Re: Possible Clang regression - object files are always mode 0600
On 2012-05-29 18:59, Dimitry Andric wrote: > On 2012-05-29 18:27, David Wolfskill wrote: >> On Tue, May 29, 2012 at 01:39:19AM -0500, James wrote: >>> Hi all. I've come across a possible Clang regression that recently >>> crept into stable/9. I'd like to check to see whether it's a true bug >>> or if I jacked something up. It appears all object files are created >>> with mode 0600 rather than honoring umask. > Just so you know, this is indeed a regression that has crept into clang, > and even into the 3.1 release. :( > > I'm working on fixing it in head, then I will merge the fix to stable/9 > in a few days. Fixed in head in r236260; the merge will follow soon. ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "freebsd-stable-unsubscr...@freebsd.org"
Re: Possible Clang regression - object files are always mode 0600
On Tue, May 29, 2012 at 11:59 AM, Dimitry Andric wrote: > Just so you know, this is indeed a regression that has crept into > clang, and even into the 3.1 release. :( > > I'm working on fixing it in head, then I will merge the fix to > stable/9 in a few days. Cool. Thanks much for fixing it, Dimitry. -- James. ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "freebsd-stable-unsubscr...@freebsd.org"
Re: Possible Clang regression - object files are always mode 0600
On Tue, May 29, 2012 at 06:59:17PM +0200, Dimitry Andric wrote: > ... > Just so you know, this is indeed a regression that has crept into clang, > and even into the 3.1 release. :( > > I'm working on fixing it in head, then I will merge the fix to stable/9 > in a few days. Very cool; should be in time for 9.1-R, then. :-) Thanks! Peace, david -- David H. Wolfskill da...@catwhisker.org Depriving a girl or boy of an opportunity for education is evil. See http://www.catwhisker.org/~david/publickey.gpg for my public key. pgpHet8QZ0fEF.pgp Description: PGP signature
Re: Possible Clang regression - object files are always mode 0600
On 2012-05-29 18:27, David Wolfskill wrote: > On Tue, May 29, 2012 at 01:39:19AM -0500, James wrote: >> Hi all. I've come across a possible Clang regression that recently >> crept into stable/9. I'd like to check to see whether it's a true bug >> or if I jacked something up. It appears all object files are created >> with mode 0600 rather than honoring umask. Just so you know, this is indeed a regression that has crept into clang, and even into the 3.1 release. :( I'm working on fixing it in head, then I will merge the fix to stable/9 in a few days. ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "freebsd-stable-unsubscr...@freebsd.org"
Re: Possible Clang regression - object files are always mode 0600
On Tue, May 29, 2012 at 01:39:19AM -0500, James wrote: > Hi all. I've come across a possible Clang regression that recently > crept into stable/9. I'd like to check to see whether it's a true bug > or if I jacked something up. It appears all object files are created > with mode 0600 rather than honoring umask. This breaks installworld > for me. I use a centralized src/obj over NFS configuration and > installing to the target system fails with a permission denied error. Hmmm... yeah, that could cause a bit of hassle for me, as well. Checking a FreeBSD/i386 stable/9 /usr/obj as of r236226, I see 16544 files with permission 0600 and 710 files that have something else (typically, 0644): d134(9.0-S)[6] find . -type f -name \*.o \! -perm 0600 -ls | wc -l 710 d134(9.0-S)[7] find . -type f -name \*.o -perm 0600 -ls | wc -l 16544 d134(9.0-S)[8] find . -type f -name \*.o \! -perm 0600 -ls | head 2075139 12 -rwxr-xr-x1 root wheel6016 May 25 05:10 ./tmp/usr/lib/dtrace/drti.o 20729944 -rwxr-xr-x1 root wheel 829 May 24 12:00 ./tmp/usr/lib/crti.o 20729954 -rwxr-xr-x1 root wheel 785 May 24 12:00 ./tmp/usr/lib/crtn.o 20729964 -rwxr-xr-x1 root wheel2014 May 24 12:00 ./tmp/usr/lib/gcrt1.o 20729974 -rwxr-xr-x1 root wheel1848 May 24 12:00 ./tmp/usr/lib/crt1.o 20729984 -rwxr-xr-x1 root wheel2004 May 24 12:00 ./tmp/usr/lib/Scrt1.o 20729994 -rwxr-xr-x1 root wheel1628 May 24 12:00 ./tmp/usr/lib/crtbegin.o 20730004 -rwxr-xr-x1 root wheel1252 May 24 12:00 ./tmp/usr/lib/crtend.o 20729804 -rwxr-xr-x1 root wheel1968 May 24 12:00 ./tmp/usr/lib/crtbeginT.o 20729814 -rwxr-xr-x1 root wheel1864 May 24 12:00 ./tmp/usr/lib/crtbeginS.o d134(9.0-S)[9] find . -type f -name \*.o -perm 0600 -ls | head 2472960 20 -rw---1 root wheel9984 May 29 04:27 ./make.i386/usr/src/usr.bin/make/cond.o 24729618 -rw---1 root wheel2456 May 29 04:27 ./make.i386/usr/src/usr.bin/make/buf.o 2472962 20 -rw---1 root wheel9296 May 29 04:27 ./make.i386/usr/src/usr.bin/make/dir.o 2472963 24 -rw---1 root wheel 11924 May 29 04:27 ./make.i386/usr/src/usr.bin/make/arch.o 24729648 -rw---1 root wheel2352 May 29 04:27 ./make.i386/usr/src/usr.bin/make/hash.o 2472966 12 -rw---1 root wheel4188 May 29 04:27 ./make.i386/usr/src/usr.bin/make/for.o 24729654 -rw---1 root wheel1588 May 29 04:27 ./make.i386/usr/src/usr.bin/make/hash_tables.o 2472967 60 -rw---1 root wheel 29844 May 29 04:27 ./make.i386/usr/src/usr.bin/make/job.o 2472968 40 -rw---1 root wheel 19964 May 29 04:27 ./make.i386/usr/src/usr.bin/make/main.o 2472970 12 -rw---1 root wheel6096 May 29 04:27 ./make.i386/usr/src/usr.bin/make/make.o d134(9.0-S)[10] So I won't claim "always mode 0600," but it looks like enough to be annoying. :-} I'm pretty sure I didn't change anything relevent from how I build with gcc (on a different slice). Peace, david -- David H. Wolfskill da...@catwhisker.org Depriving a girl or boy of an opportunity for education is evil. See http://www.catwhisker.org/~david/publickey.gpg for my public key. pgpHB9AU3KYRT.pgp Description: PGP signature
Possible Clang regression - object files are always mode 0600
Hi all. I've come across a possible Clang regression that recently crept into stable/9. I'd like to check to see whether it's a true bug or if I jacked something up. It appears all object files are created with mode 0600 rather than honoring umask. This breaks installworld for me. I use a centralized src/obj over NFS configuration and installing to the target system fails with a permission denied error. Using obj/[SNIP]/src/cddl/lib/drti as an example: Clang: -rw-r--r-- 1 jjk users 7731 May 24 15:05 .depend -rw--- 1 jjk users 8496 May 24 15:05 drti.o GCC: -rw-r--r-- 1 jjk users 7731 May 24 20:54 .depend -rw-r--r-- 1 jjk users 8808 May 24 20:54 drti.o I tested on May 24th (with 3.1 prerelease MFC) as well as the 3.1 release after it was MFC'd on Sunday. TIA! -- James. ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "freebsd-stable-unsubscr...@freebsd.org"