Hi, I'm working in a PR about pdksh in current, and I'm not really able to figure out what's going on. So that's the facts :
A breakage was reported in august about pdksh in current (ports/181438). The PR says pdksh hangs after the first command was sent ( run pdks, type 'ls' validate and you are locked) After a lookup on the code and a little debug I found that, in current, after the main process fork to perform the 'ls', the child fall in a zombi state and the father remains waiting for the SIGCHLD signal forever. Of course, in 9-STABLE things works fine. I try to use a pdksh build in 9 in current, and things works fine. I start thinking the problem commes from clang. Before any change I test GCC in current,and bang! dosn't works. Here is a little array explaining my tests. Environments: 10.32 : FreeBSD 10 Alpha 4 x32 10.64 : FreeBSD 10 Alpha 4 AMD64 9.32 : FreeBSD 9-STABLE x32 9.64 : FreeBSD 9-STABLE AMD64 build / run in| 10.32 | 10.64 | 9.32 | 9.64 | ---------------------------------------------- Build w/CLANG |KO |KO | Not tested in native env |v3.3 |v3.3 | ---------------------------------------------- Build w/GCC |KO |KO |all OK in native env |v4.6.3 |v4.2.1 |v4.2.1 ---------------------------------------------- Build w/GCC |OK |OK |all KO(Build in 10) in 9-STABLE 32|v4.2.1 |v4.2.1 |v4.6.3 ---------------------------------------------- So.... If I use a pdks build in its own environement with clang or gcc, pdksh fail. The same tes done in 9 success. If I copy 9-STABLE pdksh builds to 10, the imported pdksh works fine. If I copy the 10 pdksh builds to 9, pdksh dosn't works. It seems the problem is related to the build environement in current, without regads to the copiler used. Maybe a macro failure...... Anyone has a clue ? Best regards, - rodrigo _______________________________________________ freebsd-current@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"
