On Fri, Jun 2, 2017 at 7:44 PM, <wr...@apache.org> wrote:
> Author: wrowe
> Date: Fri Jun 2 17:44:41 2017
> New Revision: 1797413
>
> URL: http://svn.apache.org/viewvc?rev=1797413&view=rev
> Log:
> Revert to 1.5.x apr_os_proc_mutex_get() behavior on Netware with an additional
> guard against dereferencing a NULL pointer. The assignment appears to be a
> noop
> but avoiding changes in this logic from 1.5.x -> 1.6.x is the primary goal.
[]
> ==============================================================================
> --- apr/apr/branches/1.6.x/locks/netware/proc_mutex.c (original)
> +++ apr/apr/branches/1.6.x/locks/netware/proc_mutex.c Fri Jun 2 17:44:41 2017
> @@ -116,9 +116,9 @@ APR_DECLARE(apr_status_t) apr_os_proc_mu
> apr_proc_mutex_t *pmutex,
> apr_lockmech_e *mech)
> {
> - if (!pmutex->mutex) {
> - return APR_ENOLOCK;
> - }
> + if (pmutex && pmutex->mutex)
> + ospmutex = pmutex->mutex->mutex;
> + return APR_ENOLOCK;
Hmm, do you mean Netware users are doomed to get ENOLOCK whaever happens?
I don't see the issue without this change, the change was somehow a
bugfix imo...
> #if 0
> /* We need to change apr_os_proc_mutex_t to a pointer type
> * to be able to implement this function.
> @@ -127,12 +127,8 @@ APR_DECLARE(apr_status_t) apr_os_proc_mu
> if (mech) {
> *mech = APR_LOCK_DEFAULT;
> }
> -#else
> - if (mech) {
> - *mech = APR_LOCK_DEFAULT;
> - }
> -#endif
> return APR_SUCCESS;
> +#endif
> }
