This is the mail archive of the
mailing list for the Java project.
Re: libgcj cross-compilation problems (2)
- To: Chris Dornan <cdornan at arm dot com>
- Subject: Re: libgcj cross-compilation problems (2)
- From: Tom Tromey <tromey at cygnus dot com>
- Date: Thu, 20 Apr 2000 08:57:41 -0700 (PDT)
- Cc: java-discuss at sourceware dot cygnus dot com, aph at pasanda dot cygnus dot co dot uk
- References: <200004201352.OAA29170@sun14.cambridge.arm.com>
>>>>> "Chris" == Chris Dornan <firstname.lastname@example.org> writes:
Chris> Having fixed that up, it now falls over while including <signal.h>:
Chris> /export/home/cdornan/gnu/gcc-2.95.2-arm/lib/gcc-lib/arm-elf/2.95.2/../../../../arm-elf/sys-include/sys/signal.h:19: candidates are: sigaction::sigaction(const sigaction &)
Chris> /export/home/cdornan/gnu/gcc-2.95.2-arm/lib/gcc-lib/arm-elf/2.95.2/../../../../arm-elf/sys-include/sys/signal.h:19: sigaction::sigaction()
Chris> make: *** [prims.lo] Error 1
Presumably this isn't the entire text of the error message.
Chris> Now this is the kind of problem that I expected in moving over
Chris> from Solaris to ARM. I grabbed the sys-include files from an
Chris> earlier build of gcc for arm-elf and they worked fine in
Chris> building gcc. The offending code is in main_init:
Chris> // FIXME: we only want this on POSIX systems.
Chris> struct sigaction act;
Chris> act.sa_handler = SIG_IGN;
Chris> sigemptyset (&act.sa_mask);
Chris> act.sa_flags = 0;
Chris> sigaction (SIGPIPE, &act, NULL);
Chris> Does anyone know why <sys/signal.h> wouldn't support this?
Offhand, no. I think it should work fine. What operating system are
you using, if any?