On Friday 26 February 2016 15:51:28 Yuriy M. Kaminskiy wrote: > yumkam@xxxxxxxxx (Yuriy M. Kaminskiy) > > writes: > > Ruediger Meier <sweet_f_a@xxxxxx> writes: > >> From: Ruediger Meier <ruediger.meier@xxxxxxxxxxx> > > [...] > > > `O_CREAT`, etc; most notable, `O_CLOEXEC`; however, if `O_CLOEXEC` is > > missing on system, "c.h" defines it as 0, so it is silently ignored on > > those systems, instead of being emulated; so, whenever it matters, > > callers must call **both** `open(O_CLOEXEC)` and > > `fcntl(F_SETFD,FD_CLOEXEC)`]). > > ... something like attached below. In some cases it is maybe > unnecessary, but better safe than sorry. I'm not sure whether we should add all these fcntl lines just for a few old non-Linux systems. That's probably why Karel simply added quick and dirty this unsafe fallback defining O_CLOEXEC to 0. We are using this fallback already since 2011 and and now 2016 we have even less systems where O_CLOEXEC is not available. BTW we are using mkostemp (which is a real showstopper for the build!) since 3 years and nobody complained. I assume any system with mkostemp will also have O_CLOEXEC. I'm going to add another hopefully correct patch for case mkostemp only. cu, Rudi -- To unsubscribe from this list: send the line "unsubscribe util-linux" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
