Steve, All, I think smb3-encryption has been broken for quite a while (forver?) when we configure the kernel with CONFIG_DEBUG_SG="y". The problem is that with these extra checks, we can no longer pass a pointer to a buffer that is allocated off the stack safely to sg_set_buf() as this function will BUG_ON. We have a lot of places where cifs.ko does indeed pass a pointer to a stack object to this function :-( and this patch will fix all the instances I could find so far. As I think we have been broken in this regard for a long time, and since this patch is pretty big and non-trivial, I can't say if we should push this big patch right now or wait for the next merge window. At the end of the day, it does not seem that anyone has noticed that CONFIG_DEBUG_SG + smb3-encryption == crash so maybe no one is actually hurt by the presence of this bug. This is the first version I just got to work as far as I can tell. Please, careful review please. And also ideas if there are other places we end up using a stack pointer that I missed. -- To unsubscribe from this list: send the line "unsubscribe linux-cifs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
