- 19 Nov, 2019 6 commits
-
-
Ilya Leoshkevich authored
If literal pool grows past 524287 mark, it's no longer possible to use long displacement to reference literal pool entries. In JIT setting maintaining multiple literal pool registers is next to impossible, since we operate on one instruction at a time. Therefore, fall back to loading literal pool entry using PC-relative addressing, and then using a register-register form of the following machine instruction. Signed-off-by:
Ilya Leoshkevich <iii@linux.ibm.com> Signed-off-by:
Alexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20191118180340.68373-6-iii@linux.ibm.com
-
Ilya Leoshkevich authored
lg and lgrl have the same performance characteristics, but the former requires a base register and is subject to long displacement range limits, while the latter does not. Therefore, lgrl is totally superior to lg and should be used instead whenever possible. Signed-off-by:
-
Ilya Leoshkevich authored
Currently literal pool register is loaded using basr, which makes it point not to the beginning of the literal pool, but rather to the next instruction. In case JITed code is larger than 512k, this renders literal pool register absolutely useless due to long displacement range restrictions. The solution is to use larl to make literal pool register point to the very beginning of the literal pool. This makes it always possible to address 512k worth of literal pool entries using long displacement. However, for short programs, in which the entire literal pool is covered by basr-generated base, it is still beneficial to use basr, since it is 4 bytes shorter than larl. Detect situations when basr-generated base does not cover the entire literal pool, and in such cases use larl instead. Signed-off-by:
-
Ilya Leoshkevich authored
When literal pool size exceeds 512k, it's no longer possible to reference all the entries in it using a single base register and long displacement. Therefore, PC-relative lgfrl and lgrl instructions need to be used. Unfortunately, they require their arguments to be aligned to 4- and 8-byte boundaries respectively. This generates certain overhead due to necessary padding bytes. Grouping 4- and 8-byte entries together reduces the maximum overhead to 6 bytes (2 for aligning 4-byte entries and 4 for aligning 8-byte entries). While in theory it is possible to detect whether or not alignment is needed by comparing the literal pool size with 512k, in practice this leads to having two ways of emitting constants, making the code more complicated. Prefer code simplicity over trivial size saving, and always group and align literal pool entries. Signed-off-by:
-
Ilya Leoshkevich authored
Currently maximum JITed code size is limited to 64k, because JIT can emit only relative short branches, whose range is limited by 64k in both directions. Teach JIT to use relative long branches. There are no compare+branch relative long instructions, so using relative long branches consumes more space due to having to having to emit an explicit comparison instruction. Therefore do this only when relative short branch is not enough. Signed-off-by:
-
Colin Ian King authored
The error return path on when bpf_fentry_test* tests fail does not kfree 'data'. Fix this by adding the missing kfree. Addresses-Coverity: ("Resource leak") Fixes: faeb2dce ("bpf: Add kernel test functions for fentry testing") Signed-off-by:Colin Ian King <colin.king@canonical.com> Signed-off-by:
-
- 18 Nov, 2019 9 commits
-
-
Yonghong Song authored
Currently, with latest llvm trunk, selftest test_progs failed obj file test_seg6_loop.o with the following error in verifier: infinite loop detected at insn 76 The byte code sequence looks like below, and noted that alu32 has been turned off by default for better generated codes in general: 48: w3 = 100 49: *(u32 *)(r10 - 68) = r3 ... ; if (tlv.type == SR6_TLV_PADDING) { 76: if w3 == 5 goto -18 <LBB0_19> ... 85: r1 = *(u32 *)(r10 - 68) ; for (int i = 0; i < 100; i++) { 86: w1 += -1 87: if w1 == 0 goto +5 <LBB0_20> 88: *(u32 *)(r10 - 68) = r1 The main reason for verification failure is due to partial spills at r10 - 68 for induction variable "i". Current verifier only handles spills with 8-byte values. The above 4-byte value spill to stack is treated to STACK_MISC and its content is not saved. For the above example: w3 = 100 R3_w=inv100 fp-64_w=inv1086626730498 *(u32 *)(r10 - 68) = r3 R3_w=inv100 fp-64_w=inv1086626730498 ... r1 = *(u32 *)(r10 - 68) R1_w=inv(id=0,umax_value=4294967295,var_off=(0x0; 0xffffffff)) fp-64=inv1086626730498 To resolve this issue, verifier needs to be extended to track sub-registers in spilling, or llvm needs to enhanced to prevent sub-register spilling in register allocation phase. The former will increase verifier complexity and the latter will need some llvm "hacking". Let us workaround this issue by declaring the induction variable as "long" type so spilling will happen at non sub-register level. We can revisit this later if sub-register spilling causes similar or other verification issues. Signed-off-by:Yonghong Song <yhs@fb.com> Signed-off-by:
Daniel Borkmann <daniel@iogearbox.net> Acked-by:
Andrii Nakryiko <andriin@fb.com> Link: https://lore.kernel.org/bpf/20191117214036.1309510-1-yhs@fb.com
-
Jiri Benc authored
When run_kselftests.sh is run, it hangs after test_tc_tunnel.sh. The reason is test_tc_tunnel.sh ensures the server ('nc -l') is run all the time, starting it again every time it is expected to terminate. The exception is the final client_connect: the server is not started anymore, which ensures no process is kept running after the test is finished. For a sit test, though, the script is terminated prematurely without the final client_connect and the 'nc' process keeps running. This in turn causes the run_one function in kselftest/runner.sh to hang forever, waiting for the runaway process to finish. Ensure a remaining server is terminated on cleanup. Fixes: f6ad6acc ("selftests/bpf: expand test_tc_tunnel with SIT encap") Signed-off-by:Jiri Benc <jbenc@redhat.com> Signed-off-by:
Willem de Bruijn <willemb@google.com> Link: https://lore.kernel.org/bpf/60919291657a9ee89c708d8aababc28ebe1420be.1573821780.git.jbenc@redhat.com
-
Jiri Benc authored
The actual test to run is test_xdping.sh, which is already in TEST_PROGS. The xdping program alone is not runnable with 'make run_tests', it immediatelly fails due to missing arguments. Move xdping to TEST_GEN_PROGS_EXTENDED in order to be built but not run. Fixes: cd538502 ("selftests/bpf: measure RTT from xdp using xdping") Signed-off-by:
Alan Maguire <alan.maguire@oracle.com> Acked-by:
Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/bpf/4365c81198f62521344c2215909634407184387e.1573821726.git.jbenc@redhat.com
-
Daniel Borkmann authored
Andrii Nakryiko says: ==================== This patch set adds ability to memory-map BPF array maps (single- and multi-element). The primary use case is memory-mapping BPF array maps, created to back global data variables, created by libbpf implicitly. This allows for much better usability, along with avoiding syscalls to read or update data completely. Due to memory-mapping requirements, BPF array map that is supposed to be memory-mapped, has to be created with special BPF_F_MMAPABLE attribute, which triggers slightly different memory allocation strategy internally. See patch 1 for details. Libbpf is extended to detect kernel support for this flag, and if supported, will specify it for all global data maps automatically. Patch #1 refactors bpf_map_inc() and converts bpf_map's refcnt to atomic64_t to make refcounting never fail. Patch #2 does similar refactoring for bpf_prog_add()/bpf_prog_inc(). v5->v6: - add back uref counting (Daniel); v4->v5: - change bpf_prog's refcnt to atomic64_t (Daniel); v3->v4: - add mmap's open() callback to fix refcounting (Johannes); - switch to remap_vmalloc_pages() instead of custom fault handler (Johannes); - converted bpf_map's refcnt/usercnt into atomic64_t; - provide default bpf_map_default_vmops handling open/close properly; v2->v3: - change allocation strategy to avoid extra pointer dereference (Jakub); v1->v2: - fix map lookup code generation for BPF_F_MMAPABLE case; - prevent BPF_F_MMAPABLE flag for all but plain array map type; - centralize ref-counting in generic bpf_map_mmap(); - don't use uref counting (Alexei); - use vfree() directly; - print flags with %x (Song); - extend tests to verify bpf_map_{lookup,update}_elem() logic as well. ==================== Signed-off-by: -
Andrii Nakryiko authored
Add selftests validating mmap()-ing BPF array maps: both single-element and multi-element ones. Check that plain bpf_map_update_elem() and bpf_map_lookup_elem() work correctly with memory-mapped array. Also convert CO-RE relocation tests to use memory-mapped views of global data. Signed-off-by:
Song Liu <songliubraving@fb.com> Link: https://lore.kernel.org/bpf/20191117172806.2195367-6-andriin@fb.com
-
Andrii Nakryiko authored
Add detection of BPF_F_MMAPABLE flag support for arrays and add it as an extra flag to internal global data maps, if supported by kernel. This allows users to memory-map global data and use it without BPF map operations, greatly simplifying user experience. Signed-off-by:
John Fastabend <john.fastabend@gmail.com> Link: https://lore.kernel.org/bpf/20191117172806.2195367-5-andriin@fb.com
-
Andrii Nakryiko authored
Add ability to memory-map contents of BPF array map. This is extremely useful for working with BPF global data from userspace programs. It allows to avoid typical bpf_map_{lookup,update}_elem operations, improving both performance and usability. There had to be special considerations for map freezing, to avoid having writable memory view into a frozen map. To solve this issue, map freezing and mmap-ing is happening under mutex now: - if map is already frozen, no writable mapping is allowed; - if map has writable memory mappings active (accounted in map->writecnt), map freezing will keep failing with -EBUSY; - once number of writable memory mappings drops to zero, map freezing can be performed again. Only non-per-CPU plain arrays are supported right now. Maps with spinlocks can't be memory mapped either. For BPF_F_MMAPABLE array, memory allocation has to be done through vmalloc() to be mmap()'able. We also need to make sure that array data memory is page-sized and page-aligned, so we over-allocate memory in such a way that struct bpf_array is at the end of a single page of memory with array->value being aligned with the start of the second page. On deallocation we need to accomodate this memory arrangement to free vmalloc()'ed memory correctly. One important consideration regarding how memory-mapping subsystem functions. Memory-mapping subsystem provides few optional callbacks, among them open() and close(). close() is called for each memory region that is unmapped, so that users can decrease their reference counters and free up resources, if necessary. open() is *almost* symmetrical: it's called for each memory region that is being mapped, **except** the very first one. So bpf_map_mmap does initial refcnt bump, while open() will do any extra ones after that. Thus number of close() calls is equal to number of open() calls plus one more. Signed-off-by:Johannes Weiner <hannes@cmpxchg.org> Link: https://lore.kernel.org/bpf/20191117172806.2195367-4-andriin@fb.com
-
Andrii Nakryiko authored
Similarly to bpf_map's refcnt/usercnt, convert bpf_prog's refcnt to atomic64 and remove artificial 32k limit. This allows to make bpf_prog's refcounting non-failing, simplifying logic of users of bpf_prog_add/bpf_prog_inc. Validated compilation by running allyesconfig kernel build. Suggested-by:
-
Andrii Nakryiko authored
92117d84 ("bpf: fix refcnt overflow") turned refcounting of bpf_map into potentially failing operation, when refcount reaches BPF_MAX_REFCNT limit (32k). Due to using 32-bit counter, it's possible in practice to overflow refcounter and make it wrap around to 0, causing erroneous map free, while there are still references to it, causing use-after-free problems. But having a failing refcounting operations are problematic in some cases. One example is mmap() interface. After establishing initial memory-mapping, user is allowed to arbitrarily map/remap/unmap parts of mapped memory, arbitrarily splitting it into multiple non-contiguous regions. All this happening without any control from the users of mmap subsystem. Rather mmap subsystem sends notifications to original creator of memory mapping through open/close callbacks, which are optionally specified during initial memory mapping creation. These callbacks are used to maintain accurate refcount for bpf_map (see next patch in this series). The problem is that open() callback is not supposed to fail, because memory-mapped resource is set up and properly referenced. This is posing a problem for using memory-mapping with BPF maps. One solution to this is to maintain separate refcount for just memory-mappings and do single bpf_map_inc/bpf_map_put when it goes from/to zero, respectively. There are similar use cases in current work on tcp-bpf, necessitating extra counter as well. This seems like a rather unfortunate and ugly solution that doesn't scale well to various new use cases. Another approach to solve this is to use non-failing refcount_t type, which uses 32-bit counter internally, but, once reaching overflow state at UINT_MAX, stays there. This utlimately causes memory leak, but prevents use after free. But given refcounting is not the most performance-critical operation with BPF maps (it's not used from running BPF program code), we can also just switch to 64-bit counter that can't overflow in practice, potentially disadvantaging 32-bit platforms a tiny bit. This simplifies semantics and allows above described scenarios to not worry about failing refcount increment operation. In terms of struct bpf_map size, we are still good and use the same amount of space: BEFORE (3 cache lines, 8 bytes of padding at the end): struct bpf_map { const struct bpf_map_ops * ops __attribute__((__aligned__(64))); /* 0 8 */ struct bpf_map * inner_map_meta; /* 8 8 */ void * security; /* 16 8 */ enum bpf_map_type map_type; /* 24 4 */ u32 key_size; /* 28 4 */ u32 value_size; /* 32 4 */ u32 max_entries; /* 36 4 */ u32 map_flags; /* 40 4 */ int spin_lock_off; /* 44 4 */ u32 id; /* 48 4 */ int numa_node; /* 52 4 */ u32 btf_key_type_id; /* 56 4 */ u32 btf_value_type_id; /* 60 4 */ /* --- cacheline 1 boundary (64 bytes) --- */ struct btf * btf; /* 64 8 */ struct bpf_map_memory memory; /* 72 16 */ bool unpriv_array; /* 88 1 */ bool frozen; /* 89 1 */ /* XXX 38 bytes hole, try to pack */ /* --- cacheline 2 boundary (128 bytes) --- */ atomic_t refcnt __attribute__((__aligned__(64))); /* 128 4 */ atomic_t usercnt; /* 132 4 */ struct work_struct work; /* 136 32 */ char name[16]; /* 168 16 */ /* size: 192, cachelines: 3, members: 21 */ /* sum members: 146, holes: 1, sum holes: 38 */ /* padding: 8 */ /* forced alignments: 2, forced holes: 1, sum forced holes: 38 */ } __attribute__((__aligned__(64))); AFTER (same 3 cache lines, no extra padding now): struct bpf_map { const struct bpf_map_ops * ops __attribute__((__aligned__(64))); /* 0 8 */ struct bpf_map * inner_map_meta; /* 8 8 */ void * security; /* 16 8 */ enum bpf_map_type map_type; /* 24 4 */ u32 key_size; /* 28 4 */ u32 value_size; /* 32 4 */ u32 max_entries; /* 36 4 */ u32 map_flags; /* 40 4 */ int spin_lock_off; /* 44 4 */ u32 id; /* 48 4 */ int numa_node; /* 52 4 */ u32 btf_key_type_id; /* 56 4 */ u32 btf_value_type_id; /* 60 4 */ /* --- cacheline 1 boundary (64 bytes) --- */ struct btf * btf; /* 64 8 */ struct bpf_map_memory memory; /* 72 16 */ bool unpriv_array; /* 88 1 */ bool frozen; /* 89 1 */ /* XXX 38 bytes hole, try to pack */ /* --- cacheline 2 boundary (128 bytes) --- */ atomic64_t refcnt __attribute__((__aligned__(64))); /* 128 8 */ atomic64_t usercnt; /* 136 8 */ struct work_struct work; /* 144 32 */ char name[16]; /* 176 16 */ /* size: 192, cachelines: 3, members: 21 */ /* sum members: 154, holes: 1, sum holes: 38 */ /* forced alignments: 2, forced holes: 1, sum forced holes: 38 */ } __attribute__((__aligned__(64))); This patch, while modifying all users of bpf_map_inc, also cleans up its interface to match bpf_map_put with separate operations for bpf_map_inc and bpf_map_inc_with_uref (to match bpf_map_put and bpf_map_put_with_uref, respectively). Also, given there are no users of bpf_map_inc_not_zero specifying uref=true, remove uref flag and default to uref=false internally. Signed-off-by:
-
- 15 Nov, 2019 25 commits
-
-
Daniel Borkmann authored
Alexei Starovoitov says: ==================== Introduce BPF trampoline that works as a bridge between kernel functions, BPF programs and other BPF programs. The first use case is fentry/fexit BPF programs that are roughly equivalent to kprobe/kretprobe. Unlike k[ret]probe there is practically zero overhead to call a set of BPF programs before or after kernel function. The second use case is heavily influenced by pain points in XDP development. BPF trampoline allows attaching similar fentry/fexit BPF program to any networking BPF program. It's now possible to see packets on input and output of any XDP, TC, lwt, cgroup programs without disturbing them. This greatly helps BPF-based network troubleshooting. The third use case of BPF trampoline will be explored in the follow up patches. The BPF trampoline will be used to dynamicly link BPF programs. It's more generic mechanism than array and link list of programs used in tracing, networking, cgroups. In many cases it can be used as a replacement for bpf_tail_call-based program chaining. See [1] for long term design discussion. v3 -> v4: - Included Peter's "86/alternatives: Teach text_poke_bp() to emulate instructions" as a first patch. If it changes between now and merge window, I'll rebease to newer version. The patch is necessary to do s/text_poke/text_poke_bp/ in patch 3 to fix the race. - In patch 4 fixed bpf_trampoline creation race spotted by Andrii. - Added patch 15 that annotates prog->kern bpf context types. It made patches 16 and 17 cleaner and more generic. - Addressed Andrii's feedback in other patches. v2 -> v3: - Addressed Song's and Andrii's comments - Fixed few minor bugs discovered while testing - Added one more libbpf patch v1 -> v2: - Addressed Andrii's comments - Added more test for fentry/fexit to kernel functions. Including stress test for maximum number of progs per trampoline. - Fixed a race btf_resolve_helper_id() - Added a patch to compare BTF types of functions arguments with actual types. - Added support for attaching BPF program to another BPF program via trampoline - Converted to use text_poke() API. That's the only viable mechanism to implement BPF-to-BPF attach. BPF-to-kernel attach can be refactored to use register_ftrace_direct() whenever it's available. [1] https://lore.kernel.org/bpf/20191112025112.bhzmrrh2pr76ssnh@ast-mbp.dhcp.thefacebook.com/ ==================== Signed-off-by:
-
Alexei Starovoitov authored
Add a test that attaches one FEXIT program to main sched_cls networking program and two other FEXIT programs to subprograms. All three tracing programs access return values and skb->len of networking program and subprograms. Signed-off-by:
-
Alexei Starovoitov authored
The test_pkt_access.o is used by multiple tests. Fix its section name so that program type can be automatically detected by libbpf and make it call other subprograms with skb argument. Signed-off-by:
-
Alexei Starovoitov authored
Extend libbpf api to pass attach_prog_fd into bpf_object__open. Signed-off-by:
-
Alexei Starovoitov authored
Allow FENTRY/FEXIT BPF programs to attach to other BPF programs of any type including their subprograms. This feature allows snooping on input and output packets in XDP, TC programs including their return values. In order to do that the verifier needs to track types not only of vmlinux, but types of other BPF programs as well. The verifier also needs to translate uapi/linux/bpf.h types used by networking programs into kernel internal BTF types used by FENTRY/FEXIT BPF programs. In some cases LLVM optimizations can remove arguments from BPF subprograms without adjusting BTF info that LLVM backend knows. When BTF info disagrees with actual types that the verifiers sees the BPF trampoline has to fallback to conservative and treat all arguments as u64. The FENTRY/FEXIT program can still attach to such subprograms, but it won't be able to recognize pointer types like 'struct sk_buff *' and it won't be able to pass them to bpf_skb_output() for dumping packets to user space. The FENTRY/FEXIT program would need to use bpf_probe_read_kernel() instead. The BPF_PROG_LOAD command is extended with attach_prog_fd field. When it's set to zero the attach_btf_id is one vmlinux BTF type ids. When attach_prog_fd points to previously loaded BPF program the attach_btf_id is BTF type id of main function or one of its subprograms. Signed-off-by:
-
Alexei Starovoitov authored
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 authored
Annotate BPF program context types with program-side type and kernel-side type. This type information is used by the verifier. btf_get_prog_ctx_type() is used in the later patches to verify that BTF type of ctx in BPF program matches to kernel expected ctx type. For example, the XDP program type is: BPF_PROG_TYPE(BPF_PROG_TYPE_XDP, xdp, struct xdp_md, struct xdp_buff) That means that XDP program should be written as: int xdp_prog(struct xdp_md *ctx) { ... } Signed-off-by: -
Alexei Starovoitov authored
btf_resolve_helper_id() caching logic is a bit racy, since under root the verifier can verify several programs in parallel. Fix it with READ/WRITE_ONCE. Fix the type as well, since error is also recorded. Fixes: a7658e1a ("bpf: Check types of arguments passed into helpers") Signed-off-by:
-
Alexei Starovoitov authored
BPF trampoline can be made to work with existing 5 bytes of BPF program prologue, but let's add 5 bytes of NOPs to the beginning of every JITed BPF program to make BPF trampoline job easier. They can be removed in the future. Signed-off-by:
-
Alexei Starovoitov authored
Add stress test for maximum number of attached BPF programs per BPF trampoline. Signed-off-by:
-
Alexei Starovoitov authored
Add a combined fentry/fexit test. Signed-off-by:
-
Alexei Starovoitov authored
Add fexit tests for BPF trampoline that checks kernel functions with up to 6 arguments of different sizes and their return values. Signed-off-by:
-
Alexei Starovoitov authored
Add sanity test for BPF trampoline that checks kernel functions with up to 6 arguments of different sizes. Signed-off-by:
-
Alexei Starovoitov authored
Add few kernel functions with various number of arguments, their types and sizes for BPF trampoline testing to cover different calling conventions. Signed-off-by:
-
Alexei Starovoitov authored
Add simple test for fentry and fexit programs around eth_type_trans. Signed-off-by:
-
Alexei Starovoitov authored
Teach libbpf to recognize tracing programs types and attach them to fentry/fexit. Signed-off-by:
-
Alexei Starovoitov authored
Introduce btf__find_by_name_kind() helper to search BTF by name and kind, since name alone can be ambiguous. Signed-off-by:
-
Alexei Starovoitov authored
Introduce BPF trampoline concept to allow kernel code to call into BPF programs with practically zero overhead. The trampoline generation logic is architecture dependent. It's converting native calling convention into BPF calling convention. BPF ISA is 64-bit (even on 32-bit architectures). The registers R1 to R5 are used to pass arguments into BPF functions. The main BPF program accepts only single argument "ctx" in R1. Whereas CPU native calling convention is different. x86-64 is passing first 6 arguments in registers and the rest on the stack. x86-32 is passing first 3 arguments in registers. sparc64 is passing first 6 in registers. And so on. The trampolines between BPF and kernel already exist. BPF_CALL_x macros in include/linux/filter.h statically compile trampolines from BPF into kernel helpers. They convert up to five u64 arguments into kernel C pointers and integers. On 64-bit architectures this BPF_to_kernel trampolines are nops. On 32-bit architecture they're meaningful. The opposite job kernel_to_BPF trampolines is done by CAST_TO_U64 macros and __bpf_trace_##call() shim functions in include/trace/bpf_probe.h. They convert kernel function arguments into array of u64s that BPF program consumes via R1=ctx pointer. This patch set is doing the same job as __bpf_trace_##call() static trampolines, but dynamically for any kernel function. There are ~22k global kernel functions that are attachable via nop at function entry. The function arguments and types are described in BTF. The job of btf_distill_func_proto() function is to extract useful information from BTF into "function model" that architecture dependent trampoline generators will use to generate assembly code to cast kernel function arguments into array of u64s. For example the kernel function eth_type_trans has two pointers. They will be casted to u64 and stored into stack of generated trampoline. The pointer to that stack space will be passed into BPF program in R1. On x86-64 such generated trampoline will consume 16 bytes of stack and two stores of %rdi and %rsi into stack. The verifier will make sure that only two u64 are accessed read-only by BPF program. The verifier will also recognize the precise type of the pointers being accessed and will not allow typecasting of the pointer to a different type within BPF program. The tracing use case in the datacenter demonstrated that certain key kernel functions have (like tcp_retransmit_skb) have 2 or more kprobes that are always active. Other functions have both kprobe and kretprobe. So it is essential to keep both kernel code and BPF programs executing at maximum speed. Hence generated BPF trampoline is re-generated every time new program is attached or detached to maintain maximum performance. To avoid the high cost of retpoline the attached BPF programs are called directly. __bpf_prog_enter/exit() are used to support per-program execution stats. In the future this logic will be optimized further by adding support for bpf_stats_enabled_key inside generated assembly code. Introduction of preemptible and sleepable BPF programs will completely remove the need to call to __bpf_prog_enter/exit(). Detach of a BPF program from the trampoline should not fail. To avoid memory allocation in detach path the half of the page is used as a reserve and flipped after each attach/detach. 2k bytes is enough to call 40+ BPF programs directly which is enough for BPF tracing use cases. This limit can be increased in the future. BPF_TRACE_FENTRY programs have access to raw kernel function arguments while BPF_TRACE_FEXIT programs have access to kernel return value as well. Often kprobe BPF program remembers function arguments in a map while kretprobe fetches arguments from a map and analyzes them together with return value. BPF_TRACE_FEXIT accelerates this typical use case. Recursion prevention for kprobe BPF programs is done via per-cpu bpf_prog_active counter. In practice that turned out to be a mistake. It caused programs to randomly skip execution. The tracing tools missed results they were looking for. Hence BPF trampoline doesn't provide builtin recursion prevention. It's a job of BPF program itself and will be addressed in the follow up patches. BPF trampoline is intended to be used beyond tracing and fentry/fexit use cases in the future. For example to remove retpoline cost from XDP programs. Signed-off-by:
-
Alexei Starovoitov authored
Add bpf_arch_text_poke() helper that is used by BPF trampoline logic to patch nops/calls in kernel text into calls into BPF trampoline and to patch calls/nops inside BPF programs too. Signed-off-by:
-
Alexei Starovoitov authored
Refactor x86 JITing of LDX, STX, CALL instructions into separate helper functions. No functional changes in LDX and STX helpers. There is a minor change in CALL helper. It will populate target address correctly on the first pass of JIT instead of second pass. That won't reduce total number of JIT passes though. Signed-off-by:
-
Peter Zijlstra authored
In preparation for static_call and variable size jump_label support, teach text_poke_bp() to emulate instructions, namely: JMP32, JMP8, CALL, NOP2, NOP_ATOMIC5, INT3 The current text_poke_bp() takes a @handler argument which is used as a jump target when the temporary INT3 is hit by a different CPU. When patching CALL instructions, this doesn't work because we'd miss the PUSH of the return address. Instead, teach poke_int3_handler() to emulate an instruction, typically the instruction we're patching in. This fits almost all text_poke_bp() users, except arch_unoptimize_kprobe() which restores random text, and for that site we have to build an explicit emulate instruction. Tested-by:
Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by:
Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by:
Masami Hiramatsu <mhiramat@kernel.org> Reviewed-by:
Daniel Bristot de Oliveira <bristot@redhat.com> Acked-by:
Ingo Molnar <mingo@kernel.org> (cherry picked from commit 8c7eebc10687af45ac8e40ad1bac0cf7893dba9f) Signed-off-by:
-
Mao Wenan authored
The example code for the x86_64 JIT uses the wrong arguments when calling function bar(). Signed-off-by:
Mao Wenan <maowenan@huawei.com> Signed-off-by:
-
Andre Guedes authored
Commit 743e568c (samples/bpf: Add a "force" flag to XDP samples) introduced the '-F' option but missed adding it to the usage() and the 'long_option' array. Fixes: 743e568c (samples/bpf: Add a "force" flag to XDP samples) Signed-off-by:
Andre Guedes <andre.guedes@intel.com> Signed-off-by:
-
Andre Guedes authored
The '-f' option is shown twice in the usage(). This patch removes the outdated version. Signed-off-by:
-
Ilya Leoshkevich authored
The upcoming s390 branch length extension patches rely on "passes do not increase code size" property in order to consistently choose between short and long branches. Currently this property does not hold between the first and the second passes for register save/restore sequences, as well as various code fragments that depend on SEEN_* flags. Generate the code during the first pass conservatively: assume register save/restore sequences have the maximum possible length, and that all SEEN_* flags are set. Also refuse to JIT if this happens anyway (e.g. due to a bug), as this might lead to verifier bypass once long branches are introduced. Signed-off-by:
-
