There's no field for the cgroup, let's add one. To do that, users need to
specify --all-cgroup option for perf record to capture the cgroup info.
$ perf record --all-cgroups -- true
$ perf script -F comm,pid,cgroup
true 337112 /user.slice/user-657345.slice/user@657345.service/...
true 337112 /user.slice/user-657345.slice/user@657345.service/...
true 337112 /user.slice/user-657345.slice/user@657345.service/...
true 337112 /user.slice/user-657345.slice/user@657345.service/...
If it's recorded without the --all-cgroups, it'd complain.
$ perf script -F comm,pid,cgroup
Samples for 'cycles:u' event do not have CGROUP attribute set. Cannot print 'cgroup' field.
Hint: run 'perf record --all-cgroups ...'
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: https://lore.kernel.org/r/20230126213610.3381147-1-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
This problem was encountered on an arm64 system with a lot of memory.
Without kernel debug symbols installed, and with both kcore and kallsyms
available, perf managed to get confused and returned "unknown" for all
of the kernel symbols that it tried to look up.
On this system, stext fell within the vmalloc segment. The kcore symbol
matching code tries to find the first segment that contains stext and
uses that to replace the segment generated from just the kallsyms
information. In this case, however, there were two: a very large
vmalloc segment, and the text segment. This caused perf to get confused
because multiple overlapping segments were inserted into the RB tree
that holds the discovered segments. However, that alone wasn't
sufficient to cause the problem. Even when we could find the segment,
the offsets were adjusted in such a way that the newly generated symbols
didn't line up with the instruction addresses in the trace. The most
obvious solution would be to consult which segment type is text from
kcore, but this information is not exposed to users.
Instead, select the smallest matching segment that contains stext
instead of the first matching segment. This allows us to match the text
segment instead of vmalloc, if one is contained within the other.
Reviewed-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Krister Johansen <kjlx@templeofstupid.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: David Reaver <me@davidreaver.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/20230125183418.GD1963@templeofstupid.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
While parsing the tracepoint events in parse_events_add_tracepoint()
function, code checks for HAVE_LIBTRACEEVENT support. This is needed
since libtraceevent is necessary for tracepoint. But while adding probe
points, check for LIBTRACEEVENT is not done in case of perf probe.
Hence, in environment with missing libtraceevent-devel, it is observed
that adding a probe point shows below message though it can't be used
via perf record.
Example:
Adding probe point:
./perf probe 'vfs_getname=getname_flags:72 pathname=result->name:string'
Added new event:
probe:vfs_getname (on getname_flags:72 with pathname=result->name:string)
You can now use it in all perf tools, such as:
perf record -e probe:vfs_getname -aR sleep 1
But trying perf record:
./perf record -e probe:vfs_getname -aR sleep 1
event syntax error: 'probe:vfs_getname'
\___ unsupported tracepoint
libtraceevent is necessary for tracepoint support
Run 'perf list' for a list of valid events
The builtin tool like perf record needs libtraceevent to
parse tracefs. But still the probe can be used by enabling
via tracefs. Patch fixes the probe usage message to the user
based on presence of libtraceevent. With the fix,
# ./perf probe 'pmu:myprobe=schedule'
Added new event:
pmu:myprobe (on schedule)
perf is not linked with libtraceevent, to use the new probe you can use tracefs:
cd /sys/kernel/tracing/
echo 1 > events/pmu/myprobe/enable
echo 1 > tracing_on
cat trace_pipe
Before removing the probe, echo 0 > events/pmu/myprobe/enable
Signed-off-by: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Disha Goel <disgoel@linux.ibm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nageswara R Sastry <rnsastry@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: linuxppc-dev@lists.ozlabs.org
Link: https://lore.kernel.org/r/20230131134748.54567-1-atrajeev@linux.vnet.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
For x86_64, determine a symbol for .plt.got entries. That requires
computing the target offset and finding that in .rela.dyn, which in
turn means .rela.dyn needs to be sorted by offset.
Example:
In this example, the GNU C Library is using .plt.got for malloc and
free.
Before:
$ gcc --version
gcc (Ubuntu 11.3.0-1ubuntu1~22.04) 11.3.0
Copyright (C) 2021 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
$ perf record -e intel_pt//u uname
Linux
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.027 MB perf.data ]
$ perf script --itrace=be --ns -F-event,+addr,-period,-comm,-tid,-cpu > /tmp/cmp1.txt
After:
$ perf script --itrace=be --ns -F-event,+addr,-period,-comm,-tid,-cpu > /tmp/cmp2.txt
$ diff /tmp/cmp1.txt /tmp/cmp2.txt | head -12
15509,15510c15509,15510
< 27046.755390907: 7f0b2943e3ab _nl_normalize_codeset+0x5b (/usr/lib/x86_64-linux-gnu/libc.so.6) => 7f0b29428380 offset_0x28380@plt+0x0 (/usr/lib/x86_64-linux-gnu/libc.so.6)
< 27046.755390907: 7f0b29428384 offset_0x28380@plt+0x4 (/usr/lib/x86_64-linux-gnu/libc.so.6) => 7f0b294a5120 malloc+0x0 (/usr/lib/x86_64-linux-gnu/libc.so.6)
---
> 27046.755390907: 7f0b2943e3ab _nl_normalize_codeset+0x5b (/usr/lib/x86_64-linux-gnu/libc.so.6) => 7f0b29428380 malloc@plt+0x0 (/usr/lib/x86_64-linux-gnu/libc.so.6)
> 27046.755390907: 7f0b29428384 malloc@plt+0x4 (/usr/lib/x86_64-linux-gnu/libc.so.6) => 7f0b294a5120 malloc+0x0 (/usr/lib/x86_64-linux-gnu/libc.so.6)
15821,15822c15821,15822
< 27046.755394865: 7f0b2943850c _nl_load_locale_from_archive+0x5bc (/usr/lib/x86_64-linux-gnu/libc.so.6) => 7f0b29428370 offset_0x28370@plt+0x0 (/usr/lib/x86_64-linux-gnu/libc.so.6)
< 27046.755394865: 7f0b29428374 offset_0x28370@plt+0x4 (/usr/lib/x86_64-linux-gnu/libc.so.6) => 7f0b294a5460 cfree@GLIBC_2.2.5+0x0 (/usr/lib/x86_64-linux-gnu/libc.so.6)
---
> 27046.755394865: 7f0b2943850c _nl_load_locale_from_archive+0x5bc (/usr/lib/x86_64-linux-gnu/libc.so.6) => 7f0b29428370 free@plt+0x0 (/usr/lib/x86_64-linux-gnu/libc.so.6)
> 27046.755394865: 7f0b29428374 free@plt+0x4 (/usr/lib/x86_64-linux-gnu/libc.so.6) => 7f0b294a5460 cfree@GLIBC_2.2.5+0x0 (/usr/lib/x86_64-linux-gnu/libc.so.6)
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/r/20230131131625.6964-10-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Currently, timerlat top displays the timerlat tracer latency results, saving
the intuitive timerlat trace for the developer to analyze.
This patch goes a step forward in the automaton of the scheduling latency
analysis by providing a summary of the root cause of a latency higher than
the passed "stop tracing" parameter if the trace stops.
The output is intuitive enough for non-expert users to have a general idea
of the root cause by looking at each factor's contribution percentage while
keeping the technical detail in the output for more expert users to start
an in dept debug or to correlate a root cause with an existing one.
The terminology is in line with recent industry and academic publications
to facilitate the understanding of both audiences.
Here is one example of tool output:
----------------------------------------- %< -----------------------------------------------------
# taskset -c 0 timerlat -a 40 -c 1-23 -q
Timer Latency
0 00:00:12 | IRQ Timer Latency (us) | Thread Timer Latency (us)
CPU COUNT | cur min avg max | cur min avg max
1 #12322 | 0 0 1 15 | 10 3 9 31
2 #12322 | 3 0 1 12 | 10 3 9 23
3 #12322 | 1 0 1 21 | 8 2 8 34
4 #12322 | 1 0 1 17 | 10 2 11 33
5 #12322 | 0 0 1 12 | 8 3 8 25
6 #12322 | 1 0 1 14 | 16 3 11 35
7 #12322 | 0 0 1 14 | 9 2 8 29
8 #12322 | 1 0 1 22 | 9 3 9 34
9 #12322 | 0 0 1 14 | 8 2 8 24
10 #12322 | 1 0 0 12 | 9 3 8 24
11 #12322 | 0 0 0 15 | 6 2 7 29
12 #12321 | 1 0 0 13 | 5 3 8 23
13 #12319 | 0 0 1 14 | 9 3 9 26
14 #12321 | 1 0 0 13 | 6 2 8 24
15 #12321 | 1 0 1 15 | 12 3 11 27
16 #12318 | 0 0 1 13 | 7 3 10 24
17 #12319 | 0 0 1 13 | 11 3 9 25
18 #12318 | 0 0 0 12 | 8 2 8 20
19 #12319 | 0 0 1 18 | 10 2 9 28
20 #12317 | 0 0 0 20 | 9 3 8 34
21 #12318 | 0 0 0 13 | 8 3 8 28
22 #12319 | 0 0 1 11 | 8 3 10 22
23 #12320 | 28 0 1 28 | 41 3 11 41
rtla timerlat hit stop tracing
## CPU 23 hit stop tracing, analyzing it ##
IRQ handler delay: 27.49 us (65.52 %)
IRQ latency: 28.13 us
Timerlat IRQ duration: 9.59 us (22.85 %)
Blocking thread: 3.79 us (9.03 %)
objtool:49256 3.79 us
Blocking thread stacktrace
-> timerlat_irq
-> __hrtimer_run_queues
-> hrtimer_interrupt
-> __sysvec_apic_timer_interrupt
-> sysvec_apic_timer_interrupt
-> asm_sysvec_apic_timer_interrupt
-> _raw_spin_unlock_irqrestore
-> cgroup_rstat_flush_locked
-> cgroup_rstat_flush_irqsafe
-> mem_cgroup_flush_stats
-> mem_cgroup_wb_stats
-> balance_dirty_pages
-> balance_dirty_pages_ratelimited_flags
-> btrfs_buffered_write
-> btrfs_do_write_iter
-> vfs_write
-> __x64_sys_pwrite64
-> do_syscall_64
-> entry_SYSCALL_64_after_hwframe
------------------------------------------------------------------------
Thread latency: 41.96 us (100%)
The system has exit from idle latency!
Max timerlat IRQ latency from idle: 17.48 us in cpu 4
Saving trace to timerlat_trace.txt
----------------------------------------- >% -----------------------------------------------------
In this case, the major factor was the delay suffered by the IRQ handler
that handles timerlat wakeup: 65.52 %. This can be caused by the
current thread masking interrupts, which can be seen in the blocking
thread stacktrace: the current thread (objtool:49256) disabled interrupts
via raw spin lock operations inside mem cgroup, while doing write
syscall in a btrfs file system.
A simple search for the function name on Google shows that this is
a legit case for disabling the interrupts:
cgroup: Use irqsave in cgroup_rstat_flush_locked()
lore.kernel.org/linux-mm/20220301122143.1521823-2-bigeasy@linutronix.de/
The output also prints other reasons for the latency root cause, such as:
- an IRQ that happened before the IRQ handler that caused delays
- The interference from NMI, IRQ, Softirq, and Threads
The details about how these factors affect the scheduling latency
can be found here:
https://bristot.me/demystifying-the-real-time-linux-latency/
Link: https://lkml.kernel.org/r/3d45f40e630317f51ac6d678e2d96d310e495729.1675179318.git.bristot@kernel.org
Cc: Daniel Bristot de Oliveira <bristot@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
The canonical location for the tracefs filesystem is at /sys/kernel/tracing.
But, from Documentation/trace/ftrace.rst:
Before 4.1, all ftrace tracing control files were within the debugfs
file system, which is typically located at /sys/kernel/debug/tracing.
For backward compatibility, when mounting the debugfs file system,
the tracefs file system will be automatically mounted at:
/sys/kernel/debug/tracing
A few scripts in tools/power still refer to this older debugfs path, so
let's update them to avoid confusion.
Signed-off-by: Ross Zwisler <zwisler@google.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The test tool can check that the zerocopy number of completions value is
valid taking into consideration the number of datagram send calls. This can
catch the system into a state where the datagrams are still in the system
(for example in a qdisk, waiting for the network interface to return a
completion notification, etc).
This change adds a retry logic of computing the number of completions up to
a configurable (via CLI) timeout (default: 2 seconds).
Fixes: 79ebc3c260 ("net/udpgso_bench_tx: options to exercise TX CMSG")
Signed-off-by: Andrei Gherzan <andrei.gherzan@canonical.com>
Cc: Willem de Bruijn <willemb@google.com>
Cc: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://lore.kernel.org/r/20230201001612.515730-4-andrei.gherzan@canonical.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
"udpgro_bench.sh" invokes udpgso_bench_rx/udpgso_bench_tx programs
subsequently and while doing so, there is a chance that the rx one is not
ready to accept socket connections. This racing bug could fail the test
with at least one of the following:
./udpgso_bench_tx: connect: Connection refused
./udpgso_bench_tx: sendmsg: Connection refused
./udpgso_bench_tx: write: Connection refused
This change addresses this by making udpgro_bench.sh wait for the rx
program to be ready before firing off the tx one - up to a 10s timeout.
Fixes: 3a687bef14 ("selftests: udp gso benchmark")
Signed-off-by: Andrei Gherzan <andrei.gherzan@canonical.com>
Cc: Paolo Abeni <pabeni@redhat.com>
Cc: Willem de Bruijn <willemb@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://lore.kernel.org/r/20230201001612.515730-3-andrei.gherzan@canonical.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
This change fixes the following compiler warning:
/usr/include/x86_64-linux-gnu/bits/error.h:40:5: warning: ‘gso_size’ may
be used uninitialized [-Wmaybe-uninitialized]
40 | __error_noreturn (__status, __errnum, __format,
__va_arg_pack ());
|
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
udpgso_bench_rx.c: In function ‘main’:
udpgso_bench_rx.c:253:23: note: ‘gso_size’ was declared here
253 | int ret, len, gso_size, budget = 256;
Fixes: 3327a9c463 ("selftests: add functionals test for UDP GRO")
Signed-off-by: Andrei Gherzan <andrei.gherzan@canonical.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://lore.kernel.org/r/20230201001612.515730-1-andrei.gherzan@canonical.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
A statically linked executable can have a .plt due to IFUNCs, in which
case .symtab is used not .dynsym. Check the section header link to see
if that is the case, and then use symtab instead.
Example:
Before:
$ cat tstifunc.c
#include <stdio.h>
void thing1(void)
{
printf("thing1\n");
}
void thing2(void)
{
printf("thing2\n");
}
typedef void (*thing_fn_t)(void);
thing_fn_t thing_ifunc(void)
{
int x;
if (x & 1)
return thing2;
return thing1;
}
void thing(void) __attribute__ ((ifunc ("thing_ifunc")));
int main()
{
thing();
return 0;
}
$ gcc --version
gcc (Ubuntu 11.3.0-1ubuntu1~22.04) 11.3.0
Copyright (C) 2021 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
$ gcc -static -Wall -Wextra -Wno-uninitialized -o tstifuncstatic tstifunc.c
$ readelf -SW tstifuncstatic | grep 'Name\|plt\|dyn'
[Nr] Name Type Address Off Size ES Flg Lk Inf Al
[ 4] .rela.plt RELA 00000000004002e8 0002e8 000258 18 AI 29 20 8
[ 6] .plt PROGBITS 0000000000401020 001020 000190 00 AX 0 0 16
[20] .got.plt PROGBITS 00000000004c5000 0c4000 0000e0 08 WA 0 0 8
$ perf record -e intel_pt//u --filter 'filter main @ ./tstifuncstatic' ./tstifuncstatic
thing1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.008 MB perf.data ]
$ perf script --itrace=be --ns -F+flags,-event,+addr,-period,-comm,-tid,-cpu,-dso
15786.690189535: tr strt 0 [unknown] => 4017cd main+0x0
15786.690189535: tr end call 4017d5 main+0x8 => 401170 [unknown]
15786.690197660: tr strt 0 [unknown] => 4017da main+0xd
15786.690197660: tr end return 4017e0 main+0x13 => 401c1a __libc_start_call_main+0x6a
After:
$ perf script --itrace=be --ns -F+flags,-event,+addr,-period,-comm,-tid,-cpu,-dso
15786.690189535: tr strt 0 [unknown] => 4017cd main+0x0
15786.690189535: tr end call 4017d5 main+0x8 => 401170 thing_ifunc@plt+0x0
15786.690197660: tr strt 0 [unknown] => 4017da main+0xd
15786.690197660: tr end return 4017e0 main+0x13 => 401c1a __libc_start_call_main+0x6a
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/r/20230131131625.6964-8-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The section .plt.sec was originally added for MPX and was first called
.plt.bnd. While MPX has been deprecated, .plt.sec is now also used for
IBT. On x86_64, IBT may be enabled by default, but can be switched off
using gcc option -fcf-protection=none, or switched on by -z ibt or -z
ibtplt. On 32-bit, option -z ibt or -z ibtplt will enable IBT.
With .plt.sec, calls are made into .plt.sec instead of .plt, so it makes
more sense to put the symbols there instead of .plt. A notable
difference is that .plt.sec does not have a header entry.
For x86, when synthesizing symbols for plt, use offset and entry size of
.plt.sec instead of .plt when there is a .plt.sec section.
Example on Ubuntu 22.04 gcc 11.3:
Before:
$ cat tstpltlib.c
void fn1(void) {}
void fn2(void) {}
void fn3(void) {}
void fn4(void) {}
$ cat tstplt.c
void fn1(void);
void fn2(void);
void fn3(void);
void fn4(void);
int main()
{
fn4();
fn1();
fn2();
fn3();
return 0;
}
$ gcc --version
gcc (Ubuntu 11.3.0-1ubuntu1~22.04) 11.3.0
Copyright (C) 2021 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
$ gcc -Wall -Wextra -shared -o libtstpltlib.so tstpltlib.c
$ gcc -Wall -Wextra -z ibt -o tstplt tstplt.c -L . -ltstpltlib -Wl,-rpath=$(pwd)
$ readelf -SW tstplt | grep 'plt\|Name'
[Nr] Name Type Address Off Size ES Flg Lk Inf Al
[11] .rela.plt RELA 0000000000000698 000698 000060 18 AI 6 24 8
[13] .plt PROGBITS 0000000000001020 001020 000050 10 AX 0 0 16
[14] .plt.got PROGBITS 0000000000001070 001070 000010 10 AX 0 0 16
[15] .plt.sec PROGBITS 0000000000001080 001080 000040 10 AX 0 0 16
$ perf record -e intel_pt//u --filter 'filter main @ ./tstplt' ./tstplt
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.015 MB perf.data ]
$ perf script --itrace=be --ns -F+flags,-event,+addr,-period,-comm,-tid,-cpu,-dso
38970.522546686: tr strt 0 [unknown] => 55fc222a81a9 main+0x0
38970.522546686: tr end call 55fc222a81b1 main+0x8 => 55fc222a80a0 [unknown]
38970.522546687: tr strt 0 [unknown] => 55fc222a81b6 main+0xd
38970.522546687: tr end call 55fc222a81b6 main+0xd => 55fc222a8080 [unknown]
38970.522546688: tr strt 0 [unknown] => 55fc222a81bb main+0x12
38970.522546688: tr end call 55fc222a81bb main+0x12 => 55fc222a80b0 [unknown]
38970.522546688: tr strt 0 [unknown] => 55fc222a81c0 main+0x17
38970.522546688: tr end call 55fc222a81c0 main+0x17 => 55fc222a8090 [unknown]
38970.522546689: tr strt 0 [unknown] => 55fc222a81c5 main+0x1c
38970.522546894: tr end return 55fc222a81cb main+0x22 => 7f3a4dc29d90 __libc_start_call_main+0x80
After:
$ perf script --itrace=be --ns -F+flags,-event,+addr,-period,-comm,-tid,-cpu,-dso
38970.522546686: tr strt 0 [unknown] => 55fc222a81a9 main+0x0
38970.522546686: tr end call 55fc222a81b1 main+0x8 => 55fc222a80a0 fn4@plt+0x0
38970.522546687: tr strt 0 [unknown] => 55fc222a81b6 main+0xd
38970.522546687: tr end call 55fc222a81b6 main+0xd => 55fc222a8080 fn1@plt+0x0
38970.522546688: tr strt 0 [unknown] => 55fc222a81bb main+0x12
38970.522546688: tr end call 55fc222a81bb main+0x12 => 55fc222a80b0 fn2@plt+0x0
38970.522546688: tr strt 0 [unknown] => 55fc222a81c0 main+0x17
38970.522546688: tr end call 55fc222a81c0 main+0x17 => 55fc222a8090 fn3@plt+0x0
38970.522546689: tr strt 0 [unknown] => 55fc222a81c5 main+0x1c
38970.522546894: tr end return 55fc222a81cb main+0x22 => 7f3a4dc29d90 __libc_start_call_main+0x80
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/r/20230131131625.6964-3-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Test “Use vfs_getname probe to get syscall args filenames” fails in
environment with missing libtraceevent support as below:
82: Use vfs_getname probe to get syscall args filenames :
--- start ---
test child forked, pid 304726
Recording open file:
event syntax error: 'probe:vfs_getname*'
\___ unsupported tracepoint
libtraceevent is necessary for tracepoint support
Run 'perf list' for a list of valid events
Usage: perf record [<options>] [<command>]
or: perf record [<options>] -- <command> [<options>]
-e, --event <event> event selector. use 'perf list' to list available events
test child finished with -1
---- end ----
Use vfs_getname probe to get syscall args filenames: FAILED!
The environment has debuginfo but is missing the libtraceevent devel.
Hence perf is compiled without libtraceevent support. The test tries to
add probe “probe:vfs_getname” and then uses it with “perf record”. This
fails at function “parse_events_add_tracepoint" due to missing
libtraceevent.
Similarly "probe libc's inet_pton & backtrace it with ping" test slso
fails with same reason.
Add a function in 'perf test shell' library to check if perf record with
—dry-run reports any error on missing support for libtraceevent. Update
both the tests to use this new function “skip_no_probe_record_support”
before proceeding With using probe point via perf builtin record.
With the change,
82: Use vfs_getname probe to get syscall args filenames :
--- start ---
test child forked, pid 305014
Recording open file:
libtraceevent is necessary for tracepoint support
test child finished with -2
---- end ----
Use vfs_getname probe to get syscall args filenames: Skip
81: probe libc's inet_pton & backtrace it with ping :
--- start ---
test child forked, pid 305036
libtraceevent is necessary for tracepoint support
test child finished with -2
---- end ----
probe libc's inet_pton & backtrace it with ping: Skip
Signed-off-by: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Disha Goel <disgoel@linux.ibm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nageswara R Sastry <rnsastry@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: kjain@linux.ibm.com,
Cc: linuxppc-dev@lists.ozlabs.org
Link: http://lore.kernel.org/r/20230201180421.59640-2-atrajeev@linux.vnet.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The "probe libc's inet_pton & backtrace it with ping" test installs a
uprobe and uses perf record/script to check the backtrace. Currently
even if the "perf record" fails, the test reports success. Logs below:
# ./perf test -v "probe libc's inet_pton & backtrace it with ping"
81: probe libc's inet_pton & backtrace it with ping :
--- start ---
test child forked, pid 304211
failed to open /tmp/perf.data.Btf: No such file or directory
test child finished with 0
---- end ----
probe libc's inet_pton & backtrace it with ping: Ok
Fix this by adding check for presence of perf.data file
before proceeding with "perf script".
With the patch changes, test reports fail correctly.
# ./perf test -v "probe libc's inet_pton & backtrace it with ping"
81: probe libc's inet_pton & backtrace it with ping :
--- start ---
test child forked, pid 304358
FAIL: perf record failed to create "/tmp/perf.data.Uoi"
test child finished with -1
---- end ----
probe libc's inet_pton & backtrace it with ping: FAILED!
Signed-off-by: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Disha Goel <disgoel@linux.ibm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nageswara R Sastry <rnsastry@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: linuxppc-dev@lists.ozlabs.org
Link: http://lore.kernel.org/r/20230201180421.59640-1-atrajeev@linux.vnet.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
When it processes AUXTRACE_INFO, it calls to auxtrace_queue_data() to
collect AUXTRACE data first. That won't work with pipe since it needs
lseek() to read the scattered aux data.
$ perf record -o- -e intel_pt// true | perf report -i- --itrace=i100
# To display the perf.data header info, please use --header/--header-only options.
#
0x4118 [0xa0]: failed to process type: 70
Error:
failed to process sample
For the pipe mode, it can handle the aux data as it gets. But there's
no guarantee it can get the aux data in time. So the following warning
will be shown at the beginning:
WARNING: Intel PT with pipe mode is not recommended.
The output cannot relied upon. In particular,
time stamps and the order of events may be incorrect.
Fixes: dbd134322e ("perf intel-pt: Add support for decoding AUX area samples")
Reviewed-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: James Clark <james.clark@arm.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: https://lore.kernel.org/r/20230131023350.1903992-3-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Add a sub-test to verify that KVM stuffs the APIC_ID when userspace forces
a transition from x2APIC to xAPIC without first disabling the APIC. Such
a transition is architecturally disallowed (WRMSR will #GP), but needs to
be handled by KVM to allow userspace to emulate RESET (ignoring that
userspace should also stuff local APIC state on RESET).
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Link: https://lore.kernel.org/r/20230109130605.2013555-3-eesposit@redhat.com
Co-developed-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
The ifname char pointer is taken directly from the command line
as input and the string is copied directly into struct ifreq
via strcpy. This makes it easy to corrupt other members of ifreq
and generally do stack overflows.
Most often the ioctl will fail with:
./xdp_hw_metadata: ioctl(SIOCETHTOOL): Bad address
As people will likely copy-paste code for getting NIC queue
channels (rxq_num) and enabling HW timestamping (hwtstamp_ioctl)
lets make this code a bit more secure by using strncpy.
Fixes: 297a3f1241 ("selftests/bpf: Simple program to dump XDP RX metadata")
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/bpf/167527272543.937063.16993147790832546209.stgit@firesoul
The glibc error reporting function error():
void error(int status, int errnum, const char *format, ...);
The status argument should be a positive value between 0-255 as it
is passed over to the exit(3) function as the value as the shell exit
status. The least significant byte of status (i.e., status & 0xFF) is
returned to the shell parent.
Fix this by using 1 instead of -1. As 1 corresponds to C standard
constant EXIT_FAILURE.
Fixes: 297a3f1241 ("selftests/bpf: Simple program to dump XDP RX metadata")
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/bpf/167527272038.937063.9137108142012298120.stgit@firesoul
Using xdp_hw_metadata I experince Segmentation fault after seeing
"detaching bpf program....".
On my system the segfault happened when accessing bpf_obj->skeleton
in xdp_hw_metadata__destroy(bpf_obj) call. That doesn't make any sense
as this memory have not been freed by program at this point in time.
Prior to calling xdp_hw_metadata__destroy(bpf_obj) the function
close_xsk() is called for each RX-queue xsk. The real bug lays
in close_xsk() that unmap via munmap() the wrong memory pointer.
The call xsk_umem__delete(xsk->umem) will free xsk->umem, thus
the call to munmap(xsk->umem, UMEM_SIZE) will have unpredictable
behavior. And man page explain subsequent references to these
pages will generate SIGSEGV.
Unmapping xsk->umem_area instead removes the segfault.
Fixes: 297a3f1241 ("selftests/bpf: Simple program to dump XDP RX metadata")
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/bpf/167527271533.937063.5717065138099679142.stgit@firesoul
The AF_XDP userspace part of xdp_hw_metadata see non-zero as a signal of
the availability of rx_timestamp and rx_hash in data_meta area. The
kernel-side BPF-prog code doesn't initialize these members when kernel
returns an error e.g. -EOPNOTSUPP. This memory area is not guaranteed to
be zeroed, and can contain garbage/previous values, which will be read
and interpreted by AF_XDP userspace side.
Tested this on different drivers. The experiences are that for most
packets they will have zeroed this data_meta area, but occasionally it
will contain garbage data.
Example of failure tested on ixgbe:
poll: 1 (0)
xsk_ring_cons__peek: 1
0x18ec788: rx_desc[0]->addr=100000000008000 addr=8100 comp_addr=8000
rx_hash: 3697961069
rx_timestamp: 9024981991734834796 (sec:9024981991.7348)
0x18ec788: complete idx=8 addr=8000
Converting to date:
date -d @9024981991
2255-12-28T20:26:31 CET
I choose a simple fix in this patch. When kfunc fails or isn't supported
assign zero to the corresponding struct meta value.
It's up to the individual BPF-programmer to do something smarter e.g.
that fits their use-case, like getting a software timestamp and marking
a flag that gives the type of timestamp.
Fixes: 297a3f1241 ("selftests/bpf: Simple program to dump XDP RX metadata")
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/bpf/167527271027.937063.5177725618616476592.stgit@firesoul
The function close_xsk() unmap via munmap() the wrong memory pointer.
The call xsk_umem__delete(xsk->umem) have already freed xsk->umem.
Thus the call to munmap(xsk->umem, UMEM_SIZE) will have unpredictable
behavior that can lead to Segmentation fault elsewhere, as man page
explain subsequent references to these pages will generate SIGSEGV.
Fixes: e2a46d54d7 ("selftests/bpf: Verify xdp_metadata xdp->af_xdp path")
Reported-by: Martin KaFai Lau <martin.lau@kernel.org>
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/bpf/167527517464.938135.13750760520577765269.stgit@firesoul
kfuncs are allowed to be static, or not use one or more of their
arguments. For example, bpf_xdp_metadata_rx_hash() in net/core/xdp.c is
meant to be implemented by drivers, with the default implementation just
returning -EOPNOTSUPP. As described in [0], such kfuncs can have their
arguments elided, which can cause BTF encoding to be skipped. The new
__bpf_kfunc macro should address this, and this patch adds a selftest
which verifies that a static kfunc with at least one unused argument can
still be encoded and invoked by a BPF program.
Signed-off-by: David Vernet <void@manifault.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20230201173016.342758-5-void@manifault.com
Hyper-V extended hypercalls by default exit to userspace. Verify
userspace gets the call, update the result and then verify in guest
correct result is received.
Add KVM_EXIT_HYPERV to list of "known" hypercalls so errors generate
pretty strings.
Signed-off-by: Vipin Sharma <vipinsh@google.com>
Reviewed-by: David Matlack <dmatlack@google.com>
Link: https://lore.kernel.org/r/20221212183720.4062037-14-vipinsh@google.com
[sean: add KVM_EXIT_HYPERV to exit_reasons_known]
Signed-off-by: Sean Christopherson <seanjc@google.com>
In copy_bytes(), it reads the data from the (input) fd and writes it to
the output file. But it does with the read(2) unconditionally which
caused a problem of mixing buffered vs unbuffered I/O together.
You can see the problem when using pipes.
$ perf record -e intel_pt// -o- true | perf inject -b > /dev/null
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.000 MB - ]
0x45c0 [0x30]: failed to process type: 71
It should use perf_data__read() to honor the 'use_stdio' setting.
Fixes: 601366678c ("perf data: Allow to use stdio functions for pipe mode")
Reviewed-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: James Clark <james.clark@arm.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: https://lore.kernel.org/r/20230131023350.1903992-2-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Pull virtio fixes from Michael Tsirkin:
"Just small bugfixes all over the place"
* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
vdpa: ifcvf: Do proper cleanup if IFCVF init fails
vhost-scsi: unbreak any layout for response
tools/virtio: fix the vringh test for virtio ring changes
vhost/net: Clear the pending messages when the backend is removed
The current signal handling tests for SME do not account for the fact that
unlike SVE all SME vector lengths are optional so we can't guarantee that
we will encounter the minimum possible VL, they will hang enumerating VLs
on such systems. Abort enumeration when we find the lowest VL in the newly
added ssve_za_regs test.
Fixes: bc69da5ff0 ("kselftest/arm64: Verify simultaneous SSVE and ZA context generation")
Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230131-arm64-kselftest-sig-sme-no-128-v1-2-d47c13dc8e1e@kernel.org
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
By using calloc() instead of malloc() in a loop, libc does not have to
keep around bookkeeping information for each single structure.
This reduces maximum memory usage while processing vmlinux.o from
3153325 KB to 3035668 KB (-3.7%) on my notebooks "localmodconfig".
Note this introduces memory leaks, because some additional structs get
added to the lists later after reading the symbols and sections from the
original object. Luckily we don't really care about memory leaks in
objtool.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://lore.kernel.org/r/20221216-objtool-memory-v2-3-17968f85a464@weissschuh.net
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
HOSTCC is always wanted when building objtool. Setting CC to HOSTCC
happens after tools/scripts/Makefile.include is included, meaning
flags (like CFLAGS) are set assuming say CC is gcc, but then it can be
later set to HOSTCC which may be clang. tools/scripts/Makefile.include
is needed for host set up and common macros in objtool's
Makefile. Rather than override the CC variable to HOSTCC, just pass CC
as HOSTCC to the sub-makes of Makefile.build, the libsubcmd builds and
also to the linkage step.
Signed-off-by: Ian Rogers <irogers@google.com>
Link: https://lore.kernel.org/r/20230126190606.40739-4-irogers@google.com
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
