On 6/2/09, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Jeremy Kerr <jk@ozlabs.org> writes:
> > The following patch changes psecure_write to be more like psecure_read -
> > it only alters the signal mask if the connection is over SSL. It's only
> > an RFC, as I'm not entirely sure about the reasoning behind blocking
> > SIGPIPE for the non-SSL case - there may be other considerations here.
>
>
> The consideration is that the application fails completely on server
> disconnect (because it gets SIGPIPE'd). This was long ago deemed
> unacceptable, and we aren't likely to change our opinion on that.
>
> What disturbs me about your report is the suggestion that there are
> paths through that code that fail to protect against SIGPIPE. If so,
> we need to fix that.
Slightly OT, but why are we not using MSG_NOSIGNAL / SO_NOSIGPIPE
on OS'es that support them? I guess significant portion of userbase
has at least one of them available...
Thus avoiding 2 syscalls per operation plus potential locking issues.
--
marko