On Thu, Sep 20, 2012 at 09:44:00AM +0100, Ian Abbott wrote: > On 2012/09/19 08:10 PM, Dan Carpenter wrote: > > Gar. Moving the pointer around sucks. It's just not worth it. It > > makes the code more complicated. In the end, we're going to audit > > the ioctls first when we look for security bugs in this code. The > > chance that a bug is going to be introduced because of a missing > > __user tag is not very high. > > > > Of course, the memory leak fixes are good and the pass by reference > > changes are good. > > > > It's not my code and all, but the way it's done now is the simplest. > > Apart from mixing user and kernel pointers in the same type and some > ugly-looking workarounds in the CONFIG_COMPAT code? It's just mixed in do_cmd_ioct() and do_cmdtest_ioctl(). It's not so bad. The "ugly-looking workarounds" is adding a single cast. regards, dan carpenter _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/devel
