On Thu, 14 Jun 2018 14:26:55 -0600 Mathieu Poirier mathieu.poirier@linaro.org wrote:
@@ -334,6 +338,18 @@ cs_etm_decoder__buffer_range(struct cs_etm_decoder *decoder, break; }
- /*
* At the beginning of a trace session or trace discontinuity, the* tracers will generate a timestamp packet _after_ a range packet* has been generated. In order to know when traces started to be* executed we need to keep track of the amount of instructions that* were processed. That way when the timestamp packet gets generated* we can _estimate_ the start time by substracting the number of* instructions that were executed from the timestamp value.* Timestamp estimation is done in cs_etm_decoder__buffer_timestamp().*/- decoder->instr_count += elem->num_instr_range;
so.. using your branch, this line doesn't build, native, gcc 6.3, with O= set:
util/cs-etm-decoder/cs-etm-decoder.c: In function ‘cs_etm_decoder__buffer_range’: util/cs-etm-decoder/cs-etm-decoder.c:351:30: error: ‘ocsd_generic_trace_elem {aka const struct _ocsd_generic_trace_elem}’ has no member named ‘num_instr_range’ decoder->instr_count += elem->num_instr_range; ^~ mv: cannot stat '/nfsroot/linux/tools/arm64-perf/util/cs-etm-decoder/.cs-etm-decoder.o.tmp': No such file or directory linux/tools/build/Makefile.build:96: recipe for target 'linux/tools/arm64-perf/util/cs-etm-decoder/cs-etm-decoder.o' failed
Did the line adding num_instr_range get somehow lost on its way into the patch?
Kim