public inbox for java-prs@sourceware.org
help / color / mirror / Atom feed
* [Bug libgcj/49451] New: FileHandleGcTest FAILS on IRIX
@ 2011-06-17 11:47 ro at gcc dot gnu.org
  2011-06-22 17:56 ` [Bug libgcj/49451] " ro at gcc dot gnu.org
                   ` (2 more replies)
  0 siblings, 3 replies; 4+ messages in thread
From: ro at gcc dot gnu.org @ 2011-06-17 11:47 UTC (permalink / raw)
  To: java-prs

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49451

           Summary: FileHandleGcTest FAILS on IRIX
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libgcj
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: ro@gcc.gnu.org
                CC: tromey@gcc.gnu.org, aph@gcc.gnu.org, Hans.Boehm@hp.com
              Host: mips-sgi-irix6.5
            Target: mips-sgi-irix6.5
             Build: mips-sgi-irix6.5


The FileHandleGcTest FAILS on IRIX 6.5 (both N32 and N64 ABIs):

FAIL: FileHandleGcTest execution - source compiled test
FAIL: FileHandleGcTest -findirect-dispatch execution - source compiled test
FAIL: FileHandleGcTest -O3 execution - source compiled test
FAIL: FileHandleGcTest -O3 -findirect-dispatch execution - source compiled test

In the log, I find:

/proc open failed

This message is ultimately from boehm-gc/dyn_load.c
(GC_register_dynamic_libraries) or boehm-gc/os_dep.c (GC_dirty_init).  I had to
add a discriminator to the message to find it's the former.  Running the test
under par, I find

 1061mS[  3]FileHandleGcTest(69051502): open("/dev/null", O_RDONLY, 0666) =
1023
 1062mS[  3]FileHandleGcTest(69051502): open("/dev/null", O_RDONLY, 0666) errno 
= 24 (Too many open files)
 1063mS[  3]FileHandleGcTest(69051502): open("/proc/69051502", O_RDONLY, 0)
errn
o = 24 (Too many open files)
 1063mS[  3]FileHandleGcTest(69051502): write(2, "/proc open failed\n", 18) =
18
open             1056      0.07     72.78

When in gnu/java/nio/channels/natFileChannelPosix.cc (FileChannelImpl::open) an
open call fails, the code runs

      ::java::lang::System::gc ();
      ::java::lang::System::runFinalization ();

to try and free fds.  Unfortunately, in boehm-gc
(GC_register_dynamic_libraries)
an attempt is made to open /proc/<pid>, which of course fails and calls
abort().

This doesn't happen on either Solaris or Tru64 UNIX because those use their
own implementations of GC_register_dynamic_libraries which don't need
additional file descriptors.

I guess something similar can be implemented on IRIX, using __rld_obj_head 
described in rld(5).  I''ll probably give it a try.


^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2011-06-29  8:32 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-06-17 11:47 [Bug libgcj/49451] New: FileHandleGcTest FAILS on IRIX ro at gcc dot gnu.org
2011-06-22 17:56 ` [Bug libgcj/49451] " ro at gcc dot gnu.org
2011-06-29  8:20 ` ro at gcc dot gnu.org
2011-06-29  8:32 ` ro at gcc dot gnu.org

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).