On Mon, 10 Jul 2017, Rene Engelhard wrote: > On Mon, Jul 10, 2017 at 09:47:09AM -0300, Henrique de Moraes Holschuh wrote: > > On seg, 10 jul 2017, Rene Engelhard wrote: > > > On Mon, Jul 10, 2017 at 01:28:54PM +0200, Hans wrote: > > > > only once! After stopping and restarting LO again, it will not start > > > > again. > > > > > > bt? Is that the same Java crash? See > > > https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=865303 > > > and > > > https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=865866 > > > > It is related, yes. LO does something rather unexpected (and arguably > > Not Good) to its stack on i386, and that clashes with the added kernel > > protection against the "stack clash" vulnerability. > > No. Java does. LO just calls into the JVM. Other stuff having the JVM > in-process (in contrast to calling "java") is also affected. See the reports.
I did, but maybe I have misunderstood something... What I got from LKML is that LO (but not java) has a rwx weird stack-guard page (Bad Idea), while the JVM has PROT_NONE stack-guard pages between each thread-local stack. Obviously, LO with java support would *also* trigger any issues the JVM has with the stack-guard kernel changes. -- Henrique Holschuh