Suspicous code in KDE 4 trunk r764353

Christoph Bartoschek bartoschek at
Tue Jan 22 08:02:29 GMT 2008

Am Dienstag 22 Januar 2008 schrieb Allen Winter:
> I'll backport these fixes tomorrow.
> Christoph, thanks for this report.  I hope you can send them regularly.
> -Allen

Generating them is very time consuming and there has to be enough new code to 
check. Unfortunately I do not see how to directly only check changed files.


Here is what is left from the list:

- kdelibs/solid/solid/backends/fakehw/fakebattery.cpp:121

A break is missing

- kdelibs/knewstuff/knewstuff2/ui/kdxsbutton.cpp:150

Given the comment in line 146, a break might be missing.

- kdelibs/kate/completion/katecompletionconfig.cpp:176,226

The shift amount is negative on first iteration.

- kdemultimedia/juk/sortedstringlist.cpp:163

Line 153 indicates that previousNode could be NULL here.

- kdepimlibs/kioslave/imap4/imap4.cpp:359

Line 340 indicates that cache might be NULL here. 

- kdepimlibs/kblog/wordpressbuggy.cpp:129, 225
  kdepimlibs/kblog/gdata.cpp:299, 340, 396, 442
Is a warning enough?

- kdegames/kbounce/wall.cpp:256

A break seems to be missing.

- kdelibs/khtml/css/css_valueimpl.cpp:1234

docLoader remains NULL if line 1231 is false and crashes here.

- kdelibs/khtml/ecma/kjs_html.cpp:2396

A break statement or a fall-through comment should be here.

- kdebase/workspace/libs/ksysguard/processcore/processes_linux_p.cpp:365-377

There are break; statements missing.

- kdebase/workspace/libs/ksysguard/processcore/processes_remote_p.cpp:247

A break might be missing here. Or a fall-through comment.

- kdebase/workspace/libs/plasma/phase.cpp:257
Division by zero if frames is 1 or 2  (or 1 in line 295). 
Line 280 and 244 only catch the value 0. But frames is divided and
an integer in lines 255 and 293.

- kdebase/workspace/ksysguard/ksysguardd/Linux/netdev.c:420

There should be something as return here. fd is used in 423, 425,
428. I do not think that calling read or close with invalid
file descriptors is allowed.

- kdebase/workspace/ksysguard/ksysguardd/Linux/softraid.c:580
I would expect a 6 instead of 4 as the third argument to strncmp.

- kdebase/workspace/kwin/clients/plastik/plastikbutton.cpp:203

alpha is not initialized if lines 194 and 198 are not true.

- kdebase/workspace/kwin/clients/b2/b2client.cpp:168

Highly probably two breaks are missing

- kdebase/workspace/kwin/effects/thumbnailaside.cpp:146

If it is possible that windows.size() == 0 then height is 0 here.

- kdebase/workspace/khotkeys/shared/input.cpp:189

keysym is uninitialized

- kdebase/workspace/khotkeys/shared/input.cpp:217 

x_mod is uninitialized

- kdebase/apps/konqueror/src/konqviewmanager.cpp:960

Line 950 indicates that m_pMainWindow might be NULL: Is it impossible to
get here?

- kdebase/apps/konqueror/sidebar/trees/dirtree_module/dirtree_item.cpp:170

Useless "true &&".  

- kdebase/workspace/ksysguard/gui/

newDisplay is NULL here if classType == "ProcessController"  and
mLocalProcessController != NULL. I guess line 512 should move out of the
if braces starting at 509

- kdebase/workspace/kwin/events.cpp:1664 

A break is probably missing.

- kdebase/workspace/kdm/kfrontend/genkdmconf.c:2812

ftxt is NULL if the for in 2876 is not entered.

- kdebase/workspace/khotkeys/kcontrol/condition_list_widget.cpp:285

If parent2_P is NULL and parent1_P is NULL this method crashes here

- kdebase/workspace/ksplash/ksplashx/scale.cpp:194

If d is 0 problems arise later.

- kdebase/workspace/ksplash/ksplashx/qcolor.cpp:503

Lines 505 and 507 use && and not &. This one of the reasons to use &&
here too.

- kdebase/apps/konsole/src/ViewManager.cpp:399

If existingViewIter.hasNext() is always false container is NULL here and

A fall-through-comment or a real break might be missing here:


More information about the kde-core-devel mailing list