The problem isn't that there is no generated opt_netgraph.h but that of all the numerous opt_netgraph.h files, none are found during the KLD compilation. I suspect either: 1. I mistakenly deleted or lost to fs corruption the key opt_netgraph.h 2. A /usr/src/* Makefile or a bsd include mk is missing a -I path specifier 3. Whomever submitted the last netgraph.h edit forgot to tweak a conf or Makefile 4. some part of a recent cvsup failed and I didn't notice it Reason (1) is not likely since a kernel rebuild should have restored any such deleted file... My current workaround is to just touch an opt_netgraph.h into the "_at_" symlinked directory. Works fine. I just remain curious as to what's really broken. On Tuesday 13 September 2005 11:53 am, John-Mark Gurney wrote: > Lonnie VanZandt wrote this message on Tue, Sep 13, 2005 at 10:00 -0600: > > Sometime between 5.4 RC3 and 5.4 STABLE, netgraph/netgraph.h went from > > version 1.43.2.5 to 1.43.2.10. In doing so, an include of > > "opt_netgraph.h" was added to that file. This causes code including > > netgraph/netgraph.h, on my system, to fail to compile since > > opt_netgraph.h isn't in the include path. > > > > When a kernel conf file doesn't have "options NETGRAPH" the generated > > opt_netgraph.h file is empty. With "options NETGRAPH" the file simply > > contains "#define NETGRAPH 1". So, either way, the file isn't hard to > > create - but I need to know where it should reside in the "_at_" hierarchy. > > > > Does anyone (say julian_at_) know? And, just for curiousity sake, does > > anyone know which update step I probably overlooked which led to the > > breakage on my platform? > > The opt_*.h files are automaticly created by config... Sometimes you > just need to rerun config before building your kernel... > > If it fails building the module, then the module's Makefile needs to > be fixed to create the opt_*.h files as necessary...Received on Wed Sep 14 2005 - 12:09:23 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:43 UTC