On 2011-02-23 at 14:44:11 [+0100], Jérôme Duval <korli@xxxxxxxxxxxxxxxx> wrote: > 2011/2/23 Ingo Weinhold <ingo_weinhold@xxxxxx>: > >> Not sure it's exactly #5679. As I read the JamVM code, pthread_kill > >> deliver the signal to the selected thread, which should have the same > >> sigaction handler as the main thread (signal disposition is common to > >> the team). The problem I see is the sig_action array is located in > >> Thread whereas it should be in Team. > > > > That's my point. We don't have a per-team signal handling, we do > > everything > > per-thread. Signal handlers and masks exist (only) per-thread and signals > > sent to the team are actually sent to the main thread. > > OK. I just don't read explicitly in the ticket that threads except the > main thread are therefore always without signal handler. Indeed, feel free to add it, if you like. > I see what is involved in > http://dev.haiku-os.org/wiki/SignalSupportRevision. As a short term > workaround, would it be possible to inherit signal handlers from the > parent thread? Side effects wouldn't be so large I think. Yep. Should be only an additional memcpy() in the thread creation code. I haven't investigated that aspect under BeOS yet. > BTW is your signals branch aimed at providing a solution to this? Yes, it is supposed to result in a fully POSIX compliant signals implementation. CU, Ingo