You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We've had a few cases of postprocessing raw drmemtrace traces into final traces where we wanted to know whether a signal was synchronous or asynchronous. Some analyses of traces want to know what type of signal it is. The TRACE_MARKER_TYPE_KERNEL_EVENT's value is already occupied, so we'd add a new marker type.
The text was updated successfully, but these errors were encountered:
Adds a new drmemtrace marker holding the signal number which is
emitted after the kernel event marker. It is not emitted at the
signal return point; only at the transfer to the handler.
Does not add a feature bit or increase the version as this marker is
considered optional and so no mechanism is provided for users to be
guaranteed that it is present.
Adds view tool support.
Adds a test that the marker is emitted in the burst_gencode test.
Output from the test's trace:
```
52 32: 1616564 ifetch 2 byte(s) @ 0x00007f615935d023 0f 0b ud2
53 32: 1616564 <marker: kernel xfer from 0x7f615935d023 to handler>
54 32: 1616564 <marker: signal #4>
55 32: 1616564 <marker: timestamp 13365562571171519>
56 32: 1616564 <marker: tid 1616564 on core 9>
57 33: 1616564 ifetch 1 byte(s) @ 0x00005641b091170b 55 push %rbp
```
Fixes#6884
Adds a new drmemtrace marker holding the signal number which is emitted
after the kernel event marker. It is not emitted at the signal return
point; only at the transfer to the handler.
Does not add a feature bit or increase the version as this marker is
considered optional and so no mechanism is provided for users to be
guaranteed that it is present.
Adds view tool support.
Adds a test that the marker is emitted in the burst_gencode test. Output
from the test's trace:
```
52 32: 1616564 ifetch 2 byte(s) @ 0x00007f615935d023 0f 0b ud2
53 32: 1616564 <marker: kernel xfer from 0x7f615935d023 to handler>
54 32: 1616564 <marker: signal #4>
55 32: 1616564 <marker: timestamp 13365562571171519>
56 32: 1616564 <marker: tid 1616564 on core 9>
57 33: 1616564 ifetch 1 byte(s) @ 0x00005641b091170b 55 push %rbp
```
Fixes#6884
We've had a few cases of postprocessing raw drmemtrace traces into final traces where we wanted to know whether a signal was synchronous or asynchronous. Some analyses of traces want to know what type of signal it is. The TRACE_MARKER_TYPE_KERNEL_EVENT's value is already occupied, so we'd add a new marker type.
The text was updated successfully, but these errors were encountered: