This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: MicroBlaze glibc fixes
On Sat, 7 Jan 2017, Waldemar Brodkorb wrote:
> Would it be possible to reorganize the internal testsuite to allow
> the separation of compiling and running the tests?
In principle it might be possible to do
* (on the system where glibc is built) make check run-built-tests=no
* Transfer the source and build trees to the same paths on the system
where glibc itself runs.
* (on that system) make check run-built-tests=yes
Now, it's quite possible that split make check doesn't work well right
now, or that some compilations only happen in the run-built-tests=yes
stage when they should happen with run-built-tests=no, or something
otherwise depends on running the compiler with run-built-tests=yes even
when a run-built-tests=no build was run before. So there would probably
be various things to fix to get that to work. And if the generated
dependencies refer to e.g. compiler header files you might need to arrange
for those to exist at the same paths on both systems as well. But the
basic approach seems plausible.
> Would this be enough for you to get some idea about the quality of
> the next glibc release?
We want the results of the glibc testsuite reported for each release for
each architecture. (If people wish to run other testsuites and analyse
the results for regressions / glibc bugs shown by those testsuites, that
may be a useful addition, but it doesn't substitute for showing the
results of the glibc testsuite.)
--
Joseph S. Myers
joseph@codesourcery.com