Michael Hope michael.hope@linaro.org writes:
On Tue, Dec 20, 2011 at 10:00 PM, Richard Sandiford richard.sandiford@linaro.org wrote:
Michael Hope michael.hope@linaro.org writes:
Hi there. I've looked further into the intermittent gcc/testsuite/g++.dg/cdce3.C test failures. Taking Ira's vectoriser-only fix-pr51301-4.6 branch and comparing it with it's predecessor r106845: * cdce3.o itself is identical across compilers * Fault occurs in a parallel test run as part of the normal auto build * Fault occurs every time * Fault occurs with a manual 'make check-gcc RUNTESTFLAGS="dg.exp=cdce*' * Fault doesn't occur when building from the command line * Fault doesn't occur after updating binutils
I'm suspicious of the linker. The auto builders are Natty based and come with ld 2.21.0.20110327. Updating them to Oneiric's 2.21.53.20110810 clears the problem.
Hmm, FWIW, this reminds me of the problems that I had reproducing the chromium linker bug from way back (which I think was also on Natty). The bug only triggered for me when building under dpkg-buildpackage. If I ran (seemingly) the same command by hand, it worked OK.
What was the fix in the end?
I never did get to the bottom of the (seemingly) environment-dependent linker behaviour, unfortunately. Just the link took in the order of hours, and it ended up being a time sink.
There's a good possibility it was just pilot error, although I did try several times.
The bug itself was about an internal linker error, and in the end Matthias gave me a nice self-contained testcase that reproduced cross, so I fixed it that way instead.
Richard