On Thu, May 21, 2020 at 5:25 PM Paul E. McKenney <paulmck@xxxxxxxxxx> wrote:
>
> On Sun, May 17, 2020 at 12:57:21PM -0700, Andrii Nakryiko wrote:
> > This commits adds a new MPSC ring buffer implementation into BPF ecosystem,
> > which allows multiple CPUs to submit data to a single shared ring buffer. On
> > the consumption side, only single consumer is assumed.
>
> [ . . . ]
>
> Focusing just on the ring-buffer mechanism, with a question or two
> below. Looks pretty close, actually!
>
> Thanx, Paul
>
Thanks for review, Paul!
> > Signed-off-by: Andrii Nakryiko <andriin@xxxxxx>
> > ---
> > include/linux/bpf.h | 13 +
> > include/linux/bpf_types.h | 1 +
> > include/linux/bpf_verifier.h | 4 +
> > include/uapi/linux/bpf.h | 84 +++++-
> > kernel/bpf/Makefile | 2 +-
> > kernel/bpf/helpers.c | 10 +
> > kernel/bpf/ringbuf.c | 487 +++++++++++++++++++++++++++++++++
> > kernel/bpf/syscall.c | 12 +
> > kernel/bpf/verifier.c | 157 ++++++++---
> > kernel/trace/bpf_trace.c | 10 +
> > tools/include/uapi/linux/bpf.h | 90 +++++-
> > 11 files changed, 832 insertions(+), 38 deletions(-)
> > create mode 100644 kernel/bpf/ringbuf.c
>
> [ . . . ]
>
> > diff --git a/kernel/bpf/ringbuf.c b/kernel/bpf/ringbuf.c
> > new file mode 100644
> > index 000000000000..3c19f0f07726
> > --- /dev/null
> > +++ b/kernel/bpf/ringbuf.c
> > @@ -0,0 +1,487 @@
> > +#include <linux/bpf.h>
> > +#include <linux/btf.h>
> > +#include <linux/err.h>
> > +#include <linux/irq_work.h>
> > +#include <linux/slab.h>
> > +#include <linux/filter.h>
> > +#include <linux/mm.h>
> > +#include <linux/vmalloc.h>
> > +#include <linux/wait.h>
> > +#include <linux/poll.h>
> > +#include <uapi/linux/btf.h>
> > +
> > +#define RINGBUF_CREATE_FLAG_MASK (BPF_F_NUMA_NODE)
> > +
> > +/* non-mmap()'able part of bpf_ringbuf (everything up to consumer page) */
> > +#define RINGBUF_PGOFF \
> > + (offsetof(struct bpf_ringbuf, consumer_pos) >> PAGE_SHIFT)
> > +/* consumer page and producer page */
> > +#define RINGBUF_POS_PAGES 2
> > +
> > +#define RINGBUF_MAX_RECORD_SZ (UINT_MAX/4)
> > +
> > +/* Maximum size of ring buffer area is limited by 32-bit page offset within
> > + * record header, counted in pages. Reserve 8 bits for extensibility, and take
> > + * into account few extra pages for consumer/producer pages and
> > + * non-mmap()'able parts. This gives 64GB limit, which seems plenty for single
> > + * ring buffer.
> > + */
> > +#define RINGBUF_MAX_DATA_SZ \
> > + (((1ULL << 24) - RINGBUF_POS_PAGES - RINGBUF_PGOFF) * PAGE_SIZE)
> > +
> > +struct bpf_ringbuf {
> > + wait_queue_head_t waitq;
> > + struct irq_work work;
> > + u64 mask;
> > + spinlock_t spinlock ____cacheline_aligned_in_smp;
> > + /* Consumer and producer counters are put into separate pages to allow
> > + * mapping consumer page as r/w, but restrict producer page to r/o.
> > + * This protects producer position from being modified by user-space
> > + * application and ruining in-kernel position tracking.
> > + */
> > + unsigned long consumer_pos __aligned(PAGE_SIZE);
> > + unsigned long producer_pos __aligned(PAGE_SIZE);
> > + char data[] __aligned(PAGE_SIZE);
> > +};
> > +
> > +struct bpf_ringbuf_map {
> > + struct bpf_map map;
> > + struct bpf_map_memory memory;
> > + struct bpf_ringbuf *rb;
> > +};
> > +
> > +/* 8-byte ring buffer record header structure */
> > +struct bpf_ringbuf_hdr {
> > + u32 len;
> > + u32 pg_off;
> > +};
> > +
> > +static struct bpf_ringbuf *bpf_ringbuf_area_alloc(size_t data_sz, int numa_node)
> > +{
> > + const gfp_t flags = GFP_KERNEL | __GFP_RETRY_MAYFAIL | __GFP_NOWARN |
> > + __GFP_ZERO;
> > + int nr_meta_pages = RINGBUF_PGOFF + RINGBUF_POS_PAGES;
> > + int nr_data_pages = data_sz >> PAGE_SHIFT;
> > + int nr_pages = nr_meta_pages + nr_data_pages;
> > + struct page **pages, *page;
> > + size_t array_size;
> > + void *addr;
> > + int i;
> > +
> > + /* Each data page is mapped twice to allow "virtual"
> > + * continuous read of samples wrapping around the end of ring
> > + * buffer area:
> > + * ------------------------------------------------------
> > + * | meta pages | real data pages | same data pages |
> > + * ------------------------------------------------------
> > + * | | 1 2 3 4 5 6 7 8 9 | 1 2 3 4 5 6 7 8 9 |
> > + * ------------------------------------------------------
> > + * | | TA DA | TA DA |
> > + * ------------------------------------------------------
> > + * ^^^^^^^
> > + * |
> > + * Here, no need to worry about special handling of wrapped-around
> > + * data due to double-mapped data pages. This works both in kernel and
> > + * when mmap()'ed in user-space, simplifying both kernel and
> > + * user-space implementations significantly.
> > + */
> > + array_size = (nr_meta_pages + 2 * nr_data_pages) * sizeof(*pages);
> > + if (array_size > PAGE_SIZE)
> > + pages = vmalloc_node(array_size, numa_node);
> > + else
> > + pages = kmalloc_node(array_size, flags, numa_node);
> > + if (!pages)
> > + return NULL;
> > +
> > + for (i = 0; i < nr_pages; i++) {
> > + page = alloc_pages_node(numa_node, flags, 0);
> > + if (!page) {
> > + nr_pages = i;
> > + goto err_free_pages;
> > + }
> > + pages[i] = page;
> > + if (i >= nr_meta_pages)
> > + pages[nr_data_pages + i] = page;
> > + }
> > +
> > + addr = vmap(pages, nr_meta_pages + 2 * nr_data_pages,
> > + VM_ALLOC | VM_USERMAP, PAGE_KERNEL);
> > + if (addr)
> > + return addr;
> > +
> > +err_free_pages:
> > + for (i = 0; i < nr_pages; i++)
> > + free_page((unsigned long)pages[i]);
> > + kvfree(pages);
> > + return NULL;
> > +}
> > +
> > +static void bpf_ringbuf_notify(struct irq_work *work)
> > +{
> > + struct bpf_ringbuf *rb = container_of(work, struct bpf_ringbuf, work);
> > +
> > + wake_up_all(&rb->waitq);
> > +}
> > +
> > +static struct bpf_ringbuf *bpf_ringbuf_alloc(size_t data_sz, int numa_node)
> > +{
> > + struct bpf_ringbuf *rb;
> > +
> > + if (!data_sz || !PAGE_ALIGNED(data_sz))
> > + return ERR_PTR(-EINVAL);
> > +
> > + if (data_sz > RINGBUF_MAX_DATA_SZ)
> > + return ERR_PTR(-E2BIG);
> > +
> > + rb = bpf_ringbuf_area_alloc(data_sz, numa_node);
> > + if (!rb)
> > + return ERR_PTR(-ENOMEM);
> > +
> > + spin_lock_init(&rb->spinlock);
> > + init_waitqueue_head(&rb->waitq);
> > + init_irq_work(&rb->work, bpf_ringbuf_notify);
> > +
> > + rb->mask = data_sz - 1;
> > + rb->consumer_pos = 0;
> > + rb->producer_pos = 0;
> > +
> > + return rb;
> > +}
> > +
> > +static struct bpf_map *ringbuf_map_alloc(union bpf_attr *attr)
> > +{
> > + struct bpf_ringbuf_map *rb_map;
> > + u64 cost;
> > + int err;
> > +
> > + if (attr->map_flags & ~RINGBUF_CREATE_FLAG_MASK)
> > + return ERR_PTR(-EINVAL);
> > +
> > + if (attr->key_size || attr->value_size ||
> > + attr->max_entries == 0 || !PAGE_ALIGNED(attr->max_entries))
> > + return ERR_PTR(-EINVAL);
> > +
> > + rb_map = kzalloc(sizeof(*rb_map), GFP_USER);
> > + if (!rb_map)
> > + return ERR_PTR(-ENOMEM);
> > +
> > + bpf_map_init_from_attr(&rb_map->map, attr);
> > +
> > + cost = sizeof(struct bpf_ringbuf_map) +
> > + sizeof(struct bpf_ringbuf) +
> > + attr->max_entries;
> > + err = bpf_map_charge_init(&rb_map->map.memory, cost);
> > + if (err)
> > + goto err_free_map;
> > +
> > + rb_map->rb = bpf_ringbuf_alloc(attr->max_entries, rb_map->map.numa_node);
> > + if (IS_ERR(rb_map->rb)) {
> > + err = PTR_ERR(rb_map->rb);
> > + goto err_uncharge;
> > + }
> > +
> > + return &rb_map->map;
> > +
> > +err_uncharge:
> > + bpf_map_charge_finish(&rb_map->map.memory);
> > +err_free_map:
> > + kfree(rb_map);
> > + return ERR_PTR(err);
> > +}
> > +
> > +static void bpf_ringbuf_free(struct bpf_ringbuf *ringbuf)
> > +{
> > + kvfree(ringbuf);
> > +}
> > +
> > +static void ringbuf_map_free(struct bpf_map *map)
> > +{
> > + struct bpf_ringbuf_map *rb_map;
> > +
> > + /* at this point bpf_prog->aux->refcnt == 0 and this map->refcnt == 0,
> > + * so the programs (can be more than one that used this map) were
> > + * disconnected from events. Wait for outstanding critical sections in
> > + * these programs to complete
> > + */
> > + synchronize_rcu();
> > +
> > + rb_map = container_of(map, struct bpf_ringbuf_map, map);
> > + bpf_ringbuf_free(rb_map->rb);
> > + kfree(rb_map);
> > +}
> > +
> > +static void *ringbuf_map_lookup_elem(struct bpf_map *map, void *key)
> > +{
> > + return ERR_PTR(-ENOTSUPP);
> > +}
> > +
> > +static int ringbuf_map_update_elem(struct bpf_map *map, void *key, void *value,
> > + u64 flags)
> > +{
> > + return -ENOTSUPP;
> > +}
> > +
> > +static int ringbuf_map_delete_elem(struct bpf_map *map, void *key)
> > +{
> > + return -ENOTSUPP;
> > +}
> > +
> > +static int ringbuf_map_get_next_key(struct bpf_map *map, void *key,
> > + void *next_key)
> > +{
> > + return -ENOTSUPP;
> > +}
> > +
> > +static size_t bpf_ringbuf_mmap_page_cnt(const struct bpf_ringbuf *rb)
> > +{
> > + size_t data_pages = (rb->mask + 1) >> PAGE_SHIFT;
> > +
> > + /* consumer page + producer page + 2 x data pages */
> > + return RINGBUF_POS_PAGES + 2 * data_pages;
> > +}
> > +
> > +static int ringbuf_map_mmap(struct bpf_map *map, struct vm_area_struct *vma)
> > +{
> > + struct bpf_ringbuf_map *rb_map;
> > + size_t mmap_sz;
> > +
> > + rb_map = container_of(map, struct bpf_ringbuf_map, map);
> > + mmap_sz = bpf_ringbuf_mmap_page_cnt(rb_map->rb) << PAGE_SHIFT;
> > +
> > + if (vma->vm_pgoff * PAGE_SIZE + (vma->vm_end - vma->vm_start) > mmap_sz)
> > + return -EINVAL;
> > +
> > + return remap_vmalloc_range(vma, rb_map->rb,
> > + vma->vm_pgoff + RINGBUF_PGOFF);
> > +}
> > +
> > +static unsigned long ringbuf_avail_data_sz(struct bpf_ringbuf *rb)
> > +{
> > + unsigned long cons_pos, prod_pos;
> > +
> > + cons_pos = smp_load_acquire(&rb->consumer_pos);
>
> What happens if there is a delay here? (The delay might be due to
> interrupts, preemption in PREEMPT=y kernels, vCPU preemption, ...)
>
> If this is called from a producer holding the lock, then the only
> ->consumer_pos can change, and that can only decrease the amount of
> data available. Besides which, ->consumer_pos is sampled first.
> But why would a producer care how much data was queued, as opposed
> to how much free space was available?
see second point below, it's for BPF program to implement custom
notification policy/heuristics
>
> From the consumer, only ->producer_pos can change, and that can only
> increase the amount of data available. (Assuming that producers cannot
> erase old data on wrap-around before the consumer consumes it.)
right, they can't overwrite data
>
> So probably nothing bad happens.
>
> On the bit about the producer holding the lock, some lockdep assertions
> might make things easier on your future self.
So this function is currently called in two situations, both not
holding the spinlock. I'm fully aware this is a racy way to do this,
but it's ok for the applications.
So the first place is during initial poll "subscription", when we need
to figure out if there is already some unconsumed data in ringbuffer
to let poll/epoll subsystem know whether caller can do read without
waiting. If this is done from consumer thread, it's race free, because
consumer position won't change, so either we'll see that producer >
consumer, or if we race with producer, producer will see that consumer
pos == to-be-committed record pos, so will send notification. If epoll
happens from non-consumer thread, it's similarly racy to perf buffer
poll subscription implementation, but with next record subscriber will
get notification anyways.
The second place is from BPF side (so potential producer), but it's
outside of producer lock. It's called from bpf_ringbuf_query() helper
which is supposed to return various potentially stale properties of
ringbuf (e.g., consumer/producer position, amount of unconsumed data,
etc). The use case here is for BPF program to implement its own
flexible poll notification strategy (or whatever else creative
programmer will come up with, of course). E.g., batched notification,
which happens not on every record, but only once enough data is
accumulated. In this case exact amount of unconsumed data is not
critical, it's only a heuristics. And it's more convenient than amount
of data left free, IMO, but both can be calculated using the other,
provided the total size of ring buffer is known (which can be returned
by bpf_ringbuf_query() helper as well).
So I think in both cases I don't want to take spinlock. If this is not
done under spinlock, then I have to read consumer pos first, producer
pos second, otherwise I can get into a situation of having
consumer_pos > producer_pos (due to delays), which is really bad.
It's a bit wordy explanation, but I hope this makes sense.
>
> > + prod_pos = smp_load_acquire(&rb->producer_pos);
> > + return prod_pos - cons_pos;
> > +}
> > +
> > +static __poll_t ringbuf_map_poll(struct bpf_map *map, struct file *filp,
> > + struct poll_table_struct *pts)
> > +{
> > + struct bpf_ringbuf_map *rb_map;
> > +
> > + rb_map = container_of(map, struct bpf_ringbuf_map, map);
> > + poll_wait(filp, &rb_map->rb->waitq, pts);
> > +
> > + if (ringbuf_avail_data_sz(rb_map->rb))
> > + return EPOLLIN | EPOLLRDNORM;
> > + return 0;
> > +}
> > +
> > +const struct bpf_map_ops ringbuf_map_ops = {
> > + .map_alloc = ringbuf_map_alloc,
> > + .map_free = ringbuf_map_free,
> > + .map_mmap = ringbuf_map_mmap,
> > + .map_poll = ringbuf_map_poll,
> > + .map_lookup_elem = ringbuf_map_lookup_elem,
> > + .map_update_elem = ringbuf_map_update_elem,
> > + .map_delete_elem = ringbuf_map_delete_elem,
> > + .map_get_next_key = ringbuf_map_get_next_key,
> > +};
> > +
> > +/* Given pointer to ring buffer record metadata and struct bpf_ringbuf itself,
> > + * calculate offset from record metadata to ring buffer in pages, rounded
> > + * down. This page offset is stored as part of record metadata and allows to
> > + * restore struct bpf_ringbuf * from record pointer. This page offset is
> > + * stored at offset 4 of record metadata header.
> > + */
> > +static size_t bpf_ringbuf_rec_pg_off(struct bpf_ringbuf *rb,
> > + struct bpf_ringbuf_hdr *hdr)
> > +{
> > + return ((void *)hdr - (void *)rb) >> PAGE_SHIFT;
> > +}
> > +
> > +/* Given pointer to ring buffer record header, restore pointer to struct
> > + * bpf_ringbuf itself by using page offset stored at offset 4
> > + */
> > +static struct bpf_ringbuf *
> > +bpf_ringbuf_restore_from_rec(struct bpf_ringbuf_hdr *hdr)
> > +{
> > + unsigned long addr = (unsigned long)(void *)hdr;
> > + unsigned long off = (unsigned long)hdr->pg_off << PAGE_SHIFT;
> > +
> > + return (void*)((addr & PAGE_MASK) - off);
> > +}
> > +
> > +static void *__bpf_ringbuf_reserve(struct bpf_ringbuf *rb, u64 size)
> > +{
> > + unsigned long cons_pos, prod_pos, new_prod_pos, flags;
> > + u32 len, pg_off;
> > + struct bpf_ringbuf_hdr *hdr;
> > +
> > + if (unlikely(size > RINGBUF_MAX_RECORD_SZ))
> > + return NULL;
> > +
> > + len = round_up(size + BPF_RINGBUF_HDR_SZ, 8);
> > + cons_pos = smp_load_acquire(&rb->consumer_pos);
>
> There might be a longish delay acquiring the spinlock, which could mean
> that cons_pos was out of date, which might result in an unnecessary
> producer-side failure. Why not pick up cons_pos after the lock is
> acquired? After all, it is in the same cache line as the lock, so this
> should have negligible effect on lock-hold time.
Right, the goal was to minimize amount of time that spinlock is hold.
Spinlock and consumer_pos are certainly not on the same cache line,
consumer_pos is deliberately on a different *page* altogether (due to
mmap() and to avoid unnecessary sharing between consumers, who don't
touch spinlock, and producers, who do use lock to coordinate).
So I chose to potentially drop sample due to a bit stale consumer pos,
but speed up locked section.
>
> (Unless you had either really small cachelines or really big locks.
>
> > + if (in_nmi()) {
> > + if (!spin_trylock_irqsave(&rb->spinlock, flags))
> > + return NULL;
> > + } else {
> > + spin_lock_irqsave(&rb->spinlock, flags);
> > + }
> > +
> > + prod_pos = rb->producer_pos;
> > + new_prod_pos = prod_pos + len;
> > +
> > + /* check for out of ringbuf space by ensuring producer position
> > + * doesn't advance more than (ringbuf_size - 1) ahead
> > + */
> > + if (new_prod_pos - cons_pos > rb->mask) {
> > + spin_unlock_irqrestore(&rb->spinlock, flags);
> > + return NULL;
> > + }
> > +
> > + hdr = (void *)rb->data + (prod_pos & rb->mask);
> > + pg_off = bpf_ringbuf_rec_pg_off(rb, hdr);
> > + hdr->len = size | BPF_RINGBUF_BUSY_BIT;
> > + hdr->pg_off = pg_off;
> > +
> > + /* ensure header is written before updating producer positions */
> > + smp_wmb();
>
> The smp_store_release() makes this unnecessary with respect to
> ->producer_pos. So what later write is it also ordering against?
> If none, this smp_wmb() can go away.
>
> And if the later write is the xchg() in bpf_ringbuf_commit(), the
> xchg() implies full barriers before and after, so that the smp_wmb()
> could still go away.
>
> So other than the smp_store_release() and the xchg(), what later write
> is the smp_wmb() ordering against?
I *think* my intent here was to make sure that when consumer reads
producer_pos, it sees hdr->len with busy bit set. But I also think
that you are right and smp_store_release (producer_pos) ensures that
if consumer does smp_read_acquire(producer_pos) it will see up-to-date
hdr->len, for a given value of producer_pos. So yeah, I think I should
drop smp_wmb(). I dropped it in litmus tests, and they still pass,
which gives me a bit more confidence as well. :)
I say "I *think*", because this algorithm started off completely
locklessly without producer spinlock, so maybe I needed this barrier
for something there, but I honestly don't remember details of lockless
implementation by now.
So in summary, yeah, I'll drop smp_wmb().
>
> > + /* pairs with consumer's smp_load_acquire() */
> > + smp_store_release(&rb->producer_pos, new_prod_pos);
> > +
> > + spin_unlock_irqrestore(&rb->spinlock, flags);
> > +
> > + return (void *)hdr + BPF_RINGBUF_HDR_SZ;
> > +}
> > +
> > +BPF_CALL_3(bpf_ringbuf_reserve, struct bpf_map *, map, u64, size, u64, flags)
> > +{
> > + struct bpf_ringbuf_map *rb_map;
> > +
> > + if (unlikely(flags))
> > + return 0;
> > +
> > + rb_map = container_of(map, struct bpf_ringbuf_map, map);
> > + return (unsigned long)__bpf_ringbuf_reserve(rb_map->rb, size);
> > +}
> > +
> > +const struct bpf_func_proto bpf_ringbuf_reserve_proto = {
> > + .func = bpf_ringbuf_reserve,
> > + .ret_type = RET_PTR_TO_ALLOC_MEM_OR_NULL,
> > + .arg1_type = ARG_CONST_MAP_PTR,
> > + .arg2_type = ARG_CONST_ALLOC_SIZE_OR_ZERO,
> > + .arg3_type = ARG_ANYTHING,
> > +};
> > +
> > +static void bpf_ringbuf_commit(void *sample, u64 flags, bool discard)
> > +{
> > + unsigned long rec_pos, cons_pos;
> > + struct bpf_ringbuf_hdr *hdr;
> > + struct bpf_ringbuf *rb;
> > + u32 new_len;
> > +
> > + hdr = sample - BPF_RINGBUF_HDR_SZ;
> > + rb = bpf_ringbuf_restore_from_rec(hdr);
> > + new_len = hdr->len ^ BPF_RINGBUF_BUSY_BIT;
> > + if (discard)
> > + new_len |= BPF_RINGBUF_DISCARD_BIT;
> > +
> > + /* update record header with correct final size prefix */
> > + xchg(&hdr->len, new_len);
> > +
> > + /* if consumer caught up and is waiting for our record, notify about
> > + * new data availability
> > + */
> > + rec_pos = (void *)hdr - (void *)rb->data;
> > + cons_pos = smp_load_acquire(&rb->consumer_pos) & rb->mask;
> > +
> > + if (flags & BPF_RB_FORCE_WAKEUP)
> > + irq_work_queue(&rb->work);
> > + else if (cons_pos == rec_pos && !(flags & BPF_RB_NO_WAKEUP))
> > + irq_work_queue(&rb->work);
> > +}
