================
@@ -556,7 +558,12 @@ static void
get_threads_profile_data(DNBProfileDataScanType scanType,
}
// Sleep a bit and try again
- ::usleep(k_usec_delay);
+ // Use an increasing interval so that if there's a short term traffic
+ // jam, we skip past rather than exacerbating it. We see this sequence
+ // timing out occasionally on heavily loaded bots, seemingly because the
+ // syspolicyd isn't keeping up.
+ interval += k_usec_delay * i;
----------------
jimingham wrote:
I would be surprised if this is affected by address_sanitizer. We are
concerned with syspolicyd getting overwhelmed, not with a task that lldb or
debugserver is directly performing. An asanified version of the loop that's
calling task_for_pid should not be any slower that the normal one. If
anything, having address_sanitizer on is going to make everything in the test
suite happen more slowly, which should make the pileup less not more likely.
https://github.com/llvm/llvm-project/pull/217425
_______________________________________________
lldb-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
