Re: Question On External Display

[Rao G]

Hi Nico,

Thanks for your email & response.

The problem originally I had was that on PortC, DP/HDMI did not work for
the external display in the OS, In BIOS it looked good,
so I configured PortC as an eDP interface then the external display started
working in the OS.

Now at runtime when there is no external display, the VBIOS is causing
delay, so I want to disable PortC for eDP at runtime when there is no
external display connected.
So am trying to find which MMIO register gets set when there is external
Display connected to DisplayPort

Thanks
Rao

On Fri, Jul 30, 2021 at 3:29 PM Rao G  wrote:
> Hi Nico,
>
> Thanks for your response, though I enabled Bits 29,28,27 in 0x61110h, the
> values in 0x61114h returned 0,
> I have configured the port as eDP in VBIOS not as HDMI/DP, will that
> make a difference?
> For eDP which MMIO register need to be checked?
>
> Thanks
> Rao
>
> On Fri, Jul 30, 2021 at 2:47 PM Nico Huber  wrote:
> > Hello Rao,
> >
> > On 30.07.21 13:53, Rao G wrote:
> > > trying to see the MMIO or VBIOS data when a external display device is
> > > plugged/unplugged with DP/HDMI interface, which MMIO register is set or
> > > cleared
> >
> > I think you are looking at the right register (last one below).
> > > Any inputs are appreciated, some of the registers tried on Port B/Port C
> > > are given below. when Display is connected and when Display is not
> > > connected, both these instances
> > > return the same data. Looking for register/bit that sets/clears when
> > > external display is connected/disconnected
> > >
> > > printk(BIOS_DEBUG, "Intel Gfx BAR GTTMMADR\n");
> > > printk(BIOS_DEBUG, " DP_B %X\n", read32(bar+0x180000+0x64100));
> > > printk(BIOS_DEBUG, " DP_C %X\n", read32(bar+0x180000+0x64200));
> > > printk(BIOS_DEBUG, " HDMIC %X\n", read32(bar+0x180000+0x61160));
> > > printk(BIOS_DEBUG, " HDMIB %X\n", read32(bar+0x180000+0x61140));
> > > printk(BIOS_DEBUG, " Hotplug control %X\n",
> >
> > read32(bar+0x180000+0x61164));
> > > printk(BIOS_DEBUG, " pipeA control %X\n", read32(bar+0x180000+0x61204));
> > > printk(BIOS_DEBUG, " pipeB control %X\n", read32(bar+0x180000+0x61304));
> > > printk(BIOS_DEBUG, " porthot plug stat %X\n",
> >
> > read32(bar+0x180000+0x61114));
> >
> > According to the datasheet, the bits in this register should change but
> > only if enabled in PORT_HOTPLUG_EN (0x61110).
> >
> > Nico
> >
> > NB. There is also libgfxinit support under review [1]. In case you are
> > trying to do some native graphics init ;)
> >
> > [1] https://review.coreboot.org/q/topic:%2522baytrail-libgfxinit%2522

Question On External Display

[Nico Huber]

Hi,

On 30.07.21 16:47, Rao G wrote:
> The problem originally I had was that on PortC, DP/HDMI did not work for
> the external display in the OS, In BIOS it looked good,
> so I configured PortC as an eDP interface then the external display started
> working in the OS.

this sounds much like you are still facing the same problem. When you
configure the port as eDP, Linux' i915 will simply skip reading the
exact register we are looking (and just assume the display is present).

Have you tried to measure the hotplug signal when your display is
connected/disconnected? Maybe something is physically wrong with the
board design, cable, or display?

Nico

Question On External Display

[Rao G]

Thanks Nico.

HPD is active , measured the signal

Configured Ports
1.DDI0- PortB - DP with HDMI/DVI is good in BIOS & OS
2.DDI1- PortC - DP with HDMI/DVI is good in BIOS, i915 was turning off the
DP display (Never understood the reason)

So configured PortC with eDP

If there is no way BIOS to detect the external display with MMIO registers
when configured as eDP, Can i not turn off the DDI1/Port C at runtime in
BIOS,
am expecting some register should set with eDP as interface and when
display is connected

any clue why the i915 OS driver was turning off DP display in case 2?

Regards
Rao

On Fri, Jul 30, 2021 at 4:49 PM Nico Huber  wrote:
> Hi,
>
> On 30.07.21 16:47, Rao G wrote:
> > The problem originally I had was that on PortC, DP/HDMI did not work for
> > the external display in the OS, In BIOS it looked good,
> > so I configured PortC as an eDP interface then the external display
>
> started
> > working in the OS.
>
> this sounds much like you are still facing the same problem. When you
> configure the port as eDP, Linux' i915 will simply skip reading the
> exact register we are looking (and just assume the display is present).
>
> Have you tried to measure the hotplug signal when your display is
> connected/disconnected? Maybe something is physically wrong with the
> board design, cable, or display?
>
> Nico

Question On External Display

[Nico Huber]

On 30.07.21 18:59, Rao G wrote:
> Thanks Nico.
>
> HPD is active , measured the signal

Is your coreboot port public somewhere? If the hardware is fine, maybe
the firmware isn't?
> Configured Ports
> 1.DDI0- PortB - DP with HDMI/DVI is good in BIOS & OS
> 2.DDI1- PortC - DP with HDMI/DVI is good in BIOS, i915 was turning off the
> DP display (Never understood the reason)

It's very suspicious that these two behave differently. Does the HPD
signal (MMIO read) work for DDI0?
> So configured PortC with eDP
>
> If there is no way BIOS to detect the external display with MMIO registers
> when configured as eDP, Can i not turn off the DDI1/Port C at runtime in
> BIOS,

Maybe it's time to switch to an open-source solution? libgfxinit will
probably have the same issue, but it might be possible to get a rather
short delay.
> am expecting some register should set with eDP as interface and when
> display is connected
>
> any clue why the i915 OS driver was turning off DP display in case 2?

I assume the HPD signal doesn't get through to the software.

Nico

Question On External Display

[Rao G]

Thanks Nico for your response.
> am expecting some register should set with eDP as interface and when
> display is connected
>
> any clue why the i915 OS driver was turning off DP display in case 2?

I assume the HPD signal doesn't get through to the software.

[*Rao*]
You mean I need to check PORT_HOTPLUG_EN and PORT_HOTPLUG_STAT mmio offsets
for this issue?
BIOS display is fine, Windows logo is also seen but there is no signal when
"windows login" is reached

It had similar behaviour w.r.t to ubuntu 18.04, I configured DDI1 to "DP
with HDMI/DVI"

Can this issue be fixed with VBE data configured through BMP or it needs a
fix from graphics MMIO?

Regards
Rao

On Fri, Jul 30, 2021 at 10:27 PM Nico Huber  wrote:
> On 30.07.21 18:59, Rao G wrote:
> > Thanks Nico.
> >
> > HPD is active , measured the signal
>
> Is your coreboot port public somewhere? If the hardware is fine, maybe
> the firmware isn't?
> > Configured Ports
> > 1.DDI0- PortB - DP with HDMI/DVI is good in BIOS & OS
> > 2.DDI1- PortC - DP with HDMI/DVI is good in BIOS, i915 was turning off
>
> the
> > DP display (Never understood the reason)
>
> It's very suspicious that these two behave differently. Does the HPD
> signal (MMIO read) work for DDI0?
> > So configured PortC with eDP
> >
> > If there is no way BIOS to detect the external display with MMIO
>
> registers
> > when configured as eDP, Can i not turn off the DDI1/Port C at runtime in
> > BIOS,
>
> Maybe it's time to switch to an open-source solution? libgfxinit will
> probably have the same issue, but it might be possible to get a rather
> short delay.
> > am expecting some register should set with eDP as interface and when
> > display is connected
> >
> > any clue why the i915 OS driver was turning off DP display in case 2?
>
> I assume the HPD signal doesn't get through to the software.
>
> Nico

Question On External Display

[Matt DeVillier]

On Wed, Aug 11, 2021 at 5:05 AM Rao G  wrote:
> Thanks Nico for your response.
> > am expecting some register should set with eDP as interface and when
> > display is connected
> >
> > any clue why the i915 OS driver was turning off DP display in case 2?
>
> I assume the HPD signal doesn't get through to the software.
>
> [Rao]
> You mean I need to check PORT_HOTPLUG_EN and PORT_HOTPLUG_STAT mmio offsets for this issue?
> BIOS display is fine, Windows logo is also seen but there is no signal when "windows login" is reached

IME, that's almost always due to incorrect/missing/mismatched VBT data
in the ACPI opregion/mailbox
> It had similar behaviour w.r.t to ubuntu 18.04, I configured DDI1 to "DP with HDMI/DVI"
>
> Can this issue be fixed with VBE data configured through BMP or it needs a fix from graphics MMIO?
>
> Regards
> Rao
>
> On Fri, Jul 30, 2021 at 10:27 PM Nico Huber  wrote:
> > On 30.07.21 18:59, Rao G wrote:
> > > Thanks Nico.
> > >
> > > HPD is active , measured the signal
> >
> > Is your coreboot port public somewhere? If the hardware is fine, maybe
> > the firmware isn't?
> > > Configured Ports
> > > 1.DDI0- PortB - DP with HDMI/DVI is good in BIOS & OS
> > > 2.DDI1- PortC - DP with HDMI/DVI is good in BIOS, i915 was turning off the
> > > DP display (Never understood the reason)
> >
> > It's very suspicious that these two behave differently. Does the HPD
> > signal (MMIO read) work for DDI0?
> > > So configured PortC with eDP
> > >
> > > If there is no way BIOS to detect the external display with MMIO registers
> > > when configured as eDP, Can i not turn off the DDI1/Port C at runtime in
> > > BIOS,
> >
> > Maybe it's time to switch to an open-source solution? libgfxinit will
> > probably have the same issue, but it might be possible to get a rather
> > short delay.
> > > am expecting some register should set with eDP as interface and when
> > > display is connected
> > >
> > > any clue why the i915 OS driver was turning off DP display in case 2?
> >
> > I assume the HPD signal doesn't get through to the software.
> >
> > Nico
>
> coreboot mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]

Question On External Display

[Rao G]

Thank you Matt.

the VBT data in BIOS and VBT data captured in OS if they remain identical
would that mean this problem
is not due to VBT data copied into ACPINVS igd_opregion?

Attaching VBT data for this platform which is captured from OS

also Xrandr output on each Video Interface

On Wed, Aug 11, 2021 at 3:28 PM Matt DeVillier 
wrote:
> On Wed, Aug 11, 2021 at 5:05 AM Rao G  wrote:
> > Thanks Nico for your response.
> > > am expecting some register should set with eDP as interface and when
> > > display is connected
> > >
> > > any clue why the i915 OS driver was turning off DP display in case 2?
> >
> > I assume the HPD signal doesn't get through to the software.
> >
> > [Rao]
> > You mean I need to check PORT_HOTPLUG_EN and PORT_HOTPLUG_STAT mmio
>
> offsets for this issue?
> > BIOS display is fine, Windows logo is also seen but there is no signal
>
> when "windows login" is reached
>
> IME, that's almost always due to incorrect/missing/mismatched VBT data
> in the ACPI opregion/mailbox
> > It had similar behaviour w.r.t to ubuntu 18.04, I configured DDI1 to "DP
>
> with HDMI/DVI"
> > Can this issue be fixed with VBE data configured through BMP or it needs
>
> a fix from graphics MMIO?
> > Regards
> > Rao
> >
> > On Fri, Jul 30, 2021 at 10:27 PM Nico Huber  wrote:
> > > On 30.07.21 18:59, Rao G wrote:
> > > > Thanks Nico.
> > > >
> > > > HPD is active , measured the signal
> > >
> > > Is your coreboot port public somewhere? If the hardware is fine, maybe
> > > the firmware isn't?
> > > > Configured Ports
> > > > 1.DDI0- PortB - DP with HDMI/DVI is good in BIOS & OS
> > > > 2.DDI1- PortC - DP with HDMI/DVI is good in BIOS, i915 was turning
>
> off the
> > > > DP display (Never understood the reason)
> > >
> > > It's very suspicious that these two behave differently. Does the HPD
> > > signal (MMIO read) work for DDI0?
> > > > So configured PortC with eDP
> > > >
> > > > If there is no way BIOS to detect the external display with MMIO
>
> registers
> > > > when configured as eDP, Can i not turn off the DDI1/Port C at runtime
>
> in
> > > > BIOS,
> > >
> > > Maybe it's time to switch to an open-source solution? libgfxinit will
> > > probably have the same issue, but it might be possible to get a rather
> > > short delay.
> > > > am expecting some register should set with eDP as interface and when
> > > > display is connected
> > > >
> > > > any clue why the i915 OS driver was turning off DP display in case 2?
> > >
> > > I assume the HPD signal doesn't get through to the software.
> > >
> > > Nico
> >
> > coreboot mailing list -- [e-mail]
> > To unsubscribe send an email to [e-mail]

Question On External Display

[Rao G]

Hi Matt,

Compared VBT data in pre-OS and OS, the data is identical.

Likely issue due to MMIO registers not configured correctly defined in
below spec?
https://01.org/sites/default/files/documentation/intel_os_gfx_prm_vol10_-_di...

The LVDS display which is on DDI0 is fine so I could switch to OS and get
VBT data.
DP/HDMI/DVI on DDI1 in windows is causing the issue.

I expect coreboot is filling below mailbox's correctly?

INTEL_IGD_OPREGION_MBOX1  MBox1; // Mailbox 1: Public ACPI Methods
INTEL_IGD_OPREGION_MBOX2  MBox2; // Mailbox 2: Software SCI Inteface
INTEL_IGD_OPREGION_MBOX3  MBox3; // Mailbox 3: BIOS/Driver Communication
INTEL_IGD_OPREGION_VBT    VBT;   // Video BIOS Table (OEM customizable data)

Regards
Rao

On Thu, Aug 12, 2021 at 10:24 AM Rao G  wrote:
> Thank you Matt.
>
> the VBT data in BIOS and VBT data captured in OS if they remain identical
> would that mean this problem
> is not due to VBT data copied into ACPINVS igd_opregion?
>
> Attaching VBT data for this platform which is captured from OS
>
> also Xrandr output on each Video Interface
>
> On Wed, Aug 11, 2021 at 3:28 PM Matt DeVillier 
> wrote:
> > On Wed, Aug 11, 2021 at 5:05 AM Rao G  wrote:
> > > Thanks Nico for your response.
> > > > am expecting some register should set with eDP as interface and when
> > > > display is connected
> > > >
> > > > any clue why the i915 OS driver was turning off DP display in case 2?
> > >
> > > I assume the HPD signal doesn't get through to the software.
> > >
> > > [Rao]
> > > You mean I need to check PORT_HOTPLUG_EN and PORT_HOTPLUG_STAT mmio
> >
> > offsets for this issue?
> > > BIOS display is fine, Windows logo is also seen but there is no signal
> >
> > when "windows login" is reached
> >
> > IME, that's almost always due to incorrect/missing/mismatched VBT data
> > in the ACPI opregion/mailbox
> > > It had similar behaviour w.r.t to ubuntu 18.04, I configured DDI1 to
> >
> > "DP with HDMI/DVI"
> > > Can this issue be fixed with VBE data configured through BMP or it
> >
> > needs a fix from graphics MMIO?
> > > Regards
> > > Rao
> > >
> > > On Fri, Jul 30, 2021 at 10:27 PM Nico Huber  wrote:
> > > > On 30.07.21 18:59, Rao G wrote:
> > > > > Thanks Nico.
> > > > >
> > > > > HPD is active , measured the signal
> > > >
> > > > Is your coreboot port public somewhere? If the hardware is fine, maybe
> > > > the firmware isn't?
> > > > > Configured Ports
> > > > > 1.DDI0- PortB - DP with HDMI/DVI is good in BIOS & OS
> > > > > 2.DDI1- PortC - DP with HDMI/DVI is good in BIOS, i915 was turning
> >
> > off the
> > > > > DP display (Never understood the reason)
> > > >
> > > > It's very suspicious that these two behave differently. Does the HPD
> > > > signal (MMIO read) work for DDI0?
> > > > > So configured PortC with eDP
> > > > >
> > > > > If there is no way BIOS to detect the external display with MMIO
> >
> > registers
> > > > > when configured as eDP, Can i not turn off the DDI1/Port C at
> >
> > runtime in
> > > > > BIOS,
> > > >
> > > > Maybe it's time to switch to an open-source solution? libgfxinit will
> > > > probably have the same issue, but it might be possible to get a rather
> > > > short delay.
> > > > > am expecting some register should set with eDP as interface and when
> > > > > display is connected
> > > > >
> > > > > any clue why the i915 OS driver was turning off DP display in case 2?
> > > >
> > > > I assume the HPD signal doesn't get through to the software.
> > > >
> > > > Nico
> > >
> > > coreboot mailing list -- [e-mail]
> > > To unsubscribe send an email to [e-mail]