Re: [PATCH] cxl/acpi: Defer probe when ACPI0016 PCI root bridge is not ready

[Dave Jiang]

On 5/18/26 6:55 PM, Chen Pei wrote:
> On Fri, 15 May 2026 12:24:50 -0700, Dan Williams wrote:
> 
>> If the platform is defined to defer PCI root scans then this dependency
>> must be declared. Specifically firmware needs to tell Linux about the
>> dependency given it does not order PCI enumeration before ACPI0017
>> enumeration by default. Something like: 
>>
>>   Device (CXLM) { // ACPI0017
>>       Name (_HID, "ACPI0017")
>>       Name (_DEP, Package () {
>>           \_SB.CXL0, // ACPI0016 host bridge
>>           \_SB.CXL1,
>>       })
>>   }
>>
>> ...in the firmware, and then:
>>
>> acpi_dev_clear_dependencies()
>>
>> ...for each acpi_pci_root_add().
>>
>> Then ACPI0017 will naturally await all of the ACPI0016 devices that the
>> firmware knows about.
> 
> Hi Dan,
> 
> Thanks for the guidance. The _DEP approach is clearly the right way
> to handle this - it's the proper ACPI standard mechanism for declaring
> device dependencies and avoids the permanent defer risk that Dave and
> others raised with the EPROBE_DEFER approach.
> 
> I plan to abandon this patch and implement the _DEP solution instead.
> Let me confirm my understanding of the two changes needed:
> 
> 1. Firmware (QEMU DSDT): Add a _DEP package to the ACPI0017 device
>    that references all ACPI0016 CXL host bridge devices in the system.
> 
> 2. Kernel (drivers/acpi/pci_root.c): Add acpi_dev_clear_dependencies()
>    at the end of acpi_pci_root_add() so that once an ACPI0016 device
>    is bound, its consumers (ACPI0017) are notified and can proceed
>    with probing.

Please submit a patch upstream for this. Thanks!

> 
> Is my understanding correct? Any other changes or considerations
> I might be missing?
> 
> Thanks,
> Pei