|
DynamoRIO
|
#include <scheduler.h>
Public Member Functions | |
| virtual stream_status_t | next_record (RecordType &record) |
| virtual stream_status_t | next_record (RecordType &record, uint64_t cur_time) |
| virtual stream_status_t | unread_last_record () |
| virtual stream_status_t | start_speculation (addr_t start_address, bool queue_current_record) |
| virtual stream_status_t | stop_speculation () |
| virtual stream_status_t | set_active (bool active) |
| uint64_t | get_record_ordinal () const override |
| uint64_t | get_instruction_ordinal () const override |
| uint64_t | get_output_instruction_ordinal () const |
| std::string | get_stream_name () const override |
| virtual input_ordinal_t | get_input_stream_ordinal () const |
| virtual int | get_input_workload_ordinal () const |
| uint64_t | get_last_timestamp () const override |
| uint64_t | get_first_timestamp () const override |
| uint64_t | get_version () const override |
| uint64_t | get_filetype () const override |
| uint64_t | get_cache_line_size () const override |
| uint64_t | get_chunk_instr_count () const override |
| uint64_t | get_page_size () const override |
| bool | is_record_synthetic () const override |
| int64_t | get_output_cpuid () const override |
| int64_t | get_workload_id () const override |
| int64_t | get_input_id () const override |
| int64_t | get_tid () const override |
| memtrace_stream_t * | get_input_interface () const override |
| output_ordinal_t | get_output_stream_ordinal () const |
| int | get_shard_index () const override |
| bool | is_record_kernel () const override |
| double | get_schedule_statistic (schedule_statistic_t stat) const override |
 Public Member Functions inherited from dynamorio::drmemtrace::memtrace_stream_t | |
| virtual | ~memtrace_stream_t () |
Detailed Description
template<typename RecordType, typename ReaderType>
class dynamorio::drmemtrace::scheduler_tmpl_t< RecordType, ReaderType >::stream_t
Represents a stream of RecordType trace records derived from a subset of a set of input recorded traces. Provides more information about the record stream using the dynamorio::drmemtrace::memtrace_stream_t API.
Member Function Documentation
◆ get_cache_line_size()
|
inlineoverridevirtual |
Returns the cache line size from the TRACE_MARKER_TYPE_CACHE_LINE_SIZE record in the trace header.
Implements dynamorio::drmemtrace::memtrace_stream_t.
◆ get_chunk_instr_count()
|
inlineoverridevirtual |
Returns the chunk instruction count from the TRACE_MARKER_TYPE_CHUNK_INSTR_COUNT record in the trace header.
Implements dynamorio::drmemtrace::memtrace_stream_t.
◆ get_filetype()
|
inlineoverridevirtual |
Returns the OFFLINE_FILE_TYPE_* bitfields of type offline_file_type_t identifying the architecture and other key high-level attributes of the trace from the TRACE_MARKER_TYPE_FILETYPE record in the trace header. This can be queried prior to explicitly retrieving any records from output streams, unless dynamorio::drmemtrace::scheduler_tmpl_t:: scheduler_options_t.read_inputs_in_init is false.
Implements dynamorio::drmemtrace::memtrace_stream_t.
◆ get_first_timestamp()
|
overridevirtual |
Returns the value of the first seen TRACE_MARKER_TYPE_TIMESTAMP marker.
Implements dynamorio::drmemtrace::memtrace_stream_t.
◆ get_input_id()
|
inlineoverridevirtual |
Returns the ordinal for the current input stream feeding this output stream.
Reimplemented from dynamorio::drmemtrace::memtrace_stream_t.
◆ get_input_interface()
|
overridevirtual |
Returns the dynamorio::drmemtrace::memtrace_stream_t interface for the current input stream feeding this output stream.
Reimplemented from dynamorio::drmemtrace::memtrace_stream_t.
◆ get_input_stream_ordinal()
|
virtual |
Returns the ordinal for the current input stream feeding this output stream.
◆ get_input_workload_ordinal()
|
virtual |
Returns the ordinal for the workload which is the source of the current input stream feeding this output stream. This workload ordinal is the index into the vector of type dynamorio::drmemtrace::scheduler_tmpl_t::input_workload_t passed to init(). Returns -1 if there is no current input for this output stream.
◆ get_instruction_ordinal()
|
overridevirtual |
Returns the count of instructions from the start of the trace to this point. For record_scheduler_t, if any encoding records or the internal record TRACE_MARKER_TYPE_BRANCH_TARGET records are present prior to an instruction marker, the count will increase at the first of those records as they are considered part of the instruction. If SCHEDULER_USE_INPUT_ORDINALS is set, then this value matches the instruction ordinal for the current input stream (and thus might decrease or not change across records if the input changed). Otherwise, if multiple input streams fed into this output stream, this includes the records from all those streams that were presented here: thus, this may be larger than what the current input stream reports (see get_input_stream_interface() and get_input_stream_ordinal()). This does not advance across skipped records in an input stream from a region of interest (see dynamorio::drmemtrace::scheduler_tmpl_t::range_t), but it does advance if the output stream skipped ahead.
Implements dynamorio::drmemtrace::memtrace_stream_t.
◆ get_last_timestamp()
|
overridevirtual |
Returns the value of the most recently seen TRACE_MARKER_TYPE_TIMESTAMP marker.
Implements dynamorio::drmemtrace::memtrace_stream_t.
◆ get_output_cpuid()
|
overridevirtual |
Returns a unique identifier for the current output stream. For MAP_TO_RECORDED_OUTPUT, the identifier is the as-traced cpuid mapped to this output. For dynamic schedules, the identifier is the output stream ordinal, except for OFFLINE_FILE_TYPE_CORE_SHARDED inputs where the identifier is the input stream ordinal.
Reimplemented from dynamorio::drmemtrace::memtrace_stream_t.
◆ get_output_instruction_ordinal()
|
inline |
Identical to get_instruction_ordinal() but ignores the SCHEDULER_USE_INPUT_ORDINALS flag.
◆ get_output_stream_ordinal()
|
inline |
Returns the ordinal for the current output stream. If dynamorio::drmemtrace::scheduler_tmpl_t::scheduler_options_t:: single_lockstep_output is set to true, this returns the ordinal of the currently active "inner" output stream. Otherwise, this returns the constant ordinal for this output stream as there is no concept of inner or outer streams.
◆ get_page_size()
|
inlineoverridevirtual |
Returns the page size from the TRACE_MARKER_TYPE_PAGE_SIZE record in the trace header.
Implements dynamorio::drmemtrace::memtrace_stream_t.
◆ get_record_ordinal()
|
overridevirtual |
Returns the count of memref_t records from the start of the trace to this point. It does not include synthetic records (see is_record_synthetic()).
If SCHEDULER_USE_INPUT_ORDINALS is set, then this value matches the record ordinal for the current input stream (and thus might decrease or not change across records if the input changed). Otherwise, if multiple input streams fed into this output stream, this includes the records from all those streams that were presented here: thus, this may be larger than what the current input stream reports (see get_input_stream_interface() and get_input_stream_ordinal()). This does not advance across skipped records in an input stream from a region of interest (see dynamorio::drmemtrace::scheduler_tmpl_t::range_t), but it does advance if the output stream skipped ahead.
Implements dynamorio::drmemtrace::memtrace_stream_t.
◆ get_schedule_statistic()
|
overridevirtual |
Returns the value of the specified statistic for this output stream. The values for all output streams must be summed to obtain global counts. These statistics are not guaranteed to be accurate when replaying a prior schedule via MAP_TO_RECORDED_OUTPUT.
Reimplemented from dynamorio::drmemtrace::memtrace_stream_t.
◆ get_shard_index()
|
overridevirtual |
For SCHEDULER_USE_INPUT_ORDINALS or SCHEDULER_USE_SINGLE_INPUT_ORDINALS, returns the input stream ordinal, except for the case of a single combined-stream input with the passed-in thread id set to INVALID_THREAD_ID (the serial analysis mode for analyzer tools) in which case the last trace record's tid is returned; otherwise returns the output stream ordinal.
Reimplemented from dynamorio::drmemtrace::memtrace_stream_t.
◆ get_stream_name()
|
overridevirtual |
Returns a name for the current input stream feeding this output stream. For stored offline traces, this is the base name of the trace on disk. For online traces, this is the name of the pipe.
Implements dynamorio::drmemtrace::memtrace_stream_t.
◆ get_tid()
|
overridevirtual |
Returns the thread identifier for the current input stream feeding this output stream.
Reimplemented from dynamorio::drmemtrace::memtrace_stream_t.
◆ get_version()
|
inlineoverridevirtual |
Returns the trace_version_t value from the TRACE_MARKER_TYPE_VERSION record in the trace header. This can be queried prior to explicitly retrieving any records from output streams, unless dynamorio::drmemtrace::scheduler_tmpl_t:: scheduler_options_t.read_inputs_in_init is false.
Implements dynamorio::drmemtrace::memtrace_stream_t.
◆ get_workload_id()
|
inlineoverridevirtual |
Returns the ordinal for the current dynamorio::drmemtrace::scheduler_tmpl_t::input_workload_t.
Reimplemented from dynamorio::drmemtrace::memtrace_stream_t.
◆ is_record_kernel()
|
overridevirtual |
Returns whether the current record is from a part of the trace corresponding to kernel execution.
Reimplemented from dynamorio::drmemtrace::memtrace_stream_t.
◆ is_record_synthetic()
|
overridevirtual |
Returns whether the current record was synthesized and inserted into the record stream and was not present in the original stream. This is true for timestamp and cpuid headers duplicated after skipping ahead, as well as cpuid markers inserted for synthetic schedules. Such records do not cound toward the record count and get_record_ordinal() will return the value of the prior record.
Reimplemented from dynamorio::drmemtrace::memtrace_stream_t.
◆ next_record() [1/2]
|
virtual |
Advances to the next record in the stream. Returns a status code on whether and how to continue. Uses the instruction count plus idle count to this point as the time; use the variant that takes "cur_time" to instead provide a time.
◆ next_record() [2/2]
|
virtual |
Advances to the next record in the stream. Returns a status code on whether and how to continue. Supplies the current time for QUANTUM_TIME. The time should be considered to be the simulated time prior to processing the returned record. The time's units can be chosen by the caller, with dynamorio::drmemtrace::scheduler_tmpl_t::scheduler_options_t.time_units_per_us providing the conversion to simulated microseconds. STATUS_INVALID is returned if 0 or a value smaller than the start time of the current input's quantum is passed in when QUANTUM_TIME and MAP_TO_ANY_OUTPUT are specified.
◆ set_active()
|
virtual |
Disables or re-enables this output stream. If "active" is false, this stream becomes inactive and its currently assigned input is moved to the ready queue to be scheduled on other outputs. The STATUS_IDLE code is returned to next_record() for inactive streams. If "active" is true, this stream becomes active again. This is only supported for MAP_TO_ANY_OUTPUT.
◆ start_speculation()
|
virtual |
Begins a diversion from the regular inputs to a side stream of records representing speculative execution starting at 'start_address'.
Because the instruction record after a branch typically needs to be read before knowing whether a simulator is on the wrong path or not, this routine supports putting back the current record so that it will be re-provided as the first record after stop_speculation(), if "queue_current_record" is true. The same caveats on the input stream ordinals and last timestamp described under unread_last_record() apply to this record queueing. Calling start_speculation() immediately after unread_last_record() and requesting queueing will return a failure code.
This call can be nested; each call needs to be paired with a corresponding stop_speculation() call.
◆ stop_speculation()
|
virtual |
Stops speculative execution, resuming execution at the stream of records from the point at which the prior matching start_speculation() call was made, either repeating the current record at that time (if "true" was passed for "queue_current_record" to start_speculation()) or continuing on the subsequent record (if "false" was passed). Returns STATUS_INVALID if there was no prior start_speculation() call.
◆ unread_last_record()
|
virtual |
Queues the last-read record returned by next_record() such that it will be returned on the subsequent call to next_record() when this same input is active. Causes ordinal queries on the current input to be off by one until the record is re-read. Furthermore, the get_last_timestamp() query may still include this record, whether called on the input or output stream, immediately after this call. Fails if called multiple times in a row without an intervening next_record() call. Fails if called during speculation (between start_speculation() and stop_speculation() calls).
The documentation for this class was generated from the following file:
- /home/runner/work/dynamorio/dynamorio/build_release-64/clients/include/drmemtrace/scheduler.h
DynamoRIO version 11.0.20049 --- Fri Nov 22 2024 03:28:27