The following was seen with CMF enabled:
BUG: using smp_processor_id() in preemptible
code: systemd-udevd/31711
kernel: caller is lpfc_update_cmf_cmd+0x214/0x420  [lpfc]
kernel: CPU: 12 PID: 31711 Comm: systemd-udevd
kernel: Call Trace:
kernel: <TASK>
kernel: dump_stack_lvl+0x44/0x57
kernel: check_preemption_disabled+0xbf/0xe0
kernel: lpfc_update_cmf_cmd+0x214/0x420 [lpfc]
kernel: lpfc_nvme_fcp_io_submit+0x23b4/0x4df0 [lpfc]
this_cpu_ptr() calls smp_processor_id() in a preemptible context.
Fix by using per_cpu_ptr() with raw_smp_processor_id() instead.
Link: https://lore.kernel.org/r/20220412222008.126521-16-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
                else
                        time = div_u64(time + 500, 1000); /* round it */
 
-               cgs = this_cpu_ptr(phba->cmf_stat);
+               cgs = per_cpu_ptr(phba->cmf_stat, raw_smp_processor_id());
                atomic64_add(size, &cgs->rcv_bytes);
                atomic64_add(time, &cgs->rx_latency);
                atomic_inc(&cgs->rx_io_cnt);
                        atomic_set(&phba->rx_max_read_cnt, size);
        }
 
-       cgs = this_cpu_ptr(phba->cmf_stat);
+       cgs = per_cpu_ptr(phba->cmf_stat, raw_smp_processor_id());
        atomic64_add(size, &cgs->total_bytes);
        return 0;
 }