On Fri, Nov 22, 2013 at 10:04:02PM +0100, Maxime Villard wrote: > Le 22/11/2013 17:48, Ted Unangst a ??crit : > > On Fri, Nov 22, 2013 at 10:09, Stuart Henderson wrote: > >> On 2013/11/22 07:25, Maxime Villard wrote: > > > >>> If write() fails without EPIPE, d is decremented, and the function > >>> keeps looping. If write() succeeds after several loops, d will be > >>> negative, and the function will write from buf-XX. > > > > When does write() fail and do we want to keep writing? (If I fill up > > the filesystem, do I really want to spin here?) > > > I must say that I actually fail to see why EPIPE is the only case > handled; it should stop looping regardless of the errno code, > shouldn't it?
often you'll ned to retry on EAGAIN. But indeed, the code is weird. > > > > > >> Shouldn't it be something more like this? Otherwise if the write() fails, > >> we attempt writing one byte fewer for every retry. > > > > That looks better to me. > >
