Petr Ovtchenkov wrote: > Problem in kernel's (2.6.32.2) header linux/socket.h: > due to change > > -#if defined(__KERNEL__) || !defined(__GLIBC__) || (__GLIBC__ < 2) > +#ifdef __KERNEL__ > > procedure of generating of sanitized kernel headers > (i.e. 'make INSTALL_HDR_PATH=dest headers_install') produce > linux/socket.h without sa_family_t definition. > > But other headers in linux/* still expect that sa_family_t > defined in linux/socket.h. > > As a result, strace (4.5.19) can't correctly detect > linux/netlink.h and fail during compilation. > > Is this known issue?
Yes, see the thread starting at http://www.linuxfromscratch.org/pipermail/blfs-support/2009-December/066157.html and continuing at http://www.linuxfromscratch.org/pipermail/blfs-support/2010-January/066184.html during which Ken kindly points to the upstream strace patch. Current upstream kernel advice is to include sys/socket.h anywhere where you already include linux/socket.h (see http://lkml.indiana.edu/hypermail/linux/kernel/1001.0/00923.html). Regards, Matt. -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
