Adar Dembo has posted comments on this change. Change subject: Add Google Breakpad support to Kudu ......................................................................
Patch Set 24: -Code-Review (1 comment) http://gerrit.cloudera.org:8080/#/c/5620/21/src/kudu/util/minidump.cc File src/kudu/util/minidump.cc: PS21, Line 261: user_signal_handler_thread_->Join(); : } > You're correct about how sigwait() works. > Incidentally, a blocked SIGUSR1 will be inherited by all Kudu subprocesses. I > don't think that's an issue today because we don't launch subprocesses in > production, but we will soon (Todd is introducing something like that for > kerberos). Blocking SIGUSR1 for them may be innocuous if they're short-lived, > but if they're long-lived it could be problematic. We discussed this on Slack. Apparently exec() doesn't reset signal handlers for block signals[1], and Todd pointed out that the JVM reserves SIGUSR1 for internal use. So if we were to launch a JVM via our subprocess mechanism, it might encounter issues. So let's unblock (just SIGUSR1? all?) signals after the fork() but before exec() in subprocess.cc. 1. https://lwn.net/Articles/415684/ -- To view, visit http://gerrit.cloudera.org:8080/5620 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I495695cc38b75377f20b0497093a81ed5baa887f Gerrit-PatchSet: 24 Gerrit-Project: kudu Gerrit-Branch: master Gerrit-Owner: Mike Percy <mpe...@apache.org> Gerrit-Reviewer: Adar Dembo <a...@cloudera.com> Gerrit-Reviewer: Alexey Serbin <aser...@cloudera.com> Gerrit-Reviewer: Dinesh Bhat <din...@cloudera.com> Gerrit-Reviewer: Kudu Jenkins Gerrit-Reviewer: Lars Volker <l...@cloudera.com> Gerrit-Reviewer: Mike Percy <mpe...@apache.org> Gerrit-Reviewer: Tidy Bot Gerrit-Reviewer: Todd Lipcon <t...@apache.org> Gerrit-HasComments: Yes
