On Mon, Aug 26, 2019 at 05:29:49PM +0100, Al Viro wrote: > On Mon, Aug 26, 2019 at 03:48:38AM +0100, Al Viro wrote: > > > We might be able to paper over that mess by doing what /dev/st does - > > checking that file_count(file) == 1 in ->flush() instance and doing commit > > there in such case. It's not entirely reliable, though, and it's definitely > > not something I'd like to see spreading. > > This "not entirely reliable" turns out to be an understatement. > If you have /proc/*/fdinfo/* being read from at the time of final close(2), > you'll get file_count(file) > 1 the last time ->flush() is called. In other > words, we'd get the data not committed at all. How about always doing the write in ->flush instead of ->release? Yes, that means that calling close(dup(fd)) is going to flush the write, but you shouldn't be doing that. I think there'll also be extra flushes done if you fork() during one of these writes ... but, again, don't do that. It's not like these are common things. Why does the prototype of file_operations::release suggest that it can return an int? __fput doesn't pay any attention to the return value. Changing that to return void might help some future programmers avoid this mistake.
