @@ static int mirror = 0;
>  static int panel_yres = 0;
>  static int force_dfp = 0;
>  static int force_measure_pll = 0;
> +static int force_pll_calc = 0;
>  #ifdef CONFIG_MTRR
>  static int nomtrr = 0;
>  #endif
> @@ -1590,7 +1591,7 @@ static int radeonfb_set_par(struct fb_info *info)
>  
>  		pixClock = 100000000 / rinfo->panel_info.clock;
>  
> -		if (rinfo->panel_info.use_bios_dividers) {
> +		if (rinfo->panel_info.use_bios_dividers && !force_pll_calc) {
>  			nopllcalc = 1;
>  			newmode->ppll_div_3 = rinfo->panel_info.fbk_divider |
>  				(rinfo->panel_info.post_divider << 16);
> @@ -2495,6 +2496,8 @@ static int __init radeonfb_setup (char *options)
>  			nomodeset = 1;
>  		} else if (!strncmp(this_opt, "force_measure_pll", 17)) {
>  			force_measure_pll = 1;
> +		} else if (!strncmp(this_opt, "force_pll_calc", 17)) {
> +			force_pll_calc = 1;
>  		} else if (!strncmp(this_opt, "ignore_edid", 11)) {
>  			ignore_edid = 1;
>  #if defined(CONFIG_PM) && defined(CONFIG_X86)
> @@ -2550,6 +2553,8 @@ module_param(monitor_layout, charp, 0);
>  MODULE_PARM_DESC(monitor_layout, "Specify monitor mapping (like XFree86)");
>  module_param(force_measure_pll, bool, 0);
>  MODULE_PARM_DESC(force_measure_pll, "Force measurement of PLL (debug)");
> +module_param(force_pll_calc, bool, 0400);
> +MODULE_PARM_DESC(force_pll_calc, "Force calculation of PLL");
>  #ifdef CONFIG_MTRR
>  module_param(nomtrr, bool, 0);
>  MODULE_PARM_DESC(nomtrr, "bool: disable use of MTRR registers");


-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/

Re: [PATCH 2/4] radeonfb: add new module parameter to force PLL calculation

[aherrman]

On Tue, Sep 04, 2007 at 03:46:29PM +0200, Benjamin Herrenschmidt wrote:
> I don't like those tunables. First we should get a look at what values
> we obtain from the BIOS. Could be something with the parsing of ATOM
> BIOS. In any case, we might be able to detect we got wrong values or use
> subsystem vendor/device ID to blacklist.

So what should I looking for?
Here is the diff of the radeonfb kernel messages between 
(-)nopllcalc and (+)forcing pll calculation with my patch:

   vSync_width: 4
   clock: 7350
 Setting up default mode based on panel info
-hStart = 672, hEnd = 712, hTotal = 824
-vStart = 404, vEnd = 408, vTotal = 437
-h_total_disp = 0x4f0066           hsync_strt_wid = 0x5029a
-v_total_disp = 0x18f01b4          vsync_strt_wid = 0x40193
+hStart = 1312, hEnd = 1352, hTotal = 1464
+vStart = 804, vEnd = 808, vTotal = 837
+h_total_disp = 0x9f00b6           hsync_strt_wid = 0x5051a
+v_total_disp = 0x31f0344          vsync_strt_wid = 0x40323
 pixclock = 13605
 freq = 7350
-Console: switching to colour frame buffer device 80x25
+freq = 7350, PLL min = 20000, PLL max = 40000
+ref_div = 6, ref_clk = 1432, output_freq = 29400
+ref_div = 6, ref_clk = 1432, output_freq = 29400
+post div = 0x2
+fb_div = 0x7b
+ppll_div_3 = 0x2007b
+Console: switching to colour frame buffer device 160x50
 radeonfb (0000:01:05.0): ATI Radeon 5975 "Yu"
 radeonfb_pci_register END

"nopllcalc" results in a console 80x25 but forcing pll calculation gives
the expected result.

BTW, I am a little surprised that the display doesn't blank without
my patch as it used to in the past ...

  Oops, PCI ID 0x5975 was already added with commit
  b5f2f4d1a6d7efde39cfb5e1d034981c69f2214c

I guess I have to repeat some testing with both the older commit and my patch(es)
to sort out what is really needed to support my RS482/0x5975.


Regards,

Andreas


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

Re: [PATCH 2/4] radeonfb: add new module parameter to force PLL calculation

[aherrman]

On Tue, Sep 11, 2007 at 12:20:12AM +0200, aherrman@arcor.de wrote:
> On Tue, Sep 04, 2007 at 03:46:29PM +0200, Benjamin Herrenschmidt wrote:
> > I don't like those tunables. First we should get a look at what values
> > we obtain from the BIOS. Could be something with the parsing of ATOM
> > BIOS. In any case, we might be able to detect we got wrong values or use
> > subsystem vendor/device ID to blacklist.

     <snip>

> "nopllcalc" results in a console 80x25 but forcing pll calculation gives
> the expected result.
> 
> BTW, I am a little surprised that the display doesn't blank without
> my patch as it used to in the past ...
> 
>   Oops, PCI ID 0x5975 was already added with commit
>   b5f2f4d1a6d7efde39cfb5e1d034981c69f2214c
> 
> I guess I have to repeat some testing with both the older commit and my patch(es)
> to sort out what is really needed to support my RS482/0x5975.

Done that.
And meanwhile I found out that force_measure_pll gives the same results
like the new force_pll_calc knob. Hence that new tunable is really not needed.

By blacklisting you meant to add a workaround to reinit the card if subsystem vendor/devID
matches my combination of Acer Aspire+Radeon, right? Then I wouldn't have to add the
force_measure_pll option on boot to get the optimal resolution ...

Do we need a second Acer Aspire system that's not correctly working to justify
a blacklisting?
Well, and how should the reinit function look like?


Thanks and Regards,

Andreas


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/