Re: PATCH: propagate gdb host keyboard interrupts through gdbserver

[jtc at redback dot com (J.T. Conklin)]

>>>>> "Greg" == Greg McGary <greg@mcgary.org> writes:
>> ...  I noticed that code uses a NULL ptr for the timeout argument
>> instead of constructing a zeroed timeval.  I'd go for the NULL ptr
>> approach instead of the timeval.  It avoids problems with those
>> systems that don't have sys/time.h.

Greg> NULL means never timeout, i.e. block awaiting I/O.  zeroed
Greg> timeval means timeout immediately if there's no I/O.  An
Greg> alternative to select is to set NDELAY on the socket with fcntl.
Greg> Yes, i should add a comment explaining why all this rot is
Greg> necessary: because I observed NetBSD delivering redundant
Greg> SIGIOs.

Duh.  And I even read the manpage while composing my reply.  Perhaps I
need reading glasses (or is that comprehension glasses?).

It also appears that my worries about targets not having sys/time.h
are not well founded.  I surveyed the GDB sources and found that many,
including remote.c which is included in practically every config,
which include sys/time.h without special consideration.

I still think we'll need to handle unistd.h though.

        --jtc

-- 
J.T. Conklin
RedBack Networks