Review Request 128831: Check whether kwallet is enabled in Wallet::isOpen(name)

[Elvis Angelaccio]

> On Sept. 24, 2016, 12:40 p.m., Elvis Angelaccio wrote:
> > src/api/KWallet/kwallet.cpp, line 366
> > <https://git.reviewboard.kde.org/r/128831/diff/1/?file=475994#file475994line366>
> >
> >     You should probably use isEnabled() instead, in case someone is using ksecretservice (though m_walletEnabled is already used all over the place)...
> >     
> >     Anyway +1
> 
> Wolfgang Bauer wrote:
>     This is inside the non-ksecretservice codepath, so it makes no difference whether one uses m_walletEnabled or isEnabled() (not to mention that I think the distinction is not even necessary in isEnabled(), AFAICS m_walletEnabled is set accordingly in the ksecretservice case too).
>     
>     And as you write the check is like that everywhere else in the code.
>     I don't think it makes sense to deviate here in this one method, if that's desirable it should be changed everywhere in a separate commit IMHO.
>     
>     In case you mean that the check should be done using isEnabled() right at the start of the method (i.e. before the if):
>     I'm not sure this check is needed at all in the ksecretservice case, there is no ->getInterface().xxx call that could crash.
>     
>     KWalletDLauncher does check whether the wallet is enabled and just returns if not without doing anything. This means it doesn't start the KWallet service in the non-ksecretservice, that's what causes the mentioned crash in the first place because a nullptr is dereferenced.
>     
>     I don't know anything about ksecretservice, but the code here probably will return false in case the wallet is disabled anyway (at least that's my feeling after looking at this code).
> 
> Elvis Angelaccio wrote:
>     > This is inside the non-ksecretservice codepath, so it makes no difference whether one uses m_walletEnabled or isEnabled() (not to mention that I think the distinction is not even necessary in isEnabled(), AFAICS m_walletEnabled is set accordingly in the ksecretservice case too).
>     And as you write the check is like that everywhere else in the code.
>     I don't think it makes sense to deviate here in this one method, if that's desirable it should be changed everywhere in a separate commit IMHO.
>     
>     It probably makes more sense, yes. Ignore my comment then :)

Meanwhile I realized that the ksecretservice code path is broken, so there is no way that someone is using it.


- Elvis


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://git.reviewboard.kde.org/r/128831/#review99502
-----------------------------------------------------------


On Sept. 4, 2016, 9:20 p.m., Wolfgang Bauer wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://git.reviewboard.kde.org/r/128831/
> -----------------------------------------------------------
> 
> (Updated Sept. 4, 2016, 9:20 p.m.)
> 
> 
> Review request for KDE Frameworks and Valentin Rusu.
> 
> 
> Bugs: 358260
>     https://bugs.kde.org/show_bug.cgi?id=358260
> 
> 
> Repository: kwallet
> 
> 
> Description
> -------
> 
> If kwallet is disabled, walletlauncher() fails and walletLauncher()->getInterface().isOpen(name) causes a crash.
> This affects e.g. drkonqi, but probably also other applications.
> 
> Return false in this case, if kwallet is disabled a wallet cannot be open either.
> 
> 
> Diffs
> -----
> 
>   src/api/KWallet/kwallet.cpp dffebda 
> 
> Diff: https://git.reviewboard.kde.org/r/128831/diff/
> 
> 
> Testing
> -------
> 
> - disable kwallet in the settings
> - start a KF5 application (say, dolphin)
> - make it crash, e.g. via "killall -ILL dolphin"
> - try to report the crash with drkonqi
> 
> When you get to the page where you have to enter the bugzilla username/password, drkonqi crashed, with this patch it doesn't crash.
> 
> drkonqi is also still able to get the username/password from kwallet if it is enabled.
> 
> 
> Thanks,
> 
> Wolfgang Bauer
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-frameworks-devel/attachments/20160926/e35418c1/attachment-0001.html>