Any outlook on i915 backlight support

[neelchauhan]

A month ago, you committed https://github.com/FreeBSDDesktop/freebsd-base-graphics/commit/620b88921d6c6cebafa60787a991d1621335d5b4 and https://github.com/FreeBSDDesktop/freebsd-base-graphics/commit/50d21cc6cc63f6f87fffc859b87c8a64c4c48665 to add i915 backlight support, but since then, no progress has been made (from what I know) this.

On TrueOS 2016-08-31 (which uses drm-next-4.7), there is no way to set the amount of backlight on the Dell Inspiron 7352 (which unlike ThinkPad models, handles brightness through the GPU).

So is there any outlook on finishing the i915 backlight support (eg. add the acpi_video bits)?

[mattmacy]

So is there any outlook on finishing the i915 backlight support (eg. add the acpi_video bits)?

Yes, it's just taken a backseat to stability and image corruption issues.

[abishai]

Maybe this port can set backlight? http://www.freshports.org/graphics/intel-backlight/

[neelchauhan]

It does not work on my laptop. Looking at the source, it only works up to Intel Haswell.

Update: I have also tried this, without success. Looks like my laptop will be running Arch for a bit longer.

[yanko-yankulov]

Here is a quick patch to make intel_backlight work on my skylake system

If you are interested give it a try. There should be no dependency on the drm code.

diff -u intel_backlight_fbsd-93e1fe4/intel_backlight.c intel_backlight-patched/intel_backlight.c
--- intel_backlight_fbsd-93e1fe4/intel_backlight.c      2015-07-06 01:32:27.000000000 +0300
+++ intel_backlight-patched/intel_backlight.c   2017-01-21 12:32:30.000000000 +0200
@@ -69,12 +69,29 @@
 int main(int argc, char** argv)
 {
        uint32_t current, max, min;
+       uint32_t currentReg, maxReg;
        int result;
-
+       
        intel_get_mmio(intel_get_pci_device());
-
-       current = reg_read(BLC_PWM_CPU_CTL) & BACKLIGHT_DUTY_CYCLE_MASK;
-       max = reg_read(BLC_PWM_PCH_CTL2) >> 16;
+       intel_check_pch();
+       
+       switch( pch )
+       {
+           case PCH_LPT:
+           case PCH_SPT:
+           case PCH_KBP:
+               currentReg = 0xc8254;
+               maxReg = 0xc8254;
+               break;
+               
+           default:
+               currentReg = BLC_PWM_CPU_CTL;
+               maxReg = BLC_PWM_PCH_CTL2;
+               break;
+       }
+       
+       current = reg_read(currentReg) & BACKLIGHT_DUTY_CYCLE_MASK;
+       max = reg_read(maxReg) >> 16;

        min = 0.5 + 0.5 * max / 100.0;  // 0.5%
        /*
@@ -99,9 +116,9 @@
                        v = min;
                else if (v > max)
                        v = max;
-               reg_write(BLC_PWM_CPU_CTL,
-                         (reg_read(BLC_PWM_CPU_CTL) &~ BACKLIGHT_DUTY_CYCLE_MASK) | v);
-               (void) reg_read(BLC_PWM_CPU_CTL);
+               reg_write(currentReg,
+                         (reg_read(currentReg) &~ BACKLIGHT_DUTY_CYCLE_MASK) | v);
+               (void) reg_read(currentReg);
                result = 0.5 + v * 100.0 / max;
                printf ("set backlight to %d%% (%d/%d)\n", result, v, max);
        }
diff -u intel_backlight_fbsd-93e1fe4/intel_gpu_tools.h intel_backlight-patched/intel_gpu_tools.h
--- intel_backlight_fbsd-93e1fe4/intel_gpu_tools.h      2015-07-06 01:32:27.000000000 +0300
+++ intel_backlight-patched/intel_gpu_tools.h   2017-01-21 12:28:13.000000000 +0200
@@ -103,6 +103,8 @@
        PCH_IBX,
        PCH_CPT,
        PCH_LPT,
+       PCH_SPT,
+       PCH_KBP,
 };

 extern enum pch_type pch;
diff -u intel_backlight_fbsd-93e1fe4/intel_pci.c intel_backlight-patched/intel_pci.c
--- intel_backlight_fbsd-93e1fe4/intel_pci.c    2015-07-06 01:32:27.000000000 +0300
+++ intel_backlight-patched/intel_pci.c 2017-01-21 12:29:09.000000000 +0200
@@ -115,6 +115,13 @@
        case 0x9c00:
                pch = PCH_LPT;
                break;
+       case 0xA100:
+       case 0x9D00:
+               pch = PCH_SPT;
+               break;
+       case 0xA200:
+               pch = PCH_KBP;
+               break;
        default:
                pch = PCH_NONE;
                return;

[neelchauhan]

I don't know if it will work on Broadwell, but since I posted this I switched to an IvyBridge HP EliteBook 2570p.

[yanko-yankulov]

Honestly I dont know either. However if someome is willing to test it I might be able to help in case it fails. The proper way would deffinitely be to handle it in the driver, as there are multiple cases in the code. However the intel-backlight seems to do it most of the time.

[trombonehero]

Fantastic: this works on my Skylake system. I didn't know that I'd been running at something like 30% brightness all this time. Thanks!

Jon

On 21 January 2017 at 14:07, yanko-yankulov notifications@github.com wrote:

Honestly I dont know either. However if someome is willing to test it I might be able to help in case it fails. The proper way would deffinitely be to handle it in the driver, as there are multiple cases in the code. However the intel-backlight seems to do it most of the time.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/FreeBSDDesktop/freebsd-base-graphics/issues/84#issuecomment-274275929, or mute the thread https://github.com/notifications/unsubscribe-auth/AAFPSaRV2Vh6jFI6mxhcyfsdRjrVWm7Yks5rUkJcgaJpZM4J4hsR .

-- Jonathan Anderson

jonathan.anderson@ieee.org

[grahamperrin]

Might https://github.com/FreeBSDDesktop/freebsd-base-graphics/issues/84#issuecomment-274253630 work where KMS is used for Intel graphics?

[yanko-yankulov]

Yes. I am using it with kms

[martin-beran]

Your patch works perfectly on my Lenovo ThinkPad E460 with Skylake running FreeBSD 11.0. With acpi_ibm kernel module and a call of intel_backlight from /etc/devd.conf, I am even able to use brightness control buttons. Could you please propagate your patch to the graphics/intel-backlight port?

Martin Beran

[johalun]

We need to update the port to use this repo instead https://github.com/FreeBSDDesktop/intel_backlight_fbsd

It's fixed for newer systems and added some argument validation checks.

[mattmacy]

@johalun please close this if resolved

[johalun]

Still need to update the port to use our repo.

[mattmacy]

@johalun That's a bug that should be filed with the port maintainer. Not us.