On Wed, May 13, 2015 at 01:36:44PM +0000, Gujulan Elango, Hari Prasath (H.) wrote: > On Wed, May 13, 2015 at 03:02:25PM +0300, Dan Carpenter wrote: > > On Wed, May 13, 2015 at 02:51:07PM +0300, Dan Carpenter wrote: > > > On Wed, May 13, 2015 at 11:32:50AM +0000, Gujulan Elango, Hari Prasath (H.) wrote: > > > > check return value of kmalloc before accessing the memory pointer.This > > > > function returns a non-zero value on failure hence a return value of 1 > > > > is used. > > > > > > Could you do some reasearch where this thing is freed? > > > > > > Never mind. It's freed in the caller. But it does a weird thing: > > > > TODO-list: Clean up p80211_wep->data allocation in p80211knetdev_hard_start_xmit(). > > > > It's set to NULL, then memset to zero, then when we call > > skb_ether_to_p80211() we again set it to NULL then at the end we check > > if it is equal to skb->data (it is not), and we also call kzfree() > > instead of kfree(). Load of nonsense. > > > > regards, > > dan carpenter > > > > yes Dan I did see the code.Should it be corrected ? Is this driver going > to be removed as its supporting a legacy device? I couldn't see any > patches being submitted against this driver for few months. Can you just return -ENOMEM instead of 1. It's not important to match the surrounding style since the surrounding style is garbage. regards, dan carpenter _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
