[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Discuss-gnuradio] gcc 3.3
From: |
Eric Blossom |
Subject: |
Re: [Discuss-gnuradio] gcc 3.3 |
Date: |
Tue, 9 Nov 2004 10:22:48 -0800 |
User-agent: |
Mutt/1.5.6i |
On Mon, Nov 08, 2004 at 06:05:35PM -0600, David Carr wrote:
> Hi all,
>
> I have a current CVS of gnuradio-core built using:
> gcc version 3.3.4 20040623 (Gentoo Linux 3.3.4-r1, ssp-3.3.2-2, pie-8.7.6)
> glibc-2.3.4.20040808
>
> Currently, make check looks fine. What kind of bugs were you
> encountering? Can I help chase down anything and should I change
> compiler versions?
>
> Thanks,
> David Carr
Thanks for the offer. I wouldn't change anything if it's working.
One of my machines is running Mandrake 10.0 Official with is using
g++ (GCC) 3.3.2 (Mandrake Linux 10.0 3.3.2-6mdk) and it works fine.
I'm aware of three different symptoms:
(1) test_vmcircbuf and test_all were failing with a SIGBUS testing
the shm_open implementation. gnuradio-core-2.3 installs signal
handlers that detect this problem and record that shm_open
doesn't work. This problem is fixed modulo other compiler
related problems.
(2) Some versions of g++ 3.3.X, mostly on Debian, appear to have a
problem. The symptom is that the fix for (1) doesn't work and
the SIGBUS still happens.
(3) Stephane's SIGBUS/SIGSEGV somewhere in the std::string stuff
called from test_all.
My guess is that there is some version g++ 3.3.X that has a
problem. I'm not sure which values of X have the problem. Or for
that matter, which distribution patches were applied to g++, etc, etc.
Can those who are seeing either (2) or (3) please get us your compiler
versions? Ramakrishnan, I know you were seeing (2) but that it went
away when you changed compilers. Do you know the version of g++ you
were using when it wasn't working?
If we can figure out the guilty party, I'll add the configure checks
to check for the bad tool chain.
Eric