... * Building userspace cc -march=i386 -Os -pipe -fomit-frame-pointer -Wall -Wstrict-prototypes -I../include -DLinux -DNETLINK_ISCSI=8 -D_GNU_SOURCE -c -o util.o util.c cc -march=i386 -Os -pipe -fomit-frame-pointer -Wall -Wstrict-prototypes -I../include -DLinux -DNETLINK_ISCSI=8 -D_GNU_SOURCE -c -o io.o io.c cc -march=i386 -Os -pipe -fomit-frame-pointer -Wall -Wstrict-prototypes -I../include -DLinux -DNETLINK_ISCSI=8 -D_GNU_SOURCE -c -o auth.o auth.c io.c:27:21: ifaddrs.h: No such file or directory io.c: In function `get_hwaddress_from_netdev': io.c:95: warning: implicit declaration of function `getifaddrs' io.c:108: error: dereferencing pointer to incomplete type io.c:109: error: dereferencing pointer to incomplete type io.c:112: error: dereferencing pointer to incomplete type io.c:114: error: dereferencing pointer to incomplete type io.c:115: error: dereferencing pointer to incomplete type io.c:119: error: dereferencing pointer to incomplete type io.c:122: error: dereferencing pointer to incomplete type io.c:123: error: dereferencing pointer to incomplete type io.c:126: error: dereferencing pointer to incomplete type io.c:132: error: dereferencing pointer to incomplete type io.c:135: error: dereferencing pointer to incomplete type io.c:159: warning: implicit declaration of function `freeifaddrs' io.c: In function `get_netdev_from_hwaddress': io.c:190: error: dereferencing pointer to incomplete type io.c:191: error: dereferencing pointer to incomplete type io.c:194: error: dereferencing pointer to incomplete type io.c:196: error: dereferencing pointer to incomplete type io.c:197: error: dereferencing pointer to incomplete type io.c:201: error: dereferencing pointer to incomplete type io.c:204: error: dereferencing pointer to incomplete type io.c:205: error: dereferencing pointer to incomplete type io.c:208: error: dereferencing pointer to incomplete type io.c:214: error: dereferencing pointer to incomplete type io.c:234: error: dereferencing pointer to incomplete type io.c:236: error: dereferencing pointer to incomplete type io.c: At top level: io.c:86: warning: 'get_hwaddress_from_netdev' defined but not used make: *** [io.o] Error 1 make: *** Waiting for unfinished jobs.... make: *** wait: No child processes. Stop. Proper fix would be to implement getifaddrs/freeifaddrs in uclibc.
needs also versionsort(3).
I have patches to support versionsort and strverscmp in uclibc. I'm not sure we want that since its non-POSIX. alphasort could be used without problem. The only difference would be that it would parse eth10 before eth2 config files. I don't know if that matters or not. If it matters, it shouldn't be too hard to implement a simplified versionsort in the code itself. The getifaddrs can easily be worked around by using ioctl(..., SIOCGIFCONF, ...). I have reported upstream.
(In reply to comment #2) > > The getifaddrs can easily be worked around by using ioctl(..., SIOCGIFCONF, > ...). ioctl would kill ipv6 support so its probably not an option for upstream :-( We can either just skip the check that hw addr and iface correspond or we can add getifaddrs support in uclibc. It seems like the preferred way to list configured interfaces in linux is getifaddrs (which should use netlink). in other words, we probably need getifaddrs in uclibc.
Created attachment 135501 [details, diff] open-iscsi-2.0.865.15-uclibc.patch uses if_nameindex instead of getifaddrs. Also implemented a simplified versionsort. alphasort would probaly work just fine as well. (the difference is in which order the iscsi devices are managed) Commented out a big hunk of unused code.
patch is sent upstream.
(In reply to comment #5) > patch is sent upstream. If this is implemented I guess clvm will benefit too.
Patch was integrated upstream, could anyone test the newest ebuild here (maintainer is currently in devaway status): http://bugs.gentoo.org/show_bug.cgi?id=196096
*bump*
Mike, Looks like a version bump would fix this.
(In reply to comment #9) > Mike, > Looks like a version bump would fix this. > I'm on leave, so give me a hand. bump to what version? link me to upstream version? is it something more involved than a bump? all my iscsi gear is in production, my one test target is out on RMA so no testing can be done....
Mike, Sorry for the trouble. Based on comment #7 I mis-interpreted the reporters use of the word *bump* to mean that a new upstream version was available.
bah, too bad we don't have a RESO:MERGED option or I'd merge it with the bug mentioned in comment #7. I'll take a look at this when I have some iscsi gear to play with again.
(In reply to comment #7) > Patch was integrated upstream, could anyone test the newest ebuild here > (maintainer is currently in devaway status): > http://bugs.gentoo.org/show_bug.cgi?id=196096 > Yes, new version compiles just fine on uclibc. I add it as a depend so we get notified when its commited in portage and can close this bug. Thanks!
2.0.870 is in CVS finally, closing this one, too.
