make buildworld for RELENG_5 failing on RELENG_5_3 in /usr/src/usr.sbin/syslogd, _PATH_LOG_PRIV not defined

Steve Hodgson steve at acidy.com
Tue Feb 15 13:12:33 PST 2005


On Tuesday 15 February 2005 14:02, Tuure Laurinolli wrote:
> Steve Hodgson wrote:
> > On Sunday 16 January 2005 17:37, Kris Kennaway wrote:
> >>On Sun, Jan 16, 2005 at 02:22:05PM +0000, Steve Hodgson wrote:
> >>>I'm getting the following error on a RELENG_5_3 box when trying to
> >>>compile the RELENG_5 sources. I've been cvsupping now for about a week
> >>>and continuing to get the same error when we get to syslogd.
>
> I get the same error updating RELENG_5_3 to RELENG_5. It's strange
> because the constant most certainly is defined in
> /usr/src/sys/sys/syslog.h. However, when I straced the make, it seemed
> that this file was never used. Instead it used syslog.h from my already
> installed system!
>
On Tuesday 15 February 2005 14:02, Tuure Laurinolli wrote:
> Steve Hodgson wrote:
> > On Sunday 16 January 2005 17:37, Kris Kennaway wrote:
> >>On Sun, Jan 16, 2005 at 02:22:05PM +0000, Steve Hodgson wrote:
> >>>I'm getting the following error on a RELENG_5_3 box when trying to
> >>>compile the RELENG_5 sources. I've been cvsupping now for about a week
> >>>and continuing to get the same error when we get to syslogd.
>
> I get the same error updating RELENG_5_3 to RELENG_5. It's strange
> because the constant most certainly is defined in
> /usr/src/sys/sys/syslog.h. However, when I straced the make, it seemed
> that this file was never used. Instead it used syslog.h from my already
> installed system!
>
In the end I traced this down to devel/ccache which I was using at the time. I 
had copied the lines from /usr/local/share/doc/ccache/make.conf into my own 
make.conf:

.if !defined(NOCCACHE)
.if ${.CURDIR:M/usr/src*}
CC=/usr/local/libexec/ccache/cc
CXX=/usr/local/libexec/ccache/c++
.else
CC=cc
CXX=c++
.endif
.else
CC=/usr/bin/cc
CXX=/usr/bin/c++
.endif

However the check on line 2 didn't appear to have the desired effect, so 
instead of CC staying equal to "cc" and pulling in the buildworld cc and 
headers, it used the system cc and headers. Of course I then tried NOCCACHE=1 
without reading the code, which by definition will do the same thing (system 
compiler and headers). In the end I commented all these lines out and 
buildworld completed.

Can't recall why i didn't send-pr this one - probably because no one else has 
complained till now.

Steve


More information about the freebsd-stable mailing list