On Wed, Feb 7, 2024 at 9:06 AM Alexei Starovoitov
<alexei.starovoitov@xxxxxxxxx> wrote:
>
> On Tue, Feb 6, 2024 at 12:14 AM Yafang Shao <laoar.shao@xxxxxxxxx> wrote:
> >
> > Add three new kfuncs for bpf_iter_cpumask.
> > - bpf_iter_cpumask_new
> > KF_RCU is defined because the cpumask must be a RCU trusted pointer
> > such as task->cpus_ptr.
> > - bpf_iter_cpumask_next
> > - bpf_iter_cpumask_destroy
> >
> > These new kfuncs facilitate the iteration of percpu data, such as
> > runqueues, psi_cgroup_cpu, and more.
> >
> > Signed-off-by: Yafang Shao <laoar.shao@xxxxxxxxx>
> > ---
> > kernel/bpf/cpumask.c | 79 ++++++++++++++++++++++++++++++++++++++++++++
> > 1 file changed, 79 insertions(+)
> >
> > diff --git a/kernel/bpf/cpumask.c b/kernel/bpf/cpumask.c
> > index dad0fb1c8e87..ed6078cfa40e 100644
> > --- a/kernel/bpf/cpumask.c
> > +++ b/kernel/bpf/cpumask.c
> > @@ -422,6 +422,82 @@ __bpf_kfunc u32 bpf_cpumask_weight(const struct cpumask *cpumask)
> > return cpumask_weight(cpumask);
> > }
> >
> > +struct bpf_iter_cpumask {
> > + __u64 __opaque[2];
> > +} __aligned(8);
> > +
> > +struct bpf_iter_cpumask_kern {
> > + struct cpumask *mask;
> > + int cpu;
> > +} __aligned(8);
> > +
> > +/**
> > + * bpf_iter_cpumask_new() - Initialize a new CPU mask iterator for a given CPU mask
> > + * @it: The new bpf_iter_cpumask to be created.
> > + * @mask: The cpumask to be iterated over.
> > + *
> > + * This function initializes a new bpf_iter_cpumask structure for iterating over
> > + * the specified CPU mask. It assigns the provided cpumask to the newly created
> > + * bpf_iter_cpumask @it for subsequent iteration operations.
> > + *
> > + * On success, 0 is returned. On failure, ERR is returned.
> > + */
> > +__bpf_kfunc int bpf_iter_cpumask_new(struct bpf_iter_cpumask *it, const struct cpumask *mask)
> > +{
> > + struct bpf_iter_cpumask_kern *kit = (void *)it;
> > +
> > + BUILD_BUG_ON(sizeof(struct bpf_iter_cpumask_kern) > sizeof(struct bpf_iter_cpumask));
> > + BUILD_BUG_ON(__alignof__(struct bpf_iter_cpumask_kern) !=
> > + __alignof__(struct bpf_iter_cpumask));
> > +
> > + kit->mask = bpf_mem_alloc(&bpf_global_ma, cpumask_size());
> > + if (!kit->mask)
> > + return -ENOMEM;
> > +
> > + cpumask_copy(kit->mask, mask);
>
> Since it's mem_alloc plus memcpy how about we make it more
> generic ?
> Instead of cpumask specific let's pass arbitrary
> "void *unsafe_addr, u32 size"
>
> allocate that much and probe_read_kernel into the buffer?
>
>
> > +__bpf_kfunc int *bpf_iter_cpumask_next(struct bpf_iter_cpumask *it)
> > +{
> > + struct bpf_iter_cpumask_kern *kit = (void *)it;
> > + const struct cpumask *mask = kit->mask;
> > + int cpu;
> > +
> > + if (!mask)
> > + return NULL;
> > + cpu = cpumask_next(kit->cpu, mask);
>
> Instead of cpumask_next() call find_next_bit()
>
> > + if (cpu >= nr_cpu_ids)
> > + return NULL;
>
> instead of nr_cpu_ids we can check size in bits of copied bit array.
>
> > BTF_ID_FLAGS(func, bpf_iter_cpumask_new, KF_ITER_NEW | KF_RCU)
>
> KF_RCU is also not needed.
> Such iterator will be callable from anywhere and on any address.
>
> wdyt?
Good suggestion. A more generic bitmap iter is better. I will analyze it.
--
Regards
Yafang
