linux-kernel - Re: [PATCH v2 2/3] PCI/DPC: Remove CONFIG_PCIE

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <18564c1f-f676-4cb8-9549-a9b87f853a3e@linux.intel.com>
Date: Fri, 1 Mar 2024 22:42:31 -0800
From: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@...ux.intel.com>
To: Bjorn Helgaas <helgaas@...nel.org>
Cc: linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org,
 Matthew W Carlis <mattc@...estorage.com>, Keith Busch <kbusch@...nel.org>,
 Lukas Wunner <lukas@...ner.de>,
 Mika Westerberg <mika.westerberg@...ux.intel.com>,
 Jesse Brandeburg <jesse.brandeburg@...el.com>,
 Bjorn Helgaas <bhelgaas@...gle.com>
Subject: Re: [PATCH v2 2/3] PCI/DPC: Remove CONFIG_PCIE_EDR


On 3/1/24 3:06 PM, Bjorn Helgaas wrote:
> On Sun, Feb 25, 2024 at 12:05:12PM -0800, Kuppuswamy Sathyanarayanan wrote:
>> On 2/22/24 2:15 PM, Bjorn Helgaas wrote:
>>> From: Bjorn Helgaas <bhelgaas@...gle.com>
>>>
>>> Previous Kconfig allowed the possibility of enabling CONFIG_PCIE_DPC
>>> without CONFIG_PCIE_EDR.  The PCI Firmware Spec, r3.3, sec 4.5.1,
>>> table 4-5, says an ACPI OS that requests control of DPC must also
>>> advertise support for EDR.
>>>
>>> Remove CONFIG_PCIE_EDR and enable the EDR code with CONFIG_PCIE_DPC so that
>>> enabling DPC also enables EDR for ACPI systems.  Since EDR is an ACPI
>>> feature, build edr.o only when CONFIG_ACPI is enabled.  Stubs cover the
>>> case when DPC is enabled without ACPI.
>>>
>>> Signed-off-by: Bjorn Helgaas <bhelgaas@...gle.com>
>>> ---
>>>  drivers/acpi/pci_root.c   |  2 +-
>>>  drivers/pci/pcie/Kconfig  | 14 ++++----------
>>>  drivers/pci/pcie/Makefile |  5 ++++-
>>>  drivers/pci/pcie/dpc.c    | 10 ----------
>>>  include/linux/pci-acpi.h  |  4 ++--
>>>  5 files changed, 11 insertions(+), 24 deletions(-)
>>>
>>> diff --git a/drivers/acpi/pci_root.c b/drivers/acpi/pci_root.c
>>> index efc292b6214e..bcaf3d3a5e05 100644
>>> --- a/drivers/acpi/pci_root.c
>>> +++ b/drivers/acpi/pci_root.c
>>> @@ -448,7 +448,7 @@ static u32 calculate_support(void)
>>>  		support |= OSC_PCI_ASPM_SUPPORT | OSC_PCI_CLOCK_PM_SUPPORT;
>>>  	if (pci_msi_enabled())
>>>  		support |= OSC_PCI_MSI_SUPPORT;
>>> -	if (IS_ENABLED(CONFIG_PCIE_EDR))
>>> +	if (IS_ENABLED(CONFIG_PCIE_DPC))	/* DPC => EDR support */
>>>  		support |= OSC_PCI_EDR_SUPPORT;
>> Since EDR internally touches AER registers, I still think we should
>> make sure OS enables AER support before advertising the EDR support.
> I guess you're suggesting that we should make it look like this?
>
>   if (host_bridge->native_aer && IS_ENABLED(CONFIG_PCIE_DPC))
>
> That doesn't seem right to me because the implementation note in PCI
> Firmware r3.3, sec 4.6.12, shows the EDR flow when firmware maintains
> control of AER and DPC, i.e., "host_bridge->native_aer" would be
> false.

No, my idea is to check for something like below:

if (IS_ENABLED(CONFIG_PCIEAER) && IS_ENABLED(CONFIG_PCIE_DPC)) or if (pci_aer_available() && IS_ENABLED(CONFIG_PCIE_DPC) to ensure AER is not disabled by noaer command line option. Since EDR handler depends on AER routines (like pci_aer_raw_clear_status()) to |clear AER registers, we need to ensure AER is enabled in kernel before advertising suppor for EDR.


>
>>>  	return support;
>>> diff --git a/drivers/pci/pcie/Kconfig b/drivers/pci/pcie/Kconfig
>>> index 8999fcebde6a..21e98289fbe9 100644
>>> --- a/drivers/pci/pcie/Kconfig
>>> +++ b/drivers/pci/pcie/Kconfig
>>> @@ -137,6 +137,10 @@ config PCIE_DPC
>>>  	  have this capability or you do not want to use this feature,
>>>  	  it is safe to answer N.
>>>  
>>> +	  On ACPI systems, this includes Error Disconnect Recover support,
>>> +	  the hybrid model that uses both firmware and OS to implement DPC,
>>> +	  as specified in the PCI Firmware Specification r3.3.
>> Nit: Include some section reference?
> I basically copied this from the PCIE_EDR help and updated the
> revision number.  But I don't think the firmware spec is a very good
> reference because EDR is defined by ACPI.  There's very little text in
> the ACPI spec about EDR, but the firmware spec does assume you know
> what *is* there.  And the ACPI spec is available to anybody, unlike
> the PCI firmware spec.
>
> +         On ACPI systems, this includes Error Disconnect Recover support,
> +         the hybrid model that uses both firmware and OS to implement DPC,
> +         as specified in ACPI r6.5, sec 5.6.6.
>
>>>  config PCIE_PTM
>>>  	bool "PCI Express Precision Time Measurement support"
>>>  	help
>>> @@ -145,13 +149,3 @@ config PCIE_PTM
>>>  
>>>  	  This is only useful if you have devices that support PTM, but it
>>>  	  is safe to enable even if you don't.
>>> -
>>> -config PCIE_EDR
>>> -	bool "PCI Express Error Disconnect Recover support"
>>> -	depends on PCIE_DPC && ACPI
>>> -	help
>>> -	  This option adds Error Disconnect Recover support as specified
>>> -	  in the Downstream Port Containment Related Enhancements ECN to
>>> -	  the PCI Firmware Specification r3.2.  Enable this if you want to
>>> -	  support hybrid DPC model which uses both firmware and OS to
>>> -	  implement DPC.

-- 
Sathyanarayanan Kuppuswamy
Linux Kernel Developer