G73JH(probably other 5-series) Sleep/wake and shutdown issue

[maromi]

No ideas without ioreg.

If you don't want to try my suggestions, that's fine...

I did try empty ports dict and i tried with empty IOProviderMergeProperties dict still no changes on either

ioreg is IOProviderMergeProperties containing only kUSBCompanion false and DSDT _UPC patched EH01/EH02 from 00,FF to FF,FF

 

[RehabMan]

I did try empty ports dict and i tried with empty IOProviderMergeProperties dict still no changes on either

ioreg is IOProviderMergeProperties containing only kUSBCompanion false and DSDT _UPC patched EH01/EH02 from 00,FF to FF,FF

The ioreg you provide here still has active ports. Both ports on the EH01/EH02 controller and on the associated hub.

No port injector is active on either.

 

[maromi]

The ioreg you provide here still has active ports. Both ports on the EH01/EH02 controller and on the associated hub.

No port injector is active on either.

of course there are active ports i said i patched _UPC on the controller ports only. if i have 00,ff then there aren't any active ports anywhere and shutdown and sleep still don't work

 

[RehabMan]

of course there are active ports i said i patched _UPC on the controller ports only. if i have 00,ff then there aren't any active ports anywhere and shutdown and sleep still don't work

My suggestion was for an empty 'ports' dictionary using a port injector.

 

[maromi]

My suggestion was for an empty 'ports' dictionary using a port injector.

when i use only the injector empty 'ports' dictionary no hub can be present because the _UPC is 00,FF for the port it needs and in which case ill repeat no shutdown or sleep.

 

[RehabMan]

when i use only the injector empty 'ports' dictionary no hub can be present because the _UPC is 00,FF for the port it needs and in which case ill repeat no shutdown or sleep.

With port injectors, _UPC doesn't matter.... it is not used. That is the purpose of port injectors -- to override ACPI.

 

[maromi]

With port injectors, _UPC doesn't matter.... it is not used. That is the purpose of port injectors -- to override ACPI.

on real 5 series macs there isn't _UPC for those ports(PC does) only the ones on internalhub do. If i don't define the port in the injector it defaults to acpi for the missing info on connectable and/or port type.

The built-in ehc1/2 is 1 port type 255.
My dsdt is 00, FF if I remove port type from eh01/02 injector its still type 255 from my dsdt and empty ports dict will just use acpi

 

[RehabMan]

on real 5 series macs there isn't _UPC for those ports(PC does) only the ones on internalhub do. If i don't define the port in the injector it defaults to acpi for the missing info on connectable and/or port type.

My suggestion (read carefully): is to use an injector with an empty 'ports' dictionary.

With an empty 'ports' dictionary injected, the controller driver will load but no ports will work (because all ports will be powered off).

This is quite different from not injecting a ports dictionary, where the system then queries ACPI for the port information.

 

[maromi]

My suggestion (read carefully): is to use an injector with an empty 'ports' dictionary.

With an empty 'ports' dictionary injected, the controller driver will load but no ports will work (because all ports will be powered off).

This is quite different from not injecting a ports dictionary, where the system then queries ACPI for the port information.

                <key>ports</key>
                <dict>
                </dict>

results in acpi overriding it.

            <key>IOProviderMergeProperties</key>
            <dict>
                <key>kUSBCompanion</key>
                <false/>
            </dict>

results acpi overriding it

                <key>port-count</key>
                <data>AAAAAA==</data>
                <key>ports</key>
                <dict>
                </dict>

ACPI overrides again

whatever it is acpi overrrides it

 

[RehabMan]

                <key>ports</key>
                <dict>
                </dict>

results in acpi overriding it.

Not in my experience. You'd have to show ioreg to prove what you're doing.