Re: [PATCH v5 16/16] PCI/AER: Enable internal errors for CXL Upstream and Downstream Switch Ports

[Ira Weiny <ira.weiny@intel.com>]

Bowman, Terry wrote:
> 
> 
> 
> On 1/14/2025 5:45 PM, Ira Weiny wrote:
> > Bowman, Terry wrote:
> >>
> >>
> >> On 1/14/2025 5:26 PM, Ira Weiny wrote:
> >>> Terry Bowman wrote:
> >>>> The AER service driver enables PCIe Uncorrectable Internal Errors (UIE) and
> >>>> Correctable Internal errors (CIE) for CXL Root Ports. The UIE and CIE are
> >>>> used in reporting CXL Protocol Errors. The same UIE/CIE enablement is
> >>>> needed for CXL Upstream Switch Ports and CXL Downstream Switch Ports
> >>>> inorder to notify the associated Root Port and OS.[1]
> >>>>
> >>>> Export the AER service driver's pci_aer_unmask_internal_errors() function
> >>>> to CXL namespace.
> >>>>
> >>>> Remove the function's dependency on the CONFIG_PCIEAER_CXL kernel config
> >>>> because it is now an exported function.
> >>> This seems wrong to me.  As of this patch CXL_PCI requires PCIEAER_CXL for
> >>> the AER code to handle the errors which were just enabled.
> >>>
> >>> To keep PCIEAER_CXL optional pci_aer_unmask_internal_errors() should be
> >>> stubbed out in aer.h if !CONFIG_PCIEAER_CXL.
> >>>
> >>> Ira
> >> Bjorn (I believe in v1 or v2) directed me to remove
> >> pci_aer_unmask_internal_errors() dependency on PCIEAER_CXL because it is
> >> now exported. He wants the behavior for other users (and subsystems) to
> >> be consistent with/without the PCIEAER_CXL setting.
> >>
> > I see...  If PCIEAER_CXL is not enabled why even set the cxl error
> > handlers and enable these?
> >
> > I guess this is just adding some code which eventually calls
> > handles_cxl_errors() which returns false in the !PCIEAER_CXL case?
> >
> > Ira
> 
> Re-sending because I somehow sent from Outlook earlier.
> 
> cxl_dport_init_ras_reporting() and cxl_uport_init_ras_reporting() assign the error 
> handlers and are within #ifdef PCIEAER_CXL. The stubs are in cxl.h.
> 
> Correct. handles_cxl_errors() returns false in the !PCIEAER_CXL case.
> 

That is a bit convoluted...  But I'm not sure how to get the cross Kconfig
dependencies set to eliminate the set up.  :-/

So I guess it is fine.

Ira