XXX: only update rev but not hash #1
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
XXX should fail
danobi
pushed a commit
that referenced
this pull request
Jul 18, 2024
If it is dummy, then the visit() operation should not be executed. For
example, on aarch64, there is no syscalls:sys_{enter,exit}_open
tracepoint, however, opensnoop.bt tool will try to attch them and visit
args's field. We expect to get a warning, but instead we get a segmentation
fault. As follows:
$ sudo gdb bpftrace
...
(gdb) set args opensnoop.bt
(gdb) r
opensnoop.bt:22-24: WARNING: tracepoint not found: syscalls:sys_enter_open
opensnoop.bt:28-30: WARNING: tracepoint not found: syscalls:sys_exit_open
Thread 1 "bpftrace" received signal SIGSEGV, Segmentation fault.
0x000000000052bbd8 in __gnu_cxx::__normal_iterator<bpftrace::Field const*, std::vector<bpftrace::Field, std::allocator<bpftrace::Field> > >::__normal_iterator (this=0xffffffffb0b8, __i=<error reading variable: Cannot access memory at address 0x10>)
at /usr/include/c++/14/bits/stl_iterator.h:1068
1068 : _M_current(__i) { }
(gdb) bt
#0 0x000000000052bbd8 in __gnu_cxx::__normal_iterator<bpftrace::Field const*, std::vector<bpftrace::Field, std::allocator<bpftrace::Field> > >::__normal_iterator (this=0xffffffffb0b8,
__i=<error reading variable: Cannot access memory at address 0x10>) at /usr/include/c++/14/bits/stl_iterator.h:1068
#1 0x0000000000525bdc in std::vector<bpftrace::Field, std::allocator<bpftrace::Field> >::begin (this=0x10)
at /usr/include/c++/14/bits/stl_vector.h:884
bpftrace#2 0x000000000059ce8c in bpftrace::Struct::GetField (this=0x0, name="filename")
at /home/rongtao/Git/bpftrace/bpftrace/src/struct.cpp:131
bpftrace#3 0x00000000005a70a0 in bpftrace::SizedType::GetField (this=0xe9b998, name="filename")
at /home/rongtao/Git/bpftrace/bpftrace/src/types.cpp:538
bpftrace#4 0x00000000007c7178 in bpftrace::ast::CodegenLLVM::visit (this=0xffffffffc650, acc=...)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/passes/codegen_llvm.cpp:1973
bpftrace#5 0x0000000000847db0 in bpftrace::ast::FieldAccess::accept (this=0xe9be30, v=...)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/ast.cpp:31
bpftrace#6 0x00000000007d2404 in bpftrace::ast::CodegenLLVM::accept (this=0xffffffffc650, node=0xe9be30)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/passes/codegen_llvm.cpp:3778
bpftrace#7 0x00000000007c8bec in bpftrace::ast::CodegenLLVM::visit (this=0xffffffffc650, assignment=...)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/passes/codegen_llvm.cpp:2218
bpftrace#8 0x0000000000847eb4 in bpftrace::ast::AssignMapStatement::accept (this=0xffffe4429250, v=...)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/ast.cpp:36
bpftrace#9 0x00000000007d2404 in bpftrace::ast::CodegenLLVM::accept (this=0xffffffffc650, node=0xffffe4429250)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/passes/codegen_llvm.cpp:3778
bpftrace#10 0x00000000007ca4c4 in bpftrace::ast::CodegenLLVM::generateProbe (this=0xffffffffc650, probe=..., full_func_id="dummy",
name="dummy", func_type=0xff2b50, usdt_location_index=std::optional [no contained value], dummy=true)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/passes/codegen_llvm.cpp:2539
bpftrace#11 0x00000000007cb4e4 in bpftrace::ast::CodegenLLVM::visit (this=0xffffffffc650, probe=...)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/passes/codegen_llvm.cpp:2734
bpftrace#12 0x00000000008480f0 in bpftrace::ast::Probe::accept (this=0xffffe4427650, v=...)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/ast.cpp:47
bpftrace#13 0x00000000007d2404 in bpftrace::ast::CodegenLLVM::accept (this=0xffffffffc650, node=0xffffe4427650)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/passes/codegen_llvm.cpp:3778
bpftrace#14 0x00000000007cb7a4 in bpftrace::ast::CodegenLLVM::visit (this=0xffffffffc650, program=...)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/passes/codegen_llvm.cpp:2752
bpftrace#15 0x000000000084818c in bpftrace::ast::Program::accept (this=0xe90f50, v=...)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/ast.cpp:50
bpftrace#16 0x00000000007d2404 in bpftrace::ast::CodegenLLVM::accept (this=0xffffffffc650, node=0xe90f50)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/passes/codegen_llvm.cpp:3778
bpftrace#17 0x00000000007d03fc in bpftrace::ast::CodegenLLVM::generate_ir (this=0xffffffffc650)
at /home/rongtao/Git/bpftrace/bpftrace/src/ast/passes/codegen_llvm.cpp:3450
bpftrace#18 0x0000000000478138 in main (argc=2, argv=0xfffffffff4c8) at /home/rongtao/Git/bpftrace/bpftrace/src/main.cpp:898
We can simplify opensnoop.bt to:
tracepoint:syscalls:sys_enter_open_not_exist,
tracepoint:syscalls:sys_enter_openat
{
@ = args.filename;
}
This will produce the following error:
stdin:1:1-45: WARNING: tracepoint not found: syscalls:sys_enter_open_not_exist
tracepoint:syscalls:sys_enter_open_not_exist,tracepoint:syscalls:sys_enter_openat {@ = args.filename;}
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Segmentation fault
We need to generate the dummy probe only for cases when none of the probe
attach points exists.
Link: bpftrace#3274
Signed-off-by: Viktor Malik <viktor.malik@gmail.com>
Signed-off-by: Rong Tao <rongtao@cestc.cn>
danobi
added a commit
that referenced
this pull request
Jan 30, 2025
Seems like 1 clock cycle is too much for the kernel to handle - we get softlockups: ``` [ RUN ] probe.probe_builtin_scratch_buf [ 788.058977] watchdog: BUG: soft lockup - CPU#0 stuck for 26s! [swapper/0:0] [ 788.061316] Modules linked in: xfs nf_tables nfnetlink kvm [ 788.063176] CPU: 0 UID: 0 PID: 0 Comm: swapper/0 Not tainted 6.12.0 #1 [ 788.065341] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014 [ 788.069123] RIP: 0010:handle_softirqs+0x7a/0x2c0 [ 788.070694] Code: 00 01 00 00 bd 0a 00 00 00 44 89 64 24 14 45 89 f7 89 6c 24 10 40 88 7c 24 04 31 c0 65 66 89 05 74 a5 f9 46 fb bb ff ff ff ff <48> c7 c0 c0 60 40 ba 41 0f bc df 83 c3 01 49 89 c2 74 75 8d 43 ff [ 788.076737] RSP: 0018:ffff93e700003f98 EFLAGS: 00000246 [ 788.078449] RAX: 0000000000000000 RBX: 00000000ffffffff RCX: 00000000000006e0 [ 788.080828] RDX: 00000000000001b3 RSI: ffffffffba40a940 RDI: 0000000000000000 [ 788.083187] RBP: 000000000000000a R08: 000000b1cd070240 R09: 7fffffffffffffff [ 788.085542] R10: 000000b1ccf7c000 R11: 0000000000005936 R12: 0000000004200002 [ 788.087975] R13: 0000000000000000 R14: 0000000000000002 R15: 0000000000000002 [ 788.090347] FS: 0000000000000000(0000) GS:ffff9183bbc00000(0000) knlGS:0000000000000000 [ 788.093023] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 788.094954] CR2: 00007fa0843f7b8c CR3: 000000010029e003 CR4: 0000000000370ef0 ``` My uneducated guess is the high frequency is triggering some kind of race condition that's exacerbated by slow nested-virtualization VMs. I reported this to Song but he could not reproduce the issue. So let's just bump to 100,000 clock cycles to avoid bringing down CI.
danobi
added a commit
that referenced
this pull request
Feb 1, 2025
Seems like 1 clock cycle is too much for the kernel to handle - we get softlockups: ``` [ RUN ] probe.probe_builtin_scratch_buf [ 788.058977] watchdog: BUG: soft lockup - CPU#0 stuck for 26s! [swapper/0:0] [ 788.061316] Modules linked in: xfs nf_tables nfnetlink kvm [ 788.063176] CPU: 0 UID: 0 PID: 0 Comm: swapper/0 Not tainted 6.12.0 #1 [ 788.065341] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014 [ 788.069123] RIP: 0010:handle_softirqs+0x7a/0x2c0 [ 788.070694] Code: 00 01 00 00 bd 0a 00 00 00 44 89 64 24 14 45 89 f7 89 6c 24 10 40 88 7c 24 04 31 c0 65 66 89 05 74 a5 f9 46 fb bb ff ff ff ff <48> c7 c0 c0 60 40 ba 41 0f bc df 83 c3 01 49 89 c2 74 75 8d 43 ff [ 788.076737] RSP: 0018:ffff93e700003f98 EFLAGS: 00000246 [ 788.078449] RAX: 0000000000000000 RBX: 00000000ffffffff RCX: 00000000000006e0 [ 788.080828] RDX: 00000000000001b3 RSI: ffffffffba40a940 RDI: 0000000000000000 [ 788.083187] RBP: 000000000000000a R08: 000000b1cd070240 R09: 7fffffffffffffff [ 788.085542] R10: 000000b1ccf7c000 R11: 0000000000005936 R12: 0000000004200002 [ 788.087975] R13: 0000000000000000 R14: 0000000000000002 R15: 0000000000000002 [ 788.090347] FS: 0000000000000000(0000) GS:ffff9183bbc00000(0000) knlGS:0000000000000000 [ 788.093023] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 788.094954] CR2: 00007fa0843f7b8c CR3: 000000010029e003 CR4: 0000000000370ef0 ``` My uneducated guess is the high frequency is triggering some kind of race condition that's exacerbated by slow nested-virtualization VMs. I reported this to Song but he could not reproduce the issue. So let's just bump to 100,000 clock cycles to avoid bringing down CI.
danobi
pushed a commit
that referenced
this pull request
May 16, 2025
Standardized benchmarks are hard! Since we most often want to use benchmarks in an ad-hoc fashion, can build it directly into the main binary as a `--test` option (a new test mode). The benchmark mechanism is implementation totally transparently to all the internals, and operates exclusively on the `PassManager`. It executes a benchmark for each pass. This shows a structured output with every pass listed: ``` System OS: Linux 6.9.0-0_fbk3_1265_g43ac291a024d #1 SMP Wed Dec 4 07:06:17 PST 2024 Arch: x86_64 Build version: v0.21.0-713-g4969-dirty LLVM: 18.1.8 bfd: yes libdw (DWARF support): yes libsystemd (systemd notify support): no blazesym (advanced symbolization): yes Kernel helpers probe_read: no probe_read_str: no probe_read_user: no probe_read_user_str: no probe_read_kernel: no probe_read_kernel_str: no get_current_cgroup_id: no send_signal: no override_return: no get_boot_ns: no dpath: no skboutput: no get_tai_ns: no get_func_ip: no jiffies64: no for_each_map_elem: no get_ns_current_pid_tgid: no lookup_percpu_elem: no Kernel features Instruction limit: -1 btf: yes module btf: no map batch: yes uprobe refcount: yes Map types hash: yes array: yes percpu array: yes stack_trace: no perf_event_array: yes ringbuf: yes Probe types kprobe: no tracepoint: no perf_event: no fentry: no kprobe_multi: no uprobe_multi: no kprobe_session: no iter: no ast 10000 7205047 720 ± 582 ns bpftrace 10000 7724379 772 ± 576 ns parse 2103 100040389 47 ± 5 μs ConfigAnalyser 3355 100029020 29 ± 3 μs ResolveImports 1478 100015641 67 ± 9 μs ImportScripts 10000 15044124 1504 ± 878 ns UnstableFeature 10000 18684501 1868 ± 972 ns MacroExpansion 10000 19686928 1968 ± 910 ns Deprecated 10000 14804521 1480 ± 921 ns attachpoints 10000 52022211 5202 ± 1792 ns btf 10000 31688754 3168 ± 178502 ns tracepoint 10000 23252392 2325 ± 871 ns FieldAnalyser 10000 30305116 3030 ± 1106 ns ClangParser 4 105528964 26 ± 3 ms CMacroExpansion 10000 17427514 1742 ± 815 ns MapSugar 10000 36848365 3684 ± 1439 ns FoldLiterals 10000 19861081 1986 ± 833 ns PidFilter 10000 16219805 1621 ± 911 ns Semantic 10000 37571204 3757 ± 2062 ns ResourceAnalyser 10000 56294908 5629 ± 1611 ns RecursionCheck 10000 21742022 2174 ± 1016 ns ReturnPath 10000 10208386 1020 ± 581 ns Probe 10000 24246794 2424 ± 1238 ns llvm-init 10000 58527480 5852 ± 2056 ns compile 1974 100048050 50 ± 30 μs optimize 567 100030093 176 ± 58 μs object 115 100232684 871 ± 167 μs extern 10000 14664920 1466 ± 806 ns link 4511 100005119 22 ± 7 μs total 29 803344776 27 ± 3 ms PASS ``` Additional docs have been added for how to use this functionality, with appropriate caveats. Signed-off-by: Adin Scannell <amscanne@meta.com> stack-info: PR: bpftrace#3998, branch: user/amscanne/map_pipelines2/3
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
XXX should fail
Checklist
man/adoc/bpftrace.adocand if needed indocs/reference_guide.mdCHANGELOG.md