[PATCH v5 0/9] tracing: Add fprobe events

From: Masami Hiramatsu (Google)
Date: Thu Apr 20 2023 - 07:26:40 EST


Hi,

Here is the 5th version of improve fprobe and add a basic fprobe event
support for ftrace (tracefs) and perf. Here is the previous version.

https://lore.kernel.org/all/168035963900.397811.6647648816464443553.stgit@xxxxxxxxxxxxxxxxxxxxxxxxxxxxx/

I updated the first 3 patches and added 6 additional patches to support
tracepoint probe (fprobe on tracepoint) and BTF support for function
entry and tracepoint. This allows us to trace function entries with
parameter names (if BTF is available).

With this fprobe events, we can continue to trace function entry/exit
even if the CONFIG_KPROBES_ON_FTRACE is not available. Since
CONFIG_KPROBES_ON_FTRACE requires the CONFIG_DYNAMIC_FTRACE_WITH_REGS,
it is not available if the architecture only supports
CONFIG_DYNAMIC_FTRACE_WITH_ARGS (e.g. arm64). And that means kprobe
events can not probe function entry/exit effectively on such architecture.
But this problem can be solved if the dynamic events supports fprobe events
because fprobe events doesn't use kprobe but ftrace via fprobe.

Fprobe events allows user to add new events on the entry and exit of kernel
functions (which can be ftraced). Unlike kprobe events, the fprobe events
can only probe the function entry and exit, the IP address will have some
offsets from the symbol address. And it can only trace the function args,
return value, and stacks. (no registers)
For probing function body, users can continue to use the kprobe events.

The tracepoint probe events (tprobe events) also allows user to add new
events dynamically on the tracepoint. Most of the tracepoint already has
trace-events, so this feature is useful if you only want to know a
specific parameter, or trace the tracepoints which has no trace-events
(e.g. sched_*_tp tracepoints only exposes the tracepoints.)

The fprobe events syntax is;

f[:[GRP/][EVENT]] FUNCTION [FETCHARGS]
f[MAXACTIVE][:[GRP/][EVENT]] FUNCTION%return [FETCHARGS]

And tracepoint probe events syntax is;

t[:[GRP/][EVENT]] TRACEPOINT [FETCHARGS]

This series also includes BTF argument support for fprobe/tracepoint events,
and kprobe events. This allows us to fetch a specific function parameter
by name, and all parameters by '$$args'.
Note that enabling this feature, you need to enable CONFIG_BPF_SYSCALL and
confirm that your arch supports CONFIG_HAVE_FUNCTION_ARG_ACCESS_API.

E.g.

# echo 't kfree ptr' >> dynamic_events
# echo 'f kfree object' >> dynamic_events
# cat dynamic_events
t:tracepoints/kfree kfree ptr=ptr
f:fprobes/kfree__entry kfree object=object
# echo 1 > events/fprobes/enable
# echo 1 > events/tracepoints/enable
# echo > trace
# head -n 20 trace | tail
# TASK-PID CPU# ||||| TIMESTAMP FUNCTION
# | | | ||||| | |
tail-84 [000] ..... 1324.561958: kfree__entry: (kfree+0x4/0x140) object=0xffff888006383c00
tail-84 [000] ...1. 1324.561961: kfree: (__probestub_kfree+0x4/0x10) ptr=0xffff888006383c00
tail-84 [000] ..... 1324.561988: kfree__entry: (kfree+0x4/0x140) object=0x0
tail-84 [000] ...1. 1324.561988: kfree: (__probestub_kfree+0x4/0x10) ptr=0x0
tail-84 [000] ..... 1324.561989: kfree__entry: (kfree+0x4/0x140) object=0xffff88800671e600
tail-84 [000] ...1. 1324.561989: kfree: (__probestub_kfree+0x4/0x10) ptr=0xffff88800671e600
tail-84 [000] ..... 1324.562368: kfree__entry: (kfree+0x4/0x140) object=0xffff8880065e0580
tail-84 [000] ...1. 1324.562369: kfree: (__probestub_kfree+0x4/0x10) ptr=0xffff8880065e0580


Thank you,

---

Masami Hiramatsu (Google) (9):
fprobe: Pass return address to the handlers
tracing/probes: Add fprobe events for tracing function entry and exit.
selftests/ftrace: Add fprobe related testcases
tracing/probes: Add tracepoint support on fprobe_event
tracing/probes: Move event parameter fetching code to common parser
tracing/probes: Support function parameters if BTF is available
tracing/probes: Add $$args meta argument for all function args
selftests/ftrace: Add tracepoint probe test case
selftests/ftrace: Add BTF arguments test cases


include/linux/fprobe.h | 11
include/linux/rethook.h | 2
include/linux/trace_events.h | 3
include/linux/tracepoint-defs.h | 1
include/linux/tracepoint.h | 5
kernel/kprobes.c | 1
kernel/trace/Kconfig | 25
kernel/trace/Makefile | 1
kernel/trace/bpf_trace.c | 6
kernel/trace/fprobe.c | 17
kernel/trace/rethook.c | 3
kernel/trace/trace.c | 13
kernel/trace/trace.h | 11
kernel/trace/trace_eprobe.c | 44 -
kernel/trace/trace_fprobe.c | 1214 ++++++++++++++++++++
kernel/trace/trace_kprobe.c | 33 -
kernel/trace/trace_probe.c | 448 ++++++-
kernel/trace/trace_probe.h | 42 +
kernel/trace/trace_uprobe.c | 8
lib/test_fprobe.c | 10
samples/fprobe/fprobe_example.c | 6
.../ftrace/test.d/dynevent/add_remove_btfarg.tc | 54 +
.../ftrace/test.d/dynevent/add_remove_fprobe.tc | 26
.../ftrace/test.d/dynevent/add_remove_tprobe.tc | 27
.../ftrace/test.d/dynevent/fprobe_syntax_errors.tc | 98 ++
.../ftrace/test.d/dynevent/tprobe_syntax_errors.tc | 82 +
.../ftrace/test.d/kprobe/kprobe_syntax_errors.tc | 12
27 files changed, 2064 insertions(+), 139 deletions(-)
create mode 100644 kernel/trace/trace_fprobe.c
create mode 100644 tools/testing/selftests/ftrace/test.d/dynevent/add_remove_btfarg.tc
create mode 100644 tools/testing/selftests/ftrace/test.d/dynevent/add_remove_fprobe.tc
create mode 100644 tools/testing/selftests/ftrace/test.d/dynevent/add_remove_tprobe.tc
create mode 100644 tools/testing/selftests/ftrace/test.d/dynevent/fprobe_syntax_errors.tc
create mode 100644 tools/testing/selftests/ftrace/test.d/dynevent/tprobe_syntax_errors.tc

--
Masami Hiramatsu (Google) <mhiramat@xxxxxxxxxx>