How can I parse the output of trace-cmd report using awk?-CodePudding

The only thing I need to see in the output is which functions were called.

Input:

ibv_rc_pingpong-759367 [005] 8391981.416466: funcgraph_entry:                   |  ib_enum_all_devs() {
ibv_rc_pingpong-759367 [005] 8391981.416472: funcgraph_entry:        29.337 us  |    ib_get_device_fw_str();
ibv_rc_pingpong-759367 [005] 8391981.416504: funcgraph_exit:         38.787 us  |  }
ibv_rc_pingpong-759367 [005] 8391981.416543: funcgraph_entry:        1.191 us   |  ib_enum_all_devs();
ibv_rc_pingpong-759367 [005] 8391981.416621: funcgraph_entry:        1.371 us   |  ib_device_get_by_index();
ibv_rc_pingpong-759367 [005] 8391981.416624: funcgraph_entry:                   |  ib_get_client_nl_info() {
ibv_rc_pingpong-759367 [005] 8391981.416628: funcgraph_entry:        0.890 us   |    ib_uverbs_get_nl_info();
ibv_rc_pingpong-759367 [005] 8391981.416630: funcgraph_exit:         6.174 us   |  }

The output should look like this:

ib_enum_all_devs() {
 ib_get_device_fw_str();
}
ib_enum_all_devs();
ib_device_get_by_index();
ib_get_client_nl_info() {
 ib_uverbs_get_nl_info();
}

This is what I tried:

cat myfile.dat | awk '{print $7}'

However, this gives me garbage.

CodePudding user response：

Why don't you use the pipe character as a field separator?

cat myfile.dat | awk -F "|" '{print $2}'

... and as usual I've done [a useless use of cat][1]:

awk -F "|" '{print $2}' file.txt

CodePudding user response：

You may use this awk:

awk -F '  \\|  ' '{print $NF}' file

ib_enum_all_devs() {
  ib_get_device_fw_str();
}
ib_enum_all_devs();
ib_device_get_by_index();
ib_get_client_nl_info() {
  ib_uverbs_get_nl_info();
}

CodePudding user response：

I would harness GNU AWK for this task following let file.txt

ibv_rc_pingpong-759367 [005] 8391981.416466: funcgraph_entry:                   |  ib_enum_all_devs() {
ibv_rc_pingpong-759367 [005] 8391981.416472: funcgraph_entry:        29.337 us  |    ib_get_device_fw_str();
ibv_rc_pingpong-759367 [005] 8391981.416504: funcgraph_exit:         38.787 us  |  }
ibv_rc_pingpong-759367 [005] 8391981.416543: funcgraph_entry:        1.191 us   |  ib_enum_all_devs();
ibv_rc_pingpong-759367 [005] 8391981.416621: funcgraph_entry:        1.371 us   |  ib_device_get_by_index();
ibv_rc_pingpong-759367 [005] 8391981.416624: funcgraph_entry:                   |  ib_get_client_nl_info() {
ibv_rc_pingpong-759367 [005] 8391981.416628: funcgraph_entry:        0.890 us   |    ib_uverbs_get_nl_info();
ibv_rc_pingpong-759367 [005] 8391981.416630: funcgraph_exit:         6.174 us   |  }

then

awk '{print substr($0, 84)}' file.txt

gives output

ib_enum_all_devs() {
  ib_get_device_fw_str();
}
ib_enum_all_devs();
ib_device_get_by_index();
ib_get_client_nl_info() {
  ib_uverbs_get_nl_info();
}

Explanation: you have file with fixed-width column, I simply use substr function to get part of whole line ($0) starting at 84th character.

(tested in GNU Awk 5.0.1)