> A few pointers I have found out about this.

Talking to myself here, but a few more

  1. Should linux32 e_trapsignal hook be trapsignal()
     or linux_trapsignal()? Linux emul seems to use
     linux_trapsignal() but linux32 does not.

     I tried changing linux32 to use linux_trapsignal,
     but it made no difference at all.

  2. I tried debugging linux emul hooks, especially
     traps by placing printf calls in the kernel code.

     But when I add printf, uprintf or printf_nolog to
     linux_trapsignal, the first SIGSEGV causes java to
     dump core right after the message is printed on the
     console.

     When I remove the printf, java doesn't crash but goes
     into the hung up state I have described.

     Adding printf calls to linux syscall hook has the
     same effect that the process crashes with SIGSEGV.