On 2023-07-22 at 15:48 -03, David Bremner <[email protected]> wrote:
>
> I can't get this test to fail at all. I'm not sure what to conclude from
> that. This is on a 20 core system running linux.

I can't really remember what I did at the time but I believe only the
test is instrumented and not the Notmuch library.
The test fails reliably when compiled like this:
  ./configure
  make -j CFLAGS=-fsanitize=thread LDFLAGS=-fsanitize=thread
  ./test/T810-tsan.sh
  # ...
  # WARNING: ThreadSanitizer: data race (pid=35836)
  # ...

I would say it would be best to run all the tests with the different
instrumentations enabled (for the best coverage) but I don't know if
there's a CI. Or if out-of-tree compilation is supported, I guess the
ASan and TSan tests could recompile the library with the instrumentation
enabled... What do you think?
_______________________________________________
notmuch mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to