perf: arm-ni: Fix list_add() corruption in arm_ni_probe()

Commit Message

Hongbo Yao March 31, 2025, 9:10 a.m. UTC

From: Hongbo Yao <[e-mail]>

When a resource allocation fails in one clock domain of an NI device,
we need to properly roll back all previously registered perf PMUs in
other clock domains of the same device.

Otherwise, it can lead to kernel panics.

Calling arm_ni_init+0x0/0xff8 [arm_ni] @ 2374
arm-ni ARMHCB70:00: Failed to request PMU region 0x1f3c13000
arm-ni ARMHCB70:00: probe with driver arm-ni failed with error -16
list_add corruption: next->prev should be prev (fffffd01e9698a18),
but was 0000000000000000. (next=ffff10001a0decc8).
pstate: 6340009 (nZCv daif +PAN -UAO +TCO +DIT -SSBS BTYPE=--)
pc : list_add_valid_or_report+0x7c/0xb8
lr : list_add_valid_or_report+0x7c/0xb8
Call trace:
 __list_add_valid_or_report+0x7c/0xb8
 perf_pmu_register+0x22c/0x3a0
 arm_ni_probe+0x554/0x70c [arm_ni]
 platform_probe+0x70/0xe8
 really_probe+0xc6/0x4d8
 driver_probe_device+0x48/0x170
 __driver_attach+0x8e/0x1c0
 bus_for_each_dev+0x64/0xf0
 driver_add+0x138/0x260
 bus_add_driver+0x68/0x138
 __platform_driver_register+0x2c/0x40
 arm_ni_init+0x14/0x2a [arm_ni]
 do_init_module+0x36/0x298
---[ end trace 0000000000000000 ]---
Kernel panic - not syncing: Oops - BUG: Fatal exception
SMP: stopping secondary CPUs

Fixes: 4d5a7680f2b4 ("perf: Add driver for Arm NI-700 interconnect PMU")
Signed-off-by: Hongbo Yao <[e-mail]>
---
 drivers/perf/arm-ni.c | 21 ++++++++++++++++++---
 1 file changed, 18 insertions(+), 3 deletions(-)

Patch

diff --git a/drivers/perf/arm-ni.c b/drivers/perf/arm-ni.c
index fd7a5e60e963..c501d31bde68 100644
--- a/drivers/perf/arm-ni.c
+++ b/drivers/perf/arm-ni.c
@@ -102,6 +102,7 @@  struct arm_ni_unit {
 struct arm_ni_cd {
 	void __iomem *pmu_base;
 	u16 id;
+	bool pmu_registered;
 	int num_units;
 	int irq;
 	int cpu;
@@ -571,6 +572,8 @@  static int arm_ni_init_cd(struct arm_ni *ni, struct arm_ni_node *node, u64 res_s
 	err = perf_pmu_register(&cd->pmu, name, -1);
 	if (err)
 		cpuhp_state_remove_instance_nocalls(arm_ni_hp_state, &cd->cpuhp_node);
+	else
+		cd->pmu_registered = true;
 
 	return err;
 }
@@ -593,6 +596,7 @@  static int arm_ni_probe(struct platform_device *pdev)
 	void __iomem *base;
 	static atomic_t id;
 	int num_cds;
+	int ret;
 	u32 reg, part;
 
 	/*
@@ -651,18 +655,29 @@  static int arm_ni_probe(struct platform_device *pdev)
 			reg = readl_relaxed(vd.base + NI_CHILD_PTR(p));
 			arm_ni_probe_domain(base + reg, &pd);
 			for (int c = 0; c < pd.num_components; c++) {
-				int ret;
-
 				reg = readl_relaxed(pd.base + NI_CHILD_PTR(c));
 				arm_ni_probe_domain(base + reg, &cd);
 				ret = arm_ni_init_cd(ni, &cd, res->start);
 				if (ret)
-					return ret;
+					goto init_cd_cleanup;
 			}
 		}
 	}
 
 	return 0;
+
+init_cd_cleanup:
+	for (int i = 0; i < ni->num_cds; i++) {
+		struct arm_ni_cd *cd = ni->cds + i;
+
+		if (!cd->pmu_base)
+			continue;
+
+		if (cd->pmu_registered)
+			perf_pmu_unregister(&cd->pmu);
+	}
+
+	return ret;
 }
 
 static void arm_ni_remove(struct platform_device *pdev)