This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: PATCH: Remove gthr-gnat.c from Linux

On Tue, May 04, 2004 at 05:44:47PM +0200, Arnaud Charlet wrote:
> > The main difference may be I have a 4way 1.3GHz Itanium 2. I saw
> > quite a few thread bugs which I didn't see on other machines.
> > Unfortunately, I don't know anything about Ada. I can't tell if
> > the Ada library is doing the right thing with thread or not.
> Ada tasking has been used on lots of complex application on many
> different architectures and massively parallel machines (e.g. 64 proc
> machines), so it is pretty reliable.

Have you tried it on N-way Itanium 2 machines? There are some very
subtle areas in POSIX thread standards. Even some testcases in glibc
get them wrong. They run fine on other machines until I try them on
my 4way Itanium 2 machine.

I tried to debug it. The problem seems to be that there was only one
thread left and it was waiting for other thread to wake it up. My gdb
doesn't know Ada. Even if it does, I don't know Ada.

> Now, that could equally well be a libpthread bug in this case :-)

It is possible. Unfortunately, I don't have a testcase in C.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]