https://bugs.documentfoundation.org/show_bug.cgi?id=140189

--- Comment #4 from Stephan Bergmann <[email protected]> ---
I don't know if including SA_ONSTACK in sigaction calls could have any negative
consequences.  The main issue that comes to mind might be if a JVM is running
in-process alongside the LibreOffice code, and the JVM might for some reason be
picky about receiving its signals not on an altstack.  (I looked at
<https://github.com/openjdk/jdk> as an example, to see whether or not the JVM
itself would routinely include SA_ONSTACK in sigaction calls.  But it appears
to not include it in general and only use it for one specific signal on macOS,
see
<https://github.com/openjdk/jdk/commit/2fad2249a2f769677c3c94a1ed23f1fc089701af>
"8009302: Mac OS X: JVM crash on infinite recursion on Appkit Thread".)

Alternatively, LibreOfficeKit in general or that go-libreofficekit in
particular might want to not call osl_addSignalHandler (which is the only code
path that leads to that

>     act.sa_flags = SA_RESTART | SA_SIGINFO;

line) in the first place?

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to