Re: Keyboard backlight not working on my NP900X5N laptop

[Jan Vlietland]

On 13-01-19 05:02, Darren Hart wrote:
> On Fri, Jan 04, 2019 at 05:16:38PM +0100, Jan Vlietland wrote:
> > Hi Darren,
> >
> > I understand your extra workload. For me it is just being another user
> > complaining about some bug. Sorry for that :-) Good to know the response
> > time. I will keep that in mind.
> >
> > Anyway. I have changed the static variable to "0A", recompiled the module
> > and I get the same output 'no such device'.
> >
> > However I am now running in EFI mode based on another bug:
> >
> > https://bugs.freedesktop.org/show_bug.cgi?id=109209
> >
> > ...and I see in the code....
> >
> >  ÂÂÂ struct samsung_laptop *samsung;
> >  ÂÂÂ int ret;
> >
> >  ÂÂÂ if (efi_enabled(EFI_BOOT))
> >  ÂÂÂ ÂÂÂ return -ENODEV;
> >
> >  ÂÂÂ quirks = &samsung_unknown;
> >  ÂÂÂ if (!force && !dmi_check_system(samsung_dmi_table))
> >  ÂÂÂ ÂÂÂ return -ENODEV;
> >
> >  ÂÂÂ samsung = kzalloc(sizeof(*samsung), GFP_KERNEL);
> >  ÂÂÂ if (!samsung)
> >  ÂÂÂ ÂÂÂ return -ENOMEM
> >
> > Is that EFI restriction still valid. As far as I remember Samsung repaired
> > their BIOS. Or does the driver not work in EFI mode anyway?
> Hi Jan,
>
> Taking a closer look at the driver and the git log, the driver was disabled for
> EFI because it pokes at BIOS memory and that would mean poking at a completely
> different memory map when in EFI mode - it would be expected to fail - and in
> some cases it failed by bricking the laptop. See:
>
> e0094244e41c samsung-laptop: Disable on EFI hardware
I know the story. Btw in EFI there is backwards compatibility wrt to 
memory. Should work but it is not needed anyway (read on :-)
> What appears to be needed here is for someone with the hardware and some
> experience tracing ACPI calls from whatever OS it ships with to figure out the
> new interface. I suspect it is either pure ACPI or possibly WMI, and a new
> driver may be needed.

I have the hardware and I can read C code..... so understand that it is 
disabled.... However when using the EFIVars the backlight can be enabled 
with:

"The (UEFI based) kernel creates /sys/firmware/efi/efivars with the 
variable:
KBDBacklitLvl-5af56f53-985c-47d5-920c-f1c531d06852

The set immutable flag can be disabled with:
chattr -i 
/sys/firmware/efi/efivars/KBDBacklitLvl-5af56f53-985c-47d5-920c-f1c531d06852

After that the variable can be altered from 00 - 03:
echo 0700000002 | xxd -p -r > 
/sys/firmware/efi/efivars/KBDBacklitLvl-5af56f53-985c-47d5-920c-f1c531d06852 
(GUID dependent on the manufacturer)

00 - Backlight off (always)
01 - Backlight on DIM level (by low ambient light; detected by light sensor)
02 - Backlight on NORM level (by low ambient light)
03 - Backlight on FULL level (by low ambient light)

for details my posts: 
https://bugzilla.freedesktop.org/show_bug.cgi?id=109178

This need/could/should to be added to the module. In that case the 
distribution can control the maximum brightness with the FN keys. Actual 
control is done with the Embedded controller on the board.

Please let me know your view on this.

> Have you tried this driver in BIOS mode with the OA change above?
I tested in BIOS mode and it works then.
> > On 03-01-19 03:03, Darren Hart wrote:
> > > On Mon, Dec 31, 2018 at 08:40:43PM +0100, Jan Vlietland wrote:
> > > > Hi all,
> > > Hey Jan,
> > >
> > > > Greg K-H suggested to mail you guys.
> > > >
> > > > I installed Linux 4.20.0-rc7 (downloaded, compiled and installed) on a Samsung NP900X5N laptop and have noticed 3 bugs. 2 of them I found in Bugzilla and replied on them (i915 and Nouveau issues). I am currently discussing them with an intel engineer.
> > > >
> > > > On other bug I haven't found so therefore a mail directly to you guys as maintainers.
> > > >
> > > > On my other machine, a Samsung NP900X4D (just bought it in the USA, 2017 model), the samsung-laptop.ko module is enabling the use of the keyboard backlight keys.
> > > >
> > > > It is not working on my new machine NP900X5N. My samsung-laptop.ko driver isn't loading. If I try to load it manually it complains about 'no such device".
> > > >
> > > > My Linux kernel is working in CSM mode. The module is still not loaded.
> > > That's correct.
> > >
> > > > As it is weekend I did some more reading and debugging of the module. To my understanding the module checks the model and type of the laptop. The known models and types are stored in the struct:
> > > >
> > > > static struct dmi_system_id __initdata samsung_dmi_table[]
> > > >
> > > > I wondr if the NP900X5N notebook is included in this list.
> > > >
> > > > With dmidecode -t chassis it shows:
> > > > Getting SMBIOS data from sysfs.
> > > > SMBIOS 3.0.0 present.
> > > >
> > > > Handle 0x0003, DMI type 3, 22 bytes
> > > > Chassis Information
> > > >           Manufacturer: SAMSUNG ELECTRONICS CO., LTD.
> > > >           Type: Notebook
> > > >           Lock: Not Present
> > > >           Version: N/A
> > > >           Serial Number: 0F4C91CJ900346
> > > >           Asset Tag: No Asset Tag
> > > >           Boot-up State: Safe
> > > >           Power Supply State: Safe
> > > >           Thermal State: Other
> > > >           Security Status: None
> > > >           OEM Information: 0x00000000
> > > >           Height: Unspecified
> > > >           Number Of Power Cords: 1
> > > >           Contained Elements: 0
> > > >           SKU Number: Chassis
> > > >
> > > > If I use the -u flag. The notebook value is 0x0A, not 0x10!!!
> > > >
> > > > Could that be the reason for not loading?
> > > Seems likely.
> > >
> > > > 		.matches = {
> > > > 			DMI_MATCH(DMI_SYS_VENDOR,
> > > > 					"SAMSUNG ELECTRONICS CO., LTD."),
> > > > 			DMI_MATCH(DMI_CHASSIS_TYPE, "10"), /* Notebook */
> > > > 		},
> > > >
> > > > Maybe another reason could that that either the i915 and Nouveau modules are
> > > > not working well. I get black screens with the i915 and MMIO faults with the
> > > > nouveau driver. That is another issue that I need to tackle.
> > > I would expect a different error than "no such device" in that case.
> > > I think your first thought was correct.
> > >
> > > As a simple test, I'd suggest replacing "10" with "0A" in the existing
> > > DMI_CHASSIS_TYPE match, recompile, and see if it loads and works
> > > correctly.  Would you be able to test this?
> > >
> > > > Oh happy new year :-)
> > > Happy New Year!