Message ID | 20230513123314.100236-1-npiggin@gmail.com |
---|---|
State | Accepted |
Headers | show |
Series | core/init: Move opal_mpipl_init earlier | expand |
On Sat, May 13, 2023 at 10:33:14PM +1000, Nicholas Piggin wrote: >There doesn't seem to be any reason to call opal_mpipl_init so late, >after sleep, traps, interrupts, machine checks, watchdog, etc., are >all disabled. Move it earlier. Guess we'll find out! Applied to master. No issues in boot testing on what I have access to, FWIW.
diff --git a/core/init.c b/core/init.c index 005ecf31..e832a009 100644 --- a/core/init.c +++ b/core/init.c @@ -574,6 +574,8 @@ void __noreturn load_and_boot_kernel(bool is_reboot) if (!occ_sensors_init()) dts_sensor_create_nodes(sensor_node); + opal_mpipl_init(); + } else { /* fdt will be rebuilt */ free(fdt); @@ -638,10 +640,6 @@ void __noreturn load_and_boot_kernel(bool is_reboot) patch_traps(false); cpu_set_hile_mode(false); /* Clear HILE on all CPUs */ - /* init MPIPL */ - if (!is_reboot) - opal_mpipl_init(); - checksum_romem(); debug_descriptor.state_flags |= OPAL_BOOT_COMPLETE;
There doesn't seem to be any reason to call opal_mpipl_init so late, after sleep, traps, interrupts, machine checks, watchdog, etc., are all disabled. Move it earlier. Signed-off-by: Nicholas Piggin <npiggin@gmail.com> --- core/init.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-)