From: Marko Kreen Date: Thu, 27 May 2010 14:32:59 GMT Subject: Re: [PATCH] compat: Add another rudimentary poll() emulation Message-ID: In-Reply-To: On 5/27/10, Erik Faye-Lund wrote: > On Thu, May 27, 2010 at 4:11 PM, Marko Kreen wrote: > > On 5/27/10, Erik Faye-Lund wrote: > >> On Thu, May 27, 2010 at 3:58 PM, Marko Kreen wrote: > >> > On 5/27/10, Erik Faye-Lund wrote: > >> >> On Thu, May 27, 2010 at 3:29 PM, Marko Kreen wrote: > >> >> > On 5/27/10, Erik Faye-Lund wrote: > >> >> >> On Thu, May 27, 2010 at 1:00 PM, Erik Faye-Lund > >> >> >> wrote: > >> >> >> > On Thu, May 27, 2010 at 12:10 PM, Jonathan Nieder wrote: > >> >> >> >> Implement the subset of poll() semantics needed by git in terms of > >> >> >> >> select(), for use by the Interix port. Inspired by commit 6ed807f > >> >> >> >> (Windows: A rudimentary poll() emulation, 2007-12-01). > >> >> >> >> > >> >> >> > > >> >> >> > A possible problem with this approach is that the maximum number of > >> >> >> > file descriptors poll can handle limited by RLIMIT_NOFILE, whereas the > >> >> >> > maximum number of file descriptors select can handle is limited by > >> >> >> > FD_SETSIZE. > >> >> >> > > >> >> >> > I don't think this is a big problem in reality, though - both values > >> >> >> > seem to be pretty high in most implementations. And IIRC git-daemon is > >> >> >> > the only one who needs more than 2, and it doesn't even check > >> >> >> > RLIMIT_NOFILE. > >> >> >> > > >> >> >> > >> >> >> > >> >> >> To be clear: I think this strategy is the best option (at least for > >> >> >> non-Windows, where select() might be our only option). > >> >> >> > >> >> >> But perhaps you should include a check along the lines of this: > >> >> >> > >> >> >> if (nfds > FD_SETSIZE) > >> >> >> return errno = EINVAL, error("poll: nfds must be below %d", FD_SETSIZE); > >> >> >> > >> >> >> Just so we can know when the code fails :) > >> >> > > >> >> > Well, per your own FD_SET example, the FD_SETSIZE on windows > >> >> > means different thing than FD_SETSIZE on old-style bitmap-based > >> >> > select() implementation. > >> >> > > >> >> > On Unix, it's max fd number + 1, on windows it's max count. > >> >> > > >> >> > >> >> > >> >> Are you sure this applies for all Unix, not just some given Unix-y system? > >> > > >> > Not sure. Just pointing out that the above check is not > >> > universal enough. > >> > > >> > >> > >> Isn't it? How could one possibly pass more than max fd number + 1 file > >> descriptors, since they start at 0? I guess one could specify a given > >> fd more than once, but that'd be kind of redundant... and also very > >> unlikely in our case ;) > > > > Pass one fd with value 70 it. Check returns error, although > > everything would work. > > > > > No, not with the check I posted. I checked nfds, not the value of the > fds themselves. nfds is clearly the size of the fds array -- if it > wasn't, it'd be impossible for the poll-implementation to know how big > the array is! Ah, ok. I though this is the nfds for select(). In this case it can still fail, as nfds=1 for poll() can still go over FD_SETSIZE for select() if the fd value is big enough. On non-windows, that is. Dunno if this matters for git-daemon.. -- marko