Add a target module and livepatch pair that verify module function
patching via a proc entry. Two test cases cover both the
klp_enable_patch path (target loaded before livepatch) and the
klp_module_coming path (livepatch loaded before target).
Signed-off-by: Pablo Alessandro Santos Hugen <[email protected]>
Acked-by: Miroslav Benes <[email protected]>
Acked-by: Joe Lawrence <[email protected]>
Reviewed-by: Petr Mladek <[email protected]>
Tested-by: Petr Mladek <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Petr Mladek <[email protected]>
Since v4.1 kernel, a new interface for ftrace called "tracefs" was
introduced, which is usually mounted in /sys/kernel/tracing. Therefore,
tracing files can now be accessed via either the legacy path
/sys/kernel/debug/tracing or the newer path /sys/kernel/tracing.
Signed-off-by: Fushuai Wang <[email protected]>
Reviewed-by: Petr Mladek <[email protected]>
Tested-by: Petr Mladek <[email protected]>
Signed-off-by: Petr Mladek <[email protected]>
Pull livepatching updates from Petr Mladek:
- Add a selftest for tracing of a livepatched function
- Skip a selftest when kprobes are not using ftrace
- Some documentation clean up
* tag 'livepatching-for-6.15' of git://git.kernel.org/pub/scm/linux/kernel/git/livepatching/livepatching:
selftests: livepatch: test if ftrace can trace a livepatched function
selftests: livepatch: add new ftrace helpers functions
selftest/livepatch: Only run test-kprobe with CONFIG_KPROBES_ON_FTRACE
docs: livepatch: move text out of code block
livepatch: Add comment to clarify klp_add_nops()
The test proves that a function that is being kprobed and uses a
post_handler cannot be livepatched.
Only one ftrace_ops with FTRACE_OPS_FL_IPMODIFY set may be registered
to any given function at a time.
Note that the conflicting kprobe could not be created using the
tracefs interface, see Documentation/trace/kprobetrace.rst.
This interface uses only the pre_handler(), see alloc_trace_kprobe().
But FTRACE_OPS_FL_IPMODIFY is used only when the kprobe is using a
post_handler, see arm_kprobe_ftrace().
Signed-off-by: Michael Vetter <[email protected]>
Reviewed-by: Miroslav Benes <[email protected]>
Reviewed-by: Joe Lawrence <[email protected]>
Tested-by: Marcos Paulo de Souza <[email protected]>
Reviewed-by: Marcos Paulo de Souza <[email protected]>
Reviewed-by: Petr Mladek <[email protected]>
Tested-by: Petr Mladek <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Petr Mladek <[email protected]>
Pull kselftest update from Shuah Khan:
- livepatch restructuring to move the module out of lib to be built as
a out-of-tree modules during kselftest build. This makes it easier
change, debug and rebuild the tests by running make on the
selftests/livepatch directory, which is not currently possible since
the modules on lib/livepatch are build and installed using the main
makefile modules target.
- livepatch restructuring fixes for problems found by kernel test
robot. The change skips the test if kernel-devel isn't installed
(default value of KDIR), or if KDIR variable passed doesn't exists.
- resctrl test restructuring and new non-contiguous CBMs CAT test
- new ktap_helpers to print diagnostic messages, pass/fail tests based
on exit code, abort test, and finish the test.
- a new test verify power supply properties.
- a new ftrace to exercise function tracer across cpu hotplug.
- timeout increase for mqueue test to allow the test to run on i3.metal
AWS instances.
- minor spelling corrections in several tests.
- missing gitignore files and changes to existing gitignore files.
* tag 'linux_kselftest-next-6.9-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest: (57 commits)
kselftest: Add basic test for probing the rust sample modules
selftests: lib.mk: Do not process TEST_GEN_MODS_DIR
selftests: livepatch: Avoid running the tests if kernel-devel is missing
selftests: livepatch: Add initial .gitignore
selftests/resctrl: Add non-contiguous CBMs CAT test
selftests/resctrl: Add resource_info_file_exists()
selftests/resctrl: Split validate_resctrl_feature_request()
selftests/resctrl: Add a helper for the non-contiguous test
selftests/resctrl: Add test groups and name L3 CAT test L3_CAT
selftests: sched: Fix spelling mistake "hiearchy" -> "hierarchy"
selftests/mqueue: Set timeout to 180 seconds
selftests/ftrace: Add test to exercize function tracer across cpu hotplug
selftest: ftrace: fix minor typo in log
selftests: thermal: intel: workload_hint: add missing gitignore
selftests: thermal: intel: power_floor: add missing gitignore
selftests: uevent: add missing gitignore
selftests: Add test to verify power supply properties
selftests: ktap_helpers: Add a helper to finish the test
selftests: ktap_helpers: Add a helper to abort the test
selftests: ktap_helpers: Add helper to pass/fail test based on exit code
...
Ignore the binary used to test livepatching a syscall.
Signed-off-by: Marcos Paulo de Souza <[email protected]>
Signed-off-by: Shuah Khan <[email protected]>
The livepatching kselftests rely on comparing expected vs. observed
dmesg output. After each test, new dmesg entries are determined by the
'comm' utility comparing a saved, pre-test copy of dmesg to post-test
dmesg output.
Alexander reports that the 'comm --nocheck-order -13' invocation used by
the tests can be confused when dmesg entry timestamps vary in magnitude
(ie, "[ 98.820331]" vs. "[ 100.031067]"), in which case, additional
messages are reported as new. The unexpected entries then spoil the
test results.
Instead of relying on 'comm' or 'diff' to determine new testing dmesg
entries, refactor the code:
- pre-test : log a unique canary dmesg entry
- test : run tests, log messages
- post-test : filter dmesg starting from pre-test message
Reported-by: Alexander Gordeev <[email protected]>
Closes: https://lore.kernel.org/live-patching/[email protected]/
Signed-off-by: Joe Lawrence <[email protected]>
Acked-by: Alexander Gordeev <[email protected]>
Tested-by: Marcos Paulo de Souza <[email protected]>
Reviewed-by: Marcos Paulo de Souza <[email protected]>
Acked-by: Miroslav Benes <[email protected]>
Signed-off-by: Shuah Khan <[email protected]>
The test proves that a syscall can be livepatched. It is interesting
because syscalls are called a tricky way. Also the process gets
livepatched either when sleeping in the userspace or when entering
or leaving the kernel space.
The livepatch is a bit tricky:
1. The syscall function name is architecture specific. Also
ARCH_HAS_SYSCALL_WRAPPER must be taken in account.
2. The syscall must stay working the same way for other processes
on the system. It is solved by decrementing a counter only
for PIDs of the test processes. It means that the test processes
has to call the livepatched syscall at least once.
The test creates one userspace process per online cpu. The processes
are calling getpid in a busy loop. The intention is to create random
locations when the livepatch gets enabled. Nothing is guarantted.
The magic is in the randomness.
Reviewed-by: Joe Lawrence <[email protected]>
Reviewed-by: Petr Mladek <[email protected]>
Signed-off-by: Marcos Paulo de Souza <[email protected]>
Signed-off-by: Shuah Khan <[email protected]>
The modules are being moved from lib/livepatch to
tools/testing/selftests/livepatch/test_modules.
This code moving will allow writing more complex tests, like for example an
userspace C code that will call a livepatched kernel function.
The modules are now built as out-of-tree
modules, but being part of the kernel source means they will be maintained.
Another advantage of the code moving is to be able to easily change,
debug and rebuild the tests by running make on the selftests/livepatch
directory, which is not currently possible since the modules on
lib/livepatch are build and installed using the "modules" target.
The current approach also keeps the ability to execute the tests manually
by executing the scripts inside selftests/livepatch directory, as it's
currently supported. If the modules are modified, they needed to be
rebuilt before running the scripts though.
The modules are built before running the selftests when using the
kselftest invocations:
make kselftest TARGETS=livepatch
or
make -C tools/testing/selftests/livepatch run_tests
Having the modules being built as out-of-modules requires changing the
currently used 'modprobe' by 'insmod' and adapt the test scripts that
check for the kernel message buffer.
Now it is possible to only compile the modules by running:
make -C tools/testing/selftests/livepatch/
This way the test modules and other test program can be built in order
to be packaged if so desired.
As there aren't any modules being built on lib/livepatch, remove the
TEST_LIVEPATCH Kconfig and it's references.
Note: "make gen_tar" packages the pre-built binaries into the tarball.
It means that it will store the test modules pre-built for
the kernel running on the build host.
Note that these modules need not binary compatible with
the kernel built from the same sources. But the same
is true for other packaged selftest binaries.
The entire kernel sources are needed for rebuilding
the selftests on another system.
Reviewed-by: Joe Lawrence <[email protected]>
Reviewed-by: Petr Mladek <[email protected]>
Signed-off-by: Marcos Paulo de Souza <[email protected]>
Acked-by: Alexander Gordeev <[email protected]>
Signed-off-by: Shuah Khan <[email protected]>
livepatch has a set of selftests that are used to validate the behavior of
the livepatching subsystem. One of the testcases in the livepatch
testsuite is test-ftrace.sh, which among other things, validates that
livepatching gracefully fails when ftrace is disabled. In the event that
ftrace cannot be disabled using 'sysctl kernel.ftrace_enabled=0', the test
will fail later due to it unexpectedly successfully loading the
test_klp_livepatch module.
While the livepatch selftests are careful to remove any of the livepatch
test modules between testcases to avoid this situation, ftrace may still
fail to be disabled if another trace is active on the system that was
enabled with FTRACE_OPS_FL_PERMANENT. For example, any active BPF programs
that use trampolines will cause this test to fail due to the trampoline
being implemented with register_ftrace_direct(). The following is an
example of such a trace:
tcp_drop (1) R I D tramp: ftrace_regs_caller+0x0/0x58
(call_direct_funcs+0x0/0x30)
direct-->bpf_trampoline_6442550536_0+0x0/0x1000
In order to make the test more resilient to system state that is out of its
control, this patch updates set_ftrace_enabled() to detect sysctl failures,
and skip the testrun when appropriate.
Suggested-by: Petr Mladek <[email protected]>
Signed-off-by: David Vernet <[email protected]>
Acked-by: Miroslav Benes <[email protected]>
Reviewed-by: Petr Mladek <[email protected]>
Tested-by: Petr Mladek <[email protected]>
Acked-by: Joe Lawrence <[email protected]>
Signed-off-by: Petr Mladek <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
check_result() uses "comm" to check expected results of selftests output
in dmesg. Everything works fine if timestamps in dmesg are unique. If
not, like in this example
[ 86.844422] test_klp_callbacks_demo: pre_unpatch_callback: test_klp_callbacks_mod -> [MODULE_STATE_LIVE] Normal state
[ 86.844422] livepatch: 'test_klp_callbacks_demo': starting unpatching transition
, "comm" fails with "comm: file 2 is not in sorted order". Suppress the
order checking with --nocheck-order option.
Fixes: 2f3f651f37 ("selftests/livepatch: Use "comm" instead of "diff" for dmesg")
Signed-off-by: Miroslav Benes <[email protected]>
Acked-by: Joe Lawrence <[email protected]>
Signed-off-by: Jiri Kosina <[email protected]>
With procfs v3.3.16, the sysctl command doesn't print the set key and
value on error. This change breaks livepatch selftest test-ftrace.sh,
that tests the interaction of sysctl ftrace_enabled:
Make it work with all sysctl versions using '-q' option.
Explicitly print the final status on success so that it can be verified
in the log. The error message is enough on failure.
Reported-by: Kamalesh Babulal <[email protected]>
Signed-off-by: Petr Mladek <[email protected]>
Reviewed-by: Kamalesh Babulal <[email protected]>
Reviewed-by: Joe Lawrence <[email protected]>
Acked-by: Miroslav Benes <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
BusyBox diff doesn't support the GNU diff '--LTYPE-line-format' options
that were used in the selftests to filter older kernel log messages from
dmesg output.
Use "comm" which is more available in smaller boot environments.
Reported-by: Naresh Kamboju <[email protected]>
Signed-off-by: Joe Lawrence <[email protected]>
Reviewed-by: Yannick Cote <[email protected]>
Reviewed-by: Kamalesh Babulal <[email protected]>
Signed-off-by: Petr Mladek <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
The livepatch selftests currently grep on "taints" to filter out
"tainting kernel with TAINT_LIVEPATCH" messages which may be logged when
loading livepatch modules.
Further filter the log to drop "loading out-of-tree module taints
kernel" in the rare case the klp_test modules have been built
out-of-tree.
Look for the longer "taints kernel" or "tainting kernel" strings to
avoid inadvertent partial matching.
Signed-off-by: Joe Lawrence <[email protected]>
Reviewed-by: Kamalesh Babulal <[email protected]>
Reviewed-by: Yannick Cote <[email protected]>
Reviewed-by: Petr Mladek <[email protected]>
Acked-by: Miroslav Benes <[email protected]>
Signed-off-by: Petr Mladek <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
This change makes the test feel more familiar with narrowing to a
typical usage by operating on a number of identical structure instances
and populating the same two new shadow variables symmetrically while
keeping the same testing and verification criteria for the extra
variables.
Signed-off-by: Yannick Cote <[email protected]>
Reviewed-by: Kamalesh Babulal <[email protected]>
Reviewed-by: Petr Mladek <[email protected]>
Acked-by: Miroslav Benes <[email protected]>
Acked-by: Joe Lawrence <[email protected]>
Signed-off-by: Petr Mladek <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
The test-klp-callbacks script includes a few tests which rely on kernel
task timings that may not always execute as expected under system load.
These may generate out of sequence kernel log messages that result in
test failure.
Instead of using sleep timing windows to orchestrate these tests, add a
block_transition module parameter to communicate the test purpose and
utilize flush_queue() to serialize the test module's task output.
Signed-off-by: Joe Lawrence <[email protected]>
Reviewed-by: Petr Mladek <[email protected]>
Acked-by: Miroslav Benes <[email protected]>
Signed-off-by: Petr Mladek <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Commit 852c8cbf34 ("selftests/kselftest/runner.sh: Add 45 second
timeout per test") added a 45 second timeout for tests, and also added
a way for tests to customise the timeout via a settings file.
For example the ftrace tests take multiple minutes to run, so they
were given longer in commit b43e78f65b ("tracing/selftests: Turn off
timeout setting").
This works when the tests are run from the source tree. However if the
tests are installed with "make -C tools/testing/selftests install",
the settings files are not copied into the install directory. When the
tests are then run from the install directory the longer timeouts are
not applied and the tests timeout incorrectly.
So add the settings files to TEST_FILES of the appropriate Makefiles
to cause the settings files to be installed using the existing install
logic.
Fixes: 852c8cbf34 ("selftests/kselftest/runner.sh: Add 45 second timeout per test")
Signed-off-by: Michael Ellerman <[email protected]>
Signed-off-by: Shuah Khan <[email protected]>
Commit 35c9e74cff ("selftests/livepatch: Make dynamic debug setup and
restore generic") introduced setup_config() to set up the environment
for each test. It superseded set_dynamic_debug(). README still mentions
set_dynamic_debug(), so update it to setup_config() which should be used
now in every test.
Signed-off-by: Miroslav Benes <[email protected]>
Reviewed-by: Kamalesh Babulal <[email protected]>
Signed-off-by: Shuah Khan <[email protected]>
livepatch test configures the system and debug environment to run
tests. Some of these actions fail without root access and test
dumps several permission denied messages before it exits.
Fix test-state.sh to call setup_config instead of set_dynamic_debug
as suggested by Petr Mladek <[email protected]>
Fix it to check root uid and exit with skip code instead.
Signed-off-by: Shuah Khan <[email protected]>
Acked-by: Joe Lawrence <[email protected]>
Reviewed-by: Petr Mladek <[email protected]>
Signed-off-by: Shuah Khan <[email protected]>
Pull tracing updates from Steven Rostedt:
"New tracing features:
- New PERMANENT flag to ftrace_ops when attaching a callback to a
function.
As /proc/sys/kernel/ftrace_enabled when set to zero will disable
all attached callbacks in ftrace, this has a detrimental impact on
live kernel tracing, as it disables all that it patched. If a
ftrace_ops is registered to ftrace with the PERMANENT flag set, it
will prevent ftrace_enabled from being disabled, and if
ftrace_enabled is already disabled, it will prevent a ftrace_ops
with PREMANENT flag set from being registered.
- New register_ftrace_direct().
As eBPF would like to register its own trampolines to be called by
the ftrace nop locations directly, without going through the ftrace
trampoline, this function has been added. This allows for eBPF
trampolines to live along side of ftrace, perf, kprobe and live
patching. It also utilizes the ftrace enabled_functions file that
keeps track of functions that have been modified in the kernel, to
allow for security auditing.
- Allow for kernel internal use of ftrace instances.
Subsystems in the kernel can now create and destroy their own
tracing instances which allows them to have their own tracing
buffer, and be able to record events without worrying about other
users from writing over their data.
- New seq_buf_hex_dump() that lets users use the hex_dump() in their
seq_buf usage.
- Notifications now added to tracing_max_latency to allow user space
to know when a new max latency is hit by one of the latency
tracers.
- Wider spread use of generic compare operations for use of bsearch
and friends.
- More synthetic event fields may be defined (32 up from 16)
- Use of xarray for architectures with sparse system calls, for the
system call trace events.
This along with small clean ups and fixes"
* tag 'trace-v5.5' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace: (51 commits)
tracing: Enable syscall optimization for MIPS
tracing: Use xarray for syscall trace events
tracing: Sample module to demonstrate kernel access to Ftrace instances.
tracing: Adding new functions for kernel access to Ftrace instances
tracing: Fix Kconfig indentation
ring-buffer: Fix typos in function ring_buffer_producer
ftrace: Use BIT() macro
ftrace: Return ENOTSUPP when DYNAMIC_FTRACE_WITH_DIRECT_CALLS is not configured
ftrace: Rename ftrace_graph_stub to ftrace_stub_graph
ftrace: Add a helper function to modify_ftrace_direct() to allow arch optimization
ftrace: Add helper find_direct_entry() to consolidate code
ftrace: Add another check for match in register_ftrace_direct()
ftrace: Fix accounting bug with direct->count in register_ftrace_direct()
ftrace/selftests: Fix spelling mistake "wakeing" -> "waking"
tracing: Increase SYNTH_FIELDS_MAX for synthetic_events
ftrace/samples: Add a sample module that implements modify_ftrace_direct()
ftrace: Add modify_ftrace_direct()
tracing: Add missing "inline" in stub function of latency_fsnotify()
tracing: Remove stray tab in TRACE_EVAL_MAP_FILE's help text
tracing: Use seq_buf_hex_dump() to dump buffers
...
Pull livepatching updates from Petr Mladek:
- New API to track system state changes done be livepatch callbacks. It
helps to maintain compatibility between livepatches.
- Update Kconfig help text. ORC is another reliable unwinder.
- Disable generic selftest timeout. Livepatch selftests have their own
per-operation fine-grained timeouts.
* tag 'livepatching-for-5.5' of git://git.kernel.org/pub/scm/linux/kernel/git/livepatching/livepatching:
x86/stacktrace: update kconfig help text for reliable unwinders
livepatch: Selftests of the API for tracking system state changes
livepatch: Documentation of the new API for tracking system state changes
livepatch: Allow to distinguish different version of system state changes
livepatch: Basic API to track system state changes
livepatch: Keep replaced patches until post_patch callback is called
selftests/livepatch: Disable the timeout
Since livepatching depends upon ftrace handlers to implement "patched"
code functionality, verify that the ftrace_enabled sysctl value
interacts with livepatch registration as expected. At the same time,
ensure that ftrace_enabled is set and part of the test environment
configuration that is saved and restored when running the selftests.
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Joe Lawrence <[email protected]>
Signed-off-by: Miroslav Benes <[email protected]>
Signed-off-by: Steven Rostedt (VMware) <[email protected]>
Livepatch selftests currently save the current dynamic debug config and
tweak it for the selftests. The config is restored at the end. Make the
infrastructure generic, so that more variables can be saved and
restored.
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Joe Lawrence <[email protected]>
Signed-off-by: Miroslav Benes <[email protected]>
Signed-off-by: Steven Rostedt (VMware) <[email protected]>
Commit 852c8cbf34 ("selftests/kselftest/runner.sh: Add 45 second
timeout per test") introduced a timeout per test. Livepatch tests could
run longer than 45 seconds, especially on slower machines. They do not
hang and they detect if something goes awry with internal accounting.
Better than looking for an arbitrary value, just disable the timeout for
livepatch selftests.
Signed-off-by: Miroslav Benes <[email protected]>
Reviewed-by: Joe Lawrence <[email protected]>
Signed-off-by: Jiri Kosina <[email protected]>
Petr Mladek