perf: qcom_l2_pmu: Make l2_cache_pmu_probe_cluster() more robust

If an error occurs after calling list_add(), the &l2cache_pmu->clusters
list will reference some memory that will be freed when the managed
resources will be released.

Move the list_add() at the end of the function when everything is in fine.

This is harmless because if l2_cache_pmu_probe_cluster() fails, then
l2_cache_pmu_probe() will fail as well and 'l2cache_pmu' will be released
as well.
But it looks cleaner and could silence static checker warning.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/6a0f5bdb6b7b2ed4ef194fc49693e902ad5b95ea.1684397879.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Will Deacon <will@kernel.org>

diff --git a/drivers/perf/qcom_l2_pmu.c b/drivers/perf/qcom_l2_pmu.c
index aaca6db7d8f6c47f24a43a4831031e4afca1a2c9..3f9a98c17a897d39ad4bd6f3dc797d318ca57abf 100644 (file)
--- a/drivers/perf/qcom_l2_pmu.c
+++ b/drivers/perf/qcom_l2_pmu.c
@@ -857,7 +857,6 @@ static int l2_cache_pmu_probe_cluster(struct device *dev, void *data)
                return -ENOMEM;
 
        INIT_LIST_HEAD(&cluster->next);
-       list_add(&cluster->next, &l2cache_pmu->clusters);
        cluster->cluster_id = fw_cluster_id;
 
        irq = platform_get_irq(sdev, 0);
@@ -883,6 +882,7 @@ static int l2_cache_pmu_probe_cluster(struct device *dev, void *data)
 
        spin_lock_init(&cluster->pmu_lock);
 
+       list_add(&cluster->next, &l2cache_pmu->clusters);
        l2cache_pmu->num_pmus++;
 
        return 0;