Linear temporal logic monitors for RT application
Subsystems (11)
v4 cover Linear temporal logic monitors for RT application
Real-time applications may have design flaws causing them to have unexpected latency. For example, the applications may raise page faults, or may be blocked trying to take a mutex without priority inheritance.
However, while attempting to implement DA monitors for these real-time rules, deterministic automaton is found to be inappropriate as the specification language. The automaton is complicated, hard to understand, and error-prone.
For these cases, linear temporal logic is found to be more suitable. The LTL is more concise and intuitive.
This series adds support for LTL RV monitor, and use it to implement two monitors for reporting problems with real-time tasks.
Patch 1-12 cleanup and prepare the RV code for the integration of LTL monitors.
Patch 13 adds support for LTL monitors.
Patch 14 adds the container monitor "rtapp". This encapsulates the sub-monitors for real-time.
Patch 15-18 prepares the pagefault tracepoints, so that patch 19 can add the monitor which watches real-time tasks doing page faults.
Patch 20 adds the "sleep" monitor: it detects potential undesirable latency with real-time threads.
Patch 21 adds documentation on the new monitors.
Patch 22 allows the number of per-task monitors to be configurable, so that the two new monitors can be enabled simultaneously.
v3->v4 https://lore.kernel.org/lkml/[email protected]
- support deadline tasks
- rtapp_sleep: use sched_pi_setprio tracepoint instead of contention tracepoints for BLOCK_ON_RT_MUTEX, so that proxy lock is covered.
- fix the scripts generating an "slightly" incorrect verification automaton
- makes rtapp monitor depends on RV_PER_TASK_MONITORS >= 2
- make the event tracepoint output a bit more readable
- some documentation's format fixes
v2->v3 https://lore.kernel.org/lkml/[email protected]/
- fix a problem with sleep monitor's specification (around KTHREAD_SHOULD_STOP)
- merge the patches that move the dot2k/rvgen scripts around
- pull panic/printk changes into separate patches
- fixup some build errors
- fixup monitor's init function return code
- fix some flake8 warnings with the scripts
- add some references to LTL documentation
- fixup some mistakes with rtapp documentation
- fixup capitalization mistake with monitor_synthesis.rst
- remove the now-redundant macro RV_PER_TASK_MONITORS
v1->v2 https://lore.kernel.org/lkml/[email protected]/
- Integrate the LTL scripts into the existing dot2k tool, taking advantage of the existing monitor generation scripts.
- Switch the struct ltl_monitor to use bitmap instead of an array, to optimize memory usage.
- Correct the generated code to be non-deterministic state machine, instead of deterministic state machine
- Put common code for all LTL monitors into a single file (include/rv/ltl_monitor.h), reducing code duplication
- Change the LTL monitors to make user of container. Add a bug fix to container while at it.
- Make the number of per-task monitor configurable
Nam Cao (22):
rv: Add #undef TRACE_INCLUDE_FILE
printk: Make vprintk_deferred() public
panic: Add vpanic()
rv: Let the reactors take care of buffers verification/dot2k: Make a separate dot2k_templates/Kconfig_container verification/dot2k: Remove __buff_to_string() verification/dot2k: Replace is_container() hack with subparsers
rv: rename CONFIG_DA_MON_EVENTS to CONFIG_RV_MON_EVENTS verification/dot2k: Prepare the frontend for LTL inclusion Documentation/rv: Prepare monitor synthesis document for LTL inclusion verification/rvgen: Restructure the templates files verification/rvgen: Restructure the classes to prepare for LTL
inclusion
rv: Add support for LTL monitors
rv: Add rtapp container monitor
x86/tracing: Remove redundant trace_pagefault_key x86/tracing: Move page fault trace points to generic
arm64: mm: Add page fault trace points
riscv: mm: Add page fault trace points
rv: Add rtapp_pagefault monitor
rv: Add rtapp_sleep monitor
rv: Add documentation for rtapp monitor
rv: Allow to configure the number of per-task monitor
.../trace/rv/da_monitor_synthesis.rst | 147 -----
Documentation/trace/rv/index.rst | 4 +-
.../trace/rv/linear_temporal_logic.rst | 122 ++++
Documentation/trace/rv/monitor_rtapp.rst | 107 ++++
Documentation/trace/rv/monitor_synthesis.rst | 256 ++++++++
arch/arm64/mm/fault.c | 8 +
arch/riscv/mm/fault.c | 8 +
arch/x86/include/asm/trace/common.h | 12 -
arch/x86/include/asm/trace/irq_vectors.h | 1 -
arch/x86/kernel/Makefile | 1 -
arch/x86/kernel/tracepoint.c | 21 -
arch/x86/mm/fault.c | 5 +-
include/linux/panic.h | 3 +
include/linux/printk.h | 5 +
include/linux/rv.h | 74 ++-
include/linux/sched.h | 8 +-
include/rv/da_monitor.h | 45 +-
include/rv/ltl_monitor.h | 184 ++++++
.../trace/events}/exceptions.h | 27 +-
kernel/fork.c | 5 +-
kernel/panic.c | 17 +-
kernel/printk/internal.h | 1 -
kernel/trace/rv/Kconfig | 27 +-
kernel/trace/rv/Makefile | 3 +
kernel/trace/rv/monitors/pagefault/Kconfig | 11 +
.../trace/rv/monitors/pagefault/pagefault.c | 82 +++
.../trace/rv/monitors/pagefault/pagefault.h | 57 ++
.../rv/monitors/pagefault/pagefault_trace.h | 14 +
kernel/trace/rv/monitors/rtapp/Kconfig | 7 +
kernel/trace/rv/monitors/rtapp/rtapp.c | 33 ++
kernel/trace/rv/monitors/rtapp/rtapp.h | 3 +
kernel/trace/rv/monitors/sleep/Kconfig | 13 +
kernel/trace/rv/monitors/sleep/sleep.c | 201 +++++++
kernel/trace/rv/monitors/sleep/sleep.h | 287 +++++++++
kernel/trace/rv/monitors/sleep/sleep_trace.h | 14 +
kernel/trace/rv/reactor_panic.c | 8 +-
kernel/trace/rv/reactor_printk.c | 8 +-
kernel/trace/rv/rv.c | 10 +-
kernel/trace/rv/rv_reactors.c | 2 +-
kernel/trace/rv/rv_trace.h | 52 +-
tools/verification/dot2/Makefile | 26 -
tools/verification/dot2/dot2k | 53 --
tools/verification/models/rtapp/pagefault.ltl | 1 +
tools/verification/models/rtapp/sleep.ltl | 15 +
tools/verification/rvgen/.gitignore | 3 +
tools/verification/rvgen/Makefile | 27 +
tools/verification/rvgen/__main__.py | 67 +++
tools/verification/{dot2 => rvgen}/dot2c | 2 +-
.../{dot2 => rvgen/rvgen}/automata.py | 0
tools/verification/rvgen/rvgen/container.py | 22 +
.../{dot2 => rvgen/rvgen}/dot2c.py | 2 +-
tools/verification/rvgen/rvgen/dot2k.py | 129 ++++
.../dot2k.py => rvgen/rvgen/generator.py} | 249 ++------
tools/verification/rvgen/rvgen/ltl2ba.py | 558 ++++++++++++++++++
tools/verification/rvgen/rvgen/ltl2k.py | 245 ++++++++
.../rvgen/templates}/Kconfig | 0
.../rvgen/rvgen/templates/container/Kconfig | 5 +
.../rvgen/templates/container/main.c} | 0
.../rvgen/templates/container/main.h} | 0
.../rvgen/templates/dot2k}/main.c | 0
.../rvgen/templates/dot2k}/trace.h | 0
.../rvgen/rvgen/templates/ltl2k/main.c | 102 ++++
.../rvgen/rvgen/templates/ltl2k/trace.h | 14 +
63 files changed, 2863 insertions(+), 550 deletions(-)
delete mode 100644 Documentation/trace/rv/da_monitor_synthesis.rst
create mode 100644 Documentation/trace/rv/linear_temporal_logic.rst
create mode 100644 Documentation/trace/rv/monitor_rtapp.rst
create mode 100644 Documentation/trace/rv/monitor_synthesis.rst
delete mode 100644 arch/x86/include/asm/trace/common.h
delete mode 100644 arch/x86/kernel/tracepoint.c
create mode 100644 include/rv/ltl_monitor.h
rename {arch/x86/include/asm/trace => include/trace/events}/exceptions.h (55%)
create mode 100644 kernel/trace/rv/monitors/pagefault/Kconfig
create mode 100644 kernel/trace/rv/monitors/pagefault/pagefault.c
create mode 100644 kernel/trace/rv/monitors/pagefault/pagefault.h
create mode 100644 kernel/trace/rv/monitors/pagefault/pagefault_trace.h
create mode 100644 kernel/trace/rv/monitors/rtapp/Kconfig
create mode 100644 kernel/trace/rv/monitors/rtapp/rtapp.c
create mode 100644 kernel/trace/rv/monitors/rtapp/rtapp.h
create mode 100644 kernel/trace/rv/monitors/sleep/Kconfig
create mode 100644 kernel/trace/rv/monitors/sleep/sleep.c
create mode 100644 kernel/trace/rv/monitors/sleep/sleep.h
create mode 100644 kernel/trace/rv/monitors/sleep/sleep_trace.h
delete mode 100644 tools/verification/dot2/Makefile
delete mode 100644 tools/verification/dot2/dot2k
create mode 100644 tools/verification/models/rtapp/pagefault.ltl
create mode 100644 tools/verification/models/rtapp/sleep.ltl
create mode 100644 tools/verification/rvgen/.gitignore
create mode 100644 tools/verification/rvgen/Makefile
create mode 100644 tools/verification/rvgen/__main__.py
rename tools/verification/{dot2 => rvgen}/dot2c (97%)
rename tools/verification/{dot2 => rvgen/rvgen}/automata.py (100%)
create mode 100644 tools/verification/rvgen/rvgen/container.py
rename tools/verification/{dot2 => rvgen/rvgen}/dot2c.py (99%)
create mode 100644 tools/verification/rvgen/rvgen/dot2k.py
rename tools/verification/{dot2/dot2k.py => rvgen/rvgen/generator.py} (52%)
create mode 100644 tools/verification/rvgen/rvgen/ltl2ba.py
create mode 100644 tools/verification/rvgen/rvgen/ltl2k.py
rename tools/verification/{dot2/dot2k_templates => rvgen/rvgen/templates}/Kconfig (100%)
create mode 100644 tools/verification/rvgen/rvgen/templates/container/Kconfig
rename tools/verification/{dot2/dot2k_templates/main_container.c => rvgen/rvgen/templates/container/main.c} (100%)
rename tools/verification/{dot2/dot2k_templates/main_container.h => rvgen/rvgen/templates/container/main.h} (100%)
rename tools/verification/{dot2/dot2k_templates => rvgen/rvgen/templates/dot2k}/main.c (100%)
rename tools/verification/{dot2/dot2k_templates => rvgen/rvgen/templates/dot2k}/trace.h (100%)
create mode 100644 tools/verification/rvgen/rvgen/templates/ltl2k/main.c
create mode 100644 tools/verification/rvgen/rvgen/templates/ltl2k/trace.hSignature · 1 line
2.39.5
PATCH v4 1/22 Add #undef TRACE_INCLUDE_FILE
Without "#undef TRACE_INCLUDE_FILE", there could be a build error due to TRACE_INCLUDE_FILE being redefined. Therefore add it.
Also fix a typo while at it.
--- kernel/trace/rv/rv_trace.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
kernel/trace/rv/rv_trace.h +2 -1
@@ -129,8 +129,9 @@ DECLARE_EVENT_CLASS(error_da_monitor_id,#endif /* CONFIG_DA_MON_EVENTS_ID */ #endif /* _TRACE_RV_H */ -/* This part ust be outside protection */ +/* This part must be outside protection */ #undef TRACE_INCLUDE_PATH#define TRACE_INCLUDE_PATH . +#undef TRACE_INCLUDE_FILE#define TRACE_INCLUDE_FILE rv_trace#include <trace/define_trace.h>
Signature · 1 line
2.39.5
PATCH v4 2/22 printk: Make vprintk_deferred() public
In reply to Nam Caovprintk_deferred() is useful for implementing runtime verification reactors. Make it public.
---
Cc: John Ogness <[email protected]>
Cc: Sergey Senozhatsky <[email protected]>
--- include/linux/printk.h | 5 +++++ kernel/printk/internal.h | 1 - 2 files changed, 5 insertions(+), 1 deletion(-)
include/linux/printk.h +5 -0
@@ -154,6 +154,7 @@ int vprintk_emit(int facility, int level, asmlinkage __printf(1, 0) int vprintk(const char *fmt, va_list args); +__printf(1, 0) int vprintk_deferred(const char *fmt, va_list args); asmlinkage __printf(1, 2) __cold int _printk(const char *fmt, ...);
@@ -213,6 +214,10 @@ int vprintk(const char *s, va_list args) {return 0 ; } +__printf(1, 0) int vprintk_deferred(const char *fmt, va_list args) +{ +return 0 ; +} static inline __printf(1, 2) __cold int _printk(const char *s, ...) {
kernel/printk/internal.h +0 -1
@@ -71,7 +71,6 @@ int vprintk_store(int facility, int level, const char *fmt, va_list args); __printf(1, 0) int vprintk_default(const char *fmt, va_list args); -__printf(1, 0) int vprintk_deferred(const char *fmt, va_list args);void __printk_safe_enter (void );void __printk_safe_exit (void );
Signature · 1 line
2.39.5
PATCH v4 3/22 panic: Add vpanic()
In reply to Nam Caovpanic() is useful for implementing runtime verification reactors. Add it.
---
Cc: John Ogness <[email protected]>
Cc: Sergey Senozhatsky <[email protected]>
--- include/linux/panic.h | 3 +++ kernel/panic.c | 17 ++++++++++++----- 2 files changed, 15 insertions(+), 5 deletions(-)
include/linux/panic.h +3 -0
@@ -3,6 +3,7 @@#define _LINUX_PANIC_H#include <linux/compiler_attributes.h> +#include <linux/stdarg.h> #include <linux/types.h> struct pt_regs ;
@@ -10,6 +11,8 @@ struct pt_regs;extern long (*panic_blink)(int state); __printf(1, 2)void panic (const char *fmt, ...)__noreturn __cold ; +__printf(1, 0) +void vpanic (const char *fmt,va_list args)__noreturn __cold ;void nmi_panic (struct pt_regs *regs,const char *msg);void check_panic_on_warn (const char *origin);extern void oops_enter (void );
kernel/panic.c +12 -5
@@ -277,17 +277,16 @@ static void panic_other_cpus_shutdown(bool crash_kexec) }/** -* panic - halt the system +* vpanic - halt the system * @fmt: The text string to print * * Display a message, then perform cleanups. * * This function never returns. */ -void panic (const char *fmt, ...) +void vpanic (const char *fmt,va_list args) {static char buf[1024 ]; -va_list args;long i, i_next =0 , len;int state =0 ;int old_cpu, this_cpu;
@@ -338,9 +337,7 @@ void panic(const char *fmt, ...)console_verbose ();bust_spinlocks (1 ); -va_start (args, fmt); len =vscnprintf (buf,sizeof (buf), fmt, args); -va_end (args);if (len && buf[len -1 ] =='\n' ) buf[len -1 ] ='\0' ;
@@ -477,7 +474,17 @@ void panic(const char *fmt, ...)mdelay (PANIC_TIMER_STEP ); } } +EXPORT_SYMBOL (vpanic); +/* Identical to vpanic(), except it takes variadic arguments instead of va_list */ +void panic (const char *fmt, ...) +{ +va_list args; + +va_start (args, fmt); +vpanic (fmt, args); +va_end (args); +}EXPORT_SYMBOL (panic);#define TAINT_FLAG (taint, _c_true, _c_false, _module) \
Signature · 1 line
2.39.5
PATCH v4 4/22 Let the reactors take care of buffers
In reply to Nam CaoEach RV monitor has one static buffer to send to the reactors. If multiple errors are detected simultaneously, the one buffer could be overwritten.
Instead, leave it to the reactors to handle buffering.
---
Cc: Petr Mladek <[email protected]>
Cc: John Ogness <[email protected]>
Cc: Sergey Senozhatsky <[email protected]>
--- include/linux/rv.h | 9 +++++-- include/rv/da_monitor.h | 45 +++++++------------------------- kernel/trace/rv/reactor_panic.c | 8 ++++-- kernel/trace/rv/reactor_printk.c | 8 ++++-- kernel/trace/rv/rv_reactors.c | 2 +- 5 files changed, 30 insertions(+), 42 deletions(-)
include/linux/rv.h +7 -2
@@ -38,7 +38,7 @@ union rv_task_monitor {struct rv_reactor {const char *name;const char *description; -void (*react)(char *msg); + __printf(1, 2) void (*react)(const char *msg, ...); };#endif
@@ -50,7 +50,7 @@ struct rv_monitor {void (*disable)(void );void (*reset)(void );#ifdef CONFIG_RV_REACTORS -void (*react)(char *msg); + __printf(1, 2) void (*react)(const char *msg, ...);#endif };
@@ -64,6 +64,11 @@ void rv_put_task_monitor_slot(int slot);bool rv_reacting_on (void );int rv_unregister_reactor (struct rv_reactor *reactor);int rv_register_reactor (struct rv_reactor *reactor); +#else +static inline bool rv_reacting_on (void ) +{ +return false; +}#endif /* CONFIG_RV_REACTORS */ #endif /* CONFIG_RV */
include/rv/da_monitor.h +10 -35
71 lines not on this page · raw message
kernel/trace/rv/reactor_panic.c +6 -2
@@ -13,9 +13,13 @@#include <linux/init.h> #include <linux/rv.h> -static void rv_panic_reaction (char *msg) +__printf(1, 2) static void rv_panic_reaction(const char *msg, ...) { -panic (msg); +va_list args; + +va_start (args, msg); +vpanic (msg, args); +va_end (args); }static struct rv_reactor rv_panic = {
kernel/trace/rv/reactor_printk.c +6 -2
@@ -12,9 +12,13 @@#include <linux/init.h> #include <linux/rv.h> -static void rv_printk_reaction (char *msg) +__printf(1, 2) static void rv_printk_reaction(const char *msg, ...) { -printk_deferred (msg); +va_list args; + +va_start (args, msg); +vprintk_deferred (msg, args); +va_end (args); }static struct rv_reactor rv_printk = {
kernel/trace/rv/rv_reactors.c +1 -1
@@ -490,7 +490,7 @@ void reactor_cleanup_monitor(struct rv_monitor_def *mdef)/* * Nop reactor register */ -static void rv_nop_reaction (char *msg) +__printf(1, 2) static void rv_nop_reaction(const char *msg, ...) { }
Signature · 1 line
2.39.5
PATCH v4 5/22 verification/dot2k: Make a separate dot2k_templates/Kconfig_container
In reply to Nam CaoA generated container's Kconfig has an incorrect line:
select DA_MON_EVENTS_IMPLICIT
This is due to container generation uses the same template Kconfig file as deterministic automaton monitor.
Therefore, make a separate Kconfig template for container which has only the necessaries for container.
---
Alternatively, we could also modify the Python scripts. I tried both and this solution seems cleaner.
--- tools/verification/dot2/dot2k.py | 3 ++- tools/verification/dot2/dot2k_templates/Kconfig_container | 5 +++++ 2 files changed, 7 insertions(+), 1 deletion(-) create mode 100644 tools/verification/dot2/dot2k_templates/Kconfig_container
tools/verification/dot2/dot2k.py +2 -1
@@ -35,6 +35,7 @@ class dot2k(Dot2c): self.states = [] self.main_c = self.__read_file(self.monitor_templates_dir + "main_container.c") self.main_h = self.__read_file(self.monitor_templates_dir + "main_container.h") + self.kconfig = self.__read_file(self.monitor_templates_dir + "Kconfig_container") else: super().__init__(file_path, extra_params.get("model_name"))
@@ -44,7 +45,7 @@ class dot2k(Dot2c): self.monitor_type = MonitorType self.main_c = self.__read_file(self.monitor_templates_dir + "main.c") self.trace_h = self.__read_file(self.monitor_templates_dir + "trace.h") - self.kconfig = self.__read_file(self.monitor_templates_dir + "Kconfig") + self.kconfig = self.__read_file(self.monitor_templates_dir + "Kconfig") self.enum_suffix = "_%s" % self.name self.description = extra_params.get("description", self.name) or "auto-generated" self.auto_patch = extra_params.get("auto_patch")
tools/verification/dot2/dot2k_templates/Kconfig_container new file +5 -0
@@ -0,0 +1,5 @@ +config RV_MON_%%MODEL_NAME_UP%% + depends on RV + bool "%%MODEL_NAME%% monitor" + help + %%DESCRIPTION%%
Signature · 1 line
2.39.5
PATCH v4 6/22 verification/dot2k: Remove __buff_to_string()
In reply to Nam Caostr.join() can do what __buff_to_string() does. Therefore replace __buff_to_string() to make the scripts more pythonic.
--- tools/verification/dot2/dot2k.py | 21 ++++++--------------- 1 file changed, 6 insertions(+), 15 deletions(-)
tools/verification/dot2/dot2k.py +6 -15
@@ -109,15 +109,6 @@ class dot2k(Dot2c): fd.close() return content - def __buff_to_string(self, buff): - string = "" - - for line in buff: - string = string + line + "\n" - - # cut off the last \n - return string[:-1] - def fill_monitor_type(self): return self.monitor_type.upper()
@@ -148,19 +139,19 @@ class dot2k(Dot2c): buff.append("\tda_%s_%s(%s%s);" % (handle, self.name, event, self.enum_suffix)); buff.append("}") buff.append("") - return self.__buff_to_string(buff) + return '\n'.join(buff) def fill_tracepoint_attach_probe(self): buff = [] for event in self.events: buff.append("\trv_attach_trace_probe(\"%s\", /* XXX: tracepoint */, handle_%s);" % (self.name, event)) - return self.__buff_to_string(buff) + return '\n'.join(buff) def fill_tracepoint_detach_helper(self): buff = [] for event in self.events: buff.append("\trv_detach_trace_probe(\"%s\", /* XXX: tracepoint */, handle_%s);" % (self.name, event)) - return self.__buff_to_string(buff) + return '\n'.join(buff) def fill_main_c(self): main_c = self.main_c
@@ -210,7 +201,7 @@ class dot2k(Dot2c): buff = self.fill_model_h_header() buff += self.format_model() - return self.__buff_to_string(buff) + return '\n'.join(buff) def fill_monitor_class_type(self): if self.monitor_type == "per_task":
@@ -242,7 +233,7 @@ class dot2k(Dot2c): tp_args_c = ", ".join([b for a,b in tp_args]) buff.append(" TP_PROTO(%s)," % tp_proto_c) buff.append(" TP_ARGS(%s)" % tp_args_c) - return self.__buff_to_string(buff) + return '\n'.join(buff) def fill_monitor_deps(self): buff = []
@@ -250,7 +241,7 @@ class dot2k(Dot2c): if self.parent: buff.append(" depends on RV_MON_%s" % self.parent.upper()) buff.append(" default y") - return self.__buff_to_string(buff) + return '\n'.join(buff) def fill_trace_h(self): trace_h = self.trace_h
Signature · 1 line
2.39.5
PATCH v4 7/22 verification/dot2k: Replace is_container() hack with subparsers
In reply to Nam Caodot2k is used for both generating deterministic automaton (DA) monitor and generating container monitor.
Generating DA monitor and generating container requires different parameters. This is implemented by peeking at sys.argv and check whether "--container" is specified, and use that information to make some parameters optional or required.
This works, but is quite hacky and ugly.
Replace this hack with Python's built-in subparsers.
The old commands:
python3 dot2/dot2k -d wip.dot -t per_cpu python3 dot2/dot2k -n sched --container
are equivalent to the new commands:
python3 dot2/dot2k monitor -d wip.dot -t per_cpu python3 dot2/dot2k container -n sched
--- tools/verification/dot2/dot2k | 37 +++++++++++++++++--------------- tools/verification/dot2/dot2k.py | 2 +- 2 files changed, 21 insertions(+), 18 deletions(-)
tools/verification/dot2/dot2k +20 -17
@@ -13,30 +13,33 @@ if __name__ == '__main__': import argparse import sys - def is_container(): - """Should work even before parsing the arguments""" - return "-c" in sys.argv or "--container" in sys.argv - parser = argparse.ArgumentParser(description='transform .dot file into kernel rv monitor') - parser.add_argument('-d', "--dot", dest="dot_file", required=not is_container()) - parser.add_argument('-t', "--monitor_type", dest="monitor_type", required=not is_container(), - help=f"Available options: {', '.join(dot2k.monitor_types.keys())}") - parser.add_argument('-n', "--model_name", dest="model_name", required=is_container()) parser.add_argument("-D", "--description", dest="description", required=False) parser.add_argument("-a", "--auto_patch", dest="auto_patch", action="store_true", required=False, help="Patch the kernel in place") - parser.add_argument("-p", "--parent", dest="parent", - required=False, help="Create a monitor nested to parent") - parser.add_argument("-c", "--container", dest="container", - action="store_true", required=False, - help="Create an empty monitor to be used as a container") + + subparsers = parser.add_subparsers(dest="subcmd", required=True) + + monitor_parser = subparsers.add_parser("monitor") + monitor_parser.add_argument('-n', "--model_name", dest="model_name") + monitor_parser.add_argument("-p", "--parent", dest="parent", + required=False, help="Create a monitor nested to parent") + monitor_parser.add_argument('-d', "--dot", dest="dot_file") + monitor_parser.add_argument('-t', "--monitor_type", dest="monitor_type", + help=f"Available options: {', '.join(dot2k.monitor_types.keys())}") + + container_parser = subparsers.add_parser("container") + container_parser.add_argument('-n', "--model_name", dest="model_name", required=True) + params = parser.parse_args() - if not is_container(): - print("Opening and parsing the dot file %s" % params.dot_file) try: - monitor=dot2k(params.dot_file, params.monitor_type, vars(params)) + if params.subcmd == "monitor": + print("Opening and parsing the dot file %s" % params.dot_file) + monitor = dot2k(params.dot_file, params.monitor_type, vars(params)) + else: + monitor = dot2k(None, None, vars(params)) except Exception as e: print('Error: '+ str(e)) print("Sorry : :-(")
@@ -45,7 +48,7 @@ if __name__ == '__main__': print("Writing the monitor into the directory %s" % monitor.name) monitor.print_files() print("Almost done, checklist") - if not is_container(): + if params.subcmd == "monitor": print(" - Edit the %s/%s.c to add the instrumentation" % (monitor.name, monitor.name)) print(monitor.fill_tracepoint_tooltip()) print(monitor.fill_makefile_tooltip())
tools/verification/dot2/dot2k.py +1 -1
@@ -19,7 +19,7 @@ class dot2k(Dot2c): monitor_type = "per_cpu" def __init__(self, file_path, MonitorType, extra_params={}): - self.container = extra_params.get("container") + self.container = extra_params.get("subcmd") == "container" self.parent = extra_params.get("parent") self.__fill_rv_templates_dir()
Signature · 1 line
2.39.5
PATCH v4 8/22 rename CONFIG_DA_MON_EVENTS to CONFIG_RV_MON_EVENTS
In reply to Nam CaoCONFIG_DA_MON_EVENTS is not specific to deterministic automaton. It could be used for other monitor types. Therefore rename it to CONFIG_RV_MON_EVENTS.
This prepares for the introduction of linear temporal logic monitor.
--- kernel/trace/rv/Kconfig | 6 +++--- kernel/trace/rv/rv.c | 2 +- 2 files changed, 4 insertions(+), 4 deletions(-)
kernel/trace/rv/Kconfig +3 -3
@@ -1,14 +1,14 @@ # SPDX-License-Identifier: GPL-2.0-only # -config DA_MON_EVENTS +config RV_MON_EVENTS bool config DA_MON_EVENTS_IMPLICIT - select DA_MON_EVENTS + select RV_MON_EVENTS bool config DA_MON_EVENTS_ID - select DA_MON_EVENTS + select RV_MON_EVENTS bool menuconfig RV
kernel/trace/rv/rv.c +1 -1
@@ -143,7 +143,7 @@#include <linux/init.h> #include <linux/slab.h> -#ifdef CONFIG_DA_MON_EVENTS +#ifdef CONFIG_RV_MON_EVENTS #define CREATE_TRACE_POINTS #include <rv_trace.h> #endif
Signature · 1 line
2.39.5
PATCH v4 9/22 verification/dot2k: Prepare the frontend for LTL inclusion
In reply to Nam CaoThe dot2k tool has some code that can be reused for linear temporal logic monitor. Prepare its frontend for LTL inclusion:
1. Rename to be generic: rvgen
2. Replace the parameter --dot with 2 parameters:
--class: to specific the monitor class, can be 'da' or 'ltl' --spec: the monitor specification file, .dot file for DA, and .ltl
file for LTL
The old command:
python3 dot2/dot2k monitor -d wip.dot -t per_cpu
is equivalent to the new commands:
python3 rvgen monitor -c da -s wip.dot -t per_cpu
---
tools/verification/{dot2 => rvgen}/Makefile | 10 +++++-----
.../{dot2/dot2k => rvgen/__main__.py} | 18 +++++++++++++-----
tools/verification/{dot2 => rvgen}/dot2c | 2 +-
.../{dot2 => rvgen}/dot2k_templates/Kconfig | 0
.../dot2k_templates/Kconfig_container | 0
.../{dot2 => rvgen}/dot2k_templates/main.c | 0
.../dot2k_templates/main_container.c | 0
.../dot2k_templates/main_container.h | 0
.../{dot2 => rvgen}/dot2k_templates/trace.h | 0
.../{dot2 => rvgen/rvgen}/automata.py | 0
.../{dot2 => rvgen/rvgen}/dot2c.py | 2 +-
.../{dot2 => rvgen/rvgen}/dot2k.py | 10 +++++-----
12 files changed, 25 insertions(+), 17 deletions(-)
rename tools/verification/{dot2 => rvgen}/Makefile (55%)
rename tools/verification/{dot2/dot2k => rvgen/__main__.py} (72%)
rename tools/verification/{dot2 => rvgen}/dot2c (97%)
rename tools/verification/{dot2 => rvgen}/dot2k_templates/Kconfig (100%)
rename tools/verification/{dot2 => rvgen}/dot2k_templates/Kconfig_container (100%)
rename tools/verification/{dot2 => rvgen}/dot2k_templates/main.c (100%)
rename tools/verification/{dot2 => rvgen}/dot2k_templates/main_container.c (100%)
rename tools/verification/{dot2 => rvgen}/dot2k_templates/main_container.h (100%)
rename tools/verification/{dot2 => rvgen}/dot2k_templates/trace.h (100%)
rename tools/verification/{dot2 => rvgen/rvgen}/automata.py (100%)
rename tools/verification/{dot2 => rvgen/rvgen}/dot2c.py (99%)
rename tools/verification/{dot2 => rvgen/rvgen}/dot2k.py (98%)tools/verification/rvgen/Makefile renamed from tools/verification/dot2/Makefile +5 -5
@@ -3,7 +3,7 @@ INSTALL=install prefix ?= /usr bindir ?= $(prefix)/bin mandir ?= $(prefix)/share/man -miscdir ?= $(prefix)/share/dot2 +miscdir ?= $(prefix)/share/rvgen srcdir ?= $(prefix)/src PYLIB ?= $(shell python3 -c 'import sysconfig; print (sysconfig.get_path("purelib"))')
@@ -16,11 +16,11 @@ clean: .PHONY: install install: - $(INSTALL) automata.py -D -m 644 $(DESTDIR)$(PYLIB)/dot2/automata.py - $(INSTALL) dot2c.py -D -m 644 $(DESTDIR)$(PYLIB)/dot2/dot2c.py + $(INSTALL) rvgen/automata.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/automata.py + $(INSTALL) rvgen/dot2c.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/dot2c.py $(INSTALL) dot2c -D -m 755 $(DESTDIR)$(bindir)/ - $(INSTALL) dot2k.py -D -m 644 $(DESTDIR)$(PYLIB)/dot2/dot2k.py - $(INSTALL) dot2k -D -m 755 $(DESTDIR)$(bindir)/ + $(INSTALL) rvgen/dot2k.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/dot2k.py + $(INSTALL) __main__.py -D -m 755 $(DESTDIR)$(bindir)/rvgen mkdir -p ${miscdir}/ cp -rp dot2k_templates $(DESTDIR)$(miscdir)/
tools/verification/rvgen/__main__.py renamed from tools/verification/dot2/dot2k +13 -5
@@ -9,11 +9,11 @@ # Documentation/trace/rv/da_monitor_synthesis.rst if __name__ == '__main__': - from dot2.dot2k import dot2k + from rvgen.dot2k import dot2k import argparse import sys - parser = argparse.ArgumentParser(description='transform .dot file into kernel rv monitor') + parser = argparse.ArgumentParser(description='Generate kernel rv monitor') parser.add_argument("-D", "--description", dest="description", required=False) parser.add_argument("-a", "--auto_patch", dest="auto_patch", action="store_true", required=False,
@@ -25,7 +25,9 @@ if __name__ == '__main__': monitor_parser.add_argument('-n', "--model_name", dest="model_name") monitor_parser.add_argument("-p", "--parent", dest="parent", required=False, help="Create a monitor nested to parent") - monitor_parser.add_argument('-d', "--dot", dest="dot_file") + monitor_parser.add_argument('-c', "--class", dest="monitor_class", + help="Monitor class, either \"da\" or \"ltl\"") + monitor_parser.add_argument('-s', "--spec", dest="spec", help="Monitor specification file") monitor_parser.add_argument('-t', "--monitor_type", dest="monitor_type", help=f"Available options: {', '.join(dot2k.monitor_types.keys())}")
@@ -36,8 +38,14 @@ if __name__ == '__main__': try: if params.subcmd == "monitor": - print("Opening and parsing the dot file %s" % params.dot_file) - monitor = dot2k(params.dot_file, params.monitor_type, vars(params)) + print("Opening and parsing the specification file %s" % params.spec) + if params.monitor_class == "da": + monitor = dot2k(params.spec, params.monitor_type, vars(params)) + elif params.monitor_class == "ltl": + raise NotImplementedError + else: + print("Unknown monitor class:", params.monitor_class) + sys.exit(1) else: monitor = dot2k(None, None, vars(params)) except Exception as e:
tools/verification/rvgen/dot2c renamed from tools/verification/dot2/dot2c +1 -1
@@ -14,7 +14,7 @@ # Documentation/trace/rv/deterministic_automata.rst if __name__ == '__main__': - from dot2 import dot2c + from rvgen import dot2c import argparse import sys
tools/verification/rvgen/dot2k_templates/Kconfig renamed from tools/verification/dot2/dot2k_templates/Kconfig +0 -0
tools/verification/rvgen/dot2k_templates/Kconfig_container renamed from tools/verification/dot2/dot2k_templates/Kconfig_container +0 -0
tools/verification/rvgen/dot2k_templates/main.c renamed from tools/verification/dot2/dot2k_templates/main.c +0 -0
tools/verification/rvgen/dot2k_templates/main_container.c renamed from tools/verification/dot2/dot2k_templates/main_container.c +0 -0
tools/verification/rvgen/dot2k_templates/main_container.h renamed from tools/verification/dot2/dot2k_templates/main_container.h +0 -0
tools/verification/rvgen/dot2k_templates/trace.h renamed from tools/verification/dot2/dot2k_templates/trace.h +0 -0
tools/verification/rvgen/rvgen/automata.py renamed from tools/verification/dot2/automata.py +0 -0
tools/verification/rvgen/rvgen/dot2c.py renamed from tools/verification/dot2/dot2c.py +1 -1
@@ -13,7 +13,7 @@ # For further information, see: # Documentation/trace/rv/deterministic_automata.rst -from dot2.automata import Automata +from .automata import Automata class Dot2c(Automata): enum_suffix = ""
tools/verification/rvgen/rvgen/dot2k.py renamed from tools/verification/dot2/dot2k.py +5 -5
@@ -8,13 +8,13 @@ # For further information, see: # Documentation/trace/rv/da_monitor_synthesis.rst -from dot2.dot2c import Dot2c +from .dot2c import Dot2c import platform import os class dot2k(Dot2c): monitor_types = { "global" : 1, "per_cpu" : 2, "per_task" : 3 } - monitor_templates_dir = "dot2/dot2k_templates/" + monitor_templates_dir = "rvgen/dot2k_templates/" rv_dir = "kernel/trace/rv" monitor_type = "per_cpu"
@@ -60,14 +60,14 @@ class dot2k(Dot2c): if platform.system() != "Linux": raise OSError("I can only run on Linux.") - kernel_path = "/lib/modules/%s/build/tools/verification/dot2/dot2k_templates/" % (platform.release()) + kernel_path = "/lib/modules/%s/build/tools/verification/rvgen/dot2k_templates/" % (platform.release()) if os.path.exists(kernel_path): self.monitor_templates_dir = kernel_path return - if os.path.exists("/usr/share/dot2/dot2k_templates/"): - self.monitor_templates_dir = "/usr/share/dot2/dot2k_templates/" + if os.path.exists("/usr/share/rvgen/dot2k_templates/"): + self.monitor_templates_dir = "/usr/share/rvgen/dot2k_templates/" return raise FileNotFoundError("Could not find the template directory, do you have the kernel source installed?")
Signature · 1 line
2.39.5
PATCH v4 10/22 Documentation/rv: Prepare monitor synthesis document for LTL inclusion
In reply to Nam CaoMonitor synthesis from deterministic automaton and linear temporal logic have a lot in common. Therefore a single document should describe both.
Change da_monitor_synthesis.rst to monitor_synthesis.rst. LTL monitor synthesis will be added to this file by a follow-up commit.
This makes the diff far easier to read. If renaming and adding LTL info is done in a single commit, git wouldn't recognize it as a rename, but a file removal and a file addition.
While at it, correct the old dot2k commands to the new rvgen commands.
---
Documentation/trace/rv/index.rst | 2 +-
...or_synthesis.rst => monitor_synthesis.rst} | 20 +++++++++----------
2 files changed, 11 insertions(+), 11 deletions(-)
rename Documentation/trace/rv/{da_monitor_synthesis.rst => monitor_synthesis.rst} (92%)Documentation/trace/rv/index.rst +1 -1
@@ -8,7 +8,7 @@ Runtime Verification runtime-verification.rst deterministic_automata.rst - da_monitor_synthesis.rst + monitor_synthesis.rst da_monitor_instrumentation.rst monitor_wip.rst monitor_wwnr.rst
Documentation/trace/rv/monitor_synthesis.rst renamed from Documentation/trace/rv/da_monitor_synthesis.rst +10 -10
@@ -1,5 +1,5 @@ -Deterministic Automata Monitor Synthesis -======================================== +Runtime Verification Monitor Synthesis +====================================== The starting point for the application of runtime verification (RV) techniques is the *specification* or *modeling* of the desired (or undesired) behavior
@@ -36,24 +36,24 @@ below:: | +----> panic ? +-------> <user-specified> -DA monitor synthesis +RV monitor synthesis -------------------- The synthesis of automata-based models into the Linux *RV monitor* abstraction -is automated by the dot2k tool and the rv/da_monitor.h header file that +is automated by the rvgen tool and the rv/da_monitor.h header file that contains a set of macros that automatically generate the monitor's code. -dot2k +rvgen ----- -The dot2k utility leverages dot2c by converting an automaton model in +The rvgen utility leverages dot2c by converting an automaton model in the DOT format into the C representation [1] and creating the skeleton of a kernel monitor in C. For example, it is possible to transform the wip.dot model present in [1] into a per-cpu monitor with the following command:: - $ dot2k -d wip.dot -t per_cpu + $ rvgen monitor -c da -s wip.dot -t per_cpu This will create a directory named wip/ with the following files:
@@ -87,7 +87,7 @@ the second for monitors with per-cpu instances, and the third with per-task instances. In all cases, the 'name' argument is a string that identifies the monitor, and -the 'type' argument is the data type used by dot2k on the representation of +the 'type' argument is the data type used by rvgen on the representation of the model in C. For example, the wip model with two states and three events can be
@@ -134,7 +134,7 @@ Final remarks ------------- With the monitor synthesis in place using the rv/da_monitor.h and -dot2k, the developer's work should be limited to the instrumentation +rvgen, the developer's work should be limited to the instrumentation of the system, increasing the confidence in the overall approach. [1] For details about deterministic automata format and the translation
@@ -142,6 +142,6 @@ from one representation to another, see:: Documentation/trace/rv/deterministic_automata.rst -[2] dot2k appends the monitor's name suffix to the events enums to +[2] rvgen appends the monitor's name suffix to the events enums to avoid conflicting variables when exporting the global vmlinux.h use by BPF programs.
Signature · 1 line
2.39.5
PATCH v4 11/22 verification/rvgen: Restructure the templates files
In reply to Nam CaoTo simply the scripts and to allow easy integration of new monitor types, restructure the template files as followed:
1. Move the template files to be in the same directory as the rvgen package. Furthermore, the installation will now only install the templates to the package directory, not /usr/share/. This simplify templates reading, as the scripts do not need to find the templates at multiple places.
2. Move dot2k_templates/* to:
- templates/dot2k/
- templates/container/
This allows sharing templates reading code between DA monitor generation and container generation (and any future generation type).
For template files which can be shared between different generation types, support putting them in templates/
This restructure aligns with the recommendation from:
https://python-packaging.readthedocs.io/en/latest/non-code-files.html
---
tools/verification/rvgen/Makefile | 5 +-
tools/verification/rvgen/rvgen/dot2k.py | 47 ++++++++-----------
.../templates}/Kconfig | 0
.../templates/container/Kconfig} | 0
.../templates/container/main.c} | 0
.../templates/container/main.h} | 0
.../templates/dot2k}/main.c | 0
.../templates/dot2k}/trace.h | 0
8 files changed, 20 insertions(+), 32 deletions(-)
rename tools/verification/rvgen/{dot2k_templates => rvgen/templates}/Kconfig (100%)
rename tools/verification/rvgen/{dot2k_templates/Kconfig_container => rvgen/templates/container/Kconfig} (100%)
rename tools/verification/rvgen/{dot2k_templates/main_container.c => rvgen/templates/container/main.c} (100%)
rename tools/verification/rvgen/{dot2k_templates/main_container.h => rvgen/templates/container/main.h} (100%)
rename tools/verification/rvgen/{dot2k_templates => rvgen/templates/dot2k}/main.c (100%)
rename tools/verification/rvgen/{dot2k_templates => rvgen/templates/dot2k}/trace.h (100%)tools/verification/rvgen/Makefile +1 -4
@@ -3,7 +3,6 @@ INSTALL=install prefix ?= /usr bindir ?= $(prefix)/bin mandir ?= $(prefix)/share/man -miscdir ?= $(prefix)/share/rvgen srcdir ?= $(prefix)/src PYLIB ?= $(shell python3 -c 'import sysconfig; print (sysconfig.get_path("purelib"))')
@@ -21,6 +20,4 @@ install: $(INSTALL) dot2c -D -m 755 $(DESTDIR)$(bindir)/ $(INSTALL) rvgen/dot2k.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/dot2k.py $(INSTALL) __main__.py -D -m 755 $(DESTDIR)$(bindir)/rvgen - - mkdir -p ${miscdir}/ - cp -rp dot2k_templates $(DESTDIR)$(miscdir)/ + cp -rp rvgen/templates $(DESTDIR)$(PYLIB)/rvgen/
tools/verification/rvgen/rvgen/dot2k.py +19 -28
83 lines not on this page · raw message
tools/verification/rvgen/rvgen/templates/Kconfig renamed from tools/verification/rvgen/dot2k_templates/Kconfig +0 -0
tools/verification/rvgen/rvgen/templates/container/Kconfig renamed from tools/verification/rvgen/dot2k_templates/Kconfig_container +0 -0
tools/verification/rvgen/rvgen/templates/container/main.c renamed from tools/verification/rvgen/dot2k_templates/main_container.c +0 -0
tools/verification/rvgen/rvgen/templates/container/main.h renamed from tools/verification/rvgen/dot2k_templates/main_container.h +0 -0
tools/verification/rvgen/rvgen/templates/dot2k/main.c renamed from tools/verification/rvgen/dot2k_templates/main.c +0 -0
tools/verification/rvgen/rvgen/templates/dot2k/trace.h renamed from tools/verification/rvgen/dot2k_templates/trace.h +0 -0
Signature · 1 line
2.39.5
PATCH v4 12/22 verification/rvgen: Restructure the classes to prepare for LTL inclusion
In reply to Nam CaoBoth container generation and DA monitor generation is implemented in the class dot2k. That requires some ugly "if is_container ... else ...". If linear temporal logic support is added at the current state, the "if else" chain is longer and uglier.
Furthermore, container generation is irrevelant to .dot files. It is therefore illogical to be implemented in class "dot2k".
Clean it up, restructure the dot2k class into the following class hierarchy:
(RVGenerator)
/\
/ \
/ \
/ \
/ \
(Container) (Monitor)
/\
/ \
/ \
/ \
(dot2k) [ltl2k] <- intendedThis allows a simple and clean integration of LTL.
--- tools/verification/rvgen/Makefile | 2 + tools/verification/rvgen/__main__.py | 6 +- tools/verification/rvgen/rvgen/container.py | 22 ++ tools/verification/rvgen/rvgen/dot2k.py | 275 ++------------------ tools/verification/rvgen/rvgen/generator.py | 264 +++++++++++++++++++ 5 files changed, 308 insertions(+), 261 deletions(-) create mode 100644 tools/verification/rvgen/rvgen/container.py create mode 100644 tools/verification/rvgen/rvgen/generator.py
tools/verification/rvgen/Makefile +2 -0
@@ -19,5 +19,7 @@ install: $(INSTALL) rvgen/dot2c.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/dot2c.py $(INSTALL) dot2c -D -m 755 $(DESTDIR)$(bindir)/ $(INSTALL) rvgen/dot2k.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/dot2k.py + $(INSTALL) rvgen/container.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/container.py + $(INSTALL) rvgen/generator.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/generator.py $(INSTALL) __main__.py -D -m 755 $(DESTDIR)$(bindir)/rvgen cp -rp rvgen/templates $(DESTDIR)$(PYLIB)/rvgen/
tools/verification/rvgen/__main__.py +4 -2
@@ -10,6 +10,8 @@ if __name__ == '__main__': from rvgen.dot2k import dot2k + from rvgen.generator import Monitor + from rvgen.container import Container import argparse import sys
@@ -29,7 +31,7 @@ if __name__ == '__main__': help="Monitor class, either \"da\" or \"ltl\"") monitor_parser.add_argument('-s', "--spec", dest="spec", help="Monitor specification file") monitor_parser.add_argument('-t', "--monitor_type", dest="monitor_type", - help=f"Available options: {', '.join(dot2k.monitor_types.keys())}") + help=f"Available options: {', '.join(Monitor.monitor_types.keys())}") container_parser = subparsers.add_parser("container") container_parser.add_argument('-n', "--model_name", dest="model_name", required=True)
@@ -47,7 +49,7 @@ if __name__ == '__main__': print("Unknown monitor class:", params.monitor_class) sys.exit(1) else: - monitor = dot2k(None, None, vars(params)) + monitor = Container(vars(params)) except Exception as e: print('Error: '+ str(e)) print("Sorry : :-(")
tools/verification/rvgen/rvgen/container.py new file +22 -0
@@ -0,0 +1,22 @@ +#!/usr/bin/env python3 +# SPDX-License-Identifier: GPL-2.0-only +# +# Copyright (C) 2019-2022 Red Hat, Inc. Daniel Bristot de Oliveira <[email protected]> +# +# Generator for runtime verification monitor container + +from . import generator + + +class Container(generator.RVGenerator): + template_dir = "container" + + def __init__(self, extra_params={}): + super().__init__(extra_params) + self.name = extra_params.get("model_name") + self.main_h = self._read_template_file("main.h") + + def fill_model_h(self): + main_h = self.main_h + main_h = main_h.replace("%%MODEL_NAME%%", self.name) + return main_h
tools/verification/rvgen/rvgen/dot2k.py +16 -259
301 lines not on this page · raw message
tools/verification/rvgen/rvgen/generator.py new file +264 -0
264 lines not on this page · raw message
Signature · 1 line
2.39.5
PATCH v4 13/22 Add support for LTL monitors
In reply to Nam CaoWhile attempting to implement DA monitors for some complex specifications, deterministic automaton is found to be inappropriate as the specification language. The automaton is complicated, hard to understand, and error-prone.
For these cases, linear temporal logic is more suitable as the specification language.
Add support for linear temporal logic runtime verification monitor.
For all the details, see the documentations added by this commit.
--- Documentation/trace/rv/index.rst | 1 + .../trace/rv/linear_temporal_logic.rst | 122 ++++ Documentation/trace/rv/monitor_synthesis.rst | 141 ++++- include/linux/rv.h | 62 +- include/rv/ltl_monitor.h | 184 ++++++ kernel/fork.c | 5 +- kernel/trace/rv/Kconfig | 7 + kernel/trace/rv/rv_trace.h | 47 ++ tools/verification/rvgen/.gitignore | 3 + tools/verification/rvgen/Makefile | 2 + tools/verification/rvgen/__main__.py | 3 +- tools/verification/rvgen/rvgen/ltl2ba.py | 558 ++++++++++++++++++ tools/verification/rvgen/rvgen/ltl2k.py | 245 ++++++++ .../rvgen/rvgen/templates/ltl2k/main.c | 102 ++++ .../rvgen/rvgen/templates/ltl2k/trace.h | 14 + 15 files changed, 1471 insertions(+), 25 deletions(-) create mode 100644 Documentation/trace/rv/linear_temporal_logic.rst create mode 100644 include/rv/ltl_monitor.h create mode 100644 tools/verification/rvgen/.gitignore create mode 100644 tools/verification/rvgen/rvgen/ltl2ba.py create mode 100644 tools/verification/rvgen/rvgen/ltl2k.py create mode 100644 tools/verification/rvgen/rvgen/templates/ltl2k/main.c create mode 100644 tools/verification/rvgen/rvgen/templates/ltl2k/trace.h
Documentation/trace/rv/index.rst +1 -0
@@ -8,6 +8,7 @@ Runtime Verification runtime-verification.rst deterministic_automata.rst + linear_temporal_logic.rst monitor_synthesis.rst da_monitor_instrumentation.rst monitor_wip.rst
Documentation/trace/rv/linear_temporal_logic.rst new file +122 -0
122 lines not on this page · raw message
Documentation/trace/rv/monitor_synthesis.rst +125 -16
169 lines not on this page · raw message
include/linux/rv.h +58 -4
81 lines not on this page · raw message
include/rv/ltl_monitor.h new file +184 -0
184 lines not on this page · raw message
kernel/fork.c +1 -4
@@ -2127,10 +2127,7 @@ static void copy_oom_score_adj(u64 clone_flags, struct task_struct *tsk)#ifdef CONFIG_RV static void rv_task_fork (struct task_struct *p) { -int i; - -for (i =0 ; i <RV_PER_TASK_MONITORS ; i++) - p->rv[i].da_mon.monitoring = false; +memset (p->rv,0 ,sizeof (p->rv)); }#else #define rv_task_fork (p) do {} while (0)
kernel/trace/rv/Kconfig +7 -0
@@ -11,6 +11,13 @@ config DA_MON_EVENTS_ID select RV_MON_EVENTS bool +config LTL_MON_EVENTS_ID + select RV_MON_EVENTS + bool + +config RV_LTL_MONITOR + bool + menuconfig RV bool "Runtime Verification" depends on TRACING
kernel/trace/rv/rv_trace.h +47 -0
53 lines not on this page · raw message
tools/verification/rvgen/.gitignore new file +3 -0
@@ -0,0 +1,3 @@ +__pycache__/ +parser.out +parsetab.py
tools/verification/rvgen/Makefile +2 -0
@@ -21,5 +21,7 @@ install: $(INSTALL) rvgen/dot2k.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/dot2k.py $(INSTALL) rvgen/container.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/container.py $(INSTALL) rvgen/generator.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/generator.py + $(INSTALL) rvgen/ltl2ba.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/ltl2ba.py + $(INSTALL) rvgen/ltl2k.py -D -m 644 $(DESTDIR)$(PYLIB)/rvgen/ltl2k.py $(INSTALL) __main__.py -D -m 755 $(DESTDIR)$(bindir)/rvgen cp -rp rvgen/templates $(DESTDIR)$(PYLIB)/rvgen/
tools/verification/rvgen/__main__.py +2 -1
@@ -12,6 +12,7 @@ if __name__ == '__main__': from rvgen.dot2k import dot2k from rvgen.generator import Monitor from rvgen.container import Container + from rvgen.ltl2k import ltl2k import argparse import sys
@@ -44,7 +45,7 @@ if __name__ == '__main__': if params.monitor_class == "da": monitor = dot2k(params.spec, params.monitor_type, vars(params)) elif params.monitor_class == "ltl": - raise NotImplementedError + monitor = ltl2k(params.spec, params.monitor_type, vars(params)) else: print("Unknown monitor class:", params.monitor_class) sys.exit(1)
tools/verification/rvgen/rvgen/ltl2ba.py new file +558 -0
558 lines not on this page · raw message
tools/verification/rvgen/rvgen/ltl2k.py new file +245 -0
245 lines not on this page · raw message
tools/verification/rvgen/rvgen/templates/ltl2k/main.c new file +102 -0
102 lines not on this page · raw message
tools/verification/rvgen/rvgen/templates/ltl2k/trace.h new file +14 -0
@@ -0,0 +1,14 @@ +/* SPDX-License-Identifier: GPL-2.0 */ + +/* +* Snippet to be included in rv_trace.h +*/ + +#ifdef CONFIG_RV_MON_%%MODEL_NAME_UP%% +DEFINE_EVENT(event_%%MONITOR_CLASS%%, event_%%MODEL_NAME%%, + TP_PROTO(struct task_struct *task, char *states, char *atoms, char *next), +TP_ARGS (task, states, atoms, next)); +DEFINE_EVENT(error_%%MONITOR_CLASS%%, error_%%MODEL_NAME%%, + TP_PROTO(struct task_struct *task), +TP_ARGS (task)); +#endif /* CONFIG_RV_MON_%%MODEL_NAME_UP%% */
Signature · 1 line
2.39.5
PATCH v4 14/22 Add rtapp container monitor
In reply to Nam CaoAdd the container "rtapp" which is the monitor collection for detecting problems with real-time applications. The monitors will be added in the follow-up commits.
--- kernel/trace/rv/Kconfig | 1 + kernel/trace/rv/Makefile | 1 + kernel/trace/rv/monitors/rtapp/Kconfig | 6 +++++ kernel/trace/rv/monitors/rtapp/rtapp.c | 33 ++++++++++++++++++++++++++ kernel/trace/rv/monitors/rtapp/rtapp.h | 3 +++ 5 files changed, 44 insertions(+) create mode 100644 kernel/trace/rv/monitors/rtapp/Kconfig create mode 100644 kernel/trace/rv/monitors/rtapp/rtapp.c create mode 100644 kernel/trace/rv/monitors/rtapp/rtapp.h
kernel/trace/rv/Kconfig +1 -0
@@ -41,6 +41,7 @@ source "kernel/trace/rv/monitors/snroc/Kconfig" source "kernel/trace/rv/monitors/scpd/Kconfig" source "kernel/trace/rv/monitors/snep/Kconfig" source "kernel/trace/rv/monitors/sncid/Kconfig" +source "kernel/trace/rv/monitors/rtapp/Kconfig" # Add new monitors here config RV_REACTORS
kernel/trace/rv/Makefile +1 -0
@@ -12,6 +12,7 @@ obj-$(CONFIG_RV_MON_SNROC) += monitors/snroc/snroc.o obj-$(CONFIG_RV_MON_SCPD) += monitors/scpd/scpd.o obj-$(CONFIG_RV_MON_SNEP) += monitors/snep/snep.o obj-$(CONFIG_RV_MON_SNCID) += monitors/sncid/sncid.o +obj-$(CONFIG_RV_MON_RTAPP) += monitors/rtapp/rtapp.o # Add new monitors here obj-$(CONFIG_RV_REACTORS) += rv_reactors.o obj-$(CONFIG_RV_REACT_PRINTK) += reactor_printk.o
kernel/trace/rv/monitors/rtapp/Kconfig new file +6 -0
@@ -0,0 +1,6 @@ +config RV_MON_RTAPP + depends on RV + bool "rtapp monitor" + help + Collection of monitors to check for common problems with real-time application that cause + unexpected latency.
kernel/trace/rv/monitors/rtapp/rtapp.c new file +33 -0
@@ -0,0 +1,33 @@ +// SPDX-License-Identifier: GPL-2.0 +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/init.h> +#include <linux/rv.h> + +#define MODULE_NAME "rtapp" + +#include "rtapp.h" + +struct rv_monitor rv_rtapp; + +struct rv_monitor rv_rtapp = { + .name ="rtapp" , + .description ="Collection of monitors for detecting problems with real-time applications" , +}; + +static int __init register_rtapp (void ) +{ +return rv_register_monitor (&rv_rtapp,NULL ); +} + +static void __exit unregister_rtapp(void) +{ +rv_unregister_monitor (&rv_rtapp); +} + +module_init (register_rtapp); +module_exit (unregister_rtapp); + +MODULE_LICENSE ("GPL" ); +MODULE_AUTHOR ("Nam Cao <[email protected]>" ); +MODULE_DESCRIPTION ("Collection of monitors for detecting problems with real-time applications" );
kernel/trace/rv/monitors/rtapp/rtapp.h new file +3 -0
@@ -0,0 +1,3 @@ +/* SPDX-License-Identifier: GPL-2.0 */ + +extern struct rv_monitor rv_rtapp;
Signature · 1 line
2.39.5
PATCH v4 15/22 x86/tracing: Remove redundant trace_pagefault_key
In reply to Nam Caotrace_pagefault_key is used to optimize the pagefault tracepoints when it is disabled. However, tracepoints already have built-in static_key for this exact purpose.
Remove this redundant key.
---
Cc: Thomas Gleixner <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Borislav Petkov <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: [email protected]
Cc: H. Peter Anvin <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Peter Zijlstra <[email protected]>
--- arch/x86/include/asm/trace/common.h | 12 ------------ arch/x86/include/asm/trace/exceptions.h | 18 ++++++------------ arch/x86/include/asm/trace/irq_vectors.h | 1 - arch/x86/kernel/Makefile | 1 - arch/x86/kernel/tracepoint.c | 21 --------------------- arch/x86/mm/fault.c | 3 --- 6 files changed, 6 insertions(+), 50 deletions(-) delete mode 100644 arch/x86/include/asm/trace/common.h delete mode 100644 arch/x86/kernel/tracepoint.c
arch/x86/include/asm/trace/common.h deleted +0 -12
@@ -1,12 +0,0 @@ -#ifndef _ASM_TRACE_COMMON_H -#define _ASM_TRACE_COMMON_H - -#ifdef CONFIG_TRACING -DECLARE_STATIC_KEY_FALSE (trace_pagefault_key); -#define trace_pagefault_enabled () \ - static_branch_unlikely(&trace_pagefault_key) -#else -static inline bool trace_pagefault_enabled (void ) {return false; } -#endif - -#endif
arch/x86/include/asm/trace/exceptions.h +6 -12
@@ -6,10 +6,6 @@#define _TRACE_PAGE_FAULT_H#include <linux/tracepoint.h> -#include <asm/trace/common.h> - -extern int trace_pagefault_reg (void ); -extern void trace_pagefault_unreg (void ); DECLARE_EVENT_CLASS(x86_exceptions,
@@ -34,15 +30,13 @@ DECLARE_EVENT_CLASS(x86_exceptions, (void *)__entry->address, (void *)__entry->ip, __entry->error_code) ); -#define DEFINE_PAGE_FAULT_EVENT (name) \ -DEFINE_EVENT_FN(x86_exceptions, name, \ - TP_PROTO(unsigned long address, struct pt_regs *regs, \ - unsigned long error_code), \ - TP_ARGS(address, regs, error_code), \ - trace_pagefault_reg, trace_pagefault_unreg); +DEFINE_EVENT (x86_exceptions, page_fault_user, + TP_PROTO(unsigned long address, struct pt_regs *regs, unsigned long error_code), +TP_ARGS (address, regs, error_code)); -DEFINE_PAGE_FAULT_EVENT (page_fault_user); -DEFINE_PAGE_FAULT_EVENT (page_fault_kernel); +DEFINE_EVENT (x86_exceptions, page_fault_kernel, + TP_PROTO(unsigned long address, struct pt_regs *regs, unsigned long error_code), +TP_ARGS (address, regs, error_code));#undef TRACE_INCLUDE_PATH#undef TRACE_INCLUDE_FILE
arch/x86/include/asm/trace/irq_vectors.h +0 -1
@@ -6,7 +6,6 @@#define _TRACE_IRQ_VECTORS_H#include <linux/tracepoint.h> -#include <asm/trace/common.h> #ifdef CONFIG_X86_LOCAL_APIC
arch/x86/kernel/Makefile +0 -1
@@ -139,7 +139,6 @@ obj-$(CONFIG_OF) += devicetree.o obj-$(CONFIG_UPROBES) += uprobes.o obj-$(CONFIG_PERF_EVENTS) += perf_regs.o -obj-$(CONFIG_TRACING) += tracepoint.o obj-$(CONFIG_SCHED_MC_PRIO) += itmt.o obj-$(CONFIG_X86_UMIP) += umip.o
arch/x86/kernel/tracepoint.c deleted +0 -21
@@ -1,21 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0 -/* -* Copyright (C) 2013 Seiji Aguchi <[email protected]> -*/ -#include <linux/jump_label.h> -#include <linux/atomic.h> - -#include <asm/trace/exceptions.h> - -DEFINE_STATIC_KEY_FALSE (trace_pagefault_key); - -int trace_pagefault_reg (void ) -{ -static_branch_inc (&trace_pagefault_key); -return 0 ; -} - -void trace_pagefault_unreg (void ) -{ -static_branch_dec (&trace_pagefault_key); -}
arch/x86/mm/fault.c +0 -3
@@ -1455,9 +1455,6 @@ static __always_inline void trace_page_fault_entries(struct pt_regs *regs, unsigned long error_code, unsigned long address) { - if (!trace_pagefault_enabled()) - return; - if (user_mode(regs)) trace_page_fault_user(address, regs, error_code); else
Signature · 1 line
2.39.5
PATCH v4 16/22 x86/tracing: Move page fault trace points to generic
In reply to Nam CaoPage fault trace points are interesting for other architectures as well. Move them to be generic.
---
Cc: Thomas Gleixner <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Borislav Petkov <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: [email protected]
Cc: "H. Peter Anvin" <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Peter Zijlstra <[email protected]>
---
arch/x86/mm/fault.c | 2 +-
.../asm/trace => include/trace/events}/exceptions.h | 13 ++++---------
2 files changed, 5 insertions(+), 10 deletions(-)
rename {arch/x86/include/asm/trace => include/trace/events}/exceptions.h (79%)arch/x86/mm/fault.c +1 -1
@@ -38,7 +38,7 @@#include <asm/sev.h> /* snp_dump_hva_rmpentry() */ #define CREATE_TRACE_POINTS -#include <asm/trace/exceptions.h> +#include <trace/events/exceptions.h> /* * Returns 0 if mmiotrace is disabled, or if the fault is not
include/trace/events/exceptions.h renamed from arch/x86/include/asm/trace/exceptions.h +4 -9
@@ -7,7 +7,7 @@#include <linux/tracepoint.h> -DECLARE_EVENT_CLASS(x86_exceptions, +DECLARE_EVENT_CLASS(exceptions, TP_PROTO(unsigned long address, struct pt_regs *regs, unsigned long error_code),
@@ -22,7 +22,7 @@ DECLARE_EVENT_CLASS(x86_exceptions,TP_fast_assign ( __entry->address = address; - __entry->ip = regs->ip; + __entry->ip =instruction_pointer (regs); __entry->error_code = error_code; ),
@@ -30,18 +30,13 @@ DECLARE_EVENT_CLASS(x86_exceptions, (void *)__entry->address, (void *)__entry->ip, __entry->error_code) ); -DEFINE_EVENT (x86_exceptions, page_fault_user, +DEFINE_EVENT (exceptions, page_fault_user, TP_PROTO(unsigned long address, struct pt_regs *regs, unsigned long error_code),TP_ARGS (address, regs, error_code)); - -DEFINE_EVENT (x86_exceptions, page_fault_kernel, +DEFINE_EVENT (exceptions, page_fault_kernel, TP_PROTO(unsigned long address, struct pt_regs *regs, unsigned long error_code),TP_ARGS (address, regs, error_code)); -#undef TRACE_INCLUDE_PATH -#undef TRACE_INCLUDE_FILE -#define TRACE_INCLUDE_PATH . -#define TRACE_INCLUDE_FILE exceptions#endif /* _TRACE_PAGE_FAULT_H */ /* This part must be outside protection */
Signature · 1 line
2.39.5
PATCH v4 17/22 arm64: mm: Add page fault trace points
In reply to Nam CaoAdd page fault trace points, which are useful to implement RV monitor which watches page faults.
---
Cc: Catalin Marinas <[email protected]>
Cc: Will Deacon <[email protected]>
Cc: [email protected]
--- arch/arm64/mm/fault.c | 8 ++++++++ 1 file changed, 8 insertions(+)
arch/arm64/mm/fault.c +8 -0
@@ -44,6 +44,9 @@#include <asm/tlbflush.h> #include <asm/traps.h> +#define CREATE_TRACE_POINTS +#include <trace/events/exceptions.h> +struct fault_info {int (*fn)(unsigned long far,unsigned long esr,struct pt_regs *regs);
@@ -559,6 +562,11 @@ static int __kprobes do_page_fault(unsigned long far, unsigned long esr,if (kprobe_page_fault (regs, esr))return 0 ; +if (user_mode (regs)) +trace_page_fault_user (addr, regs, esr); +else +trace_page_fault_kernel (addr, regs, esr); +/* * If we're in an interrupt or have no user context, we must not take * the fault.
Signature · 1 line
2.39.5
PATCH v4 18/22 riscv: mm: Add page fault trace points
In reply to Nam CaoAdd page fault trace points, which are useful to implement RV monitor that watches page faults.
---
Cc: Paul Walmsley <[email protected]>
Cc: Palmer Dabbelt <[email protected]>
Cc: Albert Ou <[email protected]>
Cc: [email protected]
--- arch/riscv/mm/fault.c | 8 ++++++++ 1 file changed, 8 insertions(+)
arch/riscv/mm/fault.c +8 -0
@@ -20,6 +20,9 @@#include <asm/ptrace.h> #include <asm/tlbflush.h> +#define CREATE_TRACE_POINTS +#include <trace/events/exceptions.h> +#include "../kernel/head.h" static void show_pte (unsigned long addr)
@@ -291,6 +294,11 @@ void handle_page_fault(struct pt_regs *regs)if (kprobe_page_fault (regs, cause))return ; +if (user_mode (regs)) +trace_page_fault_user (addr, regs, cause); +else +trace_page_fault_kernel (addr, regs, cause); +/* * Fault-in kernel-space virtual memory on-demand. * The 'reference' page table is init_mm.pgd.
Signature · 1 line
2.39.5
PATCH v4 19/22 Add rtapp_pagefault monitor
In reply to Nam CaoUserspace real-time applications may have design flaws that they raise page faults in real-time threads, and thus have unexpected latencies.
Add an linear temporal logic monitor to detect this scenario.
--- kernel/trace/rv/Kconfig | 1 + kernel/trace/rv/Makefile | 1 + kernel/trace/rv/monitors/pagefault/Kconfig | 11 +++ .../trace/rv/monitors/pagefault/pagefault.c | 82 +++++++++++++++++++ .../trace/rv/monitors/pagefault/pagefault.h | 57 +++++++++++++ .../rv/monitors/pagefault/pagefault_trace.h | 14 ++++ kernel/trace/rv/rv_trace.h | 1 + tools/verification/models/rtapp/pagefault.ltl | 1 + 8 files changed, 168 insertions(+) create mode 100644 kernel/trace/rv/monitors/pagefault/Kconfig create mode 100644 kernel/trace/rv/monitors/pagefault/pagefault.c create mode 100644 kernel/trace/rv/monitors/pagefault/pagefault.h create mode 100644 kernel/trace/rv/monitors/pagefault/pagefault_trace.h create mode 100644 tools/verification/models/rtapp/pagefault.ltl
kernel/trace/rv/Kconfig +1 -0
@@ -42,6 +42,7 @@ source "kernel/trace/rv/monitors/scpd/Kconfig" source "kernel/trace/rv/monitors/snep/Kconfig" source "kernel/trace/rv/monitors/sncid/Kconfig" source "kernel/trace/rv/monitors/rtapp/Kconfig" +source "kernel/trace/rv/monitors/pagefault/Kconfig" # Add new monitors here config RV_REACTORS
kernel/trace/rv/Makefile +1 -0
@@ -13,6 +13,7 @@ obj-$(CONFIG_RV_MON_SCPD) += monitors/scpd/scpd.o obj-$(CONFIG_RV_MON_SNEP) += monitors/snep/snep.o obj-$(CONFIG_RV_MON_SNCID) += monitors/sncid/sncid.o obj-$(CONFIG_RV_MON_RTAPP) += monitors/rtapp/rtapp.o +obj-$(CONFIG_RV_MON_PAGEFAULT) += monitors/pagefault/pagefault.o # Add new monitors here obj-$(CONFIG_RV_REACTORS) += rv_reactors.o obj-$(CONFIG_RV_REACT_PRINTK) += reactor_printk.o
kernel/trace/rv/monitors/pagefault/Kconfig new file +11 -0
@@ -0,0 +1,11 @@ +# SPDX-License-Identifier: GPL-2.0-only +# +config RV_MON_PAGEFAULT + depends on RV + select RV_LTL_MONITOR + depends on RV_MON_RTAPP + default y + select LTL_MON_EVENTS_ID + bool "pagefault monitor" + help + Monitor that real-time tasks do not raise page faults
kernel/trace/rv/monitors/pagefault/pagefault.c new file +82 -0
82 lines not on this page · raw message
kernel/trace/rv/monitors/pagefault/pagefault.h new file +57 -0
57 lines not on this page · raw message
kernel/trace/rv/monitors/pagefault/pagefault_trace.h new file +14 -0
@@ -0,0 +1,14 @@ +/* SPDX-License-Identifier: GPL-2.0 */ + +/* +* Snippet to be included in rv_trace.h +*/ + +#ifdef CONFIG_RV_MON_PAGEFAULT +DEFINE_EVENT (event_ltl_monitor_id, event_pagefault, + TP_PROTO(struct task_struct *task, char *states, char *atoms, char *next), +TP_ARGS (task, states, atoms, next)); +DEFINE_EVENT (error_ltl_monitor_id, error_pagefault, + TP_PROTO(struct task_struct *task), +TP_ARGS (task)); +#endif /* CONFIG_RV_MON_PAGEFAULT */
kernel/trace/rv/rv_trace.h +1 -0
@@ -172,6 +172,7 @@ TRACE_EVENT(error_ltl_monitor_id,TP_printk ("%s[%d]: violation detected" ,__get_str (comm), __entry->pid) ); +#include <monitors/pagefault/pagefault_trace.h> // Add new monitors based on CONFIG_LTL_MON_EVENTS_ID here #endif /* CONFIG_LTL_MON_EVENTS_ID */ #endif /* _TRACE_RV_H */
tools/verification/models/rtapp/pagefault.ltl new file +1 -0
@@ -0,0 +1 @@ +RULE = always (RT imply not PAGEFAULT)
Signature · 1 line
2.39.5
PATCH v4 20/22 Add rtapp_sleep monitor
In reply to Nam CaoAdd a monitor for checking that real-time tasks do not go to sleep in a manner that may cause undesirable latency.
Also change
RV depends on TRACING
to
RV select TRACING
to avoid the following recursive dependency:
error: recursive dependency detected!
symbol TRACING is selected by PREEMPTIRQ_TRACEPOINTS symbol PREEMPTIRQ_TRACEPOINTS depends on TRACE_IRQFLAGS symbol TRACE_IRQFLAGS is selected by RV_MON_SLEEP symbol RV_MON_SLEEP depends on RV
symbol RV depends on TRACING
--- kernel/trace/rv/Kconfig | 3 +- kernel/trace/rv/Makefile | 1 + kernel/trace/rv/monitors/sleep/Kconfig | 13 + kernel/trace/rv/monitors/sleep/sleep.c | 201 +++++++++++++ kernel/trace/rv/monitors/sleep/sleep.h | 287 +++++++++++++++++++ kernel/trace/rv/monitors/sleep/sleep_trace.h | 14 + kernel/trace/rv/rv_trace.h | 1 + tools/verification/models/rtapp/sleep.ltl | 15 + 8 files changed, 534 insertions(+), 1 deletion(-) create mode 100644 kernel/trace/rv/monitors/sleep/Kconfig create mode 100644 kernel/trace/rv/monitors/sleep/sleep.c create mode 100644 kernel/trace/rv/monitors/sleep/sleep.h create mode 100644 kernel/trace/rv/monitors/sleep/sleep_trace.h create mode 100644 tools/verification/models/rtapp/sleep.ltl
kernel/trace/rv/Kconfig +2 -1
@@ -20,7 +20,7 @@ config RV_LTL_MONITOR menuconfig RV bool "Runtime Verification" - depends on TRACING + select TRACING help Enable the kernel runtime verification infrastructure. RV is a lightweight (yet rigorous) method that complements classical
@@ -43,6 +43,7 @@ source "kernel/trace/rv/monitors/snep/Kconfig" source "kernel/trace/rv/monitors/sncid/Kconfig" source "kernel/trace/rv/monitors/rtapp/Kconfig" source "kernel/trace/rv/monitors/pagefault/Kconfig" +source "kernel/trace/rv/monitors/sleep/Kconfig" # Add new monitors here config RV_REACTORS
kernel/trace/rv/Makefile +1 -0
@@ -14,6 +14,7 @@ obj-$(CONFIG_RV_MON_SNEP) += monitors/snep/snep.o obj-$(CONFIG_RV_MON_SNCID) += monitors/sncid/sncid.o obj-$(CONFIG_RV_MON_RTAPP) += monitors/rtapp/rtapp.o obj-$(CONFIG_RV_MON_PAGEFAULT) += monitors/pagefault/pagefault.o +obj-$(CONFIG_RV_MON_SLEEP) += monitors/sleep/sleep.o # Add new monitors here obj-$(CONFIG_RV_REACTORS) += rv_reactors.o obj-$(CONFIG_RV_REACT_PRINTK) += reactor_printk.o
kernel/trace/rv/monitors/sleep/Kconfig new file +13 -0
@@ -0,0 +1,13 @@ +# SPDX-License-Identifier: GPL-2.0-only +# +config RV_MON_SLEEP + depends on RV + select RV_LTL_MONITOR + depends on HAVE_SYSCALL_TRACEPOINTS + depends on RV_MON_RTAPP + select TRACE_IRQFLAGS + default y + select LTL_MON_EVENTS_ID + bool "sleep monitor" + help + Monitor that real-time tasks do not sleep in a manner that may cause undesirable latency.
kernel/trace/rv/monitors/sleep/sleep.c new file +201 -0
201 lines not on this page · raw message
kernel/trace/rv/monitors/sleep/sleep.h new file +287 -0
287 lines not on this page · raw message
kernel/trace/rv/monitors/sleep/sleep_trace.h new file +14 -0
@@ -0,0 +1,14 @@ +/* SPDX-License-Identifier: GPL-2.0 */ + +/* +* Snippet to be included in rv_trace.h +*/ + +#ifdef CONFIG_RV_MON_SLEEP +DEFINE_EVENT (event_ltl_monitor_id, event_sleep, + TP_PROTO(struct task_struct *task, char *states, char *atoms, char *next), +TP_ARGS (task, states, atoms, next)); +DEFINE_EVENT (error_ltl_monitor_id, error_sleep, + TP_PROTO(struct task_struct *task), +TP_ARGS (task)); +#endif /* CONFIG_RV_MON_SLEEP */
kernel/trace/rv/rv_trace.h +1 -0
@@ -173,6 +173,7 @@ TRACE_EVENT(error_ltl_monitor_id,TP_printk ("%s[%d]: violation detected" ,__get_str (comm), __entry->pid) );#include <monitors/pagefault/pagefault_trace.h> +#include <monitors/sleep/sleep_trace.h> // Add new monitors based on CONFIG_LTL_MON_EVENTS_ID here #endif /* CONFIG_LTL_MON_EVENTS_ID */ #endif /* _TRACE_RV_H */
tools/verification/models/rtapp/sleep.ltl new file +15 -0
@@ -0,0 +1,15 @@ +RULE = always (RT imply (SLEEP imply (RT_FRIENDLY_SLEEP or ALLOWLIST))) + +RT_FRIENDLY_SLEEP = (RT_VALID_SLEEP_REASON or KERNEL_THREAD) + and ((not WAKE) until RT_FRIENDLY_WAKE) + +RT_VALID_SLEEP_REASON = PI_FUTEX or NANOSLEEP + +RT_FRIENDLY_WAKE = WOKEN_BY_EQUAL_OR_HIGHER_PRIO + or WOKEN_BY_HARDIRQ + or WOKEN_BY_NMI + or KTHREAD_SHOULD_STOP + +ALLOWLIST = BLOCK_ON_RT_MUTEX + or TASK_IS_RCU + or TASK_IS_MIGRATION
Signature · 1 line
2.39.5
PATCH v4 21/22 Add documentation for rtapp monitor
In reply to Nam CaoAdd documentation describing the rtapp monitor.
--- Documentation/trace/rv/index.rst | 1 + Documentation/trace/rv/monitor_rtapp.rst | 107 +++++++++++++++++++++++ 2 files changed, 108 insertions(+) create mode 100644 Documentation/trace/rv/monitor_rtapp.rst
Documentation/trace/rv/index.rst +1 -0
@@ -14,3 +14,4 @@ Runtime Verification monitor_wip.rst monitor_wwnr.rst monitor_sched.rst + monitor_rtapp.rst
Documentation/trace/rv/monitor_rtapp.rst new file +107 -0
107 lines not on this page · raw message
Signature · 1 line
2.39.5
PATCH v4 22/22 Allow to configure the number of per-task monitor
In reply to Nam CaoNow that there are 2 monitors for real-time applications, users may want to enable both of them simultaneously. Make the number of per-task monitor configurable. Default it to 2 for now.
--- include/linux/rv.h | 9 +-------- include/linux/sched.h | 8 +++----- kernel/trace/rv/Kconfig | 9 +++++++++ kernel/trace/rv/monitors/rtapp/Kconfig | 1 + kernel/trace/rv/rv.c | 8 ++++---- 5 files changed, 18 insertions(+), 17 deletions(-)
include/linux/rv.h +1 -8
@@ -74,14 +74,7 @@ struct ltl_monitor {};#endif /* CONFIG_RV_LTL_MONITOR */ -/* -* Per-task RV monitors count. Nowadays fixed in RV_PER_TASK_MONITORS. -* If we find justification for more monitors, we can think about -* adding more or developing a dynamic method. So far, none of -* these are justified. -*/ -#define RV_PER_TASK_MONITORS 1 -#define RV_PER_TASK_MONITOR_INIT (RV_PER_TASK_MONITORS) +#define RV_PER_TASK_MONITOR_INIT (CONFIG_RV_PER_TASK_MONITORS)union rv_task_monitor {struct da_monitor da_mon;
include/linux/sched.h +3 -5
@@ -1623,12 +1623,10 @@ struct task_struct {#ifdef CONFIG_RV /* -* Per-task RV monitor. Nowadays fixed in RV_PER_TASK_MONITORS. -* If we find justification for more monitors, we can think -* about adding more or developing a dynamic method. So far, -* none of these are justified. +* Per-task RV monitor, fixed in CONFIG_RV_PER_TASK_MONITORS. +* If memory becomes a concern, we can think about a dynamic method. */ -union rv_task_monitor rv[RV_PER_TASK_MONITORS ]; +union rv_task_monitor rv[CONFIG_RV_PER_TASK_MONITORS ];#endif #ifdef CONFIG_USER_EVENTS
kernel/trace/rv/Kconfig +9 -0
@@ -32,6 +32,15 @@ menuconfig RV For further information, see: Documentation/trace/rv/runtime-verification.rst +config RV_PER_TASK_MONITORS + int "Maximum number of per-task monitor" + depends on RV + range 1 8 + default 2 + help + This option configures the maximum number of per-task RV monitors that can run + simultaneously. + source "kernel/trace/rv/monitors/wip/Kconfig" source "kernel/trace/rv/monitors/wwnr/Kconfig" source "kernel/trace/rv/monitors/sched/Kconfig"
kernel/trace/rv/monitors/rtapp/Kconfig +1 -0
@@ -1,5 +1,6 @@ config RV_MON_RTAPP depends on RV + depends on RV_PER_TASK_MONITORS >= 2 bool "rtapp monitor" help Collection of monitors to check for common problems with real-time application that cause
kernel/trace/rv/rv.c +4 -4
@@ -165,7 +165,7 @@ struct dentry *get_monitors_root(void)LIST_HEAD (rv_monitors_list);static int task_monitor_count; -static bool task_monitor_slots[RV_PER_TASK_MONITORS ]; +static bool task_monitor_slots[CONFIG_RV_PER_TASK_MONITORS ];int rv_get_task_monitor_slot (void ) {
@@ -173,12 +173,12 @@ int rv_get_task_monitor_slot(void)lockdep_assert_held (&rv_interface_lock); -if (task_monitor_count ==RV_PER_TASK_MONITORS ) +if (task_monitor_count ==CONFIG_RV_PER_TASK_MONITORS )return -EBUSY ; task_monitor_count++; -for (i =0 ; i <RV_PER_TASK_MONITORS ; i++) { +for (i =0 ; i <CONFIG_RV_PER_TASK_MONITORS ; i++) {if (task_monitor_slots[i] == false) { task_monitor_slots[i] = true;return i;
@@ -194,7 +194,7 @@ void rv_put_task_monitor_slot(int slot) {lockdep_assert_held (&rv_interface_lock); -if (slot <0 || slot >=RV_PER_TASK_MONITORS ) { +if (slot <0 || slot >=CONFIG_RV_PER_TASK_MONITORS ) {WARN_ONCE (1 ,"RV releasing an invalid slot!: %d\n" , slot);return ; }
Signature · 1 line
2.39.5
Re: [PATCH v4 19/22] rv: Add rtapp_pagefault monitor
In reply to Nam CaoQuoted 96 lines · Nam Cao
Userspace real-time applications may have design flaws that they raise
page faults in real-time threads, and thus have unexpected latencies.Add an linear temporal logic monitor to detect this scenario.
Signed-off-by Nam Cao <[email protected]>---
kernel/trace/rv/Kconfig | 1 + kernel/trace/rv/Makefile | 1 + kernel/trace/rv/monitors/pagefault/Kconfig | 11 +++ .../trace/rv/monitors/pagefault/pagefault.c | 82
+++++++++++++++++++
.../trace/rv/monitors/pagefault/pagefault.h | 57 +++++++++++++ .../rv/monitors/pagefault/pagefault_trace.h | 14 ++++ kernel/trace/rv/rv_trace.h | 1 +tools/verification/models/rtapp/pagefault.ltl| 1 +8 files changed, 168 insertions(+)create mode 100644
kernel/trace/rv/monitors/pagefault/Kconfigcreate mode 100644kernel/trace/rv/monitors/pagefault/pagefault.ccreate mode 100644kernel/trace/rv/monitors/pagefault/pagefault.hcreate mode 100644kernel/trace/rv/monitors/pagefault/pagefault_trace.hcreate mode 100644tools/verification/models/rtapp/pagefault.ltlkernel/trace/rv/Kconfig +0
-0@@ -42,6 +42,7 @@ source "kernel/trace/rv/monitors/scpd/Kconfig"source "
kernel/trace/rv/monitors/snep/Kconfig" source "kernel/trace/rv/monitors/sncid/Kconfig" source "kernel/trace/rv/monitors/rtapp/Kconfig" +source "kernel/trace/rv/monitors/pagefault/Kconfig" # Add new monitors here
configRV_REACTORSkernel/trace/rv/Makefile +0
-0@@ -13,6 +13,7 @@ obj-$(CONFIG_RV_MON_SCPD) += monitors/scpd/scpd.oobj-$(CONFIG_RV_MON_SNEP) +=
monitors/snep/snep.oobj-$(CONFIG_RV_MON_SNCID) +=monitors/sncid/sncid.oobj-$(CONFIG_RV_MON_RTAPP) +=monitors/rtapp/rtapp.o+obj-$(CONFIG_RV_MON_PAGEFAULT) +=monitors/pagefault/pagefault.o# Add new monitors here
obj-$(CONFIG_RV_REACTORS) += rv_reactors.o obj-$(CONFIG_RV_REACT_PRINTK) += reactor_printk.okernel/trace/rv/monitors/pagefault/Kconfig new file +11
-0@@ -0,0 +1,11 @@ +# SPDX-License-Identifier: GPL-2.0-only +# +config RV_MON_PAGEFAULT + depends on RV + select RV_LTL_MONITOR + depends on RV_MON_RTAPP + default y + select LTL_MON_EVENTS_ID + bool "pagefault monitor" + help + Monitor that real-time tasks do not raise page faultskernel/trace/rv/monitors/pagefault/pagefault.c new file +21
-0@@ -0,0 +1,82 @@ +// SPDX-License-Identifier: GPL-2.0 +#include <linux/ftrace.h> +#include <linux/init.h> +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/rv.h> +#include <linux/sched/deadline.h> +#include <linux/sched/rt.h> +#include <linux/tracepoint.h> +#include <rv/instrumentation.h> + +#define MODULE_NAME "pagefault" + +#include <rv_trace.h> +#include <trace/events/exceptions.h> +#include <monitors/rtapp/rtapp.h> + +#include "pagefault.h" +#include <rv/ltl_monitor.h> + +static void ltl_atoms_fetch(struct task_struct *task, struct
ltl_monitor*mon)
kernel/trace/rv/monitors/pagefault/pagefault.c line 24
Quoted 3 lines
kernel/trace/rv/monitors/pagefault/pagefault.c new file +3
-0+{ +ltl_atom_set (mon,LTL_RT ,rt_or_dl_task (task)); +}
Mmh, you probably already considered that, so ignore my comment in that case.
I just realised this function would tell you a PI boosted task is an RT task, is that acceptable in your model?
It's probably a configuration mistake on its own if an RT task following those rules shares resources with non-RT tasks not following them, but if that's something allowed, you may see this atom change more often than you'd like, not sure if that can be something worth noting.
Perhaps you could add a comment saying that this is not a problem and why (e.g. instead of using rt_or_dl_task_policy for the job).
What do you think?
Besides this detail, the monitor looks good to me
Thanks,
Gabriele
Re: [PATCH v4 19/22] rv: Add rtapp_pagefault monitor
Quoted 11 lines · Gabriele Monaco
kernel/trace/rv/monitors/pagefault/pagefault.c line 24
Quoted 5 lines · Nam Cao
+static void
ltl_atoms_fetch(struct task_struct *task, struct ltl_monitor *mon)kernel/trace/rv/monitors/pagefault/pagefault.c new file +3
-0+{ +ltl_atom_set (mon,LTL_RT ,rt_or_dl_task (task)); +}Mmh, you probably already considered that, so ignore my comment in that case.
I just realised this function would tell you a PI boosted task is an RT task, is that acceptable in your model?
Yes, that is intentional. A task being PI boosted means an "actual" real-time task is waiting for it. Therefore the PI boosted task shouldn't be delayed, otherwise the "actual" real-time task is delayed.
Quoted 2 lines
It's probably a configuration mistake on its own if an RT task following those rules shares resources with non-RT tasks not following them,
non-RT tasks do not have to follow the rules all the time. But while accessing resources shared with RT tasks, then yes the rules should be obeyed.
Quoted 5 lines
but if that's something allowed, you may see this atom change more often than you'd like, not sure if that can be something worth noting.
Perhaps you could add a comment saying that this is not a problem and why (e.g. instead of using
rt_or_dl_task_policyfor the job).
Sure, a comment doesn't hurt.
Quoted 4 lines
What do you think?
Besides this detail, the monitor looks good to me
Reviewed-by Gabriele Monaco <[email protected]>
Thanks so much for the review!
Nam
Re: [PATCH v4 20/22] rv: Add rtapp_sleep monitor
In reply to Nam CaoQuoted 17 lines · Nam Cao
Add a monitor for checking that real-time tasks do not go to sleep in a
manner that may cause undesirable latency.Also change
RV depends on TRACING
to
RV select TRACING
to avoid the following recursive dependency:error: recursive dependency detected!
symbol TRACING is selected byPREEMPTIRQ_TRACEPOINTSsymbolPREEMPTIRQ_TRACEPOINTSdepends onTRACE_IRQFLAGSsymbolTRACE_IRQFLAGSis selected byRV_MON_SLEEPsymbolRV_MON_SLEEPdepends on RV
symbol RV depends on TRACING
Quoted 3 lines
Signed-off-by Nam Cao <[email protected]>---
I've been playing with these monitors, code-wise they look good. I tested a bit and they seem to work without many surprises by doing something as simple as:
perf stat -e rv:error_sleep stress-ng --cpu-sched 1 -t 10s
-- shows several errors --
perf stat -e rv:error_sleep stress-ng --prio-inv 1 --prio-inv-policy rr
-- shows only 1 error (normal while starting the program?) --
Not quite sound, but does it look a reasonable test to you?
I quickly tried the same with the other monitor comparing the number of errors with the page_faults generated by perf, but that didn't make too much sense. Perhaps I'm doing something wrong here though (the number reported by perf for page faults feels a bit too high).
perf stat -e page-faults -e rv:error_pagefault stress-ng --cyclic 1
Anyway, the monitor looks good to me
but it'd be nice if you have tips to share how to quickly test it (e.g. without writing a custom workload).
Thanks,
Gabriele
Re: [PATCH v4 20/22] rv: Add rtapp_sleep monitor
Quoted 6 lines · Gabriele Monaco
I've been playing with these monitors, code-wise they look good. I tested a bit and they seem to work without many surprises by doing something as simple as:
perf stat -e rv:error_sleep stress-ng --cpu-sched 1 -t 10s
-- shows several errors --
This one is a monitor's bug.
The monitor mistakenly sees the task getting woken up, *then* sees it going to sleep.
This is due to trace_sched_switch() being called with a stale 'prev_state'. 'prev_state' is read at the beginning of __schedule(), but trace_sched_switch() is invoked a bit later. Therefore if task->__state is changed inbetween, 'prev_state' is not the value of task->__state.
The monitor checks (prev_state & TASK_INTERRUPTIBLE) to determine if the task is going to sleep. This can be incorrect due to the race above. The monitor sees the task going to sleep, but actually it is just preempted.
I think this also answers the race you observed with the srs monitor?
Quoted 4 lines
perf stat -e rv:error_sleep stress-ng --prio-inv 1 --prio-inv-policy rr
-- shows only 1 error (normal while starting the program?) --Not quite sound, but does it look a reasonable test to you?
The above command use mutexes with priority inheritance. That is good for real-time. The errors are due to real-time tasks being delayed by waitpid().
Priority inheritance can be disabled with "--prio-inv-type none". Then you will see lots of errors with mutexes.
Quoted 6 lines
I quickly tried the same with the other monitor comparing the number of errors with the
page_faultsgenerated by perf, but that didn't make too much sense. Perhaps I'm doing something wrong here though (the number reported by perf for page faults feels a bit too high).perf stat -e page-faults -e rv:error_pagefault stress-ng --cyclic 1
This command run a non-real-time thread to do setup, and a cyclic real-time thread. The number of pagefaults of each thread would be roughly proportional to the code size executed by each thread. As the non-real-time thread's code size is bigger, it sounds reasonable that the number of pagefaults is greater than the number of monitor's warnings.
Quoted 7 lines
Anyway, the monitor looks good to me
Reviewed-by Gabriele Monaco <[email protected]>but it'd be nice if you have tips to share how to quickly test it (e.g. without writing a custom workload).
I tested the monitor on a real system. My system has some real-time audio processing processes (pipewire, firefox running youtube), yours also should.
But thanks so much for testing with stress-ng. My testing didn't stress the system enough for the above race to happen. I will give stress-ng a few runs before the next version.
Best regards,
Nam
Gabriele Monaco
maintainerRe: [PATCH v4 20/22] rv: Add rtapp_sleep monitor
2025-04-25T06:35:09Z Nam Cao <[email protected]>:
Quoted 18 lines · Gabriele Monaco
Quoted 6 lines
I've been playing with these monitors, code-wise they look good. I tested a bit and they seem to work without many surprises by doing something as simple as:
perf stat -e rv:error_sleep stress-ng --cpu-sched 1 -t 10s -- shows several errors --
This one is a monitor's bug.
The monitor mistakenly sees the task getting woken up, *then* sees it going to sleep.
This is due to
trace_sched_switch()being called with a stale 'prev_state'. 'prev_state' is read at the beginning of__schedule(), buttrace_sched_switch()is invoked a bit later. Therefore iftask->__stateis changed inbetween, 'prev_state' is not the value oftask->__state.
Quoted 6 lines
The monitor checks (
prev_state&TASK_INTERRUPTIBLE) to determine if the task is going to sleep. This can be incorrect due to the race above. The monitor sees the task going to sleep, but actually it is just preempted.I think this also answers the race you observed with the srs monitor?
Yeah that could be the culprit.
Peter's fix [1] landed on next recently, I guess in a couple of days you'll get it on the upstream tree and you may not see the problem.
Nevertheless, I didn't check exactly what --cpu-sched does, but I'm expecting it to do any wild thing to stress the scheduler, so it may be normal to have more errors than, say, --cyclic, which only runs nanosleeps.
Quoted 12 lines
Quoted 4 lines
perf stat -e rv:error_sleep stress-ng --prio-inv 1 --prio-inv-policy rr -- shows only 1 error (normal while starting the program?) --
Not quite sound, but does it look a reasonable test to you?
The above command use mutexes with priority inheritance. That is good for real-time. The errors are due to real-time tasks being delayed by
waitpid().Priority inheritance can be disabled with "--prio-inv-type none". Then you will see lots of errors with mutexes.
Great, that's exactly what I wanted to know, thanks.
Quoted 5 lines
Quoted 5 lines
I quickly tried the same with the other monitor comparing the number of errors with the
page_faultsgenerated by perf, but that didn't make too much sense. Perhaps I'm doing something wrong here though (the number reported by perf for page faults feels a bit too high).
Quoted 7 lines
Quoted 1 line
perf stat -e page-faults -e rv:error_pagefault stress-ng --cyclic 1
This command run a non-real-time thread to do setup, and a cyclic real-time thread. The number of pagefaults of each thread would be roughly proportional to the code size executed by each thread. As the non-real-time thread's code size is bigger, it sounds reasonable that the number of pagefaults is greater than the number of monitor's warnings.
Mmh I guessed something like that, although numbers were a bit out of proportion (e.g. 500 page-faults and 8 errors), but again, I didn't check too carefully what happens under the hood.
Quoted 11 lines
Quoted 7 lines
Anyway, the monitor looks good to me
Reviewed-by Gabriele Monaco <[email protected]>but it'd be nice if you have tips to share how to quickly test it (e.g. without writing a custom workload).
I tested the monitor on a real system. My system has some real-time audio processing processes (pipewire, firefox running youtube), yours also should.
That's a good point, also I didn't mention I was running these tests in a VM (virtme-ng), so the system stress is minimal and perhaps the setup triggers some different oddities (filesystems are overlays and some other things are set up differently from a real system).
Quoted 7 lines
But thanks so much for testing with stress-ng. My testing didn't stress the system enough for the above race to happen. I will give stress-ng a few runs before the next version.
Best regards,
Nam
Thank you for the tips!
Cheers,
Gabriele
John Ogness
maintainerRe: [PATCH v4 20/22] rv: Add rtapp_sleep monitor
In reply to Nam CaoQuoted 13 lines · Nam Cao
Quoted 6 lines · Gabriele Monaco
I've been playing with these monitors, code-wise they look good. I tested a bit and they seem to work without many surprises by doing something as simple as:
perf stat -e rv:error_sleep stress-ng --cpu-sched 1 -t 10s
-- shows several errors --This one is a monitor's bug.
The monitor mistakenly sees the task getting woken up, *then* sees it going to sleep.
Quoted 8 lines
This is due to
trace_sched_switch()being called with a stale 'prev_state'. 'prev_state' is read at the beginning of__schedule(), buttrace_sched_switch()is invoked a bit later. Therefore iftask->__stateis changed inbetween, 'prev_state' is not the value oftask->__state.The monitor checks (
prev_state&TASK_INTERRUPTIBLE) to determine if the task is going to sleep. This can be incorrect due to the race above. The monitor sees the task going to sleep, but actually it is just preempted.
If I understand this correctly, trace_sched_switch() is reporting accurate state transition information, but by the time it is reported that state may have already changed (in which case another trace_sched_switch() occurs later).
So in this example, the task did go to sleep. Why do you think it was preempted instead?
John Ogness
John Ogness
maintainerRe: [PATCH v4 20/22] rv: Add rtapp_sleep monitor
Quoted 24 lines · John Ogness
Quoted 8 lines
Quoted 2 lines
perf stat -e rv:error_sleep stress-ng --cpu-sched 1 -t 10s
-- shows several errors --This one is a monitor's bug.
The monitor mistakenly sees the task getting woken up, *then* sees it going to sleep.
Quoted 8 lines
This is due to
trace_sched_switch()being called with a stale 'prev_state'. 'prev_state' is read at the beginning of__schedule(), buttrace_sched_switch()is invoked a bit later. Therefore iftask->__stateis changed inbetween, 'prev_state' is not the value oftask->__state.The monitor checks (
prev_state&TASK_INTERRUPTIBLE) to determine if the task is going to sleep. This can be incorrect due to the race above. The monitor sees the task going to sleep, but actually it is just preempted.If I understand this correctly,
trace_sched_switch()is reporting accurate state transition information, but by the time it is reported that state may have already changed (in which case anothertrace_sched_switch()occurs later).So in this example, the task did go to sleep. Why do you think it was preempted instead?
Quoted 2 lines · Gabriele Monaco
Peter's fix [1] landed on next recently, I guess in a couple of days you'll get it on the upstream tree and you may not see the problem.
Ah, thanks for pointing that out!
Re: [PATCH v4 20/22] rv: Add rtapp_sleep monitor
Quoted 8 lines · John Ogness
Quoted 7 lines · John Ogness
If I understand this correctly,
trace_sched_switch()is reporting accurate state transition information, but by the time it is reported that state may have already changed (in which case anothertrace_sched_switch()occurs later).So in this example, the task did go to sleep. Why do you think it was preempted instead?
You are right, the task did go to sleep. Scratch what I said earlier.
The monitor checks that if an RT task going to sleep, it will be woken by a "RT-friendly" source. The problem is that trace_sched_switch() may appear after trace_sched_waking(). The monitor sees the task sleeps, and waits until the task is woken and checks the waker. But the monitor doesn't see the task being woken, because it has already happened before the task sleeps.
For correct ordering, we could:
- Use trace_sched_wakeup() instead of trace_sched_waking(). This would have correct order, but information about the waker is gone at trace_sched_wakeup(), so it doesn't work.
- Use trace_set_current_state() instead of trace_sched_switch() to determine task going to sleep.
The latter option works, but then "sleep" would be defined as task doing set_current_state(TASK_INTERRUPTIBLE). This new definition is probably not precise? But for the monitor, it's fine.
Btw, while testing this, I discovered another bug. Real-time thread may "legally" sleep by a 'restart' syscall after 'nanosleep'. The monitor doesn't recognize this syscall as valid sleep reason and flags it :(
Quoted 3 lines · Gabriele Monaco
Quoted 2 lines
Peter's fix [1] landed on next recently, I guess in a couple of days you'll get it on the upstream tree and you may not see the problem.
This patch fixes stale prev_state due to signaling. It doesn't fix this case.
Best regards,
Nam
Re: [PATCH v4 20/22] rv: Add rtapp_sleep monitor
In reply to Gabriele MonacoQuoted 15 lines · Gabriele Monaco
2025-04-25T06:35:09Z Nam Cao <[email protected]>:Quoted 6 lines · Gabriele Monaco
Quoted 5 lines
I quickly tried the same with the other monitor comparing the number of errors with the
page_faultsgenerated by perf, but that didn't make too much sense. Perhaps I'm doing something wrong here though (the number reported by perf for page faults feels a bit too high).Quoted 7 lines
Quoted 1 line
perf stat -e page-faults -e rv:error_pagefault stress-ng --cyclic 1
This command run a non-real-time thread to do setup, and a cyclic real-time thread. The number of pagefaults of each thread would be roughly proportional to the code size executed by each thread. As the non-real-time thread's code size is bigger, it sounds reasonable that the number of pagefaults is greater than the number of monitor's warnings.
Quoted 3 lines
Mmh I guessed something like that, although numbers were a bit out of proportion (e.g. 500 page-faults and 8 errors), but again, I didn't check too carefully what happens under the hood.
Keep in mind that the non-real-time thread is calling into glibc. While the real-time thread is a small loop doing nanosleep.
Quoted 8 lines
Quoted 3 lines
I tested the monitor on a real system. My system has some real-time audio processing processes (pipewire, firefox running youtube), yours also should.
That's a good point, also I didn't mention I was running these tests in a VM (virtme-ng), so the system stress is minimal and perhaps the setup triggers some different oddities (filesystems are overlays and some other things are set up differently from a real system).
Oddities are good, they make some corner cases appear.
Evidently I need to torture the monitors much more, let's see what else shows up..
Best regards,
Nam
This patch was applied to riscv/linux.git (fixes) by Steven Rostedt (Google) <[email protected]>:
Hello:
This patch was applied to riscv/linux.git (fixes)
by Steven Rostedt (Google) <[email protected]>:
Quoted 10 lines · On Wed, 23 Apr 2025 08:49:55 +0200 you wrote:
Real-time applications may have design flaws causing them to have unexpected latency. For example, the applications may raise page faults, or may be blocked trying to take a mutex without priority inheritance.
However, while attempting to implement DA monitors for these real-time rules, deterministic automaton is found to be inappropriate as the specification language. The automaton is complicated, hard to understand, and error-prone.
[...]
Here is the summary with links:
- [v4,18/22] riscv: mm: Add page fault trace points
https://git.kernel.org/riscv/c/a37c71ca412d
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html