linux-kernel - Re: [v2] PCI: cadence: Add configuration space capability search API

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] [thread-next>] [day] [month] [year] [list]

Message-ID: <20250314203157.GA793598@bhelgaas>
Date: Fri, 14 Mar 2025 15:31:57 -0500
From: Bjorn Helgaas <helgaas@...nel.org>
To: Manivannan Sadhasivam <manivannan.sadhasivam@...aro.org>
Cc: Hans Zhang <18255117159@....com>,
	Siddharth Vadapalli <s-vadapalli@...com>, lpieralisi@...nel.org,
	kw@...ux.com, robh@...nel.org, bhelgaas@...gle.com,
	bwawrzyn@...co.com, thomas.richard@...tlin.com,
	wojciech.jasko-EXT@...tinental-corporation.com,
	linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [v2] PCI: cadence: Add configuration space capability search API

On Fri, Mar 14, 2025 at 06:35:11PM +0530, Manivannan Sadhasivam wrote:
> ...

> Even though this patch is mostly for an out of tree controller
> driver which is not going to be upstreamed, the patch itself is
> serving some purpose. I really like to avoid the hardcoded offsets
> wherever possible. So I'm in favor of this patch.
> 
> However, these newly introduced functions are a duplicated version
> of DWC functions. So we will end up with duplicated functions in
> multiple places. I'd like them to be moved (both this and DWC) to
> drivers/pci/pci.c if possible. The generic function
> *_find_capability() can accept the controller specific readl/ readw
> APIs and the controller specific private data.

I agree, it would be really nice to share this code.

It looks a little messy to deal with passing around pointers to
controller read ops, and we'll still end up with a lot of duplicated
code between __pci_find_next_cap() and __cdns_pcie_find_next_cap(),
etc.

Maybe someday we'll make a generic way to access non-PCI "config"
space like this host controller space and PCIe RCRBs.

Or if you add interfaces that accept read/write ops, maybe the
existing pci_find_capability() etc could be refactored on top of them
by passing in pci_bus_read_config_word() as the accessor.

Bjorn