| Message ID | 20210605014216.446867-4-npiggin@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | shoot lazy tlbs | expand |
Excerpts from Nicholas Piggin's message of June 5, 2021 11:42 am: > On big systems, the mm refcount can become highly contented when doing > a lot of context switching with threaded applications (particularly > switching between the idle thread and an application thread). > > Abandoning lazy tlb slows switching down quite a bit in the important > user->idle->user cases, so instead implement a non-refcounted scheme > that causes __mmdrop() to IPI all CPUs in the mm_cpumask and shoot down > any remaining lazy ones. > > Shootdown IPIs are some concern, but they have not been observed to be > a big problem with this scheme (the powerpc implementation generated > 314 additional interrupts on a 144 CPU system during a kernel compile). > There are a number of strategies that could be employed to reduce IPIs > if they turn out to be a problem for some workload. > > Signed-off-by: Nicholas Piggin <npiggin@gmail.com> > --- Update the comment to be clearer, and account for the improvement to MMU_LAZY_TLB_REFCOUNT comment. Signed-off-by: Nicholas Piggin <npiggin@gmail.com> --- arch/Kconfig | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-) diff --git a/arch/Kconfig b/arch/Kconfig index 2ad1a505ca55..cf468c9777d8 100644 --- a/arch/Kconfig +++ b/arch/Kconfig @@ -433,15 +433,16 @@ config MMU_LAZY_TLB_REFCOUNT def_bool y depends on !MMU_LAZY_TLB_SHOOTDOWN -# Instead of refcounting the lazy mm struct for kernel thread references -# (which can cause contention with multi-threaded apps on large multiprocessor -# systems), this option causes __mmdrop to IPI all CPUs in the mm_cpumask and -# switch to init_mm if they were using the to-be-freed mm as the lazy tlb. To -# implement this, architectures must use _lazy_tlb variants of mm refcounting -# when releasing kernel thread mm references, and mm_cpumask must include at -# least all possible CPUs in which the mm might be lazy, at the time of the -# final mmdrop. mmgrab/mmdrop in arch/ code must be switched to _lazy_tlb -# postfix as necessary. +# This option allows MMU_LAZY_TLB_REFCOUNT=n. It ensures no CPUs are using an +# mm as a lazy tlb beyond its last reference count, by shooting down these +# users before the mm is deallocated. __mmdrop() first IPIs all CPUs that may +# be using the mm as a lazy tlb, so that they may switch themselves to using +# init_mm for their active mm. mm_cpumask(mm) is used to determine which CPUs +# may be using mm as a lazy tlb mm. +# +# To implement this, an arch must ensure mm_cpumask(mm) contains at least all +# possible CPUs in which the mm is lazy, and it must meet the requirements for +# MMU_LAZY_TLB_REFCOUNT=n (see above). config MMU_LAZY_TLB_SHOOTDOWN bool
diff --git a/arch/Kconfig b/arch/Kconfig index 1cff045cdde6..f8136c893991 100644 --- a/arch/Kconfig +++ b/arch/Kconfig @@ -421,6 +421,19 @@ config ARCH_WANT_IRQS_OFF_ACTIVATE_MM # Use normal mm refcounting for MMU_LAZY_TLB kernel thread references. config MMU_LAZY_TLB_REFCOUNT def_bool y + depends on !MMU_LAZY_TLB_SHOOTDOWN + +# Instead of refcounting the lazy mm struct for kernel thread references +# (which can cause contention with multi-threaded apps on large multiprocessor +# systems), this option causes __mmdrop to IPI all CPUs in the mm_cpumask and +# switch to init_mm if they were using the to-be-freed mm as the lazy tlb. To +# implement this, architectures must use _lazy_tlb variants of mm refcounting +# when releasing kernel thread mm references, and mm_cpumask must include at +# least all possible CPUs in which the mm might be lazy, at the time of the +# final mmdrop. mmgrab/mmdrop in arch/ code must be switched to _lazy_tlb +# postfix as necessary. +config MMU_LAZY_TLB_SHOOTDOWN + bool config ARCH_HAVE_NMI_SAFE_CMPXCHG bool diff --git a/kernel/fork.c b/kernel/fork.c index dc06afd725cb..8085ff33c7f6 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -674,6 +674,53 @@ static void check_mm(struct mm_struct *mm) #define allocate_mm() (kmem_cache_alloc(mm_cachep, GFP_KERNEL)) #define free_mm(mm) (kmem_cache_free(mm_cachep, (mm))) +static void do_shoot_lazy_tlb(void *arg) +{ + struct mm_struct *mm = arg; + + if (current->active_mm == mm) { + WARN_ON_ONCE(current->mm); + current->active_mm = &init_mm; + switch_mm(mm, &init_mm, current); + } +} + +static void do_check_lazy_tlb(void *arg) +{ + struct mm_struct *mm = arg; + + WARN_ON_ONCE(current->active_mm == mm); +} + +static void shoot_lazy_tlbs(struct mm_struct *mm) +{ + if (IS_ENABLED(CONFIG_MMU_LAZY_TLB_SHOOTDOWN)) { + /* + * IPI overheads have not found to be expensive, but they could + * be reduced in a number of possible ways, for example (in + * roughly increasing order of complexity): + * - A batch of mms requiring IPIs could be gathered and freed + * at once. + * - CPUs could store their active mm somewhere that can be + * remotely checked without a lock, to filter out + * false-positives in the cpumask. + * - After mm_users or mm_count reaches zero, switching away + * from the mm could clear mm_cpumask to reduce some IPIs + * (some batching or delaying would help). + * - A delayed freeing and RCU-like quiescing sequence based on + * mm switching to avoid IPIs completely. + */ + on_each_cpu_mask(mm_cpumask(mm), do_shoot_lazy_tlb, (void *)mm, 1); + if (IS_ENABLED(CONFIG_DEBUG_VM)) + on_each_cpu(do_check_lazy_tlb, (void *)mm, 1); + } else { + /* + * In this case, lazy tlb mms are refounted and would not reach + * __mmdrop until all CPUs have switched away and mmdrop()ed. + */ + } +} + /* * Called when the last reference to the mm * is dropped: either by a lazy thread or by @@ -683,6 +730,10 @@ void __mmdrop(struct mm_struct *mm) { BUG_ON(mm == &init_mm); WARN_ON_ONCE(mm == current->mm); + + /* Ensure no CPUs are using this as their lazy tlb mm */ + shoot_lazy_tlbs(mm); + WARN_ON_ONCE(mm == current->active_mm); mm_free_pgd(mm); destroy_context(mm);
On big systems, the mm refcount can become highly contented when doing a lot of context switching with threaded applications (particularly switching between the idle thread and an application thread). Abandoning lazy tlb slows switching down quite a bit in the important user->idle->user cases, so instead implement a non-refcounted scheme that causes __mmdrop() to IPI all CPUs in the mm_cpumask and shoot down any remaining lazy ones. Shootdown IPIs are some concern, but they have not been observed to be a big problem with this scheme (the powerpc implementation generated 314 additional interrupts on a 144 CPU system during a kernel compile). There are a number of strategies that could be employed to reduce IPIs if they turn out to be a problem for some workload. Signed-off-by: Nicholas Piggin <npiggin@gmail.com> --- arch/Kconfig | 13 +++++++++++++ kernel/fork.c | 51 +++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 64 insertions(+)