[m-dev.] GCC back-end and -a / --asm ==> --target asm
Fergus Henderson
fjh at cs.mu.OZ.AU
Tue May 15 02:11:15 AEST 2001
On 13-May-2001, Fergus Henderson <fjh at cs.mu.OZ.AU> wrote:
> On 11-May-2001, Fergus Henderson <fjh at cs.mu.OZ.AU> wrote:
> > Now that I've got the GCC back-end passing all of our test suite,
>
> Actually I spoke a little too soon. The compiler build with --target asm
> passes all the tests when the tests are build with --target asm,
> as is the case e.g. when the tests are run from `tools/bootcheck --target asm'.
> However, it doesn't pass the full nightly tests yet.
> There are still a few failures when compiling the
> tests in other grades:
...
> - tests/debugger/retry fails in profiling grades;
> the generated program seg faults.
> Probably the compiler built with --target asm
> has been miscompiled somehow, and this is causing it
> to generate incorrect code for this test case.
This one is not specific to --target asm, since I can reproduce it using
other copies of the compiler installed on hg, which were not built with
--target asm (e.g. the 0.10.1 RPM distribution, installed in /usr/bin, and
the version installed in /home/hg/public/mercury-latest/i686-pc-linux-gnu/bin,
which was compiled in hlc.gc). These all generate the same C code.
The symptom of this bug is a seg fault deep in the middle of malloc(),
called from fprintf(). This may be due to writing to data which has
already been freed. The problem goes away if the same executable
is run on a machine with a more recent version of glibc, so it may be
a bug in glibc.
--
Fergus Henderson <fjh at cs.mu.oz.au> | "I have always known that the pursuit
| of excellence is a lethal habit"
WWW: <http://www.cs.mu.oz.au/~fjh> | -- the last words of T. S. Garp.
--------------------------------------------------------------------------
mercury-developers mailing list
Post messages to: mercury-developers at cs.mu.oz.au
Administrative Queries: owner-mercury-developers at cs.mu.oz.au
Subscriptions: mercury-developers-request at cs.mu.oz.au
--------------------------------------------------------------------------
More information about the developers
mailing list