On Fri, Apr 12, 2013 at 05:31:33PM +0200, Jakub Hrozek wrote:
On Thu, Apr 11, 2013 at 12:06:22PM +0200, Lukas Slebodnik wrote:
> On (11/04/13 10:29), Jakub Hrozek wrote:
> >On Thu, Apr 11, 2013 at 10:28:01AM +0200, Jakub Hrozek wrote:
> >> Just one nitpick, can you change this block:
> >>
> >> + primary_file = talloc_strdup_append(discard_const_p(char, dir),
> >> + "/primary");
> >> + if (!primary_file) {
> >> + DEBUG(SSSDBG_CRIT_FAILURE, ("talloc_strdup_append
failed.\n"));
> >> + ret = ENOMEM;
> >> + goto done;
> >> + }
> >>
> >> Change dir to be "char *" not "const char*" and use
> >> talloc_asprintf(tmp_ctx, "%s/%s", dir);
> >
> >talloc_asprintf(tmp_ctx, "%s/primary", dir);
> >
> >sorry.
>
> Patch attached.
>
> LS
Ack, works fine. I will just change one thing before pushing, no need to
re-send:
> ret = cc_residual_is_used(uid, dir, SSS_KRB5_TYPE_DIR, &active);
> - talloc_free(tmp);
> if (ret != EOK) {
> - DEBUG(SSSDBG_OP_FAILURE, ("Could not check if ccache is active.
"
> - "Will create a new one.\n"));
> + if (ret != ENOENT) {
> + DEBUG(SSSDBG_OP_FAILURE,
> + ("Could not check if ccache is active.\n"));
> + }
I will just change the debug level here to something more verbose. I
think we should let the user know why the login failed.
Pushed to master.
Lukas, can you also send a version that applies cleanly on the sssd-1-9
branch? I think we should fix this bug in 1.9 too.