user/sven/linux.git/include/linux/bpf_verifier.h, branch v5.8.10

bpf: Implement BPF ring buffer and verifier support for it

2020-06-01T21:38:22Z

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 Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20200529075424.3139988-2-andriin@fb.com Signed-off-by: Alexei Starovoitov

bpf: Implement CAP_BPF

2020-05-15T15:29:41Z

Implement permissions as stated in uapi/linux/capability.h In order to do that the verifier allow_ptr_leaks flag is split into four flags and they are set as: env->allow_ptr_leaks = bpf_allow_ptr_leaks(); env->bypass_spec_v1 = bpf_bypass_spec_v1(); env->bypass_spec_v4 = bpf_bypass_spec_v4(); env->bpf_capable = bpf_capable(); The first three currently equivalent to perfmon_capable(), since leaking kernel pointers and reading kernel memory via side channel attacks is roughly equivalent to reading kernel memory with cap_perfmon. 'bpf_capable' enables bounded loops, precision tracking, bpf to bpf calls and other verifier features. 'allow_ptr_leaks' enable ptr leaks, ptr conversions, subtraction of pointers. 'bypass_spec_v1' disables speculative analysis in the verifier, run time mitigations in bpf array, and enables indirect variable access in bpf programs. 'bypass_spec_v4' disables emission of sanitation code by the verifier. That means that the networking BPF program loaded with CAP_BPF + CAP_NET_ADMIN will have speculative checks done by the verifier and other spectre mitigation applied. Such networking BPF program will not be able to leak kernel pointers and will not be able to access arbitrary kernel memory. Signed-off-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20200513230355.7858-3-alexei.starovoitov@gmail.com

bpf: Verifier, do explicit ALU32 bounds tracking

2020-03-30T21:59:53Z

It is not possible for the current verifier to track ALU32 and JMP ops correctly. This can result in the verifier aborting with errors even though the program should be verifiable. BPF codes that hit this can work around it by changin int variables to 64-bit types, marking variables volatile, etc. But this is all very ugly so it would be better to avoid these tricks. But, the main reason to address this now is do_refine_retval_range() was assuming return values could not be negative. Once we fixed this code that was previously working will no longer work. See do_refine_retval_range() patch for details. And we don't want to suddenly cause programs that used to work to fail. The simplest example code snippet that illustrates the problem is likely this, 53: w8 = w0 // r8 <- [0, S32_MAX], // w8 <- [-S32_MIN, X] 54: w8 64-bit 2. MOV ALU64 - copy 64-bit -> 32-bit 3. op ALU32 - zext 32-bit -> 64-bit 4. op ALU64 - n/a 5. jmp ALU32 - 64-bit: var32_off | upper_32_bits(var64_off) 6. jmp ALU64 - 32-bit: (>> (<< var64_off)) Details for each case, For "MOV ALU32" BPF arch zero extends so we simply copy the bounds from 32-bit into 64-bit ensuring we truncate var_off and 64-bit bounds correctly. See zext_32_to_64. For "MOV ALU64" copy all bounds including 32-bit into new register. If the src register had 32-bit bounds the dst register will as well. For "op ALU32" zero extend 32-bit into 64-bit the same as move, see zext_32_to_64. For "op ALU64" calculate both 32-bit and 64-bit bounds no merging is done here. Except we have a special case. When RSH or ARSH is done we can't simply ignore shifting bits from 64-bit reg into the 32-bit subreg. So currently just push bounds from 64-bit into 32-bit. This will be correct in the sense that they will represent a valid state of the register. However we could lose some accuracy if an ARSH is following a jmp32 operation. We can handle this special case in a follow up series. For "jmp ALU32" mark 64-bit reg unknown and recalculate 64-bit bounds from tnum by setting var_off to ((<<(>>var_off)) | var32_off). We special case if 64-bit bounds has zero'd upper 32bits at which point we can simply copy 32-bit bounds into 64-bit register. This catches a common compiler trick where upper 32-bits are zeroed and then 32-bit ops are used followed by a 64-bit compare or 64-bit op on a pointer. See __reg_combine_64_into_32(). For "jmp ALU64" cast the bounds of the 64bit to their 32-bit counterpart. For example s32_min_value = (s32)reg->smin_value. For tnum use only the lower 32bits via, (>>(< Signed-off-by: Alexei Starovoitov Link: https://lore.kernel.org/bpf/158560419880.10843.11448220440809118343.stgit@john-Precision-5820-Tower

bpf: Introduce function-by-function verification

2020-01-10T16:20:07Z
New llvm and old llvm with libbpf help produce BTF that distinguish global and static functions. Unlike arguments of static function the arguments of global functions cannot be removed or optimized away by llvm. The compiler has to use exactly the arguments specified in a function prototype. The argument type information allows the verifier validate each global function independently. For now only supported argument types are pointer to context and scalars. In the future pointers to structures, sizes, pointer to packet data can be supported as well. Consider the following example: static int f1(int ...) { ... } int f3(int b); int f2(int a) { f1(a) + f3(a); } int f3(int b) { ... } int main(...) { f1(...) + f2(...) + f3(...); } The verifier will start its safety checks from the first global function f2(). It will recursively descend into f1() because it's static. Then it will check that arguments match for the f3() invocation inside f2(). It will not descend into f3(). It will finish f2() that has to be successfully verified for all possible values of 'a'. Then it will proceed with f3(). That function also has to be safe for all possible values of 'b'. Then it will start subprog 0 (which is main() function). It will recursively descend into f1() and will skip full check of f2() and f3(), since they are global. The order of processing global functions doesn't affect safety, since all global functions must be proven safe based on their arguments only. Such function by function verification can drastically improve speed of the verification and reduce complexity. Note that the stack limit of 512 still applies to the call chain regardless whether functions were static or global. The nested level of 8 also still applies. The same recursion prevention checks are in place as well. The type information and static/global kind is preserved after the verification hence in the above example global function f2() and f3() can be replaced later by equivalent functions with the same types that are loaded and verified later without affecting safety of this main() program. Such replacement (re-linking) of global functions is a subject of future patches. Signed-off-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann Acked-by: Song Liu Link: https://lore.kernel.org/bpf/20200110064124.1760511-3-ast@kernel.org

bpf: Constant map key tracking for prog array pokes

2019-11-25T01:04:11Z
Add tracking of constant keys into tail call maps. The signature of bpf_tail_call_proto is that arg1 is ctx, arg2 map pointer and arg3 is a index key. The direct call approach for tail calls can be enabled if the verifier asserted that for all branches leading to the tail call helper invocation, the map pointer and index key were both constant and the same. Tracking of map pointers we already do from prior work via c93552c443eb ("bpf: properly enforce index mask to prevent out-of-bounds speculation") and 09772d92cd5a ("bpf: avoid retpoline for lookup/update/ delete calls on maps"). Given the tail call map index key is not on stack but directly in the register, we can add similar tracking approach and later in fixup_bpf_calls() add a poke descriptor to the progs poke_tab with the relevant information for the JITing phase. We internally reuse insn->imm for the rewritten BPF_JMP | BPF_TAIL_CALL instruction in order to point into the prog's poke_tab, and keep insn->imm as 0 as indicator that current indirect tail call emission must be used. Note that publishing to the tracker must happen at the end of fixup_bpf_calls() since adding elements to the poke_tab reallocates its memory, so we need to wait until its in final state. Future work can generalize and add similar approach to optimize plain array map lookups. Difference there is that we need to look into the key value that sits on stack. For clarity in bpf_insn_aux_data, map_state has been renamed into map_ptr_state, so we get map_{ptr,key}_state as trackers. Signed-off-by: Daniel Borkmann Signed-off-by: Alexei Starovoitov Acked-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/e8db37f6b2ae60402fa40216c96738ee9b316c32.1574452833.git.daniel@iogearbox.net

bpf: Compare BTF types of functions arguments with actual types

2019-11-15T22:45:02Z
Make the verifier check that BTF types of function arguments match actual types passed into top-level BPF program and into BPF-to-BPF calls. If types match such BPF programs and sub-programs will have full support of BPF trampoline. If types mismatch the trampoline has to be conservative. It has to save/restore five program arguments and assume 64-bit scalars. Signed-off-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann Acked-by: Song Liu Acked-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20191114185720.1641606-17-ast@kernel.org

bpf: Implement accurate raw_tp context access via BTF

2019-10-17T14:44:35Z
libbpf analyzes bpf C program, searches in-kernel BTF for given type name and stores it into expected_attach_type. The kernel verifier expects this btf_id to point to something like: typedef void (*btf_trace_kfree_skb)(void *, struct sk_buff *skb, void *loc); which represents signature of raw_tracepoint "kfree_skb". Then btf_ctx_access() matches ctx+0 access in bpf program with 'skb' and 'ctx+8' access with 'loc' arguments of "kfree_skb" tracepoint. In first case it passes btf_id of 'struct sk_buff *' back to the verifier core and 'void *' in second case. Then the verifier tracks PTR_TO_BTF_ID as any other pointer type. Like PTR_TO_SOCKET points to 'struct bpf_sock', PTR_TO_TCP_SOCK points to 'struct bpf_tcp_sock', and so on. PTR_TO_BTF_ID points to in-kernel structs. If 1234 is btf_id of 'struct sk_buff' in vmlinux's BTF then PTR_TO_BTF_ID#1234 points to one of in kernel skbs. When PTR_TO_BTF_ID#1234 is dereferenced (like r2 = *(u64 *)r1 + 32) the btf_struct_access() checks which field of 'struct sk_buff' is at offset 32. Checks that size of access matches type definition of the field and continues to track the dereferenced type. If that field was a pointer to 'struct net_device' the r2's type will be PTR_TO_BTF_ID#456. Where 456 is btf_id of 'struct net_device' in vmlinux's BTF. Such verifier analysis prevents "cheating" in BPF C program. The program cannot cast arbitrary pointer to 'struct sk_buff *' and access it. C compiler would allow type cast, of course, but the verifier will notice type mismatch based on BPF assembly and in-kernel BTF. Signed-off-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann Acked-by: Andrii Nakryiko Acked-by: Martin KaFai Lau Link: https://lore.kernel.org/bpf/20191016032505.2089704-7-ast@kernel.org

bpf: Process in-kernel BTF

2019-10-17T14:44:35Z
If in-kernel BTF exists parse it and prepare 'struct btf *btf_vmlinux' for further use by the verifier. In-kernel BTF is trusted just like kallsyms and other build artifacts embedded into vmlinux. Yet run this BTF image through BTF verifier to make sure that it is valid and it wasn't mangled during the build. If in-kernel BTF is incorrect it means either gcc or pahole or kernel are buggy. In such case disallow loading BPF programs. Signed-off-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann Acked-by: Andrii Nakryiko Acked-by: Martin KaFai Lau Link: https://lore.kernel.org/bpf/20191016032505.2089704-4-ast@kernel.org

bpf: introduce verifier internal test flag

2019-08-27T22:30:11Z
Introduce BPF_F_TEST_STATE_FREQ flag to stress test parentage chain and state pruning. Signed-off-by: Alexei Starovoitov Acked-by: Song Liu Signed-off-by: Daniel Borkmann

Merge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next

2019-06-20T04:06:27Z
Alexei Starovoitov says: ==================== pull-request: bpf-next 2019-06-19 The following pull-request contains BPF updates for your *net-next* tree. The main changes are: 1) new SO_REUSEPORT_DETACH_BPF setsocktopt, from Martin. 2) BTF based map definition, from Andrii. 3) support bpf_map_lookup_elem for xskmap, from Jonathan. 4) bounded loops and scalar precision logic in the verifier, from Alexei. ==================== Signed-off-by: David S. Miller