On Tue, 2013-01-22 at 19:01 +0100, Samuel Thibault wrote: > Svante Signell, le Tue 22 Jan 2013 18:53:43 +0100, a écrit : > > Attached is the first patch for a 3-way split of hurdselect.c into three > > cases: DELAY, POLL, SELECT leading to a more POSIX conforming POLL. > > It is way more readable that the previous versions :) > > > + if (nfds > _hurd_dtablesize) > > + nfds = _hurd_dtablesize; > > You can't afford moving that to the general case: in the poll case, nfds > is not the maximum of the fds, but the number of elements in the array, > which can be very big, if the application puts the same fd several time > in the array.
I did update the patch but it was not included in the attachment (same problem with Pinos comment on fd < _hurd_tablesize) sorry.
