JPH02294851A - Trace information collecting system - Google Patents

Abstract

PURPOSE:To freeze the stored trace information at the time when a device fault is generated by changing the head pointer or the end pointer to a trace buffer. CONSTITUTION:When a notice of generation of a device fault is received, a trace buffer pointer changing means 2 reads out the next pointer 33 in a trace buffer pointer storage means 3. Subsequently, the contents of the next pointer 33 are checked, and when the rear half part of a trace buffer 5 is indicated, the contents of the end pointer 32 are changed so as to show an intermediate address of the trace buffer 5, and when the front part is indicated, the contents of the head pointer 33 are changed so as to show the intermediate address of the trace buffer 5. At the time point of collecting the trace information, the next pointer 33 is updated to the next address in which the read-out contents are stored by trace information collecting means 4. In such a way, in the case a fault is generated, necessary trace information is left on the trace buffer.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a trace information collection method, particularly to a trace information collection method in an information processing system.

[Conventional technology]

Traditionally, this type of 1-race information collection method collects information about the current status of hardware, firmware, and software when the hardware operation, firmware, and software programs pass through preset trace points. Data information is written sequentially from the beginning of the trace buffer, and when the trace buffer becomes full, it is overwritten again from the beginning of the trace buffer.

[Problem to be solved by the invention]

In the conventional trace information collection method described above, when the trace buffer becomes full, it is overwritten from the beginning of the trace buffer, so if the trace operation is not stopped due to a device failure, etc., trace information can be analyzed. If you try to do this, the disadvantage is that the necessary trace information may not remain in the trace buffer.

[Means to solve the problem]

The trace information collection method of the present invention is a trace information collection method in which trace information of an information processing system is circulated and stored in a trace buffer indicating a range of addresses using a start pointer and an end pointer. , if the next pointer indicating the address at which the next trace information is to be stored in the trace buffer is in the first half of the trace buffer divided into two parts, the address indicated by the first pointer is changed to the address of the two-part divided position of the trace buffer, and the next pointer is trace buffer pointer changing means that changes the address indicated by the end pointer to the address of the two-divided position of the trace buffer when the trace buffer is in the second half of the trace buffer division; trace information collection means for updating the next pointer to indicate the next address, and rewriting the address to the address indicated by the first pointer when the address indicated by the updated next pointer is the same as the address indicated by the end pointer; configured.

〔Example〕

Next, the present invention will be explained with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of the present invention, which is composed of a failure occurrence notification means 1, a trace buffer pointer changing means 2, a trace buffer pointer storage means 3, a trace information collection means 4, and a race buffer 5. The trace buffer pointer storage means 3 includes a head pointer 31.
, end pointer 32, and next pointer 33 are stored.

2 and 3 are a flowchart of the processing of the trace buffer pointer changing means 2 and the processing of the trace information collection means 4 of FIG. 1, respectively.

Next, the operation shown in FIG. 1 will be explained with reference to FIGS. 2 and 3. When a failure occurs in the device, the failure occurrence notification stage 1 notifies the trace buffer pointer changing means 2 that a device failure has occurred. When the trace buffer pointer changing means 2 is notified of the occurrence of a device failure, the trace buffer pointer changing means 2 reads out the next pointer 33 in the trace buffer pointer storage stage 3 (step (2)). Next, check the contents of the next pointer 33 (step ■), and if the contents of the next pointer point to the latter half of the trace buffer 5, change the contents of the end pointer 32 to indicate the middle address of the trace buffer 5. (Step ■). If the contents of the next pointer 33 point to the first half of the trace buffer 5, the contents of the head pointer 33 are changed to point to the middle address of the trace buffer (step 2).

Therefore, at the time of collecting trace information, the trace information collecting means 4 uses the trace buffer pump ink storage means 3 to collect the trace information.
The next pointer 33 is read out (step ■), the trace information is stored in the address of the trace buffer 5 pointed to by the contents of the next pointer 33 (step @), and the next pointer 3
The contents of 3 are updated to the next address where they are stored (step [phase]).

Furthermore, the content of this next pointer 33 is compared with the content of the last pointer 32 (step [phase]), and if they are equal, the content of the first pointer 31 is copied to the next pointer 33 (step 2).

The explanation above has been given using the case where the trace buffer is divided into two, but if you receive a failure notification again while in this state, the two divided start pointer and end pointer You can repeat the division into two in the same way as the first time.

〔Effect of the invention〕

As explained above, the present invention divides the trace buffer into two parts by changing the start pointer or end pointer for the trace buffer when a device failure occurs, thereby creating an area that can be used as a trace buffer after the device failure occurs. This has the effect of being able to freeze the trace information that has been changed and stored when a device failure occurs.

[Brief explanation of drawings]

FIG. 1 is a block diagram of one embodiment of the present invention, and FIG. 2 is a block diagram of an embodiment of the present invention.
FIG. 3 is a flowchart of the processing of the trace buffer pointer changing means in the figure, and FIG. 3 is a flowchart of the processing of the trace information collection means in FIG. DESCRIPTION OF SYMBOLS 1...Failure occurrence notification means, 2...Trace buffer pointer changing means, 3...Trace buffer pointer storage means, 31...Start pointer, 32...End pointer, 33...Next pointer. 4...Trace information collection means, 5...Trace buffer.

Claims (1)

[Claims] In a trace information collection method that circulates and stores trace information of an information processing system in a trace buffer that indicates a range of addresses using a start pointer and an end pointer, when a failure notification is received, the next trace information is stored in the trace buffer. If the next pointer indicating the address to be split into two parts of the trace buffer is in the first half of the trace buffer, the address indicated by the first pointer is changed to the address of the two parts of the trace buffer, and the next pointer is in the second half of the trace buffer.
A trace buffer pointer changing means changes the address indicated by the end pointer to the address of the two-divided position of the trace buffer when it is in the latter half of the division, and after storing the trace information at the address indicated by the next pointer, the next pointer is changed to the next address. trace information collection means for updating the address to indicate the address, and rewriting the address to the address indicated by the start pointer when the address indicated by the updated next pointer is the same as the address indicated by the end pointer. Collection method.