代码拉取完成，页面将自动刷新
同步操作将从 OpenHarmony/docs 强制同步，此操作会覆盖自 Fork 仓库以来所做的任何修改，且无法恢复！！！
确定后同步将在后台操作，完成时将刷新页面，请耐心等待。
Trace helps you learn about the kernel running process and the execution sequence of modules and tasks. With the information, you can better understand the code running process of the kernel and locate time sequence problems.
The kernel provides a hook framework to embed hooks in the main process of each module. In the initial startup phase of the kernel, the trace function is initialized and the trace handlers are registered with the hooks.
When a hook is triggered, the trace module encapsulates the input information and adds the trace frame header information, including the event type, ID of the running CPU, ID of the running task, and relative timestamp.
The trace module provides two working modes: offline mode and online mode.
In offline mode, trace frames are stored in a circular buffer. If too many frames are stored in the circular buffer, earlier frames will be overwritten to ensure that the information in the buffer is always the latest. Data in the circular buffer can be exported by running the shell command for further analysis. The exported information is sorted by timestamp.
The online mode must be used with the integrated development environment (IDE). Trace frames are sent to the IDE in real time. The IDE parses the records and displays them in a visualized manner.
The trace module of the OpenHarmony LiteOS-M kernel provides the following functions. For more details about the APIs, see the API reference.
Table 1 Trace module APIs
|
Sets the event mask to trace only events of the specified modules. |
||
|
Registers a hook to filter out events of specified interrupt IDs. |
||
You can perform function instrumentation in the source code to trace specific events. The system provides the following APIs for instrumentation:
LOS_TRACE_EASY(TYPE, IDENTITY, params...) for simple instrumentation
You only need to insert this API into the source code.
TYPE specifies the event type. The value range is 0 to 0xF. The meaning of each value is user-defined.
IDENTITY specifies the object of the event operation. The value is of the UIntPtr type.
Params specifies the event parameters. The value is of the UIntPtr type.
Example:
Perform simple instrumentation for reading and writing files fd1 and fd2.
Set TYPE to 1 for read operations and 2 for write operations.
Insert the following to the position where the fd1 file is read:
LOS_TRACE_EASY(1, fd1, flag, size);
Insert the following to the position where the fd2 file is read:
LOS_TRACE_EASY(1, fd2, flag, size);
Insert the following to the position where the fd1 file is written:
LOS_TRACE_EASY(2, fd1, flag, size);
Insert the following in the position where the fd2 file is written:
LOS_TRACE_EASY(2, fd2, flag, size);
LOS_TRACE(TYPE, IDENTITY, params...) for standard instrumentation.
Compared with simple instrumentation, standard instrumentation supports dynamic event filtering and parameter tailoring. However, you need to extend the functions based on rules.
TYPE specifies the event type. You can define the event type in enum LOS_TRACE_TYPE in the header file los_trace.h. For details about methods and rules for defining events, see other event types.
The IDENTITY and Params are the same as those of simple instrumentation.
Example:
1. Set the event mask (module-level event type) in enum LOS_TRACE_MASK.
Format: TRACE_#MOD#_FLAG (MOD indicates the module name)
Example:
TRACE_FS_FLAG = 0x4000
2. Define the event type in enum LOS_TRACE_TYPE.
Format: #TYPE# = TRACE_#MOD#_FLAG | NUMBER
Example:
FS_READ = TRACE_FS_FLAG | 0; // Read files
FS_WRITE = TRACE_FS_FLAG | 1; // Write files
3. Set event parameters in the #TYPE#_PARAMS(IDENTITY, parma1...) IDENTITY, ... format.
#TYPE# is the #TYPE# defined in step 2.
Example:
#define FS_READ_PARAMS(fp, fd, flag, size) fp, fd, flag, size
The parameters defined by the macro correspond to the event parameters recorded in the trace buffer. You can modify the parameters as required.
If no parameter is specified, events of this type are not traced.
#define FS_READ_PARAMS(fp, fd, flag, size) // File reading events are not traced.
4. Insert a code stub in a proper position.
Format: LOS_TRACE(#TYPE#, #TYPE#_PARAMS(IDENTITY, parma1...))
LOS_TRACE(FS_READ, fp, fd, flag, size); // Code stub for reading files
The parameters following #TYPE# are the input parameter of the FS_READ_PARAMS function in step 3.
NOTE: The trace event types and parameters can be modified as required. For details about the parameters, see kernel\include\los_trace.h.
For LOS_TraceEventMaskSet(UINT32 mask), only the most significant 28 bits (corresponding to the enable bit of the module in LOS_TRACE_MASK) of the mask take effect and are used only for module-based tracing. Currently, fine-grained event-based tracing is not supported. For example, in LOS_TraceEventMaskSet(0x202), the effective mask is 0x200 (TRACE_QUE_FLAG) and all events of the QUE module are collected. The recommended method is LOS_TraceEventMaskSet(TRACE_EVENT_FLAG | TRACE_MUX_FLAG | TRACE_SEM_FLAG | TRACE_QUE_FLAG);.
To enable trace of only simple instrumentation events, set Trace Mask to TRACE_MAX_FLAG.
The trace buffer has limited capacity. When the trace buffer is full, events will be overwritten. You can use LOS_TraceRecordDump to export data from the trace buffer and locate the latest records by CurEvtIndex.
The typical trace operation process includes LOS_TraceStart, LOS_TraceStop, and LOS_TraceRecordDump.
You can filter out interrupt events by interrupt ID to prevent other events from being overwritten due to frequent triggering of a specific interrupt in some scenarios. You can customize interrupt filtering rules.
The sample code is as follows:
BOOL Example_HwiNumFilter(UINT32 hwiNum)
{
if ((hwiNum == TIMER_INT) || (hwiNum == DMA_INT)) {
return TRUE;
}
return FALSE;
}
LOS_TraceHwiFilterHookReg(Example_HwiNumFilter);
The interrupt events with interrupt ID of TIMER_INT or DMA_INT are not traced.
The typical trace process is as follows:
Configure the macro related to the trace module.
Modify the configuration in the target_config.h file.
(Optional) Preset event parameters and stubs (or use the default event parameter settings and event stubs).
(Optional) Call LOS_TraceStop to stop trace and call LOS_TraceReset to clear the trace buffer. (Trace is started by default.)
(Optional) Call LOS_TraceEventMaskSet to set the event mask for trace (only the interrupts and task events are enabled by default). For details about the event mask, see LOS_TRACE_MASK in los_trace.h.
Call LOS_TraceStart at the start of the code where the event needs to be traced.
Call LOS_TraceStop at the end of the code where the event needs to be traced.
Call LOS_TraceRecordDump to output the data in the buffer. (The input parameter of the function is of the Boolean type. The value FALSE means to output data in the specified format, and the value TRUE means to output data to a Windows client.)
The methods in steps 3 to 7 are encapsulated with shell commands. After the shell is enabled, the corresponding commands can be executed. The mapping is as follows:
This example implements the following:
The sample code is as follows:
#include "los_trace.h"
UINT32 g_traceTestTaskId;
VOID Example_Trace(VOID)
{
UINT32 ret;
LOS_TaskDelay(10);
/* Start trace. */
ret = LOS_TraceStart();
if (ret != LOS_OK) {
dprintf("trace start error\n");
return;
}
/* Trigger a task switching event.*/
LOS_TaskDelay(1);
LOS_TaskDelay(1);
LOS_TaskDelay(1);
/* Stop trace.*/
LOS_TraceStop();
LOS_TraceRecordDump(FALSE);
}
UINT32 Example_Trace_test(VOID){
UINT32 ret;
TSK_INIT_PARAM_S traceTestTask;
/* Create a trace task. */
memset(&traceTestTask, 0, sizeof(TSK_INIT_PARAM_S));
traceTestTask.pfnTaskEntry = (TSK_ENTRY_FUNC)Example_Trace;
traceTestTask.pcName = "TestTraceTsk"; /* Trace task name*/
traceTestTask.uwStackSize = 0x800;
traceTestTask.usTaskPrio = 5;
traceTestTask.uwResved = LOS_TASK_STATUS_DETACHED;
ret = LOS_TaskCreate(&g_traceTestTaskId, &traceTestTask);
if(ret != LOS_OK){
dprintf("TraceTestTask create failed .\n");
return LOS_NOK;
}
/* Trace is started by default. Therefore, you can stop trace, clear the buffer, and then restart trace. */
LOS_TraceStop();
LOS_TraceReset();
/* Enable trace of the Task module events. */
LOS_TraceEventMaskSet(TRACE_TASK_FLAG);
return LOS_OK;
}
The output is as follows:
*******TraceInfo begin*******
clockFreq = 50000000
CurEvtIndex = 7
Index Time(cycles) EventType CurTask Identity params
0 0x366d5e88 0x45 0x1 0x0 0x1f 0x4 0x0
1 0x366d74ae 0x45 0x0 0x1 0x0 0x8 0x1f
2 0x36940da6 0x45 0x1 0xc 0x1f 0x4 0x9
3 0x3694337c 0x45 0xc 0x1 0x9 0x8 0x1f
4 0x36eea56e 0x45 0x1 0xc 0x1f 0x4 0x9
5 0x36eec810 0x45 0xc 0x1 0x9 0x8 0x1f
6 0x3706f804 0x45 0x1 0x0 0x1f 0x4 0x0
7 0x37070e59 0x45 0x0 0x1 0x0 0x8 0x1f
*******TraceInfo end*******
The output event information includes the occurrence time, event type, task in which the event occurs, object of the event operation, and other parameters of the event.
The following uses output No. 0 as an example.
Index Time(cycles) EventType CurTask Identity params
0 0x366d5e88 0x45 0x1 0x0 0x1f 0x4
#define TASK_SWITCH_PARAMS(taskId, oldPriority, oldTaskStatus, newPriority, newTaskStatus) \
taskId, oldPriority, oldTaskStatus, newPriority, newTaskStatus
Because of #TYPE#_PARAMS(IDENTITY, parma1...) IDENTITY, ..., Identity is taskId (0x0) and the first parameter is oldPriority (0x1f).
Task 0x1 is switched to Task 0x0. The priority of task 0x1 is 0x1f, and the state is 0x4. The priority of the task 0x0 is 0x0.
此处可能存在不合适展示的内容，页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容，可点击提交进行申诉，我们将尽快为您处理。