Re: CVS commit: src/usr.bin/make
Joerg Sonnenberger wrote: > > This improves the worst case performance (eg examining libc with > > nothing to do) by a factor of 4. > > ...except now build.sh on !NetBSD fails immediately due to missing > strlcpy. Should be fixed... Don't know if you want to get carried away with tools/make/configure* to build things like strlcpy which are available in the tree if necessary.
Re: CVS commit: src/usr.bin/make
Joerg Sonnenberger wrote: > > This improves the worst case performance (eg examining libc with > > nothing to do) by a factor of 4. > > ...except now build.sh on !NetBSD fails immediately due to missing > strlcpy. Gah, bmake provides strlcpy... sorry about that.
Re: CVS commit: src/crypto/external/bsd/openssl
In article <20160604202302.gc14...@britannica.bec.de>, Joerg Sonnenberger wrote: > >The only evidence so far presented is that automated testing. There is >no backtrace, there are other changes in the time interval, there is no >evidence that a clean build was tried. I reproduced it by building HEAD, chroot to release and running ssh both on i386 and x86_64. I did not mean to be rude :-). There is enough bleeding on head to keep us busy without opening more battlefronts. christos
Re: CVS commit: src/crypto/external/bsd/openssl
On Sun, Jun 05, 2016 at 05:56:09AM +1000, matthew green wrote: > Joerg Sonnenberger writes: > > On Sat, Jun 04, 2016 at 02:22:46PM -0400, Christos Zoulas wrote: > > > Module Name: src > > > Committed By: christos > > > Date: Sat Jun 4 18:22:46 UTC 2016 > > > > > > Modified Files: > > > src/crypto/external/bsd/openssl/dist/crypto: x86_64cpuid.pl > > > src/crypto/external/bsd/openssl/dist/crypto/perlasm: x86gas.pl > > > src/crypto/external/bsd/openssl/lib/libcrypto/arch/i386: x86cpuid.S > > > src/crypto/external/bsd/openssl/lib/libcrypto/arch/x86_64: > > > x86_64cpuid.S > > > > > > Log Message: > > > revert, everything coredumps with this change. > > > > There was a time when a revert without any evidence was considered rude. > > I have been running with those changes for month, without any coredumps. > > reverting this was the right thing. > > there was plenty of evidence on mailing lists and autoamted testing. The only evidence so far presented is that automated testing. There is no backtrace, there are other changes in the time interval, there is no evidence that a clean build was tried. Joerg
re: CVS commit: src/crypto/external/bsd/openssl
Joerg Sonnenberger writes: > On Sat, Jun 04, 2016 at 02:22:46PM -0400, Christos Zoulas wrote: > > Module Name:src > > Committed By: christos > > Date: Sat Jun 4 18:22:46 UTC 2016 > > > > Modified Files: > > src/crypto/external/bsd/openssl/dist/crypto: x86_64cpuid.pl > > src/crypto/external/bsd/openssl/dist/crypto/perlasm: x86gas.pl > > src/crypto/external/bsd/openssl/lib/libcrypto/arch/i386: x86cpuid.S > > src/crypto/external/bsd/openssl/lib/libcrypto/arch/x86_64: > > x86_64cpuid.S > > > > Log Message: > > revert, everything coredumps with this change. > > There was a time when a revert without any evidence was considered rude. > I have been running with those changes for month, without any coredumps. reverting this was the right thing. there was plenty of evidence on mailing lists and autoamted testing. .mrg.
Re: CVS commit: src/sys/ufs/ext2fs
Thank you, Hrishikesh! On Jun 3, 2016 11:35, "Christos Zoulas" wrote: > Module Name:src > Committed By: christos > Date: Fri Jun 3 15:35:48 UTC 2016 > > Modified Files: > src/sys/ufs/ext2fs: ext2fs.h ext2fs_bmap.c > Added Files: > src/sys/ufs/ext2fs: ext2fs_extents.c ext2fs_extents.h > > Log Message: > Add ext4 extent support from GSoC 2016 (Hrishikesh Goyal), from the FreeBSD > ext2 code. > > > To generate a diff of this commit: > cvs rdiff -u -r1.36 -r1.37 src/sys/ufs/ext2fs/ext2fs.h > cvs rdiff -u -r1.26 -r1.27 src/sys/ufs/ext2fs/ext2fs_bmap.c > cvs rdiff -u -r0 -r1.1 src/sys/ufs/ext2fs/ext2fs_extents.c \ > src/sys/ufs/ext2fs/ext2fs_extents.h > > Please note that diffs are not public domain; they are subject to the > copyright notices on the relevant files. > >
Re: CVS commit: src/crypto/external/bsd/openssl
On Sat, Jun 04, 2016 at 02:22:46PM -0400, Christos Zoulas wrote: > Module Name: src > Committed By: christos > Date: Sat Jun 4 18:22:46 UTC 2016 > > Modified Files: > src/crypto/external/bsd/openssl/dist/crypto: x86_64cpuid.pl > src/crypto/external/bsd/openssl/dist/crypto/perlasm: x86gas.pl > src/crypto/external/bsd/openssl/lib/libcrypto/arch/i386: x86cpuid.S > src/crypto/external/bsd/openssl/lib/libcrypto/arch/x86_64: > x86_64cpuid.S > > Log Message: > revert, everything coredumps with this change. There was a time when a revert without any evidence was considered rude. I have been running with those changes for month, without any coredumps. Joerg
Re: CVS commit: src/crypto/external/bsd/openssl/lib/libcrypto/arch
>>> Takeshi Nakayama wrote > >>> "Joerg Sonnenberger" wrote > > > Module Name:src > > Committed By: joerg > > Date: Fri Jun 3 15:42:15 UTC 2016 > > > > Modified Files: > > src/crypto/external/bsd/openssl/lib/libcrypto/arch/i386: x86cpuid.S > > src/crypto/external/bsd/openssl/lib/libcrypto/arch/x86_64: > > x86_64cpuid.S > > > > Log Message: > > Regenerate to use .ctor. > > It seems that libcrypto on amd64 and i386 is broken by this change. > > http://releng.netbsd.org/b5reports/amd64/ > http://releng.netbsd.org/b5reports/i386/ Oops, thanks to christos for reverting. -- Takeshi Nakayama
Re: CVS commit: src/crypto/external/bsd/openssl/lib/libcrypto/arch
>>> "Joerg Sonnenberger" wrote > Module Name: src > Committed By: joerg > Date: Fri Jun 3 15:42:15 UTC 2016 > > Modified Files: > src/crypto/external/bsd/openssl/lib/libcrypto/arch/i386: x86cpuid.S > src/crypto/external/bsd/openssl/lib/libcrypto/arch/x86_64: > x86_64cpuid.S > > Log Message: > Regenerate to use .ctor. It seems that libcrypto on amd64 and i386 is broken by this change. http://releng.netbsd.org/b5reports/amd64/ http://releng.netbsd.org/b5reports/i386/ -- Takeshi Nakayama
