Thread: return correct error code from pgtls_init

On Tue, Jun 01, 2021 at 10:32:59AM -0700, Zhihong Yu wrote:
> It seems error code checking in pgtls_init() should follow the same
> convention as PG codebase adopts - i.e. the non-zero error code should be
> returned (instead of hard coded -1).
>
> Please see the attached patch.

I don't see the point of changing this.  First, other areas of
fe-secure-openssl.c use a harcoded value of -1 as error codes, so the
current style is more consistent.  Second, if we were to change that,
why are you not changing one call of pthread_mutex_lock()?
--
Michael

On Tue, Jun 01, 2021 at 06:56:42PM -0700, Zhihong Yu wrote:
> Looking at the -1 return, e.g.
>
>             pq_lockarray = malloc(sizeof(pthread_mutex_t) *
> CRYPTO_num_locks());
>
> when pq_lockarray is NULL. We can return errno.
>
> I didn't change call to pthread_mutex_lock() because PGTHREAD_ERROR() is
> used which aborts.

I am not sure what you mean here, and there is nothing wrong with this
code as far as I know, as we would let the caller of pgtls_init() know
that something is wrong.
--
Michael