On Wed, 15 Jul 2020 16:30:06 PDT (-0700), Atish Patra wrote:
The trap vector is set only in trap_init which may be too late in some cases. Early ioremap/efi spits many warning messages which may be useful.Setup the trap vector early so that any warning/bug can be handled before generic code invokes trap_init. Signed-off-by: Atish Patra <atish.pa...@wdc.com> --- arch/riscv/kernel/head.S | 10 ++++++++-- arch/riscv/kernel/smpboot.c | 1 - arch/riscv/kernel/traps.c | 8 +------- 3 files changed, 9 insertions(+), 10 deletions(-) diff --git a/arch/riscv/kernel/head.S b/arch/riscv/kernel/head.S index 7ed1b22950fd..d0c5c316e9bb 100644 --- a/arch/riscv/kernel/head.S +++ b/arch/riscv/kernel/head.S @@ -77,10 +77,16 @@ relocate: csrw CSR_SATP, a0 .align 2 1: - /* Set trap vector to spin forever to help debug */ - la a0, .Lsecondary_park + /* Set trap vector to exception handler */ + la a0, handle_exception csrw CSR_TVEC, a0 + /* + * Set sup0 scratch register to 0, indicating to exception vector that + * we are presently executing in kernel. + */ + csrw CSR_SCRATCH, zero + /* Reload the global pointer */ .option push .option norelax diff --git a/arch/riscv/kernel/smpboot.c b/arch/riscv/kernel/smpboot.c index 4e9922790f6e..5a9c127a380e 100644 --- a/arch/riscv/kernel/smpboot.c +++ b/arch/riscv/kernel/smpboot.c @@ -154,7 +154,6 @@ asmlinkage __visible void smp_callin(void) mmgrab(mm); current->active_mm = mm; - trap_init(); notify_cpu_starting(smp_processor_id()); update_siblings_masks(smp_processor_id()); set_cpu_online(smp_processor_id(), 1); diff --git a/arch/riscv/kernel/traps.c b/arch/riscv/kernel/traps.c index 7d95cce5e47c..ad14f4466d92 100644 --- a/arch/riscv/kernel/traps.c +++ b/arch/riscv/kernel/traps.c @@ -174,13 +174,7 @@ int is_valid_bugaddr(unsigned long pc) } #endif /* CONFIG_GENERIC_BUG */ +/* stvec & scratch is already set from head.S */ void trap_init(void) { - /* - * Set sup0 scratch register to 0, indicating to exception vector - * that we are presently executing in the kernel - */ - csr_write(CSR_SCRATCH, 0); - /* Set the exception vector address */ - csr_write(CSR_TVEC, &handle_exception); }
While I think these are all actual fixes, it's pretty late in the cycle here so I'm going to a bit on the careful side and only take the patches that actually manifest as bugs in the current port. Assuming this doesn't manifest until early_ioremap is enabled (and we don't do that yet), I've put it on for-next. LMK if I'm wrong about this, or the following ones. Thanks!
