from:"Volodymyr Kostyrko"


23.11.2012 10:59, Beeblebrox:

9.1-PRERELEASE #2 r242867: When will buildworld + clang + ccache be
functional?
I think I read somewhere that buildworld + clang - ccache was working fine
now. Any info on date ccache will work with clang and when to try?


Define CCACHE_CPP2 as otherwise clang will yell at ccache trying to 
compile precompiled source with extra flags. This breaks libtool and a 
some other autoconf tests.


Care to not specify CCACHE_CPP2 for virtualbox and friends.


Also, clang users - do you see any speed advantage for kernel / world /
ports built with clang?


Builds faster.

--
Sphinx of black quartz, judge my vow.
___
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: Latest for buildworld + clang + ccache


23.11.2012 11:20, Beeblebrox:

Care to not specify CCACHE_CPP2 for virtualbox and friends.

Not true for Jails I assume?


No, it's just that virtualbox still doesn't build with clang and gcc 
doesn't really like some -MD (?) flags.



Define CCACHE_CPP2 as otherwise clang will yell at ccache

Is this the same reason CMAKE yells at CCACHE?


Haven't seen this before. Can you share?

--
Sphinx of black quartz, judge my vow.
___
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: Latest for buildworld + clang + ccache


23.11.2012 11:30, Beeblebrox:

Is this the same reason CMAKE yells at CCACHE?

Haven't seen this before. Can you share?

I meant for ports, not buildworld...

I use bsdadminscripts  and specify preferences in
/usr/local/etc/buildflags.conf. I tried these and did not work. What is the
setting for this, and I assume I can set CCACHE_CPP2 in that file also?
USE_GCC=clang
BUILDFLAGS_GCC= clang


CC= clang
CPP= clang-cpp
CXX= clang++

Yes, you can include CCACHE_CPP2 in this file too.

--
Sphinx of black quartz, judge my vow.
___
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: Latest for buildworld + clang + ccache


23.11.2012 11:46, Beeblebrox:

I think I made a mistake:
clang: error: linker command failed with exit code 1 (use -v to see
invocation)
*** [tblgen] Error code 1
1 error
*** [bootstrap-tools] Error code 2
1 error
*** [_bootstrap-tools] Error code 2
1 error
*** [buildworld] Error code 2

in /urs/local/buildflags.conf I have:
/usr/src | /usr/src/*{
 CC= clang
 CPP=clang
 CXX=clang
 USE_CCACHE
 USE_CCACHE_CPP2  }


What I wrote is right for ports but not world. ccache port comes with 
readme file for this setup.


--
Sphinx of black quartz, judge my vow.
___
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: buildworld with clang breaks because no cc


23.11.2012 15:45, Beeblebrox:

I am now stuck with trying to make this work. I am doing buildworld with
clang and ccache enabled.

1. I have to keep re-starting buildworld at different points of break. When
re-started the build continues past the last break-point and goes on to
break somewhere else.
2. This continues until build gets to === usr.bin/ypwhich (all) and cannot
continue past the error.
3. I disable ccache in /usr/local/etc/buildflags.conf and re-start the
build. Progresses upto:
sh /asp/src/tools/install.sh -s -o root -g wheel -m 444 libz.so.6
/usr/obj/asp/src/lib32/usr/lib32
ln -fs libz.so.6 /usr/obj/asp/src/lib32/usr/lib32/libz.so

However, examining the contents after build-break of
/usr/obj/asp/src/lib32/usr/lib32/ shows libz.so.6 and its linked alias
libz.so.
I am not able to move any further without help because I cannot figure out
where the problem may be.

Thanks for any help.


Try running make buildworld without -j.

--
Sphinx of black quartz, judge my vow.
___
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: buildworld with clang breaks because no cc


23.11.2012 16:14, Beeblebrox:

Thanks for the suggestion. Build progressed a little further then had other
problem:

=== gnu/lib/libstdc++ (all)
building shared library libstdc++.so.6
/usr/obj/asp/src/tmp/usr/bin/ld: warning: creating a DT_TEXTREL in a shared
object.
clang: error: linker command failed with exit code 1 (use -v to see
invocation)
*** [libstdc++.so.6] Error code 1
Stop in /asp/src/gnu/lib/libstdc++.
*** [all] Error code 1
Stop in /asp/src/gnu/lib.
*** [gnu/lib__L] Error code 1

Why is it building gcc4.2 ?? Should I not be using WITH_CLANG_IS_CC= yes ?
Bummer...


I'd rather say you should recheck/post your relevant make.conf / 
src.conf options. And WITHOUT_GCC is not right thing to do for now, gcc 
is still needed for a number of ports.


--
Sphinx of black quartz, judge my vow.
___
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: Logging of installed packages.

2012-11-20 Thread Volodymyr Kostyrko


20.11.2012 11:24, Peter Ankerstål wrote:

In the light of this: http://www.freebsd.org/news/2012-compromise.html

It would be nice if there was default way to log all package installations from 
both pkgs and ports.
Is there? If not, what would you recommend for doing this?


pkgng logs all installations/deinstallations. It's already default for 
CURRENT and you can switch to using it for other branches.


--
Sphinx of black quartz, judge my vow.
___
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

buildworld fails on recent stable


Hi all.

When CLANG_IS_CC build fails at sys/boot/i386/boot2:

ld -static -N --gc-sections -nostdlib -m elf_i386_fbsd -Ttext 0x2000 -o 
boot2.out /usr/obj/usr/src/sys/boot/i386/boot2/../btx/lib/crt0.o boot2.o 
sio.o

objcopy -S -O binary boot2.out boot2.bin
btxld -v -E 0x2000 -f bin -b 
/usr/obj/usr/src/sys/boot/i386/boot2/../btx/btx/btx -l boot2.ldr  -o 
boot2.ld -P 1 boot2.bin

kernel: ver=1.02 size=690 load=9000 entry=9010 map=16M pgctl=1:1
client: fmt=bin size=1575 text=0 data=0 bss=0 entry=0
output: fmt=bin size=1e05 text=200 data=1c05 org=0 entry=0
-5 bytes available
*** [boot2] Error code 1

The relevant output before r242562 looked like:

ld -static -N --gc-sections -nostdlib -m elf_i386_fbsd -Ttext 0x2000 -o 
boot2.out /usr/obj/usr/src/sys/boot/i386/boot2/../btx/lib/crt0.o boot2.o 
sio.o

objcopy -S -O binary boot2.out boot2.bin
btxld -v -E 0x2000 -f bin -b 
/usr/obj/usr/src/sys/boot/i386/boot2/../btx/btx/btx -l boot2.ldr  -o 
boot2.ld -P 1 boot2.bin

kernel: ver=1.02 size=690 load=9000 entry=9010 map=16M pgctl=1:1
client: fmt=bin size=1569 text=0 data=0 bss=0 entry=0
output: fmt=bin size=1df9 text=200 data=1bf9 org=0 entry=0
7 bytes available
dd if=boot2.ld of=boot2 obs=7680 conv=osync

--
Sphinx of black quartz, judge my vow.
___
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: buildworld fails on recent stable


05.11.2012 17:42, Andriy Gapon wrote:

on 05/11/2012 16:51 Dimitry Andric said the following:

On 2012-11-05 12:41, Volodymyr Kostyrko wrote:

When CLANG_IS_CC build fails at sys/boot/i386/boot2:

ld -static -N --gc-sections -nostdlib -m elf_i386_fbsd -Ttext 0x2000 -o
boot2.out /usr/obj/usr/src/sys/boot/i386/boot2/../btx/lib/crt0.o boot2.o
sio.o
objcopy -S -O binary boot2.out boot2.bin
btxld -v -E 0x2000 -f bin -b
/usr/obj/usr/src/sys/boot/i386/boot2/../btx/btx/btx -l boot2.ldr  -o
boot2.ld -P 1 boot2.bin
kernel: ver=1.02 size=690 load=9000 entry=9010 map=16M pgctl=1:1
client: fmt=bin size=1575 text=0 data=0 bss=0 entry=0
output: fmt=bin size=1e05 text=200 data=1c05 org=0 entry=0
-5 bytes available


This seems to be caused by r242562, which is an MFC of r241301.  The
code changes are quite trivial though.  Strangely enough, on head it
does build successfully, but head is using a newer version of clang, so
that may explain the difference.  Investigating...


I suspect that some earlier space-saving commit was not MFC-ed to stable/9.



This can be easily checked:

# cd /usr/src/sys/boot/i386/boot2/
# make clean
# svn up
Updated to revision 242616.
# setenv PATH /usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin
# unsetenv CCACHE_PATH
# /usr/local/bin/clang -v
clang version 3.2 (trunk)
Target: amd64-portbld-freebsd9.1
Thread model: posix
# echo CC=/usr/local/bin/clang  /etc/src.conf
# echo CXX=/usr/local/bin/clang++  /etc/src.conf
# make

ld -static -N --gc-sections -nostdlib -m elf_i386_fbsd -Ttext 0x2000 -o 
boot2.out /usr/obj/usr/src/sys/boot/i386/boot2/../btx/lib/crt0.o boot2.o 
sio.o

objcopy -S -O binary boot2.out boot2.bin
btxld -v -E 0x2000 -f bin -b 
/usr/obj/usr/src/sys/boot/i386/boot2/../btx/btx/btx -l boot2.ldr  -o 
boot2.ld -P 1 boot2.bin

kernel: ver=1.02 size=690 load=9000 entry=9010 map=16M pgctl=1:1
client: fmt=bin size=154d text=0 data=0 bss=0 entry=0
output: fmt=bin size=1ddd text=200 data=1bdd org=0 entry=0
35 bytes available
dd if=boot2.ld of=boot2 obs=7680 conv=osync
14+1 records in
1+0 records out
7680 bytes transferred in 0.99 secs (77433305 bytes/sec)

I bet on clang 3.2.

--
Sphinx of black quartz, judge my vow.
___
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: buildworld fails on recent stable


05.11.2012 18:00, Andriy Gapon wrote:

I bet on clang 3.2.


Thank you for checking.

So how do we proceed from here?
I could just revert the MFC-es, but the functionality could be desirable to some
users.  Are there any alternatives?


Dunno, clang 3.2 is still pending so I guess it wouldn't be MFC'ed until 
December 16th? Being a minor issue it can make some people shrug from 
using clang for testing stable...


I think reverting the working change is not good as far as building 
world with clang is still not the true and default way.


--
Sphinx of black quartz, judge my vow.
___
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: wine, gcc and clang with CPUTYPE


05.11.2012 17:47, Howard Goldstein wrote:

Volodymyr, your initial email in this thread regarding clang's being
stuck on native code generation allowed me to wrap up a week of fighting
with an updated production box with a weird wine issue - thank you!

FWIW I filed a pr on this.  You've done much more useful legwork towards
a solution of the problem and I wanted to point it out to everyone how
helpful it was.  THANK YOU!

The pr is at http://www.freebsd.org/cgi/query-pr.cgi?pr=173337


I'm still unsure of my results though. I can't reproduce gcc core dump 
anymore. I think it's shadowed by some later fix.


As for wine my results are not complete:

1. Building a world with CPUTYPE=athlon-tbird on a world built with 
CPUTYPE=k6 makes wine work.
2. Building a world with CPUTYPE=athlon-tbird on a world built with 
CPUTYPE=athlon-xp makes wine fail.


--
Sphinx of black quartz, judge my vow.
___
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

wine, gcc and clang with CPUTYPE


Hi all.

I just have taken some time to inspect CPUTYPE support for clang. It 
seems to me that clang generates incorrect code in some cases.


The first failure point I discovered was inability to build gcc from 
sources or compile something with gcc. Code produced by gcc seem to fail 
whether this was gcc compiled from bootstrap or anything else:


http://lists.freebsd.org/pipermail/freebsd-multimedia/2012-October/013469.html

I started testing by commenting out CPUTYPE in make.conf. After first 
rebuild I also updated the ports and installed new version of 
wine-devel. And to my surprise it works like a charm. Rolling back to 
the world built with CPUTYPE=native makes wine break again.


To my surprise CPUTYPE was not the cause of wine failure per se. Wine 
continues to work for k6, k6-3, athlon and athlon-tbird. But it 
completely fails when the world was built with athlon-4 and athlon-xp.


Trying to recompile gcc I also found that everything works and yet again 
up to the athlon-tbird.


My conclusion is: clang incorrectly produces code within one of core 
libraries (I haven't tested which one yet, but I suspect libgcc_s.so) 
when optimizing for athlon-4 or athlon-xp.


--
Sphinx of black quartz, judge my vow.
___
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: wine, gcc and clang with CPUTYPE


24.10.2012 11:12, Konstantin Belousov wrote:


My conclusion is: clang incorrectly produces code within one of core
libraries (I haven't tested which one yet, but I suspect libgcc_s.so)
when optimizing for athlon-4 or athlon-xp.


I am not versed in the AMD marketing monikers. I guess that athlon-{4,xp}
turns on SSE and might be SSE2, while previous selections turn it off.
Can you confirm/deny this ?


As http://en.wikipedia.org/wiki/Athlon states athlon-xp was first to 
include SSE support. I don't know whether athlon-xp actually differs 
from athlon-4. None of them include SSE2 support.



BTW, did you tested on i386 or amd64 ?


i386

--
Sphinx of black quartz, judge my vow.
___
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: wine, gcc and clang with CPUTYPE

24.10.2012 13:05, Dimitry Andric wrote:

I just have taken some time to inspect CPUTYPE support for clang. It
seems to me that clang generates incorrect code in some cases.

The first failure point I discovered was inability to build gcc from
sources or compile something with gcc. Code produced by gcc seem to fail
whether this was gcc compiled from bootstrap or anything else:

http://lists.freebsd.org/pipermail/freebsd-multimedia/2012-October/013469.html

Can you attempt to figure out what the illegal instruction was, in that
case?

How can I do that? I'm not very familiar with gdb.

I started testing by commenting out CPUTYPE in make.conf. After first
rebuild I also updated the ports and installed new version of
wine-devel. And to my surprise it works like a charm. Rolling back to
the world built with CPUTYPE=native makes wine break again.

To my surprise CPUTYPE was not the cause of wine failure per se. Wine
continues to work for k6, k6-3, athlon and athlon-tbird. But it
completely fails when the world was built with athlon-4 and athlon-xp.

Trying to recompile gcc I also found that everything works and yet again
up to the athlon-tbird.

My conclusion is: clang incorrectly produces code within one of core
libraries (I haven't tested which one yet, but I suspect libgcc_s.so)
when optimizing for athlon-4 or athlon-xp.

On the problematic athlons, can you please post the exact CPUIDs from
dmesg? If you have WITH_CLANG_EXTRAS enabled, please also post the
output of opt -version.

Oct 24 01:47:20 limbo kernel: CPU: AMD Athlon(tm) XP 2500+ (1833.95-MHz
686-class CPU)
Oct 24 01:47:20 limbo kernel: Origin = AuthenticAMD Id = 0x6a0
Family = 0x6 Model = 0xa Stepping = 0
Oct 24 01:47:20 limbo kernel:
Features=0x383fbffFPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,MMX,FXSR,SSE
Oct 24 01:47:20 limbo kernel: AMD
Features=0xc0400800SYSCALL,MMX+,3DNow!+,3DNow!

I will rebuild WITH_CLANG_EXTRAS.

--
Sphinx of black quartz, judge my vow.
___
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: wine, gcc and clang with CPUTYPE