Replace direct modifications to vma->vm_flags with calls to modifier
functions to be able to track flag changes and to keep vma locking
correctness.
[akpm@linux-foundation.org: fix drivers/misc/open-dice.c, per Hyeonggon Yoo]
Link: https://lkml.kernel.org/r/20230126193752.297968-5-surenb@google.com
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Acked-by: Sebastian Reichel <sebastian.reichel@collabora.com>
Reviewed-by: Liam R. Howlett <Liam.Howlett@Oracle.com>
Reviewed-by: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Arjun Roy <arjunroy@google.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: David Rientjes <rientjes@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jann Horn <jannh@google.com>
Cc: Joel Fernandes <joelaf@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Cc: Laurent Dufour <ldufour@linux.ibm.com>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Minchan Kim <minchan@google.com>
Cc: Paul E. McKenney <paulmck@kernel.org>
Cc: Peter Oskolkov <posk@google.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Punit Agrawal <punit.agrawal@bytedance.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Soheil Hassas Yeganeh <soheil@google.com>
Cc: Song Liu <songliubraving@fb.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Currently, verifier uses MEM_ALLOC type tag to specially tag memory
returned from bpf_ringbuf_reserve helper. However, this is currently
only used for this purpose and there is an implicit assumption that it
only refers to ringbuf memory (e.g. the check for ARG_PTR_TO_ALLOC_MEM
in check_func_arg_reg_off).
Hence, rename MEM_ALLOC to MEM_RINGBUF to indicate this special
relationship and instead open the use of MEM_ALLOC for more generic
allocations made for user types.
Also, since ARG_PTR_TO_ALLOC_MEM_OR_NULL is unused, simply drop it.
Finally, update selftests using 'alloc_' verifier string to 'ringbuf_'.
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Link: https://lore.kernel.org/r/20221114191547.1694267-7-memxor@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
In a prior change, we added a new BPF_MAP_TYPE_USER_RINGBUF map type which
will allow user-space applications to publish messages to a ring buffer
that is consumed by a BPF program in kernel-space. In order for this
map-type to be useful, it will require a BPF helper function that BPF
programs can invoke to drain samples from the ring buffer, and invoke
callbacks on those samples. This change adds that capability via a new BPF
helper function:
bpf_user_ringbuf_drain(struct bpf_map *map, void *callback_fn, void *ctx,
u64 flags)
BPF programs may invoke this function to run callback_fn() on a series of
samples in the ring buffer. callback_fn() has the following signature:
long callback_fn(struct bpf_dynptr *dynptr, void *context);
Samples are provided to the callback in the form of struct bpf_dynptr *'s,
which the program can read using BPF helper functions for querying
struct bpf_dynptr's.
In order to support bpf_ringbuf_drain(), a new PTR_TO_DYNPTR register
type is added to the verifier to reflect a dynptr that was allocated by
a helper function and passed to a BPF program. Unlike PTR_TO_STACK
dynptrs which are allocated on the stack by a BPF program, PTR_TO_DYNPTR
dynptrs need not use reference tracking, as the BPF helper is trusted to
properly free the dynptr before returning. The verifier currently only
supports PTR_TO_DYNPTR registers that are also DYNPTR_TYPE_LOCAL.
Note that while the corresponding user-space libbpf logic will be added
in a subsequent patch, this patch does contain an implementation of the
.map_poll() callback for BPF_MAP_TYPE_USER_RINGBUF maps. This
.map_poll() callback guarantees that an epoll-waiting user-space
producer will receive at least one event notification whenever at least
one sample is drained in an invocation of bpf_user_ringbuf_drain(),
provided that the function is not invoked with the BPF_RB_NO_WAKEUP
flag. If the BPF_RB_FORCE_WAKEUP flag is provided, a wakeup
notification is sent even if no sample was drained.
Signed-off-by: David Vernet <void@manifault.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20220920000100.477320-3-void@manifault.com
We want to support a ringbuf map type where samples are published from
user-space, to be consumed by BPF programs. BPF currently supports a
kernel -> user-space circular ring buffer via the BPF_MAP_TYPE_RINGBUF
map type. We'll need to define a new map type for user-space -> kernel,
as none of the helpers exported for BPF_MAP_TYPE_RINGBUF will apply
to a user-space producer ring buffer, and we'll want to add one or
more helper functions that would not apply for a kernel-producer
ring buffer.
This patch therefore adds a new BPF_MAP_TYPE_USER_RINGBUF map type
definition. The map type is useless in its current form, as there is no
way to access or use it for anything until we one or more BPF helpers. A
follow-on patch will therefore add a new helper function that allows BPF
programs to run callbacks on samples that are published to the ring
buffer.
Signed-off-by: David Vernet <void@manifault.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20220920000100.477320-2-void@manifault.com
Some of the bpf maps are created with __GFP_NOWARN, i.e. arraymap,
bloom_filter, bpf_local_storage, bpf_struct_ops, lpm_trie,
queue_stack_maps, reuseport_array, stackmap and xskmap, while others are
created without __GFP_NOWARN, i.e. cpumap, devmap, hashtab,
local_storage, offload, ringbuf and sock_map. But there are not key
differences between the creation of these maps. So let make this
allocation flag consistent in all bpf maps creation. Then we can use a
generic helper to alloc all bpf maps.
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Link: https://lore.kernel.org/r/20220810151840.16394-4-laoar.shao@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Currently, our only way of writing dynamically-sized data into a ring
buffer is through bpf_ringbuf_output but this incurs an extra memcpy
cost. bpf_ringbuf_reserve + bpf_ringbuf_commit avoids this extra
memcpy, but it can only safely support reservation sizes that are
statically known since the verifier cannot guarantee that the bpf
program won’t access memory outside the reserved space.
The bpf_dynptr abstraction allows for dynamically-sized ring buffer
reservations without the extra memcpy.
There are 3 new APIs:
long bpf_ringbuf_reserve_dynptr(void *ringbuf, u32 size, u64 flags, struct bpf_dynptr *ptr);
void bpf_ringbuf_submit_dynptr(struct bpf_dynptr *ptr, u64 flags);
void bpf_ringbuf_discard_dynptr(struct bpf_dynptr *ptr, u64 flags);
These closely follow the functionalities of the original ringbuf APIs.
For example, all ringbuffer dynptrs that have been reserved must be
either submitted or discarded before the program exits.
Signed-off-by: Joanne Koong <joannelkoong@gmail.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: David Vernet <void@manifault.com>
Link: https://lore.kernel.org/bpf/20220523210712.3641569-4-joannelkoong@gmail.com
For now, the field 'map_btf_id' in 'struct bpf_map_ops' for all map
types are computed during vmlinux-btf init:
btf_parse_vmlinux() -> btf_vmlinux_map_ids_init()
It will lookup the btf_type according to the 'map_btf_name' field in
'struct bpf_map_ops'. This process can be done during build time,
thanks to Jiri's resolve_btfids.
selftest of map_ptr has passed:
$96 map_ptr:OK
Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Menglong Dong <imagedong@tencent.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add a new type flag for bpf_arg_type that when set tells verifier that
for a release function, that argument's register will be the one for
which meta.ref_obj_id will be set, and which will then be released
using release_reference. To capture the regno, introduce a new field
release_regno in bpf_call_arg_meta.
This would be required in the next patch, where we may either pass NULL
or a refcounted pointer as an argument to the release function
bpf_kptr_xchg. Just releasing only when meta.ref_obj_id is set is not
enough, as there is a case where the type of argument needed matches,
but the ref_obj_id is set to 0. Hence, we must enforce that whenever
meta.ref_obj_id is zero, the register that is to be released can only
be NULL for a release function.
Since we now indicate whether an argument is to be released in
bpf_func_proto itself, is_release_function helper has lost its utitlity,
hence refactor code to work without it, and just rely on
meta.release_regno to know when to release state for a ref_obj_id.
Still, the restriction of one release argument and only one ref_obj_id
passed to BPF helper or kfunc remains. This may be lifted in the future.
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20220424214901.2743946-3-memxor@gmail.com
After commit 2fd3fb0be1d1 ("kasan, vmalloc: unpoison VM_ALLOC pages
after mapping"), non-VM_ALLOC mappings will be marked as accessible
in __get_vm_area_node() when KASAN is enabled. But now the flag for
ringbuf area is VM_ALLOC, so KASAN will complain out-of-bound access
after vmap() returns. Because the ringbuf area is created by mapping
allocated pages, so use VM_MAP instead.
After the change, info in /proc/vmallocinfo also changes from
[start]-[end] 24576 ringbuf_map_alloc+0x171/0x290 vmalloc user
to
[start]-[end] 24576 ringbuf_map_alloc+0x171/0x290 vmap user
Fixes: 457f44363a ("bpf: Implement BPF ring buffer and verifier support for it")
Reported-by: syzbot+5ad567a418794b9b5983@syzkaller.appspotmail.com
Signed-off-by: Hou Tao <houtao1@huawei.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20220202060158.6260-1-houtao1@huawei.com
Some helper functions may modify its arguments, for example,
bpf_d_path, bpf_get_stack etc. Previously, their argument types
were marked as ARG_PTR_TO_MEM, which is compatible with read-only
mem types, such as PTR_TO_RDONLY_BUF. Therefore it's legitimate,
but technically incorrect, to modify a read-only memory by passing
it into one of such helper functions.
This patch tags the bpf_args compatible with immutable memory with
MEM_RDONLY flag. The arguments that don't have this flag will be
only compatible with mutable memory types, preventing the helper
from modifying a read-only memory. The bpf_args that have
MEM_RDONLY are compatible with both mutable memory and immutable
memory.
Signed-off-by: Hao Luo <haoluo@google.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20211217003152.48334-9-haoluo@google.com
kmemleak scans struct page, but it does not scan the page content. If we
allocate some memory with kmalloc(), then allocate page with alloc_page(),
and if we put kmalloc pointer somewhere inside that page, kmemleak will
report kmalloc pointer as a false positive.
We can instruct kmemleak to scan the memory area by calling kmemleak_alloc()
and kmemleak_free(), but part of struct bpf_ringbuf is mmaped to user space,
and if struct bpf_ringbuf changes we would have to revisit and review size
argument in kmemleak_alloc(), because we do not want kmemleak to scan the
user space memory. Let's simplify things and use kmemleak_not_leak() here.
For posterity, also adding additional prior analysis from Andrii:
I think either kmemleak or syzbot are misreporting this. I've added a
bunch of printks around all allocations performed by BPF ringbuf. [...]
On repro side I get these two warnings:
[vmuser@archvm bpf]$ sudo ./repro
BUG: memory leak
unreferenced object 0xffff88810d538c00 (size 64):
comm "repro", pid 2140, jiffies 4294692933 (age 14.540s)
hex dump (first 32 bytes):
00 af 19 04 00 ea ff ff c0 ae 19 04 00 ea ff ff ................
80 ae 19 04 00 ea ff ff c0 29 2e 04 00 ea ff ff .........)......
backtrace:
[<0000000077bfbfbd>] __bpf_map_area_alloc+0x31/0xc0
[<00000000587fa522>] ringbuf_map_alloc.cold.4+0x48/0x218
[<0000000044d49e96>] __do_sys_bpf+0x359/0x1d90
[<00000000f601d565>] do_syscall_64+0x2d/0x40
[<0000000043d3112a>] entry_SYSCALL_64_after_hwframe+0x44/0xae
BUG: memory leak
unreferenced object 0xffff88810d538c80 (size 64):
comm "repro", pid 2143, jiffies 4294699025 (age 8.448s)
hex dump (first 32 bytes):
80 aa 19 04 00 ea ff ff 00 ab 19 04 00 ea ff ff ................
c0 ab 19 04 00 ea ff ff 80 44 28 04 00 ea ff ff .........D(.....
backtrace:
[<0000000077bfbfbd>] __bpf_map_area_alloc+0x31/0xc0
[<00000000587fa522>] ringbuf_map_alloc.cold.4+0x48/0x218
[<0000000044d49e96>] __do_sys_bpf+0x359/0x1d90
[<00000000f601d565>] do_syscall_64+0x2d/0x40
[<0000000043d3112a>] entry_SYSCALL_64_after_hwframe+0x44/0xae
Note that both reported leaks (ffff88810d538c80 and ffff88810d538c00)
correspond to pages array bpf_ringbuf is allocating and tracking properly
internally. Note also that syzbot repro doesn't close FD of created BPF
ringbufs, and even when ./repro itself exits with error, there are still
two forked processes hanging around in my system. So clearly ringbuf maps
are alive at that point. So reporting any memory leak looks weird at that
point, because that memory is being used by active referenced BPF ringbuf.
It's also a question why repro doesn't clean up its forks. But if I do a
`pkill repro`, I do see that all the allocated memory is /properly/ cleaned
up [and the] "leaks" are deallocated properly.
BTW, if I add close() right after bpf() syscall in syzbot repro, I see that
everything is immediately deallocated, like designed. And no memory leak
is reported. So I don't think the problem is anywhere in bpf_ringbuf code,
rather in the leak detection and/or repro itself.
Reported-by: syzbot+5d895828587f49e7fe9b@syzkaller.appspotmail.com
Signed-off-by: Rustam Kovhaev <rkovhaev@gmail.com>
[ Daniel: also included analysis from Andrii to the commit log ]
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: syzbot+5d895828587f49e7fe9b@syzkaller.appspotmail.com
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/CAEf4BzYk+dqs+jwu6VKXP-RttcTEGFe+ySTGWT9CRNkagDiJVA@mail.gmail.com
Link: https://lore.kernel.org/lkml/YNTAqiE7CWJhOK2M@nuc10
Link: https://lore.kernel.org/lkml/20210615101515.GC26027@arm.com
Link: https://syzkaller.appspot.com/bug?extid=5d895828587f49e7fe9b
Link: https://lore.kernel.org/bpf/20210626181156.1873604-1-rkovhaev@gmail.com
Only the very first page of BPF ringbuf that contains consumer position
counter is supposed to be mapped as writeable by user-space. Producer
position is read-only and can be modified only by the kernel code. BPF ringbuf
data pages are read-only as well and are not meant to be modified by
user-code to maintain integrity of per-record headers.
This patch allows to map only consumer position page as writeable and
everything else is restricted to be read-only. remap_vmalloc_range()
internally adds VM_DONTEXPAND, so all the established memory mappings can't be
extended, which prevents any future violations through mremap()'ing.
Fixes: 457f44363a ("bpf: Implement BPF ring buffer and verifier support for it")
Reported-by: Ryota Shiga (Flatt Security)
Reported-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
A BPF program might try to reserve a buffer larger than the ringbuf size.
If the consumer pointer is way ahead of the producer, that would be
successfully reserved, allowing the BPF program to read or write out of
the ringbuf allocated area.
Reported-by: Ryota Shiga (Flatt Security)
Fixes: 457f44363a ("bpf: Implement BPF ring buffer and verifier support for it")
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Do not use rlimit-based memory accounting for bpf ringbuffer.
It has been replaced with the memcg-based memory accounting.
bpf_ringbuf_alloc() can't return anything except ERR_PTR(-ENOMEM)
and a valid pointer, so to simplify the code make it return NULL
in the first case. This allows to drop a couple of lines in
ringbuf_map_alloc() and also makes it look similar to other memory
allocating function like kmalloc().
Signed-off-by: Roman Gushchin <guro@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Song Liu <songliubraving@fb.com>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Link: https://lore.kernel.org/bpf/20201201215900.3569844-28-guro@fb.com
Some properties of the inner map is used in the verification time.
When an inner map is inserted to an outer map at runtime,
bpf_map_meta_equal() is currently used to ensure those properties
of the inserting inner map stays the same as the verification
time.
In particular, the current bpf_map_meta_equal() checks max_entries which
turns out to be too restrictive for most of the maps which do not use
max_entries during the verification time. It limits the use case that
wants to replace a smaller inner map with a larger inner map. There are
some maps do use max_entries during verification though. For example,
the map_gen_lookup in array_map_ops uses the max_entries to generate
the inline lookup code.
To accommodate differences between maps, the map_meta_equal is added
to bpf_map_ops. Each map-type can decide what to check when its
map is used as an inner map during runtime.
Also, some map types cannot be used as an inner map and they are
currently black listed in bpf_map_meta_alloc() in map_in_map.c.
It is not unusual that the new map types may not aware that such
blacklist exists. This patch enforces an explicit opt-in
and only allows a map to be used as an inner map if it has
implemented the map_meta_equal ops. It is based on the
discussion in [1].
All maps that support inner map has its map_meta_equal points
to bpf_map_meta_equal in this patch. A later patch will
relax the max_entries check for most maps. bpf_types.h
counts 28 map types. This patch adds 23 ".map_meta_equal"
by using coccinelle. -5 for
BPF_MAP_TYPE_PROG_ARRAY
BPF_MAP_TYPE_(PERCPU)_CGROUP_STORAGE
BPF_MAP_TYPE_STRUCT_OPS
BPF_MAP_TYPE_ARRAY_OF_MAPS
BPF_MAP_TYPE_HASH_OF_MAPS
The "if (inner_map->inner_map_meta)" check in bpf_map_meta_alloc()
is moved such that the same error is returned.
[1]: https://lore.kernel.org/bpf/20200522022342.899756-1-kafai@fb.com/
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200828011806.1970400-1-kafai@fb.com
bpf_free_used_maps() or close(map_fd) will trigger map_free callback.
bpf_free_used_maps() is called after bpf prog is no longer executing:
bpf_prog_put->call_rcu->bpf_prog_free->bpf_free_used_maps.
Hence there is no need to call synchronize_rcu() to protect map elements.
Note that hash_of_maps and array_of_maps update/delete inner maps via
sys_bpf() that calls maybe_wait_bpf_programs() and synchronize_rcu().
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Link: https://lore.kernel.org/bpf/20200630043343.53195-2-alexei.starovoitov@gmail.com
BPF ringbuf assumes the size to be a multiple of page size and the power of
2 value. The latter is important to avoid division while calculating position
inside the ring buffer and using (N-1) mask instead. This patch fixes omission
to enforce power-of-2 size rule.
Fixes: 457f44363a ("bpf: Implement BPF ring buffer and verifier support for it")
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200630061500.1804799-1-andriin@fb.com
Set map_btf_name and map_btf_id for all map types so that map fields can
be accessed by bpf programs.
Signed-off-by: Andrey Ignatov <rdna@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Link: https://lore.kernel.org/bpf/a825f808f22af52b018dbe82f1c7d29dab5fc978.1592600985.git.rdna@fb.com
This commit 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.
Motivation
----------
There are two distinctive motivators for this work, which are not satisfied by
existing perf buffer, which prompted creation of a new ring buffer
implementation.
- more efficient memory utilization by sharing ring buffer across CPUs;
- preserving ordering of events that happen sequentially in time, even
across multiple CPUs (e.g., fork/exec/exit events for a task).
These two problems are independent, but perf buffer fails to satisfy both.
Both are a result of a choice to have per-CPU perf ring buffer. Both can be
also solved by having an MPSC implementation of ring buffer. The ordering
problem could technically be solved for perf buffer with some in-kernel
counting, but given the first one requires an MPSC buffer, the same solution
would solve the second problem automatically.
Semantics and APIs
------------------
Single ring buffer is presented to BPF programs as an instance of BPF map of
type BPF_MAP_TYPE_RINGBUF. Two other alternatives considered, but ultimately
rejected.
One way would be to, similar to BPF_MAP_TYPE_PERF_EVENT_ARRAY, make
BPF_MAP_TYPE_RINGBUF could represent an array of ring buffers, but not enforce
"same CPU only" rule. This would be more familiar interface compatible with
existing perf buffer use in BPF, but would fail if application needed more
advanced logic to lookup ring buffer by arbitrary key. HASH_OF_MAPS addresses
this with current approach. Additionally, given the performance of BPF
ringbuf, many use cases would just opt into a simple single ring buffer shared
among all CPUs, for which current approach would be an overkill.
Another approach could introduce a new concept, alongside BPF map, to
represent generic "container" object, which doesn't necessarily have key/value
interface with lookup/update/delete operations. This approach would add a lot
of extra infrastructure that has to be built for observability and verifier
support. It would also add another concept that BPF developers would have to
familiarize themselves with, new syntax in libbpf, etc. But then would really
provide no additional benefits over the approach of using a map.
BPF_MAP_TYPE_RINGBUF doesn't support lookup/update/delete operations, but so
doesn't few other map types (e.g., queue and stack; array doesn't support
delete, etc).
The approach chosen has an advantage of re-using existing BPF map
infrastructure (introspection APIs in kernel, libbpf support, etc), being
familiar concept (no need to teach users a new type of object in BPF program),
and utilizing existing tooling (bpftool). For common scenario of using
a single ring buffer for all CPUs, it's as simple and straightforward, as
would be with a dedicated "container" object. On the other hand, by being
a map, it can be combined with ARRAY_OF_MAPS and HASH_OF_MAPS map-in-maps to
implement a wide variety of topologies, from one ring buffer for each CPU
(e.g., as a replacement for perf buffer use cases), to a complicated
application hashing/sharding of ring buffers (e.g., having a small pool of
ring buffers with hashed task's tgid being a look up key to preserve order,
but reduce contention).
Key and value sizes are enforced to be zero. max_entries is used to specify
the size of ring buffer and has to be a power of 2 value.
There are a bunch of similarities between perf buffer
(BPF_MAP_TYPE_PERF_EVENT_ARRAY) and new BPF ring buffer semantics:
- variable-length records;
- if there is no more space left in ring buffer, reservation fails, no
blocking;
- memory-mappable data area for user-space applications for ease of
consumption and high performance;
- epoll notifications for new incoming data;
- but still the ability to do busy polling for new data to achieve the
lowest latency, if necessary.
BPF ringbuf provides two sets of APIs to BPF programs:
- bpf_ringbuf_output() allows to *copy* data from one place to a ring
buffer, similarly to bpf_perf_event_output();
- bpf_ringbuf_reserve()/bpf_ringbuf_commit()/bpf_ringbuf_discard() APIs
split the whole process into two steps. First, a fixed amount of space is
reserved. If successful, a pointer to a data inside ring buffer data area
is returned, which BPF programs can use similarly to a data inside
array/hash maps. Once ready, this piece of memory is either committed or
discarded. Discard is similar to commit, but makes consumer ignore the
record.
bpf_ringbuf_output() has disadvantage of incurring extra memory copy, because
record has to be prepared in some other place first. But it allows to submit
records of the length that's not known to verifier beforehand. It also closely
matches bpf_perf_event_output(), so will simplify migration significantly.
bpf_ringbuf_reserve() avoids the extra copy of memory by providing a memory
pointer directly to ring buffer memory. In a lot of cases records are larger
than BPF stack space allows, so many programs have use extra per-CPU array as
a temporary heap for preparing sample. bpf_ringbuf_reserve() avoid this needs
completely. But in exchange, it only allows a known constant size of memory to
be reserved, such that verifier can verify that BPF program can't access
memory outside its reserved record space. bpf_ringbuf_output(), while slightly
slower due to extra memory copy, covers some use cases that are not suitable
for bpf_ringbuf_reserve().
The difference between commit and discard is very small. Discard just marks
a record as discarded, and such records are supposed to be ignored by consumer
code. Discard is useful for some advanced use-cases, such as ensuring
all-or-nothing multi-record submission, or emulating temporary malloc()/free()
within single BPF program invocation.
Each reserved record is tracked by verifier through existing
reference-tracking logic, similar to socket ref-tracking. It is thus
impossible to reserve a record, but forget to submit (or discard) it.
bpf_ringbuf_query() helper allows to query various properties of ring buffer.
Currently 4 are supported:
- BPF_RB_AVAIL_DATA returns amount of unconsumed data in ring buffer;
- BPF_RB_RING_SIZE returns the size of ring buffer;
- BPF_RB_CONS_POS/BPF_RB_PROD_POS returns current logical possition of
consumer/producer, respectively.
Returned values are momentarily snapshots of ring buffer state and could be
off by the time helper returns, so this should be used only for
debugging/reporting reasons or for implementing various heuristics, that take
into account highly-changeable nature of some of those characteristics.
One such heuristic might involve more fine-grained control over poll/epoll
notifications about new data availability in ring buffer. Together with
BPF_RB_NO_WAKEUP/BPF_RB_FORCE_WAKEUP flags for output/commit/discard helpers,
it allows BPF program a high degree of control and, e.g., more efficient
batched notifications. Default self-balancing strategy, though, should be
adequate for most applications and will work reliable and efficiently already.
Design and implementation
-------------------------
This reserve/commit schema allows a natural way for multiple producers, either
on different CPUs or even on the same CPU/in the same BPF program, to reserve
independent records and work with them without blocking other producers. This
means that if BPF program was interruped by another BPF program sharing the
same ring buffer, they will both get a record reserved (provided there is
enough space left) and can work with it and submit it independently. This
applies to NMI context as well, except that due to using a spinlock during
reservation, in NMI context, bpf_ringbuf_reserve() might fail to get a lock,
in which case reservation will fail even if ring buffer is not full.
The ring buffer itself internally is implemented as a power-of-2 sized
circular buffer, with two logical and ever-increasing counters (which might
wrap around on 32-bit architectures, that's not a problem):
- consumer counter shows up to which logical position consumer consumed the
data;
- producer counter denotes amount of data reserved by all producers.
Each time a record is reserved, producer that "owns" the record will
successfully advance producer counter. At that point, data is still not yet
ready to be consumed, though. Each record has 8 byte header, which contains
the length of reserved record, as well as two extra bits: busy bit to denote
that record is still being worked on, and discard bit, which might be set at
commit time if record is discarded. In the latter case, consumer is supposed
to skip the record and move on to the next one. Record header also encodes
record's relative offset from the beginning of ring buffer data area (in
pages). This allows bpf_ringbuf_commit()/bpf_ringbuf_discard() to accept only
the pointer to the record itself, without requiring also the pointer to ring
buffer itself. Ring buffer memory location will be restored from record
metadata header. This significantly simplifies verifier, as well as improving
API usability.
Producer counter increments are serialized under spinlock, so there is
a strict ordering between reservations. Commits, on the other hand, are
completely lockless and independent. All records become available to consumer
in the order of reservations, but only after all previous records where
already committed. It is thus possible for slow producers to temporarily hold
off submitted records, that were reserved later.
Reservation/commit/consumer protocol is verified by litmus tests in
Documentation/litmus-test/bpf-rb.
One interesting implementation bit, that significantly simplifies (and thus
speeds up as well) implementation of both producers and consumers is how data
area is mapped twice contiguously back-to-back in the virtual memory. This
allows to not take any special measures for samples that have to wrap around
at the end of the circular buffer data area, because the next page after the
last data page would be first data page again, and thus the sample will still
appear completely contiguous in virtual memory. See comment and a simple ASCII
diagram showing this visually in bpf_ringbuf_area_alloc().
Another feature that distinguishes BPF ringbuf from perf ring buffer is
a self-pacing notifications of new data being availability.
bpf_ringbuf_commit() implementation will send a notification of new record
being available after commit only if consumer has already caught up right up
to the record being committed. If not, consumer still has to catch up and thus
will see new data anyways without needing an extra poll notification.
Benchmarks (see tools/testing/selftests/bpf/benchs/bench_ringbuf.c) show that
this allows to achieve a very high throughput without having to resort to
tricks like "notify only every Nth sample", which are necessary with perf
buffer. For extreme cases, when BPF program wants more manual control of
notifications, commit/discard/output helpers accept BPF_RB_NO_WAKEUP and
BPF_RB_FORCE_WAKEUP flags, which give full control over notifications of data
availability, but require extra caution and diligence in using this API.
Comparison to alternatives
--------------------------
Before considering implementing BPF ring buffer from scratch existing
alternatives in kernel were evaluated, but didn't seem to meet the needs. They
largely fell into few categores:
- per-CPU buffers (perf, ftrace, etc), which don't satisfy two motivations
outlined above (ordering and memory consumption);
- linked list-based implementations; while some were multi-producer designs,
consuming these from user-space would be very complicated and most
probably not performant; memory-mapping contiguous piece of memory is
simpler and more performant for user-space consumers;
- io_uring is SPSC, but also requires fixed-sized elements. Naively turning
SPSC queue into MPSC w/ lock would have subpar performance compared to
locked reserve + lockless commit, as with BPF ring buffer. Fixed sized
elements would be too limiting for BPF programs, given existing BPF
programs heavily rely on variable-sized perf buffer already;
- specialized implementations (like a new printk ring buffer, [0]) with lots
of printk-specific limitations and implications, that didn't seem to fit
well for intended use with BPF programs.
[0] https://lwn.net/Articles/779550/
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200529075424.3139988-2-andriin@fb.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>