Re: [PATCH 0/3] fio: Avoid errno and errno string mismatch
Vincent Fu <[email protected]> Fri, 13 Feb 2026 21:39:32 -0500
| Newsgroups | org.kernel.vger.fio |
|---|---|
| Message-ID | <CAOp=CXm86A_sG0SokEUsLJV8_7wJ8A15t0ugV16tGDp8Ef1FJA@mail.gmail.com> |
On Fri, Feb 6, 2026 at 11:46 AM Niklas Cassel <[email protected]> wrote: > > Hello all, > > This series fixes a problem where errno and error string representing that > errno could be out of sync. > > Another possible solution is to remove td->verror everywhere and make sure > that the function wants the error string calls strerror() with the errno > when they actually need it. However, that would be a larger change. > > Another optimization that could also be done is to look at the places where > we call update_error_count() and td_clear_error(), if we don't set > td->error for a non-fatal error, there is no reason to unconditionally call > td_clear_error() for a non-fatal error. (update_error_count() still has to > be called, as it sets td->first_error.) However, I do not really understand > why update_error_count() + td_clear_error() is called by both: > io_u.c:io_completed() and backend.c:break_on_this_error(), so I avoided to > do this optimization. > > > Niklas Cassel (3): > fio: Fix error string not matching errno > io_u: Fix inconsistent handling of non-fatal errors with option > error_dump > stat: Remove duplicate space in __show_run_stats() > > fio.h | 7 +++---- > io_u.c | 10 ++++++++-- > stat.c | 2 +- > 3 files changed, 12 insertions(+), 7 deletions(-) > > -- > 2.53.0 Applied. Thanks. Vincent