On 2/21/07, Mulyadi Santosa <mulyadi.santosa@xxxxxxxxx> wrote:
Hi Aneesh... > Hi, > > Is __copy_from_user_inatomic and __copy_to_user_inatomic really atomic > ?. Looking from the name and the code, I think the operation is clearly non atomic, especially if it copies larger than data bus (am I right?) size e.g 32 bit for 32 bit CPU. > What is the implication of zeroing the tail ? I don't get it, which tail were you referring to?
__copy_from_user differs from __copy_from_user_inatomic only int hat the former put zero in the area that it failed to copy. And then it adds a might_sleep();. I am confused how adding zero to the tail results in sleep. In my understanding we can sleep due to page in from swap and that would happen in case of __copy_from_user_inatomic also. -aneesh -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
