On 10/04/19 10:29, Kurt Van Dijck wrote: > I don't know a system where (a) or (b) are valid. My point is that such > system could could exist, so I learned not to inspect the header files > looking for a type, but inspect man-pages or specifications when looking > for a type, and so time_t is defined in time.h. I didn't just go trolling a grepping for time_t. sys/time.h is pretty well-established in UNIX, and I think you're punting when you point to ANSI C alone as the authority here. As for documentation, how does SUSv2 seem? https://pubs.opengroup.org/onlinepubs/7908799/xsh/systime.h.html > Now that I know that that file is used as include for kernel code, I'd > rather include time.h in the userspace c-files. My point is that include/net/ isn't strictly userspace. If you feel the need, then go ahead and include <time.h> in user level files. This just isn't one of those. If you must do this in ppp_def.h, then it needs to be guarded against *all* of the systems where including a top-level header file inside a kernel module is the wrong thing to do, not just "ifndef SOLARIS". Do you know which systems those are? I can tell you that Solaris/Illumos is at least one such system, but I can't tell you that it's *all* of them. I think this include is out of place here. -- James Carlson 42.703N 71.076W <carlsonj@xxxxxxxxxxxxxxx>
