Re: Smoke [blead] v5.27.8-546-g8d8484540d FAIL(XM) os/390 25.00 (2964/)

[Karl Williamson]

On 03/21/2018 07:09 AM, Yaroslav Kuzmin wrote:
> Smoke log at available https://drive.google.com/file/d/1XHUAwq74zUMfS3Ps3WnMAvpSagcXFWxJ
>
>
> Automated smoke report for branch blead 5.27.10 patch 8d8484540d380c09fb40616135be00825189f578 v5.27.8-546-g8d8484540d
> RS12: 2964 (2964/)
> on os/390 - 25.00
> using c99 version
> smoketime 2 hours 53 minutes (average 1 hour 26 minutes)
>
> Summary: FAIL(XM)
>

The many failures in this report indicates that something fundamental
got broken. So to start finding it, I think we need to go back to the
basics.

Try this one-liner, and send me the results:

./perl -le 'my $a = "\x{587}"; my $b = uc($a); use bytes; print map {
sprintf " %x", ord } split "", $a; print map { sprintf " %x", ord }
split "", $b'

[Karl Williamson]

On 03/28/2018 05:44 AM, Yaroslav Kuzmin wrote:
> Output :
>
> (16:42) RS12 : PDKUZM | ~/ussport/perl/perl-build
> :> ./perl -V
> Summary of my perl5 (revision 5 version 27 subversion 11) configuration:
> Snapshot of: 3a7f897ad3c330a4e99ba46255bb72e89c43cd08
> Platform:
> osname=os390
> osvers=25.00
> archname=os390
> uname='os390 rs12 25.00 04 2964 '
> config_args='-Dusedevel -Dusedl -d -s'
> hint=recommended
> useposix=true
> d_sigaction=define
> useithreads=undef
> usemultiplicity=undef
> use64bitint=undef
> use64bitall=undef
> uselongdouble=undef
> usemymalloc=n
> default_inc_excludes_dot=define
> bincompat5005=undef
> Compiler:
> cc='c99'
> ccflags ='-qlanglvl=extended:extc89:extc99 -qlongname -qxplink -qdll -qfloat=ieee -qhaltonmsg=3296:4108 -DMAXSIG=39 -DOEMVS -D_OE_SOCKETS -D_XOPEN_SOURCE_EXTENDED -D_ALL_SOURCE -DYYDYNAMIC
> -D_POSIX_SOURCE=1 -D_SHR_ENVIRON -DPERL_EXTERNAL_GLOB -qexportall -qdll -qxplink'
> optimize=' '
> cppflags='-DMAXSIG=39 -DOEMVS -D_OE_SOCKETS -D_XOPEN_SOURCE_EXTENDED -D_ALL_SOURCE -DYYDYNAMIC -D_POSIX_SOURCE=1 -D_SHR_ENVIRON -DPERL_EXTERNAL_GLOB'
> ccversion=''
> gccversion=''
> gccosandvers=''
> intsize=4
> longsize=4
> ptrsize=4
> doublesize=8
> byteorder=4321
> doublekind=4
> d_longlong=define
> longlongsize=8
> d_longdbl=define
> longdblsize=16
> longdblkind=2
> ivtype='long'
> ivsize=4
> nvtype='double'
> nvsize=8
> Off_t='off_t'
> lseeksize=4
> alignbytes=8
> prototype=define
> Linker and Libraries:
> ld='c99'
> ldflags ='-qxplink -qdll'
> libpth=/lib /usr/lib
> libs=-lm -lc
> perllibs=-lm -lc
> libc=
> so=so
> useshrplib=true
> libperl=libperl.so
> gnulibc_version=''
> Dynamic Linking:
> dlsrc=dl_dllload.xs
> dlext=so
> d_dlsymun=undef
> ccdlflags='-qxplink -qdll /u/pdkuzm/ussport/perl/perl-build/libperl.x'
> cccdlflags='-c -qexportall -qxplink -qdll'
> lddlflags='-qxplink -qdll /u/pdkuzm/ussport/perl/perl-build/libperl.x'
>
>
> Characteristics of this binary (from libperl):
> Compile-time options:
> HAS_TIMES
> PERL_COPY_ON_WRITE
> PERL_DONT_CREATE_GVSV
> PERL_EXTERNAL_GLOB
> PERL_MALLOC_WRAP
> PERL_OP_PARENT
> PERL_PRESERVE_IVUV
> PERL_USE_DEVEL
> PERLIO_LAYERS
> USE_LARGE_FILES
> USE_LOCALE
> USE_LOCALE_COLLATE
> USE_LOCALE_CTYPE
> USE_LOCALE_NUMERIC
> USE_LOCALE_TIME
> USE_PERL_ATOF
> USE_PERLIO
> Locally applied patches:
> SMOKE3a7f897ad3c330a4e99ba46255bb72e89c43cd08
> Built under os390
> Compiled at Mar 28 2018 12:12:09
> %ENV:
> PERL5LIB="/u/pdkuzm/ussport/perl/perl-build/lib"
> @INC:
> /u/pdkuzm/ussport/perl/perl-build/lib
> /usr/local/lib/perl5/site_perl/5.27.11/os390
> /usr/local/lib/perl5/site_perl/5.27.11
> /usr/local/lib/perl5/5.27.11/os390
> /usr/local/lib/perl5/5.27.11
> (16:43) RS12 : PDKUZM | ~/ussport/perl/perl-build
> :> ./perl -le 'my $a = "\x{587}"; my $b = uc($a); use bytes; print map { ^Jsprintf " %x", ord } split "", $a; print map { sprintf " %x", ord } ^Jsplit "", $b'
> b8 53 48
> b8 53 48

The first sequence is the correct UTF-EBCDIC for \x{587}, but it has an
uppercase value that is different.

I think the first thing is for you to package up the directory
lib/unicore and send a copy to me.

If you have the machine available, you might want to try testing
https://perl5.git.perl.org/perl.git/shortlog/refs/heads/smoke-me/khw-core

which has a somewhat different casing implementation


> В Вс., 25/03/2018 в 16:05 -0600, Karl Williamson пишет:

>> On 03/21/2018 07:09 AM, Yaroslav Kuzmin wrote:
>>>

>>> Smoke log at available https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdrive.google.com%2Ffile%2Fd%2F1XHUAwq74zUMfS3Ps3WnMAvpSagcXFWxJ&data=02%7C01%7Cykuzmin%40rocketsoftware.com%
>>> 7C797e757095254fbcdbc408d5929c90fa%7C79544c1eed224879a082b67a9a672aae%7C0%7C1%7C636576123523851610&sdata=n0upDMJ9%2B5AW5XqESiqK2lWUP3FXk%2BcsVa0exwoztSE%3D&reserved=0

>>>
>>>
>>> Automated smoke report for branch blead 5.27.10 patch 8d8484540d380c09fb40616135be00825189f578 v5.27.8-546-g8d8484540d
>>> RS12: 2964 (2964/)
>>> on os/390 - 25.00
>>> using c99 version
>>> smoketime 2 hours 53 minutes (average 1 hour 26 minutes)
>>>
>>> Summary: FAIL(XM)
>>>
>> The many failures in this report indicates that something fundamental
>> got broken. So to start finding it, I think we need to go back to the
>> basics.
>>
>> Try this one-liner, and send me the results:
>>
>> ./perl -le 'my $a = "\x{587}"; my $b = uc($a); use bytes; print map {
>> sprintf " %x", ord } split "", $a; print map { sprintf " %x", ord }
>> split "", $b'

> ================================
> Rocket Software, Inc. and subsidiaries ■ 77 Fourth Avenue, Waltham MA 02451 ■ Main Office Toll Free Number: +1 855.577.4323
> Contact Customer Support: https://my.rocketsoftware.com/RocketCommunity/RCEmailSupport
> Unsubscribe from Marketing Messages/Manage Your Subscription Preferences - http://www.rocketsoftware.com/manage-your-email-preferences
> Privacy Policy - http://www.rocketsoftware.com/company/legal/privacy-policy
> ================================
>
> This communication and any attachments may contain confidential information of Rocket Software, Inc. All unauthorized use, disclosure or distribution is prohibited. If you are not the intended recipient, please notify Rocket Software immediately and destroy all copies of this communication. Thank you.
>

[Karl Williamson]

For example, a function that is segfaulting in your report is completely
removed.

I also looked through the .25 million lines of code that got changed
between successful smokes. The likely perpetrators are changes to
multiconcat and/or regular expression pattern matching.