Hi Igor,

I have some worries about the timeval struct definition handling that
you mention in the patch. Exactly related with which headers are
actually required to get the timeval struct definition.

The following is what makes think that something is not yet fine, but
please correct me in whatever is wrong, and try to find out needed
info...

sockLib has a function named connectWithTimeout() in which its fourth
argument is a pointer to a timeval struct. Could you verify if the
prototype in sockLib.h actually uses a pointer to timeval struct or a
pointer to void?

If it actually is a pointer to a timeval struct, isn't it enough to
include sockLib.h to get the timeval struct definition, or should
other header be included before sockLib.h to be able to use the
connectWithTimeout() function?

Another thing...

It seems that for VxWorks 6.0 the definition of timeval was in time.h
but for VxWorks 6.2 it is now in sys/time.h. Existing both time.h and
sys/time.h for VxWorks 6.X.

Could you verify if both sys/time.h and time.h can be included in a
program without compilation errors?

Could you verify what actually happens with these two header files on
VxWorks 5.X?

Isn't it enough to include both time.h and sys/time.h on all VxWorks
versions to get the timeval struct definition?

Please double check if the only way to get the timeval struct
definition is including <net/if.h>, I find it a little bit strange.

-- 
-=[Yang]=-