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
Sometimes it's hard to navigate code without having line numbers in source. This is especially relevant for Vala-based apps, as they tend to mangle function names in various ways. So I'm trying to add srcline column to perf script, so that I'd have in flamegraph line numbers, but this breaks flamegraph.
Steps to reproduce (in terms of terminal commands)
$ perf record -g --sample-cpu ls
$ perf script -F+srcline | stackcollapse-perf.pl
Expected
Script finishes with no warnings or errors
Actual
Lots of
Unrecognized line: ld-2.29.so[11634] at /usr/bin/stackcollapse-perf.pl line 339, <> line 50.
The text was updated successfully, but these errors were encountered:
I recognize this is actually not a failing of FlameGraph, so maybe it belongs somewhere else... but I too would love to know if there is a way to get perf (or some other profiler) to record stack traces with line numbers in a way that FlameGraph can use them. The -F+srcline above doesn't really give me line numbers anyway, otherwise I would be able to massage the output to send to stackcollapse-perf.pl.
I see this feature as very handy! The @stefantalpalaru's patch works fine, but I would prefer to display the location only on hover, it makes the report less readable.
Sometimes it's hard to navigate code without having line numbers in source. This is especially relevant for Vala-based apps, as they tend to mangle function names in various ways. So I'm trying to add
srcline
column toperf script
, so that I'd have in flamegraph line numbers, but this breaks flamegraph.Steps to reproduce (in terms of terminal commands)
Expected
Script finishes with no warnings or errors
Actual
Lots of
The text was updated successfully, but these errors were encountered: