]> git.ipfire.org Git - thirdparty/linux.git/commitdiff
powercap: intel_rapl: Fix possible recursive lock warning
authorSrinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Wed, 17 Dec 2025 15:34:55 +0000 (07:34 -0800)
committerRafael J. Wysocki <rafael.j.wysocki@intel.com>
Wed, 17 Dec 2025 16:24:28 +0000 (17:24 +0100)
With the RAPL PMU addition, there is a recursive locking when CPU online
callback function calls rapl_package_add_pmu(). Here cpu_hotplug_lock
is already acquired by cpuhp_thread_fun() and rapl_package_add_pmu()
tries to acquire again.

<4>[ 8.197433] ============================================
<4>[ 8.197437] WARNING: possible recursive locking detected
<4>[ 8.197440] 6.19.0-rc1-lgci-xe-xe-4242-05b7c58b3367dca84+ #1 Not tainted
<4>[ 8.197444] --------------------------------------------
<4>[ 8.197447] cpuhp/0/20 is trying to acquire lock:
<4>[ 8.197450] ffffffff83487870 (cpu_hotplug_lock){++++}-{0:0}, at:
rapl_package_add_pmu+0x37/0x370 [intel_rapl_common]
<4>[ 8.197463]
but task is already holding lock:
<4>[ 8.197466] ffffffff83487870 (cpu_hotplug_lock){++++}-{0:0}, at:
cpuhp_thread_fun+0x6d/0x290
<4>[ 8.197477]
other info that might help us debug this:
<4>[ 8.197480] Possible unsafe locking scenario:

<4>[ 8.197483] CPU0
<4>[ 8.197485] ----
<4>[ 8.197487] lock(cpu_hotplug_lock);
<4>[ 8.197490] lock(cpu_hotplug_lock);
<4>[ 8.197493]
*** DEADLOCK ***
..
..
<4>[ 8.197542] __lock_acquire+0x146e/0x2790
<4>[ 8.197548] lock_acquire+0xc4/0x2c0
<4>[ 8.197550] ? rapl_package_add_pmu+0x37/0x370 [intel_rapl_common]
<4>[ 8.197556] cpus_read_lock+0x41/0x110
<4>[ 8.197558] ? rapl_package_add_pmu+0x37/0x370 [intel_rapl_common]
<4>[ 8.197561] rapl_package_add_pmu+0x37/0x370 [intel_rapl_common]
<4>[ 8.197565] rapl_cpu_online+0x85/0x87 [intel_rapl_msr]
<4>[ 8.197568] ? __pfx_rapl_cpu_online+0x10/0x10 [intel_rapl_msr]
<4>[ 8.197570] cpuhp_invoke_callback+0x41f/0x6c0
<4>[ 8.197573] ? cpuhp_thread_fun+0x6d/0x290
<4>[ 8.197575] cpuhp_thread_fun+0x1e2/0x290
<4>[ 8.197578] ? smpboot_thread_fn+0x26/0x290
<4>[ 8.197581] smpboot_thread_fn+0x12f/0x290
<4>[ 8.197584] ? __pfx_smpboot_thread_fn+0x10/0x10
<4>[ 8.197586] kthread+0x11f/0x250
<4>[ 8.197589] ? __pfx_kthread+0x10/0x10
<4>[ 8.197592] ret_from_fork+0x344/0x3a0
<4>[ 8.197595] ? __pfx_kthread+0x10/0x10
<4>[ 8.197597] ret_from_fork_asm+0x1a/0x30
<4>[ 8.197604] </TASK>

Fix this issue in the same way as rapl powercap package domain is added
from the same CPU online callback by introducing another interface which
doesn't call cpus_read_lock(). Add rapl_package_add_pmu_locked() and
rapl_package_remove_pmu_locked() which don't call cpus_read_lock().

Fixes: 748d6ba43afd ("powercap: intel_rapl: Enable MSR-based RAPL PMU support")
Reported-by: Borah, Chaitanya Kumar <chaitanya.kumar.borah@intel.com>
Closes: https://lore.kernel.org/linux-pm/5427ede1-57a0-43d1-99f3-8ca4b0643e82@intel.com/T/#u
Tested-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com>
Tested-by: RavitejaX Veesam <ravitejax.veesam@intel.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Link: https://patch.msgid.link/20251217153455.3560176-1-srinivas.pandruvada@linux.intel.com
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
drivers/powercap/intel_rapl_common.c
drivers/powercap/intel_rapl_msr.c
include/linux/intel_rapl.h

index b9d87e56cbbc8af63961975cbd722d7b538b1c06..3ff6da3bf4e63000f29ba326a8979823a5c993df 100644 (file)
@@ -2032,7 +2032,7 @@ end:
        return ret;
 }
 
-int rapl_package_add_pmu(struct rapl_package *rp)
+int rapl_package_add_pmu_locked(struct rapl_package *rp)
 {
        struct rapl_package_pmu_data *data = &rp->pmu_data;
        int idx;
@@ -2040,8 +2040,6 @@ int rapl_package_add_pmu(struct rapl_package *rp)
        if (rp->has_pmu)
                return -EEXIST;
 
-       guard(cpus_read_lock)();
-
        for (idx = 0; idx < rp->nr_domains; idx++) {
                struct rapl_domain *rd = &rp->domains[idx];
                int domain = rd->id;
@@ -2091,17 +2089,23 @@ int rapl_package_add_pmu(struct rapl_package *rp)
 
        return rapl_pmu_update(rp);
 }
+EXPORT_SYMBOL_GPL(rapl_package_add_pmu_locked);
+
+int rapl_package_add_pmu(struct rapl_package *rp)
+{
+       guard(cpus_read_lock)();
+
+       return rapl_package_add_pmu_locked(rp);
+}
 EXPORT_SYMBOL_GPL(rapl_package_add_pmu);
 
-void rapl_package_remove_pmu(struct rapl_package *rp)
+void rapl_package_remove_pmu_locked(struct rapl_package *rp)
 {
        struct rapl_package *pos;
 
        if (!rp->has_pmu)
                return;
 
-       guard(cpus_read_lock)();
-
        list_for_each_entry(pos, &rapl_packages, plist) {
                /* PMU is still needed */
                if (pos->has_pmu && pos != rp)
@@ -2111,6 +2115,14 @@ void rapl_package_remove_pmu(struct rapl_package *rp)
        perf_pmu_unregister(&rapl_pmu.pmu);
        memset(&rapl_pmu, 0, sizeof(struct rapl_pmu));
 }
+EXPORT_SYMBOL_GPL(rapl_package_remove_pmu_locked);
+
+void rapl_package_remove_pmu(struct rapl_package *rp)
+{
+       guard(cpus_read_lock)();
+
+       rapl_package_remove_pmu_locked(rp);
+}
 EXPORT_SYMBOL_GPL(rapl_package_remove_pmu);
 #endif
 
index 0ce1096b63145d078ade376aa8dcd2de8557cdc3..9a7e150b3536b9e0d1e0caf1d0c32d8f79d2c290 100644 (file)
@@ -82,7 +82,7 @@ static int rapl_cpu_online(unsigned int cpu)
                if (IS_ERR(rp))
                        return PTR_ERR(rp);
                if (rapl_msr_pmu)
-                       rapl_package_add_pmu(rp);
+                       rapl_package_add_pmu_locked(rp);
        }
        cpumask_set_cpu(cpu, &rp->cpumask);
        return 0;
@@ -101,7 +101,7 @@ static int rapl_cpu_down_prep(unsigned int cpu)
        lead_cpu = cpumask_first(&rp->cpumask);
        if (lead_cpu >= nr_cpu_ids) {
                if (rapl_msr_pmu)
-                       rapl_package_remove_pmu(rp);
+                       rapl_package_remove_pmu_locked(rp);
                rapl_remove_package_cpuslocked(rp);
        } else if (rp->lead_cpu == cpu) {
                rp->lead_cpu = lead_cpu;
index e9ade2ff4af6646b4c3b9fc0574e3b3ab05746a1..f479ef5b3341cff3a59518811285893993828476 100644 (file)
@@ -214,10 +214,14 @@ void rapl_remove_package(struct rapl_package *rp);
 
 #ifdef CONFIG_PERF_EVENTS
 int rapl_package_add_pmu(struct rapl_package *rp);
+int rapl_package_add_pmu_locked(struct rapl_package *rp);
 void rapl_package_remove_pmu(struct rapl_package *rp);
+void rapl_package_remove_pmu_locked(struct rapl_package *rp);
 #else
 static inline int rapl_package_add_pmu(struct rapl_package *rp) { return 0; }
+static inline int rapl_package_add_pmu_locked(struct rapl_package *rp) { return 0; }
 static inline void rapl_package_remove_pmu(struct rapl_package *rp) { }
+static inline void rapl_package_remove_pmu_locked(struct rapl_package *rp) { }
 #endif
 
 #endif /* __INTEL_RAPL_H__ */