[Intel-gfx] Intel Arc A370M vs Linux 5.19

[Intel-gfx] Intel Arc A370M vs Linux 5.19

[Daniel J Blueman]

Dear Intel et al,

With a HP Spectre x360 16 16-f1xxx/891D (Intel i7-1260P) with an Arc
A370M GPU [1] running the latest Ubuntu 22.10 5.19.0-15-generic
kernel, we see:

i915 0000:03:00.0: Your graphics device 5693 is not properly supported
by the driver in this kernel version. To force driver probe anyway,
use i915.force_probe=5693

Since the GPU is unmanaged, battery life is around 30% of what it
could be. Unsurprisingly, adding i915.force_probe=5693 causes
additional  issues. Given a lack of BIOS option to disable the GPU, is
there any advice for Linux support or at least putting the GPU into
D3? I see only Windows drivers on the official support page [2], and
Linux 6.0-rc5 isn't buildable [3].

Thanks,
  Daniel

-- [1] 03:00.0 Display controller: Intel Corporation DG2 [Arc A370M] (rev 05)
    Subsystem: Hewlett-Packard Company Device 891d
    Control: I/O- Mem- BusMaster- SpecCycle- MemWINV- VGASnoop-
ParErr- Stepping- SERR- FastB2B- DisINTx-
    Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort-
<TAbort- <MAbort- >SERR- <PERR- INTx-
    IOMMU group: 23
    Region 0: Memory at 5f000000 (64-bit, non-prefetchable) [disabled]
[size=16M]
    Region 2: Memory at 6000000000 (64-bit, prefetchable) [disabled] [size=4G]
    Expansion ROM at <ignored> [disabled]
    Capabilities: [40] Vendor Specific Information: Len=0c <?>
    Capabilities: [70] Express (v2) Endpoint, MSI 00
        DevCap:    MaxPayload 128 bytes, PhantFunc 0, Latency L0s <64ns, L1 <1us
            ExtTag+ AttnBtn- AttnInd- PwrInd- RBE+ FLReset+ SlotPowerLimit 0W
        DevCtl:    CorrErr- NonFatalErr- FatalErr- UnsupReq-
            RlxdOrd+ ExtTag+ PhantFunc- AuxPwr- NoSnoop+ FLReset-
            MaxPayload 128 bytes, MaxReadReq 128 bytes
        DevSta:    CorrErr- NonFatalErr- FatalErr- UnsupReq- AuxPwr- TransPend-
        LnkCap:    Port #0, Speed 2.5GT/s, Width x1, ASPM L0s L1, Exit
Latency L0s <64ns, L1 <1us
            ClockPM- Surprise- LLActRep- BwNot- ASPMOptComp+
        LnkCtl:    ASPM Disabled; RCB 64 bytes, Disabled- CommClk-
            ExtSynch- ClockPM- AutWidDis- BWInt- AutBWInt-
        LnkSta:    Speed 2.5GT/s, Width x1
            TrErr- Train- SlotClk- DLActive- BWMgmt- ABWMgmt-
        DevCap2: Completion Timeout: Range B, TimeoutDis+ NROPrPrP- LTR+
             10BitTagComp+ 10BitTagReq+ OBFF Not Supported, ExtFmt+ EETLPPrefix-
             EmergencyPowerReduction Not Supported, EmergencyPowerReductionInit-
             FRS- TPHComp- ExtTPHComp-
             AtomicOpsCap: 32bit- 64bit- 128bitCAS-
        DevCtl2: Completion Timeout: 50us to 50ms, TimeoutDis- LTR+
10BitTagReq- OBFF Disabled,
             AtomicOpsCtl: ReqEn-
        LnkCap2: Supported Link Speeds: 2.5GT/s, Crosslink- Retimer-
2Retimers- DRS-
        LnkCtl2: Target Link Speed: 2.5GT/s, EnterCompliance- SpeedDis-
             Transmit Margin: Normal Operating Range,
EnterModifiedCompliance- ComplianceSOS-
             Compliance Preset/De-emphasis: -6dB de-emphasis, 0dB preshoot
        LnkSta2: Current De-emphasis Level: -6dB,
EqualizationComplete- EqualizationPhase1-
             EqualizationPhase2- EqualizationPhase3- LinkEqualizationRequest-
             Retimer- 2Retimers- CrosslinkRes: unsupported
    Capabilities: [ac] MSI: Enable- Count=1/1 Maskable+ 64bit+
        Address: 0000000000000000  Data: 0000
        Masking: 00000000  Pending: 00000000
    Capabilities: [d0] Power Management version 3
        Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA
PME(D0+,D1-,D2-,D3hot+,D3cold-)
        Status: D0 NoSoftRst+ PME-Enable- DSel=0 DScale=0 PME-
    Capabilities: [100 v1] Alternative Routing-ID Interpretation (ARI)
        ARICap:    MFVC- ACS-, Next Function: 0
        ARICtl:    MFVC- ACS-, Function Group: 0
    Capabilities: [420 v1] Physical Resizable BAR
        BAR 2: current size: 4GB, supported: 256MB 512MB 1GB 2GB 4GB
    Capabilities: [400 v1] Latency Tolerance Reporting
        Max snoop latency: 15728640ns
        Max no snoop latency: 15728640ns
    Kernel modules: i915

-- [2] https://www.intel.com/content/www/us/en/support/products/228342/graphics/intel-arc-dedicated-graphics-family/intel-arc-a-series-graphics/intel-arc-a370m-graphics.html
-- [3] https://kernel.ubuntu.com/~kernel-ppa/mainline/v6.0-rc5/
-- 
Daniel J Blueman

Re: [Intel-gfx] Intel Arc A370M vs Linux 5.19

[Rodrigo Vivi]

On Thu, Sep 15, 2022 at 09:08:08PM +0800, Daniel J Blueman wrote:
> Dear Intel et al,
> 
> With a HP Spectre x360 16 16-f1xxx/891D (Intel i7-1260P) with an Arc
> A370M GPU [1] running the latest Ubuntu 22.10 5.19.0-15-generic
> kernel, we see:
> 
> i915 0000:03:00.0: Your graphics device 5693 is not properly supported
> by the driver in this kernel version. To force driver probe anyway,
> use i915.force_probe=5693
> 
> Since the GPU is unmanaged, battery life is around 30% of what it
> could be. Unsurprisingly, adding i915.force_probe=5693 causes
> additional  issues. Given a lack of BIOS option to disable the GPU, is
> there any advice for Linux support or at least putting the GPU into
> D3? I see only Windows drivers on the official support page [2], and
> Linux 6.0-rc5 isn't buildable [3].

I believe this is what you are looking for:

echo auto | sudo tee /sys/bus/pci/devices/0000\:03\:00.0/power/control

In Linux the default is to keep the unmanaged devices in D0.
But changing the rpm to auto should transition the device to D3.

You can go further and check with the lspci -vv if there are other
unmanaged devices in the same pci root tree and also add them to the
'auto' rpm so you can even achieve D3cold in that whole device, what
gives you extra power savings.

I hope this helps for now.

> 
> Thanks,
>   Daniel
> 
> -- [1] 03:00.0 Display controller: Intel Corporation DG2 [Arc A370M] (rev 05)
>     Subsystem: Hewlett-Packard Company Device 891d
>     Control: I/O- Mem- BusMaster- SpecCycle- MemWINV- VGASnoop-
> ParErr- Stepping- SERR- FastB2B- DisINTx-
>     Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort-
> <TAbort- <MAbort- >SERR- <PERR- INTx-
>     IOMMU group: 23
>     Region 0: Memory at 5f000000 (64-bit, non-prefetchable) [disabled]
> [size=16M]
>     Region 2: Memory at 6000000000 (64-bit, prefetchable) [disabled] [size=4G]
>     Expansion ROM at <ignored> [disabled]
>     Capabilities: [40] Vendor Specific Information: Len=0c <?>
>     Capabilities: [70] Express (v2) Endpoint, MSI 00
>         DevCap:    MaxPayload 128 bytes, PhantFunc 0, Latency L0s <64ns, L1 <1us
>             ExtTag+ AttnBtn- AttnInd- PwrInd- RBE+ FLReset+ SlotPowerLimit 0W
>         DevCtl:    CorrErr- NonFatalErr- FatalErr- UnsupReq-
>             RlxdOrd+ ExtTag+ PhantFunc- AuxPwr- NoSnoop+ FLReset-
>             MaxPayload 128 bytes, MaxReadReq 128 bytes
>         DevSta:    CorrErr- NonFatalErr- FatalErr- UnsupReq- AuxPwr- TransPend-
>         LnkCap:    Port #0, Speed 2.5GT/s, Width x1, ASPM L0s L1, Exit
> Latency L0s <64ns, L1 <1us
>             ClockPM- Surprise- LLActRep- BwNot- ASPMOptComp+
>         LnkCtl:    ASPM Disabled; RCB 64 bytes, Disabled- CommClk-
>             ExtSynch- ClockPM- AutWidDis- BWInt- AutBWInt-
>         LnkSta:    Speed 2.5GT/s, Width x1
>             TrErr- Train- SlotClk- DLActive- BWMgmt- ABWMgmt-
>         DevCap2: Completion Timeout: Range B, TimeoutDis+ NROPrPrP- LTR+
>              10BitTagComp+ 10BitTagReq+ OBFF Not Supported, ExtFmt+ EETLPPrefix-
>              EmergencyPowerReduction Not Supported, EmergencyPowerReductionInit-
>              FRS- TPHComp- ExtTPHComp-
>              AtomicOpsCap: 32bit- 64bit- 128bitCAS-
>         DevCtl2: Completion Timeout: 50us to 50ms, TimeoutDis- LTR+
> 10BitTagReq- OBFF Disabled,
>              AtomicOpsCtl: ReqEn-
>         LnkCap2: Supported Link Speeds: 2.5GT/s, Crosslink- Retimer-
> 2Retimers- DRS-
>         LnkCtl2: Target Link Speed: 2.5GT/s, EnterCompliance- SpeedDis-
>              Transmit Margin: Normal Operating Range,
> EnterModifiedCompliance- ComplianceSOS-
>              Compliance Preset/De-emphasis: -6dB de-emphasis, 0dB preshoot
>         LnkSta2: Current De-emphasis Level: -6dB,
> EqualizationComplete- EqualizationPhase1-
>              EqualizationPhase2- EqualizationPhase3- LinkEqualizationRequest-
>              Retimer- 2Retimers- CrosslinkRes: unsupported
>     Capabilities: [ac] MSI: Enable- Count=1/1 Maskable+ 64bit+
>         Address: 0000000000000000  Data: 0000
>         Masking: 00000000  Pending: 00000000
>     Capabilities: [d0] Power Management version 3
>         Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA
> PME(D0+,D1-,D2-,D3hot+,D3cold-)
>         Status: D0 NoSoftRst+ PME-Enable- DSel=0 DScale=0 PME-
>     Capabilities: [100 v1] Alternative Routing-ID Interpretation (ARI)
>         ARICap:    MFVC- ACS-, Next Function: 0
>         ARICtl:    MFVC- ACS-, Function Group: 0
>     Capabilities: [420 v1] Physical Resizable BAR
>         BAR 2: current size: 4GB, supported: 256MB 512MB 1GB 2GB 4GB
>     Capabilities: [400 v1] Latency Tolerance Reporting
>         Max snoop latency: 15728640ns
>         Max no snoop latency: 15728640ns
>     Kernel modules: i915
> 
> -- [2] https://www.intel.com/content/www/us/en/support/products/228342/graphics/intel-arc-dedicated-graphics-family/intel-arc-a-series-graphics/intel-arc-a370m-graphics.html
> -- [3] https://kernel.ubuntu.com/~kernel-ppa/mainline/v6.0-rc5/
> -- 
> Daniel J Blueman

Re: [Intel-gfx] Intel Arc A370M vs Linux 5.19

[Daniel J Blueman]

On Thu, 15 Sept 2022 at 22:09, Rodrigo Vivi <rodrigo.vivi@intel.com> wrote:
> On Thu, Sep 15, 2022 at 09:08:08PM +0800, Daniel J Blueman wrote:
> > Dear Intel et al,
> >
> > With a HP Spectre x360 16 16-f1xxx/891D (Intel i7-1260P) with an Arc
> > A370M GPU [1] running the latest Ubuntu 22.10 5.19.0-15-generic
> > kernel, we see:
> >
> > i915 0000:03:00.0: Your graphics device 5693 is not properly supported
> > by the driver in this kernel version. To force driver probe anyway,
> > use i915.force_probe=5693
> >
> > Since the GPU is unmanaged, battery life is around 30% of what it
> > could be. Unsurprisingly, adding i915.force_probe=5693 causes
> > additional  issues. Given a lack of BIOS option to disable the GPU, is
> > there any advice for Linux support or at least putting the GPU into
> > D3? I see only Windows drivers on the official support page [2], and
> > Linux 6.0-rc5 isn't buildable [3].
>
> I believe this is what you are looking for:
>
> echo auto | sudo tee /sys/bus/pci/devices/0000\:03\:00.0/power/control
>
> In Linux the default is to keep the unmanaged devices in D0.
> But changing the rpm to auto should transition the device to D3.
>
> You can go further and check with the lspci -vv if there are other
> unmanaged devices in the same pci root tree and also add them to the
> 'auto' rpm so you can even achieve D3cold in that whole device, what
> gives you extra power savings.
>
> I hope this helps for now.

Yes, I was also hoping this would work as we see D3hot is supported:

# echo auto > /sys/bus/pci/devices/0000\:03\:00.0/power/control
# lspci -vvvs 03:00.0
...
Capabilities: [d0] Power Management version 3
        Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA
PME(D0+,D1-,D2-,D3hot+,D3cold-)
        Status: D0 NoSoftRst+ PME-Enable- DSel=0 DScale=0 PME-
                ^^

However it stays in D0 with PME disabled as we see. "Kernel modules:
i915" may suggest the i915 driver holds a reference to it, preventing
the transition.

Dan
-- 
Daniel J Blueman

Re: [Intel-gfx] Intel Arc A370M vs Linux 5.19

[Rodrigo Vivi]

On Thu, Sep 15, 2022 at 10:40:59PM +0800, Daniel J Blueman wrote:
> On Thu, 15 Sept 2022 at 22:09, Rodrigo Vivi <rodrigo.vivi@intel.com> wrote:
> > On Thu, Sep 15, 2022 at 09:08:08PM +0800, Daniel J Blueman wrote:
> > > Dear Intel et al,
> > >
> > > With a HP Spectre x360 16 16-f1xxx/891D (Intel i7-1260P) with an Arc
> > > A370M GPU [1] running the latest Ubuntu 22.10 5.19.0-15-generic
> > > kernel, we see:
> > >
> > > i915 0000:03:00.0: Your graphics device 5693 is not properly supported
> > > by the driver in this kernel version. To force driver probe anyway,
> > > use i915.force_probe=5693
> > >
> > > Since the GPU is unmanaged, battery life is around 30% of what it
> > > could be. Unsurprisingly, adding i915.force_probe=5693 causes
> > > additional  issues. Given a lack of BIOS option to disable the GPU, is
> > > there any advice for Linux support or at least putting the GPU into
> > > D3? I see only Windows drivers on the official support page [2], and
> > > Linux 6.0-rc5 isn't buildable [3].
> >
> > I believe this is what you are looking for:
> >
> > echo auto | sudo tee /sys/bus/pci/devices/0000\:03\:00.0/power/control
> >
> > In Linux the default is to keep the unmanaged devices in D0.
> > But changing the rpm to auto should transition the device to D3.
> >
> > You can go further and check with the lspci -vv if there are other
> > unmanaged devices in the same pci root tree and also add them to the
> > 'auto' rpm so you can even achieve D3cold in that whole device, what
> > gives you extra power savings.
> >
> > I hope this helps for now.
> 
> Yes, I was also hoping this would work as we see D3hot is supported:
> 
> # echo auto > /sys/bus/pci/devices/0000\:03\:00.0/power/control
> # lspci -vvvs 03:00.0
> ...
> Capabilities: [d0] Power Management version 3
>         Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA
> PME(D0+,D1-,D2-,D3hot+,D3cold-)
>         Status: D0 NoSoftRst+ PME-Enable- DSel=0 DScale=0 PME-
>                 ^^
> 
> However it stays in D0 with PME disabled as we see. "Kernel modules:
> i915" may suggest the i915 driver holds a reference to it, preventing
> the transition.

Oh, yes. I was thinking more on using the command line I sent when
the i915 is not probed. i.e. without using the force probe. your first
scenario.

With the i915 loaded I'd like to see the dmesg and a few of the debugfs
files under: /sys/kernel/debug/dri/0

like: i915_runtime_pm_status, i915_power_domain_info

> 
> Dan
> -- 
> Daniel J Blueman

Re: [Intel-gfx] Intel Arc A370M vs Linux 5.19

[Tvrtko Ursulin]

On 15/09/2022 15:54, Rodrigo Vivi wrote:
> On Thu, Sep 15, 2022 at 10:40:59PM +0800, Daniel J Blueman wrote:
>> On Thu, 15 Sept 2022 at 22:09, Rodrigo Vivi <rodrigo.vivi@intel.com> wrote:
>>> On Thu, Sep 15, 2022 at 09:08:08PM +0800, Daniel J Blueman wrote:
>>>> Dear Intel et al,
>>>>
>>>> With a HP Spectre x360 16 16-f1xxx/891D (Intel i7-1260P) with an Arc
>>>> A370M GPU [1] running the latest Ubuntu 22.10 5.19.0-15-generic
>>>> kernel, we see:
>>>>
>>>> i915 0000:03:00.0: Your graphics device 5693 is not properly supported
>>>> by the driver in this kernel version. To force driver probe anyway,
>>>> use i915.force_probe=5693
>>>>
>>>> Since the GPU is unmanaged, battery life is around 30% of what it
>>>> could be. Unsurprisingly, adding i915.force_probe=5693 causes
>>>> additional  issues. Given a lack of BIOS option to disable the GPU, is
>>>> there any advice for Linux support or at least putting the GPU into
>>>> D3? I see only Windows drivers on the official support page [2], and
>>>> Linux 6.0-rc5 isn't buildable [3].
>>>
>>> I believe this is what you are looking for:
>>>
>>> echo auto | sudo tee /sys/bus/pci/devices/0000\:03\:00.0/power/control
>>>
>>> In Linux the default is to keep the unmanaged devices in D0.
>>> But changing the rpm to auto should transition the device to D3.
>>>
>>> You can go further and check with the lspci -vv if there are other
>>> unmanaged devices in the same pci root tree and also add them to the
>>> 'auto' rpm so you can even achieve D3cold in that whole device, what
>>> gives you extra power savings.
>>>
>>> I hope this helps for now.
>>
>> Yes, I was also hoping this would work as we see D3hot is supported:
>>
>> # echo auto > /sys/bus/pci/devices/0000\:03\:00.0/power/control
>> # lspci -vvvs 03:00.0
>> ...
>> Capabilities: [d0] Power Management version 3
>>          Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA
>> PME(D0+,D1-,D2-,D3hot+,D3cold-)
>>          Status: D0 NoSoftRst+ PME-Enable- DSel=0 DScale=0 PME-
>>                  ^^
>>
>> However it stays in D0 with PME disabled as we see. "Kernel modules:
>> i915" may suggest the i915 driver holds a reference to it, preventing
>> the transition.
> 
> Oh, yes. I was thinking more on using the command line I sent when
> the i915 is not probed. i.e. without using the force probe. your first
> scenario.

Could it help to bind DG2 to vfio-pci and so prevent i915 touching it?

Regards,

Tvrtko

Re: [Intel-gfx] Intel Arc A370M vs Linux 5.19

[Daniel J Blueman]

On Thu, 15 Sept 2022 at 22:55, Rodrigo Vivi <rodrigo.vivi@intel.com> wrote:
>
> On Thu, Sep 15, 2022 at 10:40:59PM +0800, Daniel J Blueman wrote:
> > On Thu, 15 Sept 2022 at 22:09, Rodrigo Vivi <rodrigo.vivi@intel.com> wrote:
> > > On Thu, Sep 15, 2022 at 09:08:08PM +0800, Daniel J Blueman wrote:
> > > > Dear Intel et al,
> > > >
> > > > With a HP Spectre x360 16 16-f1xxx/891D (Intel i7-1260P) with an Arc
> > > > A370M GPU [1] running the latest Ubuntu 22.10 5.19.0-15-generic
> > > > kernel, we see:
> > > >
> > > > i915 0000:03:00.0: Your graphics device 5693 is not properly supported
> > > > by the driver in this kernel version. To force driver probe anyway,
> > > > use i915.force_probe=5693
> > > >
> > > > Since the GPU is unmanaged, battery life is around 30% of what it
> > > > could be. Unsurprisingly, adding i915.force_probe=5693 causes
> > > > additional  issues. Given a lack of BIOS option to disable the GPU, is
> > > > there any advice for Linux support or at least putting the GPU into
> > > > D3? I see only Windows drivers on the official support page [2], and
> > > > Linux 6.0-rc5 isn't buildable [3].
> > >
> > > I believe this is what you are looking for:
> > >
> > > echo auto | sudo tee /sys/bus/pci/devices/0000\:03\:00.0/power/control
> > >
> > > In Linux the default is to keep the unmanaged devices in D0.
> > > But changing the rpm to auto should transition the device to D3.
> > >
> > > You can go further and check with the lspci -vv if there are other
> > > unmanaged devices in the same pci root tree and also add them to the
> > > 'auto' rpm so you can even achieve D3cold in that whole device, what
> > > gives you extra power savings.
> > >
> > > I hope this helps for now.
> >
> > Yes, I was also hoping this would work as we see D3hot is supported:
> >
> > # echo auto > /sys/bus/pci/devices/0000\:03\:00.0/power/control
> > # lspci -vvvs 03:00.0
> > ...
> > Capabilities: [d0] Power Management version 3
> >         Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA
> > PME(D0+,D1-,D2-,D3hot+,D3cold-)
> >         Status: D0 NoSoftRst+ PME-Enable- DSel=0 DScale=0 PME-
> >                 ^^
> >
> > However it stays in D0 with PME disabled as we see. "Kernel modules:
> > i915" may suggest the i915 driver holds a reference to it, preventing
> > the transition.
>
> Oh, yes. I was thinking more on using the command line I sent when
> the i915 is not probed. i.e. without using the force probe. your first
> scenario.

My bad - a reference is held when lspci shows "Kernel driver in use:
..." which is not the case here. Accordingly, we see:
# ls /sys/bus/pci/drivers/i915
0000:00:02.0 bind  module  new_id  remove_id  uevent  unbind
...ie no "0000:03:00.0" to unbind.

> With the i915 loaded I'd like to see the dmesg and a few of the debugfs
> files under: /sys/kernel/debug/dri/0

Sure, see:
https://quora.org/a370m/{dmesg.txt,i915-debug.txt}
https://quora.org/a370m/{dmesg-forceprobe.txt,i915-debug-forceprobe.txt}

Thanks,
  Dan
--
Daniel J Blueman

Re: [Intel-gfx] Intel Arc A370M vs Linux 5.19

[Daniel J Blueman]

On Fri, 16 Sept 2022 at 00:35, Tvrtko Ursulin
<tvrtko.ursulin@linux.intel.com> wrote:
>> On 15/09/2022 15:54, Rodrigo Vivi wrote:
> > On Thu, Sep 15, 2022 at 10:40:59PM +0800, Daniel J Blueman wrote:
> >> On Thu, 15 Sept 2022 at 22:09, Rodrigo Vivi <rodrigo.vivi@intel.com> wrote:
> >>> On Thu, Sep 15, 2022 at 09:08:08PM +0800, Daniel J Blueman wrote:
> >>>> Dear Intel et al,
> >>>>
> >>>> With a HP Spectre x360 16 16-f1xxx/891D (Intel i7-1260P) with an Arc
> >>>> A370M GPU [1] running the latest Ubuntu 22.10 5.19.0-15-generic
> >>>> kernel, we see:
> >>>>
> >>>> i915 0000:03:00.0: Your graphics device 5693 is not properly supported
> >>>> by the driver in this kernel version. To force driver probe anyway,
> >>>> use i915.force_probe=5693
> >>>>
> >>>> Since the GPU is unmanaged, battery life is around 30% of what it
> >>>> could be. Unsurprisingly, adding i915.force_probe=5693 causes
> >>>> additional  issues. Given a lack of BIOS option to disable the GPU, is
> >>>> there any advice for Linux support or at least putting the GPU into
> >>>> D3? I see only Windows drivers on the official support page [2], and
> >>>> Linux 6.0-rc5 isn't buildable [3].
> >>>
> >>> I believe this is what you are looking for:
> >>>
> >>> echo auto | sudo tee /sys/bus/pci/devices/0000\:03\:00.0/power/control
> >>>
> >>> In Linux the default is to keep the unmanaged devices in D0.
> >>> But changing the rpm to auto should transition the device to D3.
> >>>
> >>> You can go further and check with the lspci -vv if there are other
> >>> unmanaged devices in the same pci root tree and also add them to the
> >>> 'auto' rpm so you can even achieve D3cold in that whole device, what
> >>> gives you extra power savings.
> >>>
> >>> I hope this helps for now.
> >>
> >> Yes, I was also hoping this would work as we see D3hot is supported:
> >>
> >> # echo auto > /sys/bus/pci/devices/0000\:03\:00.0/power/control
> >> # lspci -vvvs 03:00.0
> >> ...
> >> Capabilities: [d0] Power Management version 3
> >>          Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA
> >> PME(D0+,D1-,D2-,D3hot+,D3cold-)
> >>          Status: D0 NoSoftRst+ PME-Enable- DSel=0 DScale=0 PME-
> >>                  ^^
> >>
> >> However it stays in D0 with PME disabled as we see. "Kernel modules:
> >> i915" may suggest the i915 driver holds a reference to it, preventing
> >> the transition.
> >
> > Oh, yes. I was thinking more on using the command line I sent when
> > the i915 is not probed. i.e. without using the force probe. your first
> > scenario.
>
> Could it help to bind DG2 to vfio-pci and so prevent i915 touching it?

Bingo! Booting with vfio-pci.ids=8086:5693 in my case left the dGPU in
D3, giving the expected battery runtime.

Many thanks Tvrtko,
  Dan
-- 
Daniel J Blueman