The code was restructured where the function graph notrace code, that
would not trace a function and all its children is done by setting a
NOTRACE flag when the function that is not to be traced is hit.
There's a TRACE_GRAPH_NOTRACE_BIT which defines the bit in the flags and a
TRACE_GRAPH_NOTRACE which is the mask with that bit set. But the
restructuring used TRACE_GRAPH_NOTRACE_BIT when it should have used
TRACE_GRAPH_NOTRACE.
For example:
# cd /sys/kernel/tracing
# echo set_track_prepare stack_trace_save > set_graph_notrace
# echo function_graph > current_tracer
# cat trace
[..]
0) | __slab_free() {
0) | free_to_partial_list() {
0) | arch_stack_walk() {
0) | __unwind_start() {
0) 0.501 us | get_stack_info();
Where a non filter trace looks like:
# echo > set_graph_notrace
# cat trace
0) | free_to_partial_list() {
0) | set_track_prepare() {
0) | stack_trace_save() {
0) | arch_stack_walk() {
0) | __unwind_start() {
Where the filter should look like:
# cat trace
0) | free_to_partial_list() {
0) | _raw_spin_lock_irqsave() {
0) 0.350 us | preempt_count_add();
0) 0.351 us | do_raw_spin_lock();
0) 2.440 us | }
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: https://lore.kernel.org/20250208001511.535be150@batman.local.home
Fixes: b84214890a ("function_graph: Move graph notrace bit to shadow stack global var")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Add the const qualifier to all the ctl_tables in the tree except for
watchdog_hardlockup_sysctl, memory_allocation_profiling_sysctls,
loadpin_sysctl_table and the ones calling register_net_sysctl (./net,
drivers/inifiniband dirs). These are special cases as they use a
registration function with a non-const qualified ctl_table argument or
modify the arrays before passing them on to the registration function.
Constifying ctl_table structs will prevent the modification of
proc_handler function pointers as the arrays would reside in .rodata.
This is made possible after commit 78eb4ea25c ("sysctl: treewide:
constify the ctl_table argument of proc_handlers") constified all the
proc_handlers.
Created this by running an spatch followed by a sed command:
Spatch:
virtual patch
@
depends on !(file in "net")
disable optional_qualifier
@
identifier table_name != {
watchdog_hardlockup_sysctl,
iwcm_ctl_table,
ucma_ctl_table,
memory_allocation_profiling_sysctls,
loadpin_sysctl_table
};
@@
+ const
struct ctl_table table_name [] = { ... };
sed:
sed --in-place \
-e "s/struct ctl_table .table = &uts_kern/const struct ctl_table *table = \&uts_kern/" \
kernel/utsname_sysctl.c
Reviewed-by: Song Liu <song@kernel.org>
Acked-by: Steven Rostedt (Google) <rostedt@goodmis.org> # for kernel/trace/
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com> # SCSI
Reviewed-by: Darrick J. Wong <djwong@kernel.org> # xfs
Acked-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Corey Minyard <cminyard@mvista.com>
Acked-by: Wei Liu <wei.liu@kernel.org>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Bill O'Donnell <bodonnel@redhat.com>
Acked-by: Baoquan He <bhe@redhat.com>
Acked-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Acked-by: Anna Schumaker <anna.schumaker@oracle.com>
Signed-off-by: Joel Granados <joel.granados@kernel.org>
Pull rv and tools/rtla updates from Steven Rostedt:
- Add a test suite to test the tool
Add a small test suite that can be used to test rtla's basic features
to at least have something to test when applying changes.
- Automate manual steps in monitor creation
While creating a new monitor in RV, besides generating code from
dot2k, there are a few manual steps which can be tedious and error
prone, like adding the tracepoints, makefile lines and kconfig, or
selecting events that start the monitor in the initial state.
Updates were made to try and automate as much as possible among those
steps to make creating a new RV monitor much quicker. It is still
requires to select proper tracepoints, this step is harder to
automate in a general way and, in several cases, would still need
user intervention.
- Have rtla timerlat hist and top set OSNOISE_WORKLOAD flag
Have both rtla-timerlat-hist and rtla-timerlat-top set
OSNOISE_WORKLOAD to the proper value ("on" when running with -k,
"off" when running with -u) every time the option is available
instead of setting it only when running with -u.
This prevents rtla timerlat -k from giving no results when
NO_OSNOISE_WORKLOAD is set, either manually or by an abnormally
exited earlier run of rtla timerlat -u.
- Stop rtla timerlat on signal properly when overloaded
There is an issue where if rtla is run on machines with a high number
of CPUs (100+), timerlat can generate more samples than rtla is able
to process via tracefs_iterate_raw_events. This is especially common
when the interval is set to 100us (rteval and cyclictest default) as
opposed to the rtla default of 1000us, but also happens with the rtla
default.
Currently, this leads to rtla hanging and having to be terminated
with SIGTERM. SIGINT setting stop_tracing is not enough, since more
and more events are coming and tracefs_iterate_raw_events never
exits.
To fix this: Stop the timerlat tracer on SIGINT/SIGALRM to ensure no
more events are generated when rtla is supposed to exit.
Also on receiving SIGINT/SIGALRM twice, abort iteration immediately
with tracefs_iterate_stop, making rtla exit right away instead of
waiting for all events to be processed.
- Account for missed events
Due to tracefs buffer overflow, it can happen that rtla misses
events, making the tracing results inaccurate.
Count both the number of missed events and the total number of
processed events, and display missed events as well as their
percentage. The numbers are displayed for both osnoise and timerlat,
even though for the earlier, missed events are generally not
expected.
For hist, the number is displayed at the end of the run; for top, it
is displayed on each printing of the top table.
- Changes to make osnoise more robust
There was a dependency in the code that the first field of the
osnoise_tool structure was the trace field. If that that ever
changed, then the code work break. Change the code to encapsulate
this dependency where the code that uses the structure does not have
this dependency.
* tag 'trace-tools-v6.14' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace: (22 commits)
rtla: Report missed event count
rtla: Add function to report missed events
rtla: Count all processed events
rtla: Count missed trace events
tools/rtla: Add osnoise_trace_is_off()
rtla/timerlat_top: Set OSNOISE_WORKLOAD for kernel threads
rtla/timerlat_hist: Set OSNOISE_WORKLOAD for kernel threads
rtla/osnoise: Distinguish missing workload option
rtla/timerlat_top: Abort event processing on second signal
rtla/timerlat_hist: Abort event processing on second signal
rtla/timerlat_top: Stop timerlat tracer on signal
rtla/timerlat_hist: Stop timerlat tracer on signal
rtla: Add trace_instance_stop
tools/rtla: Add basic test suite
verification/dot2k: Implement event type detection
verification/dot2k: Auto patch current kernel source
verification/dot2k: Simplify manual steps in monitor creation
rv: Simplify manual steps in monitor creation
verification/dot2k: Add support for name and description options
verification/dot2k: More robust template variables
...
Pull runtime verifier and osnoise fixes from Steven Rostedt:
- Reset idle tasks on reset for runtime verifier
When the runtime verifier is reset, it resets the task's data that is
being monitored. But it only iterates for_each_process() which does
not include the idle tasks. As the idle tasks can be monitored, they
need to be reset as well.
- Fix the enabling and disabling of tracepoints in osnoise
If timerlat is enabled and the WORKLOAD flag is not set, then the
osnoise tracer will enable the migrate task tracepoint to monitor it
for its own workload. The test to enable the tracepoint is done
against user space modifiable parameters. On disabling of the tracer,
those same parameters are used to determine if the tracepoint should
be disabled. The problem is if user space were to modify the
parameters after it enables the tracer then it may not disable the
tracepoint.
Instead, a static variable is used to keep track if the tracepoint
was enabled or not. Then when the tracer shuts down, it will use this
variable to decide to disable the tracepoint or not, instead of
looking at the user space parameters.
* tag 'trace-rv-v6.14' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
tracing/osnoise: Fix resetting of tracepoints
rv: Reset per-task monitors also for idle tasks
If a timerlat tracer is started with the osnoise option OSNOISE_WORKLOAD
disabled, but then that option is enabled and timerlat is removed, the
tracepoints that were enabled on timerlat registration do not get
disabled. If the option is disabled again and timelat is started, then it
triggers a warning in the tracepoint code due to registering the
tracepoint again without ever disabling it.
Do not use the same user space defined options to know to disable the
tracepoints when timerlat is removed. Instead, set a global flag when it
is enabled and use that flag to know to disable the events.
~# echo NO_OSNOISE_WORKLOAD > /sys/kernel/tracing/osnoise/options
~# echo timerlat > /sys/kernel/tracing/current_tracer
~# echo OSNOISE_WORKLOAD > /sys/kernel/tracing/osnoise/options
~# echo nop > /sys/kernel/tracing/current_tracer
~# echo NO_OSNOISE_WORKLOAD > /sys/kernel/tracing/osnoise/options
~# echo timerlat > /sys/kernel/tracing/current_tracer
Triggers:
------------[ cut here ]------------
WARNING: CPU: 6 PID: 1337 at kernel/tracepoint.c:294 tracepoint_add_func+0x3b6/0x3f0
Modules linked in:
CPU: 6 UID: 0 PID: 1337 Comm: rtla Not tainted 6.13.0-rc4-test-00018-ga867c441128e-dirty #73
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2 04/01/2014
RIP: 0010:tracepoint_add_func+0x3b6/0x3f0
Code: 48 8b 53 28 48 8b 73 20 4c 89 04 24 e8 23 59 11 00 4c 8b 04 24 e9 36 fe ff ff 0f 0b b8 ea ff ff ff 45 84 e4 0f 84 68 fe ff ff <0f> 0b e9 61 fe ff ff 48 8b 7b 18 48 85 ff 0f 84 4f ff ff ff 49 8b
RSP: 0018:ffffb9b003a87ca0 EFLAGS: 00010202
RAX: 00000000ffffffef RBX: ffffffff92f30860 RCX: 0000000000000000
RDX: 0000000000000000 RSI: ffff9bf59e91ccd0 RDI: ffffffff913b6410
RBP: 000000000000000a R08: 00000000000005c7 R09: 0000000000000002
R10: ffffb9b003a87ce0 R11: 0000000000000002 R12: 0000000000000001
R13: ffffb9b003a87ce0 R14: ffffffffffffffef R15: 0000000000000008
FS: 00007fce81209240(0000) GS:ffff9bf6fdd00000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 000055e99b728000 CR3: 00000001277c0002 CR4: 0000000000172ef0
Call Trace:
<TASK>
? __warn.cold+0xb7/0x14d
? tracepoint_add_func+0x3b6/0x3f0
? report_bug+0xea/0x170
? handle_bug+0x58/0x90
? exc_invalid_op+0x17/0x70
? asm_exc_invalid_op+0x1a/0x20
? __pfx_trace_sched_migrate_callback+0x10/0x10
? tracepoint_add_func+0x3b6/0x3f0
? __pfx_trace_sched_migrate_callback+0x10/0x10
? __pfx_trace_sched_migrate_callback+0x10/0x10
tracepoint_probe_register+0x78/0xb0
? __pfx_trace_sched_migrate_callback+0x10/0x10
osnoise_workload_start+0x2b5/0x370
timerlat_tracer_init+0x76/0x1b0
tracing_set_tracer+0x244/0x400
tracing_set_trace_write+0xa0/0xe0
vfs_write+0xfc/0x570
? do_sys_openat2+0x9c/0xe0
ksys_write+0x72/0xf0
do_syscall_64+0x79/0x1c0
entry_SYSCALL_64_after_hwframe+0x76/0x7e
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Tomas Glozar <tglozar@redhat.com>
Cc: Gabriele Monaco <gmonaco@redhat.com>
Cc: Luis Goncalves <lgoncalv@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
Link: https://lore.kernel.org/20250123204159.4450c88e@gandalf.local.home
Fixes: e88ed227f6 ("tracing/timerlat: Add user-space interface")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Pull trace fing buffer fix from Steven Rostedt:
"Fix atomic64 operations on some architectures for the tracing ring
buffer:
- Have emulating atomic64 use arch_spin_locks instead of
raw_spin_locks
The tracing ring buffer events have a small timestamp that holds
the delta between itself and the event before it. But this can be
tricky to update when interrupts come in. It originally just set
the deltas to zero for events that interrupted the adding of
another event which made all the events in the interrupt have the
same timestamp as the event it interrupted. This was not suitable
for many tools, so it was eventually fixed. But that fix required
adding an atomic64 cmpxchg on the timestamp in cases where an event
was added while another event was in the process of being added.
Originally, for 32 bit architectures, the manipulation of the 64
bit timestamp was done by a structure that held multiple 32bit
words to hold parts of the timestamp and a counter. But as updates
to the ring buffer were done, maintaining this became too complex
and was replaced by the atomic64 generic operations which are now
used by both 64bit and 32bit architectures. Shortly after that, it
was reported that riscv32 and other 32 bit architectures that just
used the generic atomic64 were locking up. This was because the
generic atomic64 operations defined in lib/atomic64.c uses a
raw_spin_lock() to emulate an atomic64 operation. The problem here
was that raw_spin_lock() can also be traced by the function tracer
(which is commonly used for debugging raw spin locks). Since the
function tracer uses the tracing ring buffer, which now is being
traced internally, this was triggering a recursion and setting off
a warning that the spin locks were recusing.
There's no reason for the code that emulates atomic64 operations to
be using raw_spin_locks which have a lot of debugging
infrastructure attached to them (depending on the config options).
Instead it should be using the arch_spin_lock() which does not have
any infrastructure attached to them and is used by low level
infrastructure like RCU locks, lockdep and of course tracing. Using
arch_spin_lock()s fixes this issue.
- Do not trace in NMI if the architecture uses emulated atomic64
operations
Another issue with using the emulated atomic64 operations that uses
spin locks to emulate the atomic64 operations is that they cannot
be used in NMI context. As an NMI can trigger while holding the
atomic64 spin locks it can try to take the same lock and cause a
deadlock.
Have the ring buffer fail recording events if in NMI context and
the architecture uses the emulated atomic64 operations"
* tag 'trace-ringbuffer-v6.14-2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
atomic64: Use arch_spin_locks instead of raw_spin_locks
ring-buffer: Do not allow events in NMI with generic atomic64 cmpxchg()
Pull fgraph updates from Steven Rostedt:
"Remove calltime and rettime from fgraph infrastructure
The calltime and rettime were used by the function graph tracer to
calculate the timings of functions where it traced their entry and
exit. The calltime and rettime were stored in the generic structures
that were used for the mechanisms to add an entry and exit callback.
Now that function graph infrastructure is used by other subsystems
than just the tracer, the calltime and rettime are not needed for
them. Remove the calltime and rettime from the generic fgraph
infrastructure and have the callers that require them handle them"
* tag 'ftrace-v6.14-2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
fgraph: Remove calltime and rettime from generic operations
Pull tracing updates from Steven Rostedt:
- Cleanup with guard() and free() helpers
There were several places in the code that had a lot of "goto out" in
the error paths to either unlock a lock or free some memory that was
allocated. But this is error prone. Convert the code over to use the
guard() and free() helpers that let the compiler unlock locks or free
memory when the function exits.
- Update the Rust tracepoint code to use the C code too
There was some duplication of the tracepoint code for Rust that did
the same logic as the C code. Add a helper that makes it possible for
both algorithms to use the same logic in one place.
- Add poll to trace event hist files
It is useful to know when an event is triggered, or even with some
filtering. Since hist files of events get updated when active and the
event is triggered, allow applications to poll the hist file and wake
up when an event is triggered. This will let the application know
that the event it is waiting for happened.
- Add :mod: command to enable events for current or future modules
The function tracer already has a way to enable functions to be
traced in modules by writing ":mod:<module>" into set_ftrace_filter.
That will enable either all the functions for the module if it is
loaded, or if it is not, it will cache that command, and when the
module is loaded that matches <module>, its functions will be
enabled. This also allows init functions to be traced. But currently
events do not have that feature.
Add the command where if ':mod:<module>' is written into set_event,
then either all the modules events are enabled if it is loaded, or
cache it so that the module's events are enabled when it is loaded.
This also works from the kernel command line, where
"trace_event=:mod:<module>", when the module is loaded at boot up,
its events will be enabled then.
* tag 'trace-v6.14-3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace: (26 commits)
tracing: Fix output of set_event for some cached module events
tracing: Fix allocation of printing set_event file content
tracing: Rename update_cache() to update_mod_cache()
tracing: Fix #if CONFIG_MODULES to #ifdef CONFIG_MODULES
selftests/ftrace: Add test that tests event :mod: commands
tracing: Cache ":mod:" events for modules not loaded yet
tracing: Add :mod: command to enabled module events
selftests/tracing: Add hist poll() support test
tracing/hist: Support POLLPRI event for poll on histogram
tracing/hist: Add poll(POLLIN) support on hist file
tracing: Fix using ret variable in tracing_set_tracer()
tracepoint: Reduce duplication of __DO_TRACE_CALL
tracing/string: Create and use __free(argv_free) in trace_dynevent.c
tracing: Switch trace_stat.c code over to use guard()
tracing: Switch trace_stack.c code over to use guard()
tracing: Switch trace_osnoise.c code over to use guard() and __free()
tracing: Switch trace_events_synth.c code over to use guard()
tracing: Switch trace_events_filter.c code over to use guard()
tracing: Switch trace_events_trigger.c code over to use guard()
tracing: Switch trace_events_hist.c code over to use guard()
...
Pull probes updates from Masami Hiramatsu:
- kprobes: Cleanups using guard() and __free(): Use cleanup.h macros to
cleanup code and remove all gotos from kprobes code.
- tracing/probes: Also cleanups tracing/*probe events code with guard()
and __free(). These patches are just to simplify the parser codes.
- kprobes: Reduce preempt disable scope in check_kprobe_access_safe()
This reduces preempt disable time to only when getting the module
refcount in check_kprobe_access_safe().
Previously it disabled preempt needlessly for other checks including
jump_label_text_reserved(), which took a long time because of the
linear search.
* tag 'probes-v6.14' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
tracing/kprobes: Simplify __trace_kprobe_create() by removing gotos
tracing: Use __free() for kprobe events to cleanup
tracing: Use __free() in trace_probe for cleanup
kprobes: Remove remaining gotos
kprobes: Remove unneeded goto
kprobes: Use guard for rcu_read_lock
kprobes: Use guard() for external locks
jump_label: Define guard() for jump_label_lock
tracing/eprobe: Adopt guard() and scoped_guard()
tracing/uprobe: Adopt guard() and scoped_guard()
tracing/kprobe: Adopt guard() and scoped_guard()
kprobes: Adopt guard() and scoped_guard()
kprobes: Reduce preempt disable scope in check_kprobe_access_safe()
Pull bpf updates from Alexei Starovoitov:
"A smaller than usual release cycle.
The main changes are:
- Prepare selftest to run with GCC-BPF backend (Ihor Solodrai)
In addition to LLVM-BPF runs the BPF CI now runs GCC-BPF in compile
only mode. Half of the tests are failing, since support for
btf_decl_tag is still WIP, but this is a great milestone.
- Convert various samples/bpf to selftests/bpf/test_progs format
(Alexis Lothoré and Bastien Curutchet)
- Teach verifier to recognize that array lookup with constant
in-range index will always succeed (Daniel Xu)
- Cleanup migrate disable scope in BPF maps (Hou Tao)
- Fix bpf_timer destroy path in PREEMPT_RT (Hou Tao)
- Always use bpf_mem_alloc in bpf_local_storage in PREEMPT_RT (Martin
KaFai Lau)
- Refactor verifier lock support (Kumar Kartikeya Dwivedi)
This is a prerequisite for upcoming resilient spin lock.
- Remove excessive 'may_goto +0' instructions in the verifier that
LLVM leaves when unrolls the loops (Yonghong Song)
- Remove unhelpful bpf_probe_write_user() warning message (Marco
Elver)
- Add fd_array_cnt attribute for prog_load command (Anton Protopopov)
This is a prerequisite for upcoming support for static_branch"
* tag 'bpf-next-6.14' of git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next: (125 commits)
selftests/bpf: Add some tests related to 'may_goto 0' insns
bpf: Remove 'may_goto 0' instruction in opt_remove_nops()
bpf: Allow 'may_goto 0' instruction in verifier
selftests/bpf: Add test case for the freeing of bpf_timer
bpf: Cancel the running bpf_timer through kworker for PREEMPT_RT
bpf: Free element after unlock in __htab_map_lookup_and_delete_elem()
bpf: Bail out early in __htab_map_lookup_and_delete_elem()
bpf: Free special fields after unlock in htab_lru_map_delete_node()
tools: Sync if_xdp.h uapi tooling header
libbpf: Work around kernel inconsistently stripping '.llvm.' suffix
bpf: selftests: verifier: Add nullness elision tests
bpf: verifier: Support eliding map lookup nullness
bpf: verifier: Refactor helper access type tracking
bpf: tcp: Mark bpf_load_hdr_opt() arg2 as read-write
bpf: verifier: Add missing newline on verbose() call
selftests/bpf: Add distilled BTF test about marking BTF_IS_EMBEDDED
libbpf: Fix incorrect traversal end type ID when marking BTF_IS_EMBEDDED
libbpf: Fix return zero when elf_begin failed
selftests/bpf: Fix btf leak on new btf alloc failure in btf_distill test
veristat: Load struct_ops programs only once
...
The function graph infrastructure is now generic so that kretprobes,
fprobes and BPF can use it. But there is still some leftover logic that
only the function graph tracer itself uses. This is the calculation of the
calltime and return time of the functions. The calculation of the calltime
has been moved into the function graph tracer and those users that need it
so that it doesn't cause overhead to the other users. But the return
function timestamp was still called.
Instead of just moving the taking of the timestamp into the function graph
trace remove the calltime and rettime completely from the ftrace_graph_ret
structure. Instead, move it into the function graph return entry event
structure and this also moves all the calltime and rettime logic out of
the generic fgraph.c code and into the tracing code that uses it.
This has been reported to decrease the overhead by ~27%.
Link: https://lore.kernel.org/all/Z3aSuql3fnXMVMoM@krava/
Link: https://lore.kernel.org/all/173665959558.1629214.16724136597211810729.stgit@devnote2/
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: https://lore.kernel.org/20250121194436.15bdf71a@gandalf.local.home
Reported-by: Jiri Olsa <olsajiri@gmail.com>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Pull ftrace updates from Steven Rostedt:
- Have fprobes built on top of function graph infrastructure
The fprobe logic is an optimized kprobe that uses ftrace to attach to
functions when a probe is needed at the start or end of the function.
The fprobe and kretprobe logic implements a similar method as the
function graph tracer to trace the end of the function. That is to
hijack the return address and jump to a trampoline to do the trace
when the function exits. To do this, a shadow stack needs to be
created to store the original return address. Fprobes and function
graph do this slightly differently. Fprobes (and kretprobes) has
slots per callsite that are reserved to save the return address. This
is fine when just a few points are traced. But users of fprobes, such
as BPF programs, are starting to add many more locations, and this
method does not scale.
The function graph tracer was created to trace all functions in the
kernel. In order to do this, when function graph tracing is started,
every task gets its own shadow stack to hold the return address that
is going to be traced. The function graph tracer has been updated to
allow multiple users to use its infrastructure. Now have fprobes be
one of those users. This will also allow for the fprobe and kretprobe
methods to trace the return address to become obsolete. With new
technologies like CFI that need to know about these methods of
hijacking the return address, going toward a solution that has only
one method of doing this will make the kernel less complex.
- Cleanup with guard() and free() helpers
There were several places in the code that had a lot of "goto out" in
the error paths to either unlock a lock or free some memory that was
allocated. But this is error prone. Convert the code over to use the
guard() and free() helpers that let the compiler unlock locks or free
memory when the function exits.
- Remove disabling of interrupts in the function graph tracer
When function graph tracer was first introduced, it could race with
interrupts and NMIs. To prevent that race, it would disable
interrupts and not trace NMIs. But the code has changed to allow NMIs
and also interrupts. This change was done a long time ago, but the
disabling of interrupts was never removed. Remove the disabling of
interrupts in the function graph tracer is it is not needed. This
greatly improves its performance.
- Allow the :mod: command to enable tracing module functions on the
kernel command line.
The function tracer already has a way to enable functions to be
traced in modules by writing ":mod:<module>" into set_ftrace_filter.
That will enable either all the functions for the module if it is
loaded, or if it is not, it will cache that command, and when the
module is loaded that matches <module>, its functions will be
enabled. This also allows init functions to be traced. But currently
events do not have that feature.
Because enabling function tracing can be done very early at boot up
(before scheduling is enabled), the commands that can be done when
function tracing is started is limited. Having the ":mod:" command to
trace module functions as they are loaded is very useful. Update the
kernel command line function filtering to allow it.
* tag 'ftrace-v6.14' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace: (26 commits)
ftrace: Implement :mod: cache filtering on kernel command line
tracing: Adopt __free() and guard() for trace_fprobe.c
bpf: Use ftrace_get_symaddr() for kprobe_multi probes
ftrace: Add ftrace_get_symaddr to convert fentry_ip to symaddr
Documentation: probes: Update fprobe on function-graph tracer
selftests/ftrace: Add a test case for repeating register/unregister fprobe
selftests: ftrace: Remove obsolate maxactive syntax check
tracing/fprobe: Remove nr_maxactive from fprobe
fprobe: Add fprobe_header encoding feature
fprobe: Rewrite fprobe on function-graph tracer
s390/tracing: Enable HAVE_FTRACE_GRAPH_FUNC
ftrace: Add CONFIG_HAVE_FTRACE_GRAPH_FUNC
bpf: Enable kprobe_multi feature if CONFIG_FPROBE is enabled
tracing/fprobe: Enable fprobe events with CONFIG_DYNAMIC_FTRACE_WITH_ARGS
tracing: Add ftrace_fill_perf_regs() for perf event
tracing: Add ftrace_partial_regs() for converting ftrace_regs to pt_regs
fprobe: Use ftrace_regs in fprobe exit handler
fprobe: Use ftrace_regs in fprobe entry handler
fgraph: Pass ftrace_regs to retfunc
fgraph: Replace fgraph_ret_regs with ftrace_regs
...
Pull trace ring-buffer updates from Steven Rostedt:
- Clean up the __rb_map_vma() logic
The logic of __rb_map_vma() has a error check with WARN_ON() that
makes sure that the index does not go past the end of the array of
buffers. The test in the loop pretty much guarantees that it will
never happen, but since the relation of the variables used is a
little complex, the WARN_ON() check was added. It was noticed that
the array was dereferenced before this check and if the logic does
break and for some reason the logic goes past the array, there will
be an out of bounds access here. Move the access to after the
WARN_ON().
- Consolidate how the ring buffer is determined to be empty
Currently there's two ways that are used to determine if the ring
buffer is empty. One relies on the status of the commit and reader
pages and what was read, and the other is on what was written vs what
was read. By using the number of entries (written) method, it can be
used for reading events that are out of the kernel's control (what
pKVM will use). Move to this method to make it easier to implement a
pKVM ring buffer that the kernel can read.
* tag 'trace-ringbuffer-v6.14' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
ring-buffer: Make reading page consistent with the code logic
ring-buffer: Check for empty ring-buffer with rb_num_of_entries()
The following works fine:
~# echo ':mod:trace_events_sample' > /sys/kernel/tracing/set_event
~# cat /sys/kernel/tracing/set_event
*:*:mod:trace_events_sample
~#
But if a name is given without a ':' where it can match an event name or
system name, the output of the cached events does not include a new line:
~# echo 'foo_bar:mod:trace_events_sample' > /sys/kernel/tracing/set_event
~# cat /sys/kernel/tracing/set_event
foo_bar:mod:trace_events_sample~#
Add the '\n' to that as well.
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: https://lore.kernel.org/20250121151336.6c491844@gandalf.local.home
Fixes: b355247df1 ("tracing: Cache ":mod:" events for modules not loaded yet")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Pull performance events updates from Ingo Molnar:
"Seqlock optimizations that arose in a perf context and were merged
into the perf tree:
- seqlock: Add raw_seqcount_try_begin (Suren Baghdasaryan)
- mm: Convert mm_lock_seq to a proper seqcount (Suren Baghdasaryan)
- mm: Introduce mmap_lock_speculate_{try_begin|retry} (Suren
Baghdasaryan)
- mm/gup: Use raw_seqcount_try_begin() (Peter Zijlstra)
Core perf enhancements:
- Reduce 'struct page' footprint of perf by mapping pages in advance
(Lorenzo Stoakes)
- Save raw sample data conditionally based on sample type (Yabin Cui)
- Reduce sampling overhead by checking sample_type in
perf_sample_save_callchain() and perf_sample_save_brstack() (Yabin
Cui)
- Export perf_exclude_event() (Namhyung Kim)
Uprobes scalability enhancements: (Andrii Nakryiko)
- Simplify find_active_uprobe_rcu() VMA checks
- Add speculative lockless VMA-to-inode-to-uprobe resolution
- Simplify session consumer tracking
- Decouple return_instance list traversal and freeing
- Ensure return_instance is detached from the list before freeing
- Reuse return_instances between multiple uretprobes within task
- Guard against kmemdup() failing in dup_return_instance()
AMD core PMU driver enhancements:
- Relax privilege filter restriction on AMD IBS (Namhyung Kim)
AMD RAPL energy counters support: (Dhananjay Ugwekar)
- Introduce topology_logical_core_id() (K Prateek Nayak)
- Remove the unused get_rapl_pmu_cpumask() function
- Remove the cpu_to_rapl_pmu() function
- Rename rapl_pmu variables
- Make rapl_model struct global
- Add arguments to the init and cleanup functions
- Modify the generic variable names to *_pkg*
- Remove the global variable rapl_msrs
- Move the cntr_mask to rapl_pmus struct
- Add core energy counter support for AMD CPUs
Intel core PMU driver enhancements:
- Support RDPMC 'metrics clear mode' feature (Kan Liang)
- Clarify adaptive PEBS processing (Kan Liang)
- Factor out functions for PEBS records processing (Kan Liang)
- Simplify the PEBS records processing for adaptive PEBS (Kan Liang)
Intel uncore driver enhancements: (Kan Liang)
- Convert buggy pmu->func_id use to pmu->registered
- Support more units on Granite Rapids"
* tag 'perf-core-2025-01-20' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (33 commits)
perf: map pages in advance
perf/x86/intel/uncore: Support more units on Granite Rapids
perf/x86/intel/uncore: Clean up func_id
perf/x86/intel: Support RDPMC metrics clear mode
uprobes: Guard against kmemdup() failing in dup_return_instance()
perf/x86: Relax privilege filter restriction on AMD IBS
perf/core: Export perf_exclude_event()
uprobes: Reuse return_instances between multiple uretprobes within task
uprobes: Ensure return_instance is detached from the list before freeing
uprobes: Decouple return_instance list traversal and freeing
uprobes: Simplify session consumer tracking
uprobes: add speculative lockless VMA-to-inode-to-uprobe resolution
uprobes: simplify find_active_uprobe_rcu() VMA checks
mm: introduce mmap_lock_speculate_{try_begin|retry}
mm: convert mm_lock_seq to a proper seqcount
mm/gup: Use raw_seqcount_try_begin()
seqlock: add raw_seqcount_try_begin
perf/x86/rapl: Add core energy counter support for AMD CPUs
perf/x86/rapl: Move the cntr_mask to rapl_pmus struct
perf/x86/rapl: Remove the global variable rapl_msrs
...
Pull block updates from Jens Axboe:
- NVMe pull requests via Keith:
- Target support for PCI-Endpoint transport (Damien)
- TCP IO queue spreading fixes (Sagi, Chaitanya)
- Target handling for "limited retry" flags (Guixen)
- Poll type fix (Yongsoo)
- Xarray storage error handling (Keisuke)
- Host memory buffer free size fix on error (Francis)
- MD pull requests via Song:
- Reintroduce md-linear (Yu Kuai)
- md-bitmap refactor and fix (Yu Kuai)
- Replace kmap_atomic with kmap_local_page (David Reaver)
- Quite a few queue freeze and debugfs deadlock fixes
Ming introduced lockdep support for this in the 6.13 kernel, and it
has (unsurprisingly) uncovered quite a few issues
- Use const attributes for IO schedulers
- Remove bio ioprio wrappers
- Fixes for stacked device atomic write support
- Refactor queue affinity helpers, in preparation for better supporting
isolated CPUs
- Cleanups of loop O_DIRECT handling
- Cleanup of BLK_MQ_F_* flags
- Add rotational support for null_blk
- Various fixes and cleanups
* tag 'for-6.14/block-20250118' of git://git.kernel.dk/linux: (106 commits)
block: Don't trim an atomic write
block: Add common atomic writes enable flag
md/md-linear: Fix a NULL vs IS_ERR() bug in linear_add()
block: limit disk max sectors to (LLONG_MAX >> 9)
block: Change blk_stack_atomic_writes_limits() unit_min check
block: Ensure start sector is aligned for stacking atomic writes
blk-mq: Move more error handling into blk_mq_submit_bio()
block: Reorder the request allocation code in blk_mq_submit_bio()
nvme: fix bogus kzalloc() return check in nvme_init_effects_log()
md/md-bitmap: move bitmap_{start, end}write to md upper layer
md/raid5: implement pers->bitmap_sector()
md: add a new callback pers->bitmap_sector()
md/md-bitmap: remove the last parameter for bimtap_ops->endwrite()
md/md-bitmap: factor behind write counters out from bitmap_{start/end}write()
md: Replace deprecated kmap_atomic() with kmap_local_page()
md: reintroduce md-linear
partitions: ldm: remove the initial kernel-doc notation
blk-cgroup: rwstat: fix kernel-doc warnings in header file
blk-cgroup: fix kernel-doc warnings in header file
nbd: fix partial sending
...
Pull pid_max namespacing update from Christian Brauner:
"The pid_max sysctl is a global value. For a long time the default
value has been 65535 and during the pidfd dicussions Linus proposed to
bump pid_max by default. Based on this discussion systemd started
bumping pid_max to 2^22. So all new systems now run with a very high
pid_max limit with some distros having also backported that change.
The decision to bump pid_max is obviously correct. It just doesn't
make a lot of sense nowadays to enforce such a low pid number. There's
sufficient tooling to make selecting specific processes without typing
really large pid numbers available.
In any case, there are workloads that have expections about how large
pid numbers they accept. Either for historical reasons or
architectural reasons. One concreate example is the 32-bit version of
Android's bionic libc which requires pid numbers less than 65536.
There are workloads where it is run in a 32-bit container on a 64-bit
kernel. If the host has a pid_max value greater than 65535 the libc
will abort thread creation because of size assumptions of
pthread_mutex_t.
That's a fairly specific use-case however, in general specific
workloads that are moved into containers running on a host with a new
kernel and a new systemd can run into issues with large pid_max
values. Obviously making assumptions about the size of the allocated
pid is suboptimal but we have userspace that does it.
Of course, giving containers the ability to restrict the number of
processes in their respective pid namespace indepent of the global
limit through pid_max is something desirable in itself and comes in
handy in general.
Independent of motivating use-cases the existence of pid namespaces
makes this also a good semantical extension and there have been prior
proposals pushing in a similar direction. The trick here is to
minimize the risk of regressions which I think is doable. The fact
that pid namespaces are hierarchical will help us here.
What we mostly care about is that when the host sets a low pid_max
limit, say (crazy number) 100 that no descendant pid namespace can
allocate a higher pid number in its namespace. Since pid allocation is
hierarchial this can be ensured by checking each pid allocation
against the pid namespace's pid_max limit. This means if the
allocation in the descendant pid namespace succeeds, the ancestor pid
namespace can reject it. If the ancestor pid namespace has a higher
limit than the descendant pid namespace the descendant pid namespace
will reject the pid allocation. The ancestor pid namespace will
obviously not care about this.
All in all this means pid_max continues to enforce a system wide limit
on the number of processes but allows pid namespaces sufficient leeway
in handling workloads with assumptions about pid values and allows
containers to restrict the number of processes in a pid namespace
through the pid_max interface"
* tag 'kernel-6.14-rc1.pid' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs:
tests/pid_namespace: add pid_max tests
pid: allow pid_max to be set per pid namespace
Module functions can be set to set_ftrace_filter before the module is
loaded.
# echo :mod:snd_hda_intel > set_ftrace_filter
This will enable all the functions for the module snd_hda_intel. If that
module is not loaded, it is "cached" in the trace array for when the
module is loaded, its functions will be traced.
But this is not implemented in the kernel command line. That's because the
kernel command line filtering is added very early in boot up as it is
needed to be done before boot time function tracing can start, which is
also available very early in boot up. The code used by the
"set_ftrace_filter" file can not be used that early as it depends on some
other initialization to occur first. But some of the functions can.
Implement the ":mod:" feature of "set_ftrace_filter" in the kernel command
line parsing. Now function tracing on just a single module that is loaded
at boot up can be done.
Adding:
ftrace=function ftrace_filter=:mod:sna_hda_intel
To the kernel command line will only enable the sna_hda_intel module
functions when the module is loaded, and it will start tracing.
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: https://lore.kernel.org/20250116175832.34e39779@gandalf.local.home
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
When the :mod: command is written into /sys/kernel/tracing/set_event (or
that file within an instance), if the module specified after the ":mod:"
is not yet loaded, it will store that string internally. When the module
is loaded, it will enable the events as if the module was loaded when the
string was written into the set_event file.
This can also be useful to enable events that are in the init section of
the module, as the events are enabled before the init section is executed.
This also works on the kernel command line:
trace_event=:mod:<module>
Will enable the events for <module> when it is loaded.
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Link: https://lore.kernel.org/20250116143533.514730995@goodmis.org
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Add a :mod: command to enable only events from a given module from the
set_events file.
echo '*:mod:<module>' > set_events
Or
echo ':mod:<module>' > set_events
Will enable all events for that module. Specific events can also be
enabled via:
echo '<event>:mod:<module>' > set_events
Or
echo '<system>:<event>:mod:<module>' > set_events
Or
echo '*:<event>:mod:<module>' > set_events
The ":mod:" keyword is consistent with the function tracing filter to
enable functions from a given module.
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Link: https://lore.kernel.org/20250116143533.214496360@goodmis.org
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
The function graph tracer has become generic so that kretprobes and BPF
can use it along with function graph tracing itself. Some of the
infrastructure was specific for function graph tracing such as recording
the calltime and return time of the functions. Calling the clock code on a
high volume function does add overhead. The calculation of the calltime
was removed from the generic code and placed into the function graph
tracer itself so that the other users did not incur this overhead as they
did not need that timestamp.
The calltime field was still kept in the generic return entry structure
and the function graph return entry callback filled it as that structure
was passed to other code.
But this broke both irqsoff and wakeup latency tracer as they still
depended on the trace structure containing the calltime when the option
display-graph is set as it used some of those same functions that the
function graph tracer used. But now the calltime was not set and was just
zero. This caused the calculation of the function time to be the absolute
value of the return timestamp and not the length of the function.
# cd /sys/kernel/tracing
# echo 1 > options/display-graph
# echo irqsoff > current_tracer
The tracers went from:
# REL TIME CPU TASK/PID |||| DURATION FUNCTION CALLS
# | | | | |||| | | | | | |
0 us | 4) <idle>-0 | d..1. | 0.000 us | irqentry_enter();
3 us | 4) <idle>-0 | d..2. | | irq_enter_rcu() {
4 us | 4) <idle>-0 | d..2. | 0.431 us | preempt_count_add();
5 us | 4) <idle>-0 | d.h2. | | tick_irq_enter() {
5 us | 4) <idle>-0 | d.h2. | 0.433 us | tick_check_oneshot_broadcast_this_cpu();
6 us | 4) <idle>-0 | d.h2. | 2.426 us | ktime_get();
9 us | 4) <idle>-0 | d.h2. | | tick_nohz_stop_idle() {
10 us | 4) <idle>-0 | d.h2. | 0.398 us | nr_iowait_cpu();
11 us | 4) <idle>-0 | d.h1. | 1.903 us | }
11 us | 4) <idle>-0 | d.h2. | | tick_do_update_jiffies64() {
12 us | 4) <idle>-0 | d.h2. | | _raw_spin_lock() {
12 us | 4) <idle>-0 | d.h2. | 0.360 us | preempt_count_add();
13 us | 4) <idle>-0 | d.h3. | 0.354 us | do_raw_spin_lock();
14 us | 4) <idle>-0 | d.h2. | 2.207 us | }
15 us | 4) <idle>-0 | d.h3. | 0.428 us | calc_global_load();
16 us | 4) <idle>-0 | d.h3. | | _raw_spin_unlock() {
16 us | 4) <idle>-0 | d.h3. | 0.380 us | do_raw_spin_unlock();
17 us | 4) <idle>-0 | d.h3. | 0.334 us | preempt_count_sub();
18 us | 4) <idle>-0 | d.h1. | 1.768 us | }
18 us | 4) <idle>-0 | d.h2. | | update_wall_time() {
[..]
To:
# REL TIME CPU TASK/PID |||| DURATION FUNCTION CALLS
# | | | | |||| | | | | | |
0 us | 5) <idle>-0 | d.s2. | 0.000 us | _raw_spin_lock_irqsave();
0 us | 5) <idle>-0 | d.s3. | 312159583 us | preempt_count_add();
2 us | 5) <idle>-0 | d.s4. | 312159585 us | do_raw_spin_lock();
3 us | 5) <idle>-0 | d.s4. | | _raw_spin_unlock() {
3 us | 5) <idle>-0 | d.s4. | 312159586 us | do_raw_spin_unlock();
4 us | 5) <idle>-0 | d.s4. | 312159587 us | preempt_count_sub();
4 us | 5) <idle>-0 | d.s2. | 312159587 us | }
5 us | 5) <idle>-0 | d.s3. | | _raw_spin_lock() {
5 us | 5) <idle>-0 | d.s3. | 312159588 us | preempt_count_add();
6 us | 5) <idle>-0 | d.s4. | 312159589 us | do_raw_spin_lock();
7 us | 5) <idle>-0 | d.s3. | 312159590 us | }
8 us | 5) <idle>-0 | d.s4. | 312159591 us | calc_wheel_index();
9 us | 5) <idle>-0 | d.s4. | | enqueue_timer() {
9 us | 5) <idle>-0 | d.s4. | | wake_up_nohz_cpu() {
11 us | 5) <idle>-0 | d.s4. | | native_smp_send_reschedule() {
11 us | 5) <idle>-0 | d.s4. | 312171987 us | default_send_IPI_single_phys();
12408 us | 5) <idle>-0 | d.s3. | 312171990 us | }
12408 us | 5) <idle>-0 | d.s3. | 312171991 us | }
12409 us | 5) <idle>-0 | d.s3. | 312171991 us | }
Where the calculation of the time for each function was the return time
minus zero and not the time of when the function returned.
Have these tracers also save the calltime in the fgraph data section and
retrieve it again on the return to get the correct timings again.
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Link: https://lore.kernel.org/20250113183124.61767419@gandalf.local.home
Fixes: f1f36e22be ("ftrace: Have calltime be saved in the fgraph storage")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
In the loop of __rb_map_vma(), the 's' variable is calculated from the
same logic that nr_pages is and they both come from nr_subbufs. But the
relationship is not obvious and there's a WARN_ON_ONCE() around the 's'
variable to make sure it never becomes equal to nr_subbufs within the
loop. If that happens, then the code is buggy and needs to be fixed.
The 'page' variable is calculated from cpu_buffer->subbuf_ids[s] which is
an array of 'nr_subbufs' entries. If the code becomes buggy and 's'
becomes equal to or greater than 'nr_subbufs' then this will be an out of
bounds hit before the WARN_ON() is triggered and the code exiting safely.
Make the 'page' initialization consistent with the code logic and assign
it after the out of bounds check.
Link: https://lore.kernel.org/20250110162612.13983-1-aha310510@gmail.com
Signed-off-by: Jeongjun Park <aha310510@gmail.com>
[ sdr: rewrote change log ]
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Currently there are two ways of identifying an empty ring-buffer. One
relying on the current status of the commit / reader page
(rb_per_cpu_empty()) and the other on the write and read counters
(rb_num_of_entries() used in rb_get_reader_page()).
with rb_num_of_entries(). This intends to ease later
introduction of ring-buffer writers which are out of the kernel control
and with whom, the only information available is through the meta-page
counters.
Link: https://lore.kernel.org/20250108114536.627715-2-vdonnefort@google.com
Signed-off-by: Vincent Donnefort <vdonnefort@google.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
When kprobe multi bpf program can't be executed due to recursion check,
we currently return 0 (success) to fprobe layer where it's ignored for
standard kprobe multi probes.
For kprobe session the success return value will make fprobe layer to
install return probe and try to execute it as well.
But the return session probe should not get executed, because the entry
part did not run. FWIW the return probe bpf program most likely won't get
executed, because its recursion check will likely fail as well, but we
don't need to run it in the first place.. also we can make this clear
and obvious.
It also affects missed counts for kprobe session program execution, which
are now doubled (extra count for not executed return probe).
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Link: https://lore.kernel.org/r/20250106175048.1443905-1-jolsa@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Commit ef1b808e3b ("bpf: Fix UAF via mismatching bpf_prog/attachment
RCU flavors") resolved a possible UAF issue in uprobes that attach
non-sleepable bpf prog by explicitly waiting for a tasks-trace-RCU grace
period. But, in the current implementation, synchronize_rcu_tasks_trace
is included within the mutex critical section, which increases the
length of the critical section and may affect performance. So let's move
out synchronize_rcu_tasks_trace from mutex CS.
Signed-off-by: Pu Lehui <pulehui@huawei.com>
Reviewed-by: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/r/20250104013946.1111785-1-pulehui@huaweicloud.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Since POLLIN will not be flushed until the hist file is read, the user
needs to repeatedly read() and poll() on the hist file for monitoring the
event continuously. But the read() is somewhat redundant when the user is
only monitoring for event updates.
Add POLLPRI poll event on the hist file so the event returns when a
histogram is updated after open(), poll() or read(). Thus it is possible
to wait for the next event without having to issue a read().
Cc: Shuah Khan <shuah@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: https://lore.kernel.org/173527248770.464571.2536902137325258133.stgit@devnote2
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Reviewed-by: Tom Zanussi <zanussi@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Add poll syscall support on the `hist` file. The Waiter will be waken
up when the histogram is updated with POLLIN.
Currently, there is no way to wait for a specific event in userspace.
So user needs to peek the `trace` periodicaly, or wait on `trace_pipe`.
But it is not a good idea to peek at the `trace` for an event that
randomly happens. And `trace_pipe` is not coming back until a page is
filled with events.
This allows a user to wait for a specific event on the `hist` file. User
can set a histogram trigger on the event which they want to monitor
and poll() on its `hist` file. Since this poll() returns POLLIN, the next
poll() will return soon unless a read() happens on that hist file.
NOTE: To read the hist file again, you must set the file offset to 0,
but just for monitoring the event, you may not need to read the
histogram.
Cc: Shuah Khan <shuah@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: https://lore.kernel.org/173527247756.464571.14236296701625509931.stgit@devnote2
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Reviewed-by: Tom Zanussi <zanussi@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
When the function tracing_set_tracer() switched over to using the guard()
infrastructure, it did not need to save the 'ret' variable and would just
return the value when an error arised, instead of setting ret and jumping
to an out label.
When CONFIG_TRACER_SNAPSHOT is enabled, it had code that expected the
"ret" variable to be initialized to zero and had set 'ret' while holding
an arch_spin_lock() (not used by guard), and then upon releasing the lock
it would check 'ret' and exit if set. But because ret was only set when an
error occurred while holding the locks, 'ret' would be used uninitialized
if there was no error. The code in the CONFIG_TRACER_SNAPSHOT block should
be self contain. Make sure 'ret' is also set when no error occurred.
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: https://lore.kernel.org/20250106111143.2f90ff65@gandalf.local.home
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Closes: https://lore.kernel.org/r/202412271654.nJVBuwmF-lkp@intel.com/
Fixes: d33b10c0c7 ("tracing: Switch trace.c code over to use guard()")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Pull ftrace fixes from Steven Rostedt:
- Add needed READ_ONCE() around access to the fgraph array element
The updates to the fgraph array can happen when callbacks are
registered and unregistered. The __ftrace_return_to_handler() can
handle reading either the old value or the new value. But once it
reads that value it must stay consistent otherwise the check that
looks to see if the value is a stub may show false, but if the
compiler decides to re-read after that check, it can be true which
can cause the code to crash later on.
- Make function profiler use the top level ops for filtering again
When function graph became available for instances, its filter ops
became independent from the top level set_ftrace_filter. In the
process the function profiler received its own filter ops as well.
But the function profiler uses the top level set_ftrace_filter file
and does not have one of its own. In giving it its own filter ops, it
lost any user interface it once had. Make it use the top level
set_ftrace_filter file again. This fixes a regression.
* tag 'ftrace-v6.13-rc5-2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
ftrace: Fix function profiler's filtering functionality
fgraph: Add READ_ONCE() when accessing fgraph_array[]
Commit c132be2c4f ("function_graph: Have the instances use their own
ftrace_ops for filtering"), function profiler (enabled via
function_profile_enabled) has been showing statistics for all functions,
ignoring set_ftrace_filter settings.
While tracers are instantiated, the function profiler is not. Therefore, it
should use the global set_ftrace_filter for consistency. This patch
modifies the function profiler to use the global filter, fixing the
filtering functionality.
Before (filtering not working):
```
root@localhost:~# echo 'vfs*' > /sys/kernel/tracing/set_ftrace_filter
root@localhost:~# echo 1 > /sys/kernel/tracing/function_profile_enabled
root@localhost:~# sleep 1
root@localhost:~# echo 0 > /sys/kernel/tracing/function_profile_enabled
root@localhost:~# head /sys/kernel/tracing/trace_stat/*
Function Hit Time Avg
s^2
-------- --- ---- ---
---
schedule 314 22290594 us 70989.15 us
40372231 us
x64_sys_call 1527 8762510 us 5738.382 us
3414354 us
schedule_hrtimeout_range 176 8665356 us 49234.98 us
405618876 us
__x64_sys_ppoll 324 5656635 us 17458.75 us
19203976 us
do_sys_poll 324 5653747 us 17449.83 us
19214945 us
schedule_timeout 67 5531396 us 82558.15 us
2136740827 us
__x64_sys_pselect6 12 3029540 us 252461.7 us
63296940171 us
do_pselect.constprop.0 12 3029532 us 252461.0 us
63296952931 us
```
After (filtering working):
```
root@localhost:~# echo 'vfs*' > /sys/kernel/tracing/set_ftrace_filter
root@localhost:~# echo 1 > /sys/kernel/tracing/function_profile_enabled
root@localhost:~# sleep 1
root@localhost:~# echo 0 > /sys/kernel/tracing/function_profile_enabled
root@localhost:~# head /sys/kernel/tracing/trace_stat/*
Function Hit Time Avg
s^2
-------- --- ---- ---
---
vfs_write 462 68476.43 us 148.217 us
25874.48 us
vfs_read 641 9611.356 us 14.994 us
28868.07 us
vfs_fstat 890 878.094 us 0.986 us
1.667 us
vfs_fstatat 227 757.176 us 3.335 us
18.928 us
vfs_statx 226 610.610 us 2.701 us
17.749 us
vfs_getattr_nosec 1187 460.919 us 0.388 us
0.326 us
vfs_statx_path 297 343.287 us 1.155 us
11.116 us
vfs_rename 6 291.575 us 48.595 us
9889.236 us
```
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/20250101190820.72534-1-enjuk@amazon.com
Fixes: c132be2c4f ("function_graph: Have the instances use their own ftrace_ops for filtering")
Signed-off-by: Kohei Enju <enjuk@amazon.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
In __ftrace_return_to_handler(), a loop iterates over the fgraph_array[]
elements, which are fgraph_ops. The loop checks if an element is a
fgraph_stub to prevent using a fgraph_stub afterward.
However, if the compiler reloads fgraph_array[] after this check, it might
race with an update to fgraph_array[] that introduces a fgraph_stub. This
could result in the stub being processed, but the stub contains a null
"func_hash" field, leading to a NULL pointer dereference.
To ensure that the gops compared against the fgraph_stub matches the gops
processed later, add a READ_ONCE(). A similar patch appears in commit
63a8dfb ("function_graph: Add READ_ONCE() when accessing fgraph_array[]").
Cc: stable@vger.kernel.org
Fixes: 37238abe3c ("ftrace/function_graph: Pass fgraph_ops to function graph callbacks")
Link: https://lore.kernel.org/20241231113731.277668-1-zilin@seu.edu.cn
Signed-off-by: Zilin Guan <zilin@seu.edu.cn>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
In order to catch a common bug where a TRACE_EVENT() TP_fast_assign()
assigns an address of an allocated string to the ring buffer and then
references it in TP_printk(), which can be executed hours later when the
string is free, the function test_event_printk() runs on all events as
they are registered to make sure there's no unwanted dereferencing.
It calls process_string() to handle cases in TP_printk() format that has
"%s". It returns whether or not the string is safe. But it can have some
false positives.
For instance, xe_bo_move() has:
TP_printk("move_lacks_source:%s, migrate object %p [size %zu] from %s to %s device_id:%s",
__entry->move_lacks_source ? "yes" : "no", __entry->bo, __entry->size,
xe_mem_type_to_name[__entry->old_placement],
xe_mem_type_to_name[__entry->new_placement], __get_str(device_id))
Where the "%s" references into xe_mem_type_to_name[]. This is an array of
pointers that should be safe for the event to access. Instead of flagging
this as a bad reference, if a reference points to an array, where the
record field is the index, consider it safe.
Link: https://lore.kernel.org/all/9dee19b6185d325d0e6fa5f7cbba81d007d99166.camel@sapience.com/
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: https://lore.kernel.org/20241231000646.324fb5f7@gandalf.local.home
Fixes: 65a25d9f7a ("tracing: Add "%s" check in test_event_printk()")
Reported-by: Genes Lists <lists@sapience.com>
Tested-by: Gene C <arch@sapience.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
dot2k suggests a list of changes to the kernel tree while adding a
monitor: edit tracepoints header, Makefile, Kconfig and moving the
monitor folder. Those changes can be easily run automatically.
Add a flag to dot2k to alter the kernel source.
The kernel source directory can be either assumed from the PWD, or from
the running kernel, if installed.
This feature works best if the kernel tree is a git repository, so that
its easier to make sure there are no unintended changes.
The main RV files (e.g. Makefile) have now a comment placeholder that
can be useful for manual editing (e.g. to know where to add new
monitors) and it is used by the script to append the required lines.
We also slightly adapt the file handling functions in dot2k: __open_file
is now called __read_file and also closes the file before returning the
content; __create_file is now a more general __write_file, we no longer
return on FileExistsError (not thrown while opening), a new
__create_file simply calls __write_file specifying the monitor folder in
the path.
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
Link: https://lore.kernel.org/20241227144752.362911-8-gmonaco@redhat.com
Signed-off-by: Gabriele Monaco <gmonaco@redhat.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
While creating a new monitor in RV, besides generating code from dot2k,
there are a few manual steps which can be tedious and error prone, like
adding the tracepoints, makefile lines and kconfig.
This patch restructures the existing monitors to keep some files in the
monitor's folder itself, which can be automatically generated by future
versions of dot2k.
Monitors have now their own Kconfig and tracepoint snippets. For
simplicity, the main tracepoint definition, is moved to the RV
directory, it defines only the tracepoint classes and includes the
monitor-specific tracepoints, which reside in the monitor directory.
Tracepoints and Kconfig no longer need to be copied and adapted from
existing ones but only need to be included in the main files.
The Makefile remains untouched since there's little advantage in having
a separated Makefile for each monitor with a single line and including
it in the main RV Makefile.
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
Link: https://lore.kernel.org/20241227144752.362911-6-gmonaco@redhat.com
Signed-off-by: Gabriele Monaco <gmonaco@redhat.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Pull probes fix from Masami Hiramatsu:
"Change the priority of the module callback of kprobe events so that it
is called after the jump label list on the module is updated.
This ensures the kprobe can check whether it is not on the jump label
address correctly"
* tag 'probes-fixes-v6.13-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
tracing/kprobe: Make trace_kprobe's module callback called after jump_label update
