- 01 Jul, 2020 7 commits
-
-
Song Liu authored
The new test is similar to other bpf_iter tests. It dumps all /proc/<pid>/stack to a seq_file. Here is some example output: pid: 2873 num_entries: 3 [<0>] worker_thread+0xc6/0x380 [<0>] kthread+0x135/0x150 [<0>] ret_from_fork+0x22/0x30 pid: 2874 num_entries: 9 [<0>] __bpf_get_stack+0x15e/0x250 [<0>] bpf_prog_22a400774977bb30_dump_task_stack+0x4a/0xb3c [<0>] bpf_iter_run_prog+0x81/0x170 [<0>] __task_seq_show+0x58/0x80 [<0>] bpf_seq_read+0x1c3/0x3b0 [<0>] vfs_read+0x9e/0x170 [<0>] ksys_read+0xa7/0xe0 [<0>] do_syscall_64+0x4c/0xa0 [<0>] entry_SYSCALL_64_after_hwframe+0x44/0xa9 Note: bpf_iter test as-is doesn't print the contents of the seq_file. To see the example above, it is necessary to add printf() to do_dummy_read. Signed-off-by:
Song Liu <songliubraving@fb.com> Signed-off-by:
Alexei Starovoitov <ast@kernel.org> Acked-by:
Andrii Nakryiko <andriin@fb.com> Acked-by:
Yonghong Song <yhs@fb.com> Link: https://lore.kernel.org/bpf/20200630062846.664389-5-songliubraving@fb.com
-
Song Liu authored
This makes it easy to dump stack trace in text. Signed-off-by:
-
Song Liu authored
Introduce helper bpf_get_task_stack(), which dumps stack trace of given task. This is different to bpf_get_stack(), which gets stack track of current task. One potential use case of bpf_get_task_stack() is to call it from bpf_iter__task and dump all /proc/<pid>/stack to a seq_file. bpf_get_task_stack() uses stack_trace_save_tsk() instead of get_perf_callchain() for kernel stack. The benefit of this choice is that stack_trace_save_tsk() doesn't require changes in arch/. The downside of using stack_trace_save_tsk() is that stack_trace_save_tsk() dumps the stack trace to unsigned long array. For 32-bit systems, we need to translate it to u64 array. Signed-off-by:
-
Song Liu authored
Sanitize and expose get/put_callchain_entry(). This would be used by bpf stack map. Suggested-by:
Peter Zijlstra <peterz@infradead.org> Signed-off-by:
-
Alexei Starovoitov authored
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:
Paul E. McKenney <paulmck@kernel.org> Link: https://lore.kernel.org/bpf/20200630043343.53195-2-alexei.starovoitov@gmail.com
-
Andrii Nakryiko authored
Add simple selftest validating byte swap built-ins and compile-time macros. Signed-off-by:
Daniel Borkmann <daniel@iogearbox.net> Acked-by:
-
Andrii Nakryiko authored
Make bpf_endian.h compatible with vmlinux.h. It is a frequent request from users wanting to use bpf_endian.h in their BPF applications using CO-RE and vmlinux.h. To achieve that, re-implement byte swap macros and drop all the header includes. This way it can be used both with linux header includes, as well as with a vmlinux.h. Signed-off-by:
-
- 30 Jun, 2020 2 commits
-
-
Andrii Nakryiko authored
Similarly to bpftool Makefile, allow to specify custom location of vmlinux.h to be used during the build. This allows simpler testing setups with checked-in pre-generated vmlinux.h. Signed-off-by:
-
Andrii Nakryiko authored
In some build contexts (e.g., Travis CI build for outdated kernel), vmlinux.h, generated from available kernel, doesn't contain all the types necessary for BPF program compilation. For such set up, the most maintainable way to deal with this problem is to keep pre-generated (almost up-to-date) vmlinux.h checked in and use it for compilation purposes. bpftool after that can deal with kernel missing some of the features in runtime with no problems. To that effect, allow to specify path to custom vmlinux.h to bpftool's Makefile with VMLINUX_H variable. Signed-off-by:
-
- 28 Jun, 2020 3 commits
-
-
Alexei Starovoitov authored
Andrii Nakryiko says: ==================== Add ability to turn off default auto-loading of each BPF program by libbpf on BPF object load. This is the feature that allows BPF applications to have optional functionality, which is only excercised on kernel that support necessary features, while falling back to reduced/less performant functionality, if kernel is outdated. ==================== Acked-by:
Martin KaFai Lau <kafai@fb.com> Signed-off-by:
-
Andrii Nakryiko authored
Validate that BPF object with broken (in multiple ways) BPF program can still be successfully loaded, if that broken BPF program is disabled. Signed-off-by:
-
Andrii Nakryiko authored
Currently, bpf_object__load() (and by induction skeleton's load), will always attempt to prepare, relocate, and load into kernel every single BPF program found inside the BPF object file. This is often convenient and the right thing to do and what users expect. But there are plenty of cases (especially with BPF development constantly picking up the pace), where BPF application is intended to work with old kernels, with potentially reduced set of features. But on kernels supporting extra features, it would like to take a full advantage of them, by employing extra BPF program. This could be a choice of using fentry/fexit over kprobe/kretprobe, if kernel is recent enough and is built with BTF. Or BPF program might be providing optimized bpf_iter-based solution that user-space might want to use, whenever available. And so on. With libbpf and BPF CO-RE in particular, it's advantageous to not have to maintain two separate BPF object files to achieve this. So to enable such use cases, this patch adds ability to request not auto-loading chosen BPF programs. In such case, libbpf won't attempt to perform relocations (which might fail due to old kernel), won't try to resolve BTF types for BTF-aware (tp_btf/fentry/fexit/etc) program types, because BTF might not be present, and so on. Skeleton will also automatically skip auto-attachment step for such not loaded BPF programs. Overall, this feature allows to simplify development and deployment of real-world BPF applications with complicated compatibility requirements. Signed-off-by:
-
- 25 Jun, 2020 18 commits
-
-
Tobias Klauser authored
Define attach_type_name in common.c instead of main.h so it is only defined once. This leads to a slight decrease in the binary size of bpftool. Before: text data bss dec hex filename 399024 11168 1573160 1983352 1e4378 bpftool After: text data bss dec hex filename 398256 10880 1573160 1982296 1e3f58 bpftool Signed-off-by:
Tobias Klauser <tklauser@distanz.ch> Signed-off-by:
Quentin Monnet <quentin@isovalent.com> Link: https://lore.kernel.org/bpf/20200624143154.13145-1-tklauser@distanz.ch
-
Tobias Klauser authored
Define prog_type_name in prog.c instead of main.h so it is only defined once. This leads to a slight decrease in the binary size of bpftool. Before: text data bss dec hex filename 401032 11936 1573160 1986128 1e4e50 bpftool After: text data bss dec hex filename 399024 11168 1573160 1983352 1e4378 bpftool Signed-off-by:
-
Alexei Starovoitov authored
Yonghong Song says: ==================== bpf iterator implments traversal of kernel data structures and these data structures are passed to a bpf program for processing. This gives great flexibility for users to examine kernel data structure without using e.g. /proc/net which has limited and fixed format. Commit 138d0be3 ("net: bpf: Add netlink and ipv6_route bpf_iter targets") implemented bpf iterators for netlink and ipv6_route. This patch set intends to implement bpf iterators for tcp and udp. Currently, /proc/net/tcp is used to print tcp4 stats and /proc/net/tcp6 is used to print tcp6 stats. /proc/net/udp[6] have similar usage model. In contrast, only one tcp iterator is implemented and it is bpf program resposibility to filter based on socket family. The same is for udp. This will avoid another unnecessary traversal pass if users want to check both tcp4 and tcp6. Several helpers are also implemented in this patch bpf_skc_to_{tcp, tcp6, tcp_timewait, tcp_request, udp6}_sock The argument for these helpers is not a fixed btf_id. For example, bpf_skc_to_tcp(struct sock_common *), or bpf_skc_to_tcp(struct sock *), or bpf_skc_to_tcp(struct inet_sock *), ... are all valid. At runtime, the helper will check whether pointer cast is legal or not. Please see Patch #5 for details. Since btf_id's for both arguments and return value are known at build time, the btf_id's are pre-computed once vmlinux btf becomes valid. Jiri's "adding d_path helper" patch set https://lore.kernel.org/bpf/20200616100512.2168860-1-jolsa@kernel.org/T/ provides a way to pre-compute btf id during vmlinux build time. This can be applied here as well. A followup patch can convert to build time btf id computation after Jiri's patch landed. Changelogs: v4 -> v5: - fix bpf_skc_to_udp6_sock helper as besides sk_protocol, sk_family, sk_type == SOCK_DGRAM is also needed to differentiate from SOCK_RAW (Eric) v3 -> v4: - fix bpf_skc_to_{tcp_timewait, tcp_request}_sock helper implementation as just checking sk->sk_state is not enough (Martin) - fix a few kernel test robot reported failures - move bpf_tracing_net.h from libbpf to selftests (Andrii) - remove __weak attribute from selftests CONFIG_HZ variables (Andrii) v2 -> v3: - change sock_cast*/SOCK_CAST* names to btf_sock* names for generality (Martin) - change gpl_license to false (Martin) - fix helper to cast to tcp timewait/request socket. (Martin) v1 -> v2: - guard init_sock_cast_types() defination properly with CONFIG_NET (Martin) - reuse the btf_ids, computed for new helper argument, for return values (Martin) - using BTF_TYPE_EMIT to express intent of btf type generation (Andrii) - abstract out common net macros into bpf_tracing_net.h (Andrii) ==================== Signed-off-by:
-
Yonghong Song authored
Added tcp{4,6} and udp{4,6} bpf programs into test_progs selftest so that they at least can load successfully. $ ./test_progs -n 3 ... #3/7 tcp4:OK #3/8 tcp6:OK #3/9 udp4:OK #3/10 udp6:OK ... #3 bpf_iter:OK Summary: 1/16 PASSED, 0 SKIPPED, 0 FAILED Signed-off-by: -
Yonghong Song authored
On my VM, I got identical results between /proc/net/udp[6] and the udp{4,6} bpf iterator. For udp6: $ cat /sys/fs/bpf/p1 sl local_address remote_address st tx_queue rx_queue tr tm->when retrnsmt uid timeout inode ref pointer drops 1405: 000080FE00000000FF7CC4D0D9EFE4FE:0222 00000000000000000000000000000000:0000 07 00000000:00000000 00:00000000 00000000 193 0 19183 2 0000000029eab111 0 $ cat /proc/net/udp6 sl local_address remote_address st tx_queue rx_queue tr tm->when retrnsmt uid timeout inode ref pointer drops 1405: 000080FE00000000FF7CC4D0D9EFE4FE:0222 00000000000000000000000000000000:0000 07 00000000:00000000 00:00000000 00000000 193 0 19183 2 0000000029eab111 0 For udp4: $ cat /sys/fs/bpf/p4 sl local_address rem_address st tx_queue rx_queue tr tm->when retrnsmt uid timeout inode ref pointer drops 2007: 00000000:1F90 00000000:0000 07 00000000:00000000 00:00000000 00000000 0 0 72540 2 000000004ede477a 0 $ cat /proc/net/udp sl local_address rem_address st tx_queue rx_queue tr tm->when retrnsmt uid timeout inode ref pointer drops 2007: 00000000:1F90 00000000:0000 07 00000000:00000000 00:00000000 00000000 0 0 72540 2 000000004ede477a 0 Signed-off-by: -
Yonghong Song authored
In my VM, I got identical result compared to /proc/net/{tcp,tcp6}. For tcp6: $ cat /proc/net/tcp6 sl local_address remote_address st tx_queue rx_queue tr tm->when retrnsmt uid timeout inode 0: 00000000000000000000000000000000:0016 00000000000000000000000000000000:0000 0A 00000000:00000000 00:00000001 00000000 0 0 17955 1 000000003eb3102e 100 0 0 10 0 $ cat /sys/fs/bpf/p1 sl local_address remote_address st tx_queue rx_queue tr tm->when retrnsmt uid timeout inode 0: 00000000000000000000000000000000:0016 00000000000000000000000000000000:0000 0A 00000000:00000000 00:00000000 00000000 0 0 17955 1 000000003eb3102e 100 0 0 10 0 For tcp: $ cat /proc/net/tcp sl local_address rem_address st tx_queue rx_queue tr tm->when retrnsmt uid timeout inode 0: 00000000:0016 00000000:0000 0A 00000000:00000000 00:00000000 00000000 0 0 2666 1 000000007152e43f 100 0 0 10 0 $ cat /sys/fs/bpf/p2 sl local_address remote_address st tx_queue rx_queue tr tm->when retrnsmt uid timeout inode 1: 00000000:0016 00000000:0000 0A 00000000:00000000 00:00000000 00000000 0 0 2666 1 000000007152e43f 100 0 0 10 0 Signed-off-by: -
Yonghong Song authored
These newly added macros will be used in subsequent bpf iterator tcp{4,6} and udp{4,6} programs. Signed-off-by: -
Yonghong Song authored
Refactor bpf_iter_ipv6_route.c and bpf_iter_netlink.c so net macros, originally from various include/linux header files, are moved to a new header file bpf_tracing_net.h. The goal is to improve reuse so networking tracing programs do not need to copy these macros every time they use them. Signed-off-by:
-
Yonghong Song authored
Commit b9f4c01f ("selftest/bpf: Make bpf_iter selftest compilable against old vmlinux.h") and Commit dda18a5c ("selftests/bpf: Convert bpf_iter_test_kern{3, 4}.c to define own bpf_iter_meta") redefined newly introduced types in bpf programs so the bpf program can still compile properly with old kernels although loading may fail. Since this patch set introduced new types and the same workaround is needed, so let us move the workaround to a separate header file so they do not clutter bpf programs. Signed-off-by:
-
Yonghong Song authored
The helper is used in tracing programs to cast a socket pointer to a udp6_sock pointer. The return value could be NULL if the casting is illegal. Signed-off-by:
-
Yonghong Song authored
The bpf iterator for udp is implemented. Both udp4 and udp6 sockets will be traversed. It is up to bpf program to filter for udp4 or udp6 only, or both families of sockets. Signed-off-by:
-
Yonghong Song authored
Similar to tcp_iter_state, a new field bpf_seq_afinfo is added to udp_iter_state to provide bpf udp iterator afinfo. This does not change /proc/net/{udp, udp6} behavior. But it enables bpf iterator to avoid get afinfo from PDE_DATA and iterate through all udp and udp6 sockets in one pass. Signed-off-by: -
Yonghong Song authored
Three more helpers are added to cast a sock_common pointer to an tcp_sock, tcp_timewait_sock or a tcp_request_sock for tracing programs. Signed-off-by:
-
Yonghong Song authored
The helper is used in tracing programs to cast a socket pointer to a tcp6_sock pointer. The return value could be NULL if the casting is illegal. A new helper return type RET_PTR_TO_BTF_ID_OR_NULL is added so the verifier is able to deduce proper return types for the helper. Different from the previous BTF_ID based helpers, the bpf_skc_to_tcp6_sock() argument can be several possible btf_ids. More specifically, all possible socket data structures with sock_common appearing in the first in the memory layout. This patch only added socket types related to tcp and udp. All possible argument btf_id and return value btf_id for helper bpf_skc_to_tcp6_sock() are pre-calculcated and cached. In the future, it is even possible to precompute these btf_id's at kernel build time. Signed-off-by:
-
Yonghong Song authored
/proc/net/tcp{4,6} uses jiffies for various computations. Let us add bpf_jiffies64() helper to tracing program so bpf_iter and other programs can use it. Signed-off-by: -
Yonghong Song authored
'X' tells kernel to print hex with upper case letters. /proc/net/tcp{4,6} seq_file show() used this, and supports it in bpf_seq_printf() helper too. Signed-off-by: -
Yonghong Song authored
The bpf iterator for tcp is implemented. Both tcp4 and tcp6 sockets will be traversed. It is up to bpf program to filter for tcp4 or tcp6 only, or both families of sockets. Signed-off-by:
-
Yonghong Song authored
A new field bpf_seq_afinfo is added to tcp_iter_state to provide bpf tcp iterator afinfo. There are two reasons on why we did this. First, the current way to get afinfo from PDE_DATA does not work for bpf iterator as its seq_file inode does not conform to /proc/net/{tcp,tcp6} inode structures. More specifically, anonymous bpf iterator will use an anonymous inode which is shared in the system and we cannot change inode private data structure at all. Second, bpf iterator for tcp/tcp6 wants to traverse all tcp and tcp6 sockets in one pass and bpf program can control whether they want to skip one sk_family or not. Having a different afinfo with family AF_UNSPEC make it easier to understand in the code. This patch does not change /proc/net/{tcp,tcp6} behavior as the bpf_seq_afinfo will be NULL for these two proc files. Signed-off-by:
-
- 24 Jun, 2020 7 commits
-
-
Dmitry Yakunin authored
This patch adds support of SO_KEEPALIVE flag and TCP related options to bpf_setsockopt() routine. This is helpful if we want to enable or tune TCP keepalive for applications which don't do it in the userspace code. v3: - update kernel-doc in uapi (Nikita Vetoshkin <nekto0n@yandex-team.ru>) v4: - update kernel-doc in tools too (Alexei Starovoitov) - add test to selftests (Alexei Starovoitov) Signed-off-by:
Dmitry Yakunin <zeil@yandex-team.ru> Signed-off-by:
-
Dmitry Yakunin authored
This is preparation for usage in bpf_setsockopt. v2: - remove redundant EXPORT_SYMBOL (Alexei Starovoitov) Signed-off-by:
-
Dmitry Yakunin authored
This is preparation for usage in bpf_setsockopt. Signed-off-by:
-
Alexei Starovoitov authored
./test_progs-no_alu32 -t get_stack_raw_tp fails due to: 52: (85) call bpf_get_stack#67 53: (bf) r8 = r0 54: (bf) r1 = r8 55: (67) r1 <<= 32 56: (c7) r1 s>>= 32 ; if (usize < 0) 57: (c5) if r1 s< 0x0 goto pc+26 R0=inv(id=0,smax_value=800) R1_w=inv(id=0,umax_value=800,var_off=(0x0; 0x3ff)) R6=ctx(id=0,off=0,imm=0) R7=map_value(id=0,off=0,ks=4,vs=1600,imm=0) R8_w=inv(id=0,smax_value=800) R9=inv800 ; ksize = bpf_get_stack(ctx, raw_data + usize, max_len - usize, 0); 58: (1f) r9 -= r8 ; ksize = bpf_get_stack(ctx, raw_data + usize, max_len - usize, 0); 59: (bf) r2 = r7 60: (0f) r2 += r1 regs=1 stack=0 before 52: (85) call bpf_get_stack#67 ; ksize = bpf_get_stack(ctx, raw_data + usize, max_len - usize, 0); 61: (bf) r1 = r6 62: (bf) r3 = r9 63: (b7) r4 = 0 64: (85) call bpf_get_stack#67 R0=inv(id=0,smax_value=800) R1_w=ctx(id=0,off=0,imm=0) R2_w=map_value(id=0,off=0,ks=4,vs=1600,umax_value=800,var_off=(0x0; 0x3ff),s32_max_value=1023,u32_max_value=1023) R3_w=inv(id=0,umax_value=9223372036854776608) R3 unbounded memory access, use 'var &= const' or 'if (var < const)' In the C code: usize = bpf_get_stack(ctx, raw_data, max_len, BPF_F_USER_STACK); if (usize < 0) return 0; ksize = bpf_get_stack(ctx, raw_data + usize, max_len - usize, 0); if (ksize < 0) return 0; We used to have problem with pointer arith in R2. Now it's a problem with two integers in R3. 'if (usize < 0)' is comparing R1 and makes it [0,800], but R8 stays [-inf,800]. Both registers represent the same 'usize' variable. Then R9 -= R8 is doing 800 - [-inf, 800] so the result of "max_len - usize" looks unbounded to the verifier while it's obvious in C code that "max_len - usize" should be [0, 800]. To workaround the problem convert ksize and usize variables from int to long. Signed-off-by: -
Andrii Nakryiko authored
Prevent loading/parsing vmlinux BTF twice in some cases: for CO-RE relocations and for BTF-aware hooks (tp_btf, fentry/fexit, etc). Fixes: a6ed02ca ("libbpf: Load btf_vmlinux only once per object.") Signed-off-by:
-
Colin Ian King authored
There is a spelling mistake in a pr_warn message. Fix it. Signed-off-by:
Colin Ian King <colin.king@canonical.com> Signed-off-by:
-
Quentin Monnet authored
Building bpftool yields the following complaint: pids.c: In function 'emit_obj_refs_json': pids.c:175:80: warning: declaration of 'json_wtr' shadows a global declaration [-Wshadow] 175 | void emit_obj_refs_json(struct obj_refs_table *table, __u32 id, json_writer_t *json_wtr) | ~~~~~~~~~~~~~~~^~~~~~~~ In file included from pids.c:11: main.h:141:23: note: shadowed declaration is here 141 | extern json_writer_t *json_wtr; | ^~~~~~~~ Let's rename the variable. v2: - Rename the variable instead of calling the global json_wtr directly. Signed-off-by:
-
- 23 Jun, 2020 3 commits
-
-
Alexei Starovoitov authored
-
Tobias Klauser authored
Currently, if the clang-bpf-co-re feature is not available, the build fails with e.g. CC prog.o prog.c:1462:10: fatal error: profiler.skel.h: No such file or directory 1462 | #include "profiler.skel.h" | ^~~~~~~~~~~~~~~~~ This is due to the fact that the BPFTOOL_WITHOUT_SKELETONS macro is not defined, despite BUILD_BPF_SKELS not being set. Fix this by correctly evaluating $(BUILD_BPF_SKELS) when deciding on whether to add -DBPFTOOL_WITHOUT_SKELETONS to CFLAGS. Fixes: 05aca6da ("tools/bpftool: Generalize BPF skeleton support and generate vmlinux.h") Signed-off-by: -
John Fastabend authored
Extend original variable-length tests with a case to catch a common existing pattern of testing for < 0 for errors. Note because verifier also tracks upper bounds and we know it can not be greater than MAX_LEN here we can skip upper bound check. In ALU64 enabled compilation converting from long->int return types in probe helpers results in extra instruction pattern, <<= 32, s >>= 32. The trade-off is the non-ALU64 case works. If you really care about every extra insn (XDP case?) then you probably should be using original int type. In addition adding a sext insn to bpf might help the verifier in the general case to avoid these types of tricks. Signed-off-by:
John Fastabend <john.fastabend@gmail.com> Signed-off-by:
-
