linux-kernel - [PATCH v2 00/13] PCI: xgene: Fix and simplify the MSI driver

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

Message-Id: <20250708173404.1278635-1-maz@kernel.org>
Date: Tue,  8 Jul 2025 18:33:51 +0100
From: Marc Zyngier <maz@...nel.org>
To: linux-pci@...r.kernel.org,
	linux-arm-kernel@...ts.infradead.org,
	linux-kernel@...r.kernel.org
Cc: Toan Le <toan@...amperecomputing.com>,
	Lorenzo Pieralisi <lpieralisi@...nel.org>,
	Krzysztof Wilczyński <kwilczynski@...nel.org>,
	Manivannan Sadhasivam <mani@...nel.org>,
	Rob Herring <robh@...nel.org>,
	Bjorn Helgaas <bhelgaas@...gle.com>,
	Thomas Gleixner <tglx@...utronix.de>
Subject: [PATCH v2 00/13] PCI: xgene: Fix and simplify the MSI driver

Having recently dipped into the xgene-msi driver to bring it to use
the MSI-parent concept, I have realised that some of it was slightly
sub-par (read: downright broken).

The driver is playing horrible tricks behind the core code, missing
proper affinity management, is terribly over-designed for no good
reason, and despite what MAINTAINERS says, completely unmaintained.

This series is an attempt to fix most of the issues, and effectively
results more or less in a full rewrite of the driver, removing a lot
of cruft and fixing the interactions with the PCI host driver in the
process (there really isn't any reason to rely on initcall ordering
anymore).

I've stopped short of repainting the MAINTAINERS file, but given how
reactive Toan Le has been, maybe that's on the cards. Patches on top
of -rc3, tested on a Mustang board. I'd really like this to hit 6.17!

* From v1:

  - Killed CPUHP_PCI_XGENE_DEAD altogether

  - Added a couple of definitions and helpers to make the hwirq/frame/group
    mapping a bit less awkward

  - More comments about the weird and wonderful behaviour of MSInIRx
    registerse

  - Collected RB from tglx, with thanks

Marc Zyngier (13):
  genirq: Teach handle_simple_irq() to resend an in-progress interrupt
  PCI: xgene: Defer probing if the MSI widget driver hasn't probed yet
  PCI: xgene: Drop useless conditional compilation
  PCI: xgene: Drop XGENE_PCIE_IP_VER_UNKN
  PCI: xgene-msi: Make per-CPU interrupt setup robust
  PCI: xgene-msi: Drop superfluous fields from xgene_msi structure
  PCI: xgene-msi: Use device-managed memory allocations
  PCI: xgene-msi: Get rid of intermediate tracking structure
  PCI: xgene-msi: Sanitise MSI allocation and affinity setting
  PCI: xgene-msi: Resend an MSI racing with itself on a different CPU
  PCI: xgene-msi: Probe as a standard platform driver
  PCI: xgene-msi: Restructure handler setup/teardown
  cpu/hotplug: Remove unused cpuhp_state CPUHP_PCI_XGENE_DEAD

 drivers/pci/controller/pci-xgene-msi.c | 428 +++++++++----------------
 drivers/pci/controller/pci-xgene.c     |  33 +-
 include/linux/cpuhotplug.h             |   1 -
 kernel/irq/chip.c                      |   8 +-
 4 files changed, 187 insertions(+), 283 deletions(-)

-- 
2.39.2