nsz libc-test

nsz libc-test

[stephen Turner]

[-- Attachment #1: Type: text/plain, Size: 227 bytes --]

hey i was trying to clone the libc test files from git://
nsz.repo.hu:45100/repo/libc-test but im getting a timeout. is that the
right address? I was running "git clone git://
nsz.repo.hu:45100/repo/libc-test"

thanks,
stephen

[-- Attachment #2: Type: text/html, Size: 400 bytes --]

Re: nsz libc-test

[Szabolcs Nagy]

* stephen Turner <stephen.n.turner@gmail.com> [2015-03-25 16:37:10 -0400]:
> hey i was trying to clone the libc test files from git://
> nsz.repo.hu:45100/repo/libc-test but im getting a timeout. is that the
> right address? I was running "git clone git://

works from here, but that's not a very reliable server,
so the project is mirrored at repo.or.cz:
http://repo.or.cz/w/libc-test.git
there are various git clone options there

Re: nsz libc-test

[stephen Turner]

[-- Attachment #1: Type: text/plain, Size: 738 bytes --]

Let me know if this didnt go through. I got a mailer error.
On Mar 25, 2015 7:07 PM, "stephen Turner" <stephen.n.turner@gmail.com>
wrote:

> ok i found a tarball i could grab without needing git. I unpacked and ran
> the build but it looks like everything fails. I found a REPORT file in the
> src directory i have attached as well as the output from make. Have i run
> your test correctly?
>
> On Wed, Mar 25, 2015 at 4:37 PM, stephen Turner <
> stephen.n.turner@gmail.com> wrote:
>
>> hey i was trying to clone the libc test files from git://
>> nsz.repo.hu:45100/repo/libc-test but im getting a timeout. is that the
>> right address? I was running "git clone git://
>> nsz.repo.hu:45100/repo/libc-test"
>>
>> thanks,
>> stephen
>>
>
>

[-- Attachment #2: Type: text/html, Size: 1453 bytes --]

Re: nsz libc-test

[stephen Turner]

[-- Attachment #1.1: Type: text/plain, Size: 868 bytes --]

>
> On Mar 25, 2015 7:07 PM, "stephen Turner" <stephen.n.turner@gmail.com>
> wrote:
>
>> ok i found a tarball i could grab without needing git. I unpacked and ran
>> the build but it looks like everything fails. I found a REPORT file in the
>> src directory i have attached as well as the output from make. Have i run
>> your test correctly?
>>
>
I ran these tests on glibc and produced a lot of errors as well so im
assuming im running it correctly and this is standard output.

Rich, this output is from my pcc + musl system that has been producing
corrupted libc.so files. if anything stands out to anyone i would be happy
to hear it so i can look into it further.

there is the make.log2 which is the make output from "CC="pcc -g" make" and
of course the REPORT from the src folder.

thanks for all the support and i appreciate any input you have about this
test.

[-- Attachment #1.2: Type: text/html, Size: 1754 bytes --]

[-- Attachment #2: logs.tar.bz2 --]
[-- Type: application/x-bzip2, Size: 17507 bytes --]

Re: Re: nsz libc-test

[Rich Felker]

On Wed, Mar 25, 2015 at 08:58:12PM -0400, stephen Turner wrote:
> >
> > On Mar 25, 2015 7:07 PM, "stephen Turner" <stephen.n.turner@gmail.com>
> > wrote:
> >
> >> ok i found a tarball i could grab without needing git. I unpacked and ran
> >> the build but it looks like everything fails. I found a REPORT file in the
> >> src directory i have attached as well as the output from make. Have i run
> >> your test correctly?
> >>
> >
> I ran these tests on glibc and produced a lot of errors as well so im
> assuming im running it correctly and this is standard output.
> 
> Rich, this output is from my pcc + musl system that has been producing
> corrupted libc.so files. if anything stands out to anyone i would be happy
> to hear it so i can look into it further.
> 
> there is the make.log2 which is the make output from "CC="pcc -g" make" and
> of course the REPORT from the src folder.
> 
> thanks for all the support and i appreciate any input you have about this
> test.

You're not going to get any meaningful results without removing
-pedantic-errors from the config.mak. pcc is erroring out due to the
unrecognized option without building any tests.

Rich

Re: nsz libc-test

[DELOGET, Emmanuel]

Hello, 

Le mercredi 25 mars 2015 à 23:59 +0100, Szabolcs Nagy a écrit :
> * stephen Turner <stephen.n.turner@gmail.com> [2015-03-25 16:37:10 -0400]:
> > hey i was trying to clone the libc test files from git://
> > nsz.repo.hu:45100/repo/libc-test but im getting a timeout. is that the
> > right address? I was running "git clone git://
> 
> works from here, but that's not a very reliable server,
> so the project is mirrored at repo.or.cz:
> http://repo.or.cz/w/libc-test.git
> there are various git clone options there

I'm interested as well, and having tests like these for a libc is a good
thing. Would you mind setting up a github repo ? (or any other similar
solution) That would enable some of us to propose some evolutions or
corrections here and there. 

BR, 

-- Emmanuel Deloget

Re: Re: nsz libc-test

[stephen Turner]

[-- Attachment #1.1: Type: text/plain, Size: 557 bytes --]

>
> You're not going to get any meaningful results without removing
> -pedantic-errors from the config.mak. pcc is erroring out due to the
> unrecognized option without building any tests.
>
> Rich
>

Thanks Rich, I found and removed the setting in the makefile and re-ran the
test.

I have been using some shell code to identify pcc and make adjustments, its
super crude but here it is. Perhaps something along these lines could be
added to the nsz libc test.

if command -v pcc >/dev/null 2>&1; then
export CC=pcc
else
export CFLAGS="$CFLAGS --static"
fi

[-- Attachment #1.2: Type: text/html, Size: 1210 bytes --]

[-- Attachment #2: logs.tar.bz2 --]
[-- Type: application/x-bzip2, Size: 17090 bytes --]

Re: Re: nsz libc-test

[stephen Turner]

[-- Attachment #1: Type: text/plain, Size: 853 bytes --]

so ideally with a musl system what kind of report output can i expect? the
report i have is from a broken system, would someone happen to have a
report from a working musl system?

On Thu, Mar 26, 2015 at 11:28 AM, stephen Turner <stephen.n.turner@gmail.com
> wrote:

> You're not going to get any meaningful results without removing
>> -pedantic-errors from the config.mak. pcc is erroring out due to the
>> unrecognized option without building any tests.
>>
>> Rich
>>
>
> Thanks Rich, I found and removed the setting in the makefile and re-ran
> the test.
>
> I have been using some shell code to identify pcc and make adjustments,
> its super crude but here it is. Perhaps something along these lines could
> be added to the nsz libc test.
>
> if command -v pcc >/dev/null 2>&1; then
> export CC=pcc
> else
> export CFLAGS="$CFLAGS --static"
> fi
>

[-- Attachment #2: Type: text/html, Size: 1800 bytes --]

Re: Re: nsz libc-test

[Rich Felker]

[-- Attachment #1: Type: text/plain, Size: 407 bytes --]

On Thu, Mar 26, 2015 at 04:54:03PM -0400, stephen Turner wrote:
> so ideally with a musl system what kind of report output can i expect? the
> report i have is from a broken system, would someone happen to have a
> report from a working musl system?

Attached is a report of what you should expect to fail with current
git master. A few more tests (the recently added ones) will fail with
musl 1.1.7.

Rich

[-- Attachment #2: REPORT --]
[-- Type: text/plain, Size: 4086 bytes --]

src/api/fcntl.c: In function 'f':
src/api/fcntl.c:30:1: error: 'O_TTY_INIT' undeclared (first use in this function)
src/api/fcntl.c:30:1: note: each undeclared identifier is reported only once for each function it appears in
src/common/runtest.c:29: ./src/api/main.exe exec failed: No such file or directory
FAIL ./src/api/main.exe [status 1]
cc: error: src/api/fcntl.o: No such file or directory
cc: error: src/api/unistd.o: No such file or directory
src/api/unistd.c: In function 'f':
src/api/unistd.c:87:1: error: '_CS_POSIX_V7_THREADS_CFLAGS' undeclared (first use in this function)
src/api/unistd.c:87:1: note: each undeclared identifier is reported only once for each function it appears in
src/api/unistd.c:88:1: error: '_CS_POSIX_V7_THREADS_LDFLAGS' undeclared (first use in this function)
src/api/unistd.c:90:1: error: '_CS_V7_ENV' undeclared (first use in this function)
src/api/unistd.c:117:1: error: '_PC_TIMESTAMP_RESOLUTION' undeclared (first use in this function)
src/api/unistd.c:238:1: error: '_SC_XOPEN_UUCP' undeclared (first use in this function)
src/math/sanity/expm1l.h:2: RN expm1l(0x1.161868e18bc67782p+2) want 0x1.30712ed238c064a2p+6 got 0x1.30712ed238c064a6p+6 ulperr 2.355 = 0x1.ffffffp+0 + 0x1.6b4c44p-2
src/math/sanity/expm1l.h:5: RN expm1l(0x1.288bbb0d6a1e5bdap+3) want 0x1.4ab477496e07b24ap+13 got 0x1.4ab477496e07b24ep+13 ulperr 1.583 = 0x1.0000004p+1 + -0x1.aa91d2p-2
FAIL ./src/math/expm1l.exe [status 1]
src/math/j0.c: known to be broken near zeros
FAIL ./src/math/j0.exe [status 1]
src/math/sanity/jnf.h:1: RN jnf(-2, -0x1.0223ap+3) want -0x1.863726p-4 got -0x1.86372ap-4, ulperr -2.009 = -0x1.ffffff6p+0 + -0x1.28885p-7
FAIL ./src/math/jnf.exe [status 1]
src/math/special/lgamma.h:145: RN lgamma(-0x1.4p+1) want -0x1.ccbf9f5ed0f16p-5,-1 got -0x1.ccbf9f5ed0f1bp-5,-1 ulperr -5.465 = -0x1.3fffffcp+2 + -0x1.dc4f24p-2
FAIL ./src/math/lgamma.exe [status 1]
src/math/sanity/sinhl.h:1: RN sinhl(-0x1.02239f3c6a8f13dep+3) want -0x1.8e6152d2b112dcb8p+10 got -0x1.8e6152d2b112dcc2p+10 ulperr -4.567 = -0x1.4p+2 + 0x1.bb71ep-2
src/math/sanity/sinhl.h:2: RN sinhl(0x1.161868e18bc67782p+2) want 0x1.3463e73bcdaac0bep+5 got 0x1.3463e73bcdaac0c4p+5 ulperr 2.856 = 0x1.7fffff6p+1 + -0x1.25fab6p-3
src/math/sanity/sinhl.h:4: RN sinhl(-0x1.a206f0a19dcc3948p+2) want -0x1.5745bb86e8aecf3ep+8 got -0x1.5745bb86e8aecf44p+8 ulperr -3.322 = -0x1.8000004p+1 + -0x1.49a5aep-2
FAIL ./src/math/sinhl.exe [status 1]
src/math/special/tgamma.h:47: RN tgamma(0x1p-53) want 0x1.fffffffffffffp+52 got 0x1.ffffffffffffdp+52 ulperr -2.423 = -0x1p+1 + -0x1.b0ee6p-2
FAIL ./src/math/tgamma.exe [status 1]
src/math/y0.c: known to be broken near zeros
FAIL ./src/math/y0.exe [status 1]
src/math/y0f.c: known to be broken near zeros
FAIL ./src/math/y0f.exe [status 1]
src/musl/pleval.c: In function 'main':
src/musl/pleval.c:52:2: warning: suggest parentheses around comparison in operand of '==' [-Wparentheses]
src/musl/pleval.c:53:2: warning: comparisons like 'X<=Y<=Z' do not have their mathematical meaning [-Wparentheses]
src/musl/pleval.c:55:2: warning: suggest parentheses around '&&' within '||' [-Wparentheses]
src/regression/malloc-brk-fail.c:41: malloc(10000) failed (eventhough 64k is available to mmap): Out of memory
FAIL ./src/regression/malloc-brk-fail-static.exe [status 1]
src/regression/regex-backref-0.c:22: regexec(/a\0/ ~ "a0") returned 1 (No match), wanted 0
FAIL ./src/regression/regex-backref-0-static.exe [status 1]
src/regression/regex-backref-0.c:22: regexec(/a\0/ ~ "a0") returned 1 (No match), wanted 0
FAIL ./src/regression/regex-backref-0.exe [status 1]
src/regression/strverscmp.c:14: strverscmp("000", "00") < 0 failed
src/regression/strverscmp.c:15: strverscmp("00", "000") > 0 failed
src/regression/strverscmp.c:20: strverscmp("09", "0") < 0 failed
FAIL ./src/regression/strverscmp-static.exe [status 1]
src/regression/strverscmp.c:14: strverscmp("000", "00") < 0 failed
src/regression/strverscmp.c:15: strverscmp("00", "000") > 0 failed
src/regression/strverscmp.c:20: strverscmp("09", "0") < 0 failed
FAIL ./src/regression/strverscmp.exe [status 1]

Re: Re: nsz libc-test

[stephen Turner]

[-- Attachment #1: Type: text/plain, Size: 602 bytes --]

> Attached is a report of what you should expect to fail with current
> git master. A few more tests (the recently added ones) will fail with
> musl 1.1.7.
>
> Rich

Thanks. Theres a noticable diference here. Hopefully after a diff of the
two I can easily pick out which tests fail and focus on them.

Each test actually compiles into a file right? And what is the best thing I
can do to identify the compilers corruption, is gdb the best and just keep
running the disassemble? I noticed strace+ which apparently was merged
mainline as of version 4.9 as -k option and it looks like almost code
output.

[-- Attachment #2: Type: text/html, Size: 695 bytes --]

Re: nsz libc-test

[Szabolcs Nagy]

* DELOGET, Emmanuel <emmanuel.deloget@sfr.com> [2015-03-26 13:15:27 +0000]:
> Le mercredi 25 mars 2015 à 23:59 +0100, Szabolcs Nagy a écrit :
> > * stephen Turner <stephen.n.turner@gmail.com> [2015-03-25 16:37:10 -0400]:
> > > hey i was trying to clone the libc test files from git://
> > > nsz.repo.hu:45100/repo/libc-test but im getting a timeout. is that the
> > > right address? I was running "git clone git://
> > 
> > works from here, but that's not a very reliable server,
> > so the project is mirrored at repo.or.cz:
> > http://repo.or.cz/w/libc-test.git
> > there are various git clone options there
> 
> I'm interested as well, and having tests like these for a libc is a good
> thing. Would you mind setting up a github repo ? (or any other similar
> solution) That would enable some of us to propose some evolutions or
> corrections here and there. 
> 

you are welcome to mirror the repo on github

you can propose changes on this mailing list i think

(the current repo was supposed to be a temporary place,
it is a server under my control, i don't use github or any
other us based hosting service since one of my projects got
deleted from sourceforge by a dmca takedown request)