[PATCH v8 0/2] perf: Add Raspberry Pi AXI PMU driver

[Ian Rogers <irogers@google.com>]

This patch series adds an uncore Performance Monitoring Unit (PMU) driver
for Broadcom AXI system and VideoCore VPU performance monitors found on
Raspberry Pi SoCs (BCM2835 through BCM2712 / Raspberry Pi 1 through 5).

Motivation & Background
=======================
Currently, Linux lacks a standard perf-API compatible driver for the
Broadcom AXI performance counter blocks on Raspberry Pi platforms. Prior
out-of-tree vendor solutions relied on custom debugfs nodes and ad-hoc
kthreads, preventing integration with standard Linux perf tooling.

This driver implements standard `struct pmu` hardware uncore callbacks
under `drivers/perf/`, exposing human-readable sysfs event aliases, unit
scaling (`Bytes`), and bus filtering directly to user space.

Key Architectural Improvements & Features
=========================================
1. Standard Linux Perf Integration:
   - Exposes uncore AXI interconnect events via sysfs boundaries.
   - Supports event sampling and hardware counter accumulation,
     automatically managing 31-bit hardware counter wraparound.

2. CPU Hotplug Support (`cpuhp`):
   - Registers dynamic CPU hotplug notifiers (`CPUHP_AP_ONLINE_DYN`).
   - Automatically migrates PMU context to an online CPU core when a
     designated CPU goes offline, avoiding stale uncore state.

3. Hybrid Memory-Mapped & Mailbox Work Queue Architecture:
   - System Monitor (MMIO): Performs fast atomic-safe memory reads
     directly mapped over ARM physical memory space.
   - VPU Monitor (Mailbox IPC): For Broadcom BCM2835-BCM2711 platforms,
     VideoCore VPU monitor IPC calls are offloaded to process context via
     a dedicated workqueue (`vpu_work`) and serialized.

4. PREEMPT_RT & Safety Hardening:
   - Uses HRTIMER_MODE_REL_SOFT for timer callbacks to execute in softirq
     context, avoiding local hard interrupt nesting priority deadlocks.

5. SoC Generation Support:
   - Patch 1 adds core driver support for Broadcom BCM2835-BCM2711.
   - Patch 2 expands support for Broadcom BCM2712 (Raspberry Pi 5).

Changes in v8
=============
- Fully transitioned locking architecture from 'spinlock_t' globally to
  'raw_spinlock_t' to mathematically guarantee fastpath safety against
  "scheduling while atomic" crashes during PREEMPT_RT deployment where
  the perf core unconditionally disables hard IRQs prior to PMU callbacks.
- Re-architected 'RPI_AXI_PMU_TIMER_INTERVAL' significantly downward from
  2000ms natively to 100ms. This acts as a structural baseline compromise
  that prevents excessive VideoCore Mailbox IPC context-switch wakeovers
  on older Pi 1/2 boards, while safely minimizing uncollected residuals
  causing 0-byte dropouts during time multiplexing/undersubscribed traces.
- Enforced strict 100-column sysfs string macro breakages globally.
- Added proper .scale and .unit bytes aliases across all AXI transaction
  events.
- Fixed a fatal use-after-free initialization race condition catching cases
  where cpuhp_state instances were registered prior to the PMU backend,
  which allowed concurrent CPU hotplug migrations to panic aggressively
  against unregistered core structures.

Ian Rogers (2):
  perf: Add Raspberry Pi BCM2835 AXI PMU driver
  perf: Add Raspberry Pi 5 (BCM2712) AXI PMU support

 drivers/perf/rpi_axi_pmu.c | 2884 ++++++++++++++++++++++++++++++++++++
 1 file changed, 2884 insertions(+)
 create mode 100644 drivers/perf/rpi_axi_pmu.c

-- 
2.47.0