Re: [PATCH v2 1/8] RISC-V: Do not wait indefinitely in __cpu_up
On Thu, Jan 17, 2019 at 06:35:39PM -0800, Atish Patra wrote: > On 1/15/19 5:51 AM, Christoph Hellwig wrote: > > > void *__cpu_up_stack_pointer[NR_CPUS]; > > > void *__cpu_up_task_pointer[NR_CPUS]; > > > +static DECLARE_COMPLETION(cpu_running); > > > void __init smp_prepare_boot_cpu(void) > > > { > > > @@ -81,6 +82,7 @@ void __init setup_smp(void) > > > int __cpu_up(unsigned int cpu, struct task_struct *tidle) > > > { > > > + int ret = 0; > > > int hartid = cpuid_to_hartid_map(cpu); > > > tidle->thread_info.cpu = cpu; > > > @@ -96,10 +98,15 @@ int __cpu_up(unsigned int cpu, struct task_struct > > > *tidle) > > > task_stack_page(tidle) + THREAD_SIZE); > > > WRITE_ONCE(__cpu_up_task_pointer[hartid], tidle); > > > - while (!cpu_online(cpu)) > > > - cpu_relax(); > > > + wait_for_completion_timeout(&cpu_running, > > > + msecs_to_jiffies(1000)); > > > > Having a global completion here worries me. I bet we have some higher > > level serialization, but can we comment or even better lockdep assert on > > that? > > > > Yes. It is serialized from smp.c in smp_init(). It brings one cpu online > at a time for preset_cpu mask. > > Do we still need a lockdep assert ? I guess the real lock is through cpu_hotplug_lock. And yes, a comment or even better lockdep assert would be good.
Re: [PATCH v2 1/8] RISC-V: Do not wait indefinitely in __cpu_up
On 1/15/19 5:51 AM, Christoph Hellwig wrote: void *__cpu_up_stack_pointer[NR_CPUS]; void *__cpu_up_task_pointer[NR_CPUS]; +static DECLARE_COMPLETION(cpu_running); void __init smp_prepare_boot_cpu(void) { @@ -81,6 +82,7 @@ void __init setup_smp(void) int __cpu_up(unsigned int cpu, struct task_struct *tidle) { + int ret = 0; int hartid = cpuid_to_hartid_map(cpu); tidle->thread_info.cpu = cpu; @@ -96,10 +98,15 @@ int __cpu_up(unsigned int cpu, struct task_struct *tidle) task_stack_page(tidle) + THREAD_SIZE); WRITE_ONCE(__cpu_up_task_pointer[hartid], tidle); - while (!cpu_online(cpu)) - cpu_relax(); + wait_for_completion_timeout(&cpu_running, + msecs_to_jiffies(1000)); Having a global completion here worries me. I bet we have some higher level serialization, but can we comment or even better lockdep assert on that? Yes. It is serialized from smp.c in smp_init(). It brings one cpu online at a time for preset_cpu mask. Do we still need a lockdep assert ? Regards, Atish Also please use up your available lines (72 in commit logs, 80 in source files) instead of adding spurious line wraps. ___ linux-riscv mailing list linux-ri...@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-riscv
Re: [PATCH v2 1/8] RISC-V: Do not wait indefinitely in __cpu_up
> > void *__cpu_up_stack_pointer[NR_CPUS]; > void *__cpu_up_task_pointer[NR_CPUS]; > +static DECLARE_COMPLETION(cpu_running); > > void __init smp_prepare_boot_cpu(void) > { > @@ -81,6 +82,7 @@ void __init setup_smp(void) > > int __cpu_up(unsigned int cpu, struct task_struct *tidle) > { > + int ret = 0; > int hartid = cpuid_to_hartid_map(cpu); > tidle->thread_info.cpu = cpu; > > @@ -96,10 +98,15 @@ int __cpu_up(unsigned int cpu, struct task_struct *tidle) > task_stack_page(tidle) + THREAD_SIZE); > WRITE_ONCE(__cpu_up_task_pointer[hartid], tidle); > > - while (!cpu_online(cpu)) > - cpu_relax(); > + wait_for_completion_timeout(&cpu_running, > + msecs_to_jiffies(1000)); Having a global completion here worries me. I bet we have some higher level serialization, but can we comment or even better lockdep assert on that? Also please use up your available lines (72 in commit logs, 80 in source files) instead of adding spurious line wraps.
[PATCH v2 1/8] RISC-V: Do not wait indefinitely in __cpu_up
In SMP path, __cpu_up waits for other CPU to come online indefinitely. This is wrong as other CPU might be disabled in machine mode and possible CPU is set to the cpus present in DT. Introduce a completion variable and waits only for a second. Signed-off-by: Atish Patra Reviewed-by: Anup Patel --- arch/riscv/kernel/smpboot.c | 14 +++--- 1 file changed, 11 insertions(+), 3 deletions(-) diff --git a/arch/riscv/kernel/smpboot.c b/arch/riscv/kernel/smpboot.c index fc185eca..32e14572 100644 --- a/arch/riscv/kernel/smpboot.c +++ b/arch/riscv/kernel/smpboot.c @@ -39,6 +39,7 @@ void *__cpu_up_stack_pointer[NR_CPUS]; void *__cpu_up_task_pointer[NR_CPUS]; +static DECLARE_COMPLETION(cpu_running); void __init smp_prepare_boot_cpu(void) { @@ -81,6 +82,7 @@ void __init setup_smp(void) int __cpu_up(unsigned int cpu, struct task_struct *tidle) { + int ret = 0; int hartid = cpuid_to_hartid_map(cpu); tidle->thread_info.cpu = cpu; @@ -96,10 +98,15 @@ int __cpu_up(unsigned int cpu, struct task_struct *tidle) task_stack_page(tidle) + THREAD_SIZE); WRITE_ONCE(__cpu_up_task_pointer[hartid], tidle); - while (!cpu_online(cpu)) - cpu_relax(); + wait_for_completion_timeout(&cpu_running, + msecs_to_jiffies(1000)); - return 0; + if (!cpu_online(cpu)) { + pr_crit("CPU%u: failed to come online\n", cpu); + ret = -EIO; + } + + return ret; } void __init smp_cpus_done(unsigned int max_cpus) @@ -125,6 +132,7 @@ asmlinkage void __init smp_callin(void) * a local TLB flush right now just in case. */ local_flush_tlb_all(); + complete(&cpu_running); /* * Disable preemption before enabling interrupts, so we don't try to * schedule a CPU that hasn't actually started yet. -- 2.7.4