https://bugs.kde.org/show_bug.cgi?id=438099
David Edmundson <k...@davidedmundson.co.uk> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |k...@davidedmundson.co.uk
--- Comment #3 from David Edmundson <k...@davidedmundson.co.uk> ---
Patch itself looks sound, but it is super important to understand the full
context to make sure we're fixing things at the root.
We only get into the abort flag if we go through:
checkpass_pam.c:84
```
if (!repl[count].resp) {
pd->abort = 1;
goto conv_err;
}
```
Which means we were in a handled: msg[count]->msg_style, but not one that
provided a reply.
>From the code surrounding it, that implies we were in the first path of this
ternary operator and putting a null pointer here.
case PAM_PROMPT_ECHO_OFF:
repl[count].resp = pd->conv(ConvGetHidden, pd->classic ? 0 :
msg[count]->msg);
FWITC pd->classic is uninitialised?! Does setting that to false explicitly also
fix it?
--
You are receiving this mail because:
You are watching all bug changes.
