GenericExperiment: A standardized fault injection experiment

The GenericExperiment is a standard campaign+experiment pair. It is
derived from the DatabaseCampaign+DatabaseExperiment. Its experiment
endpoints are set on the command line, therefore it can be used to give
users a first impression of FAIL*. Currently it supports different
endpoints:

--trap: Catch all traps that occur and end the experiment
--timeout <N>: kill the experiment after N microseconds
--catch-write-text: detect writes on the text segment
--catch-write-outerspace: detect writes into nirvana
--{ok,fail,detected}-marker: groups of ELF symbols that are used as
     execution breakpoints

Change-Id: Idc7fcf8875953f1007e1a37bacb086eddd29cd10
This commit is contained in:
Christian Dietrich
2014-06-05 16:01:20 +02:00
parent d53c83d7c8
commit 491312bb91
9 changed files with 524 additions and 0 deletions

View File

@ -0,0 +1,111 @@
#ifndef __GENERIC_EXPERIMENT_EXPERIMENT_HPP__
#define __GENERIC_EXPERIMENT_EXPERIMENT_HPP__
#include "sal/SALInst.hpp"
#include "efw/DatabaseExperiment.hpp"
#include "sal/Listener.hpp"
#include "efw/JobClient.hpp"
#include "util/Logger.hpp"
#include "util/ElfReader.hpp"
#include <string>
#include <stdlib.h>
#include <map>
#include <set>
class GenericExperiment : public fail::DatabaseExperiment {
fail::ElfReader m_elf;
std::string m_state_dir;
bool enabled_mem_text;
fail::MemAccessListener l_mem_text;
bool enabled_mem_outerspace;
fail::MemAccessListener l_mem_outerspace;
bool enabled_trap;
fail::TrapListener l_trap;
bool enabled_timeout;
unsigned m_Timeout;
fail::TimerListener l_timeout;
std::map<fail::BaseListener *, const fail::ElfSymbol *> listener_to_symbol;
typedef std::set<fail::BaseListener *> ListenerSet;
ListenerSet end_markers;
ListenerSet OK_marker;
ListenerSet FAIL_marker;
ListenerSet DETECTED_marker;
ListenerSet GROUP1_marker;
ListenerSet GROUP2_marker;
ListenerSet GROUP3_marker;
ListenerSet GROUP4_marker;
std::map<std::string, ListenerSet * > end_marker_groups;
void parseSymbols(const std::string &args, std::set<fail::BaseListener *> *into);
public:
GenericExperiment() : DatabaseExperiment("GenericExperiment"),
m_state_dir("state"),
l_trap(fail::ANY_TRAP), l_timeout(0) {
enabled_mem_text = false;
enabled_mem_outerspace = false;
enabled_trap = false;
enabled_timeout = false;
end_marker_groups["ok-marker"] = &OK_marker;
end_marker_groups["fail-marker"] = &FAIL_marker;
end_marker_groups["detected-marker"] = &FAIL_marker;
end_marker_groups["group1-marker"] = &GROUP1_marker;
end_marker_groups["group2-marker"] = &GROUP2_marker;
end_marker_groups["group3-marker"] = &GROUP3_marker;
end_marker_groups["group4-marker"] = &GROUP4_marker;
}
virtual ~GenericExperiment();
/**
* Get path to the state directory
*/
virtual std::string cb_state_directory() { return m_state_dir; }
/**
* Allocate enough space to hold the incoming ExperimentData message.
*/
virtual fail::ExperimentData* cb_allocate_experiment_data();
/**
* Allocate a new result slot in the given experiment data
*/
virtual google::protobuf::Message* cb_new_result(fail::ExperimentData* data);
/**
* Callback that is called, before the actual experiment
* starts. Simulation is terminated on false.
* @param The current result message
* @return \c true on success, \c false otherwise
*/
virtual bool cb_start_experiment();
/**
* Callback that is called before the resuming till crash has
* started. This is called after the fault was injected. Here the
* end listeners should be installed. Returns true on
* success. Otherwise the experiment is canceled.
* @return \c true on success, \c false otherwise
*/
virtual bool cb_before_resume();
/**
* Callback that is called after the resume-till-crash phase with
* the last triggered listener. This callback should collect all
* data and fill up the result message.
*/
virtual void cb_after_resume(fail::BaseListener *event);
};
#endif // __GENERIC_EXPERIMENT_EXPERIMENT_HPP__