On Mon, 21 May 2007, Ratnadeep Joshi wrote: > my two cents: > > * No process context. So, data transfer from/to user space is invalid. > > * not associated with any process, so not allowed to "sleep" in any > > way > > So, no calling schedule, no taking any semaphores, no calling wait_ > functions. Also, allocating memory only with GFP_ATOMIC. as a kernel janitors experiment, it might be amusing to scan the tree to see if there are any interrupt handlers that are actually trying to allocate memory in a non-GFP_ATOMIC mode, but i'm not sure how to look for stuff like that easily. > * since sleeping is not allowed, kernel preemption is automatically > > disabled > > * when servicing a given interrupt, that interrupt is masked out on > > all processors so interrupt handlers don't have to be re-entrant > > * however, when servicing a given interrupt, *other* interrupts are > > still possible unless you explicitly disable them > > This explicit disabling of interrupts can only be done for the local > processor. great, thanks. rday -- ======================================================================== Robert P. J. Day Linux Consulting, Training and Annoying Kernel Pedantry Waterloo, Ontario, CANADA http://fsdev.net/wiki/index.php?title=Main_Page ======================================================================== -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
