[ANNOUNCE] util-linux v2.30-rc2

[ANNOUNCE] util-linux v2.30-rc2

[Karel Zak]

The util-linux release v2.30-rc2 is available at

  http://www.kernel.org/pub/linux/utils/util-linux/v2.30/

Feedback and bug reports, as always, are welcomed.

    Karel

-- 
 Karel Zak  <kzak@redhat.com>
 http://karelzak.blogspot.com

Re: [ANNOUNCE] util-linux v2.30-rc2

[Assaf Gordon]

Hello

> On May 23, 2017, at 06:50, Karel Zak <kzak@redhat.com> wrote:
>=20
> The util-linux release v2.30-rc2 is available at
>=20
>  http://www.kernel.org/pub/linux/utils/util-linux/v2.30/

Few results from various system:

1.
On several systems which do not have ncurses development files
installed, the "configure" script does not detect it,
and completion later fails.
Example with Debian 8.1:

=3D=3D=3D
  CC       disk-utils/cfdisk-cfdisk.o
disk-utils/cfdisk.c: In function =E2=80=98ui_get_string=E2=80=99:
disk-utils/cfdisk.c:1757:3: warning: implicit declaration of function =
=E2=80=98get_wch=E2=80=99 [-Wimplicit-function-declaration]
   if (get_wch(&c) =3D=3D ERR) {
   ^
disk-utils/cfdisk.c:1757:3: warning: nested extern declaration of =
=E2=80=98get_wch=E2=80=99 [-Wnested-externs]
  CCLD     cfdisk
/usr/bin/ld: cannot find -lncursesw
collect2: error: ld returned 1 exit status
make[2]: *** [cfdisk] Error 1
=3D=3D=3D


2.
On ppc64 machine running Fedora 20, kernel 3.17.2-200.fc20.ppc64p7, gcc =
4.8.3,
the following tests fail:
   misc: flock: [04] exclusive                         ... FAILED =
(misc/flock-exclusive)
   misc: flock                                         ... FAILED (1 =
from 6 sub-tests)
Log here:
   https://pretest.housegordon.org/g/4806/logs/make-check.log?inlined=3D1
Build details here:
   https://pretest.housegordon.org/d/4806


3.
On a X86-64 QEMU virtual machine running Fedora 25, Fedora 24, the =
following test failed:
   fincore: count file contents in core    ... FAILED (fincore/count)
Log here:
   https://pretest.housegordon.org/g/4824/logs/make-check.log?inlined=3D1
Build details here:
   https://pretest.housegordon.org/d/4824



more exotic cases:

4.
Building with TinyCC compiler (tcc) fails with:
=3D=3D=3D
CC       sys-utils/hwclock-cmos.o
sys-utils/hwclock-cmos.c:139: warning: implicit declaration of function =
'outb'
sys-utils/hwclock-cmos.c:140: warning: implicit declaration of function =
'inb'
CC       sys-utils/hwclock-rtc.o
CCLD     hwclock
tcc: error: undefined symbol 'outb'
tcc: error: undefined symbol 'inb'
make[2]: *** [hwclock] Error 1
=3D=3D=3D
(this is not a problem in util-linux, but perhaps can be detected and =
avoided?)



5.
On Alpine-Linux 3.5.2 (musl-libc instead of glibc), compilation fails =
with:
=3D=3D=3D
CC       lib/libcommon_la-loopdev.lo
lib/loopdev.c:35:31: fatal error: linux/posix_types.h: No such file or =
directory
#include <linux/posix_types.h>
^
compilation terminated.
make[2]: *** [Makefile:8466: lib/libcommon_la-loopdev.lo] Error 1
=3D=3D=3D
The problem is that the alpine-linux package "linux-headers" [1]
is not installed. It might be helpful to detect it in 'configure'.
[1] https://pkgs.alpinelinux.org/package/v3.5/main/x86_64/linux-headers


Still on Alpine-Linux, after installing "linux-headers" package,
one of the binaries in "make check" fails to build due to another =
missing
header:
=3D=3D=3D
CC       misc-utils/test_uuidd-test_uuidd.o
misc-utils/test_uuidd.c:26:19: fatal error: error.h: No such file or =
directory
#include <error.h>
^
compilation terminated.
make[3]: *** [Makefile:11056: misc-utils/test_uuidd-test_uuidd.o] Error =
1
=3D=3D=3D
I guess "error.h" is glibc's header for error(3) - but alpine uses =
musl-libc
and does not have it.





6.
Building on GNU Hurd 0.7 (not linux), the following test fail:
   fdisk: invalid input tests            ... FAILED (fdisk/oddinput)
   ipcs: headers                        ... FAILED (ipcs/headers)
   mountpoint                  ... FAILED (libmount/utils-mountpoint)
   mountpoint-subdir           ... FAILED =
(libmount/utils-mountpoint-subdir)
   misc: flock                          ...ps: unrecognized option =
'--ppid'
     Try `ps --help' or `ps --usage' for more information.
     ps: unrecognized option '--ppid'
   FAILED (misc/flock)

Log here:
   https://pretest.housegordon.org/g/4834/logs/make-check.log?inlined=3D1
Build details here:
   https://pretest.housegordon.org/d/4834


7.
For the truly curious, more build reports
from non-linux systems (e.g. OpenBSD, FreeBSD, SunOS, Darwin):
  https://pretest.housegordon.org/q/util-linux-2.30-rc2



regards,
 - assaf

Re: [ANNOUNCE] util-linux v2.30-rc2

[Karel Zak]

On Tue, May 30, 2017 at 01:03:53AM -0400, Assaf Gordon wrote:
> Hello
> 
> > On May 23, 2017, at 06:50, Karel Zak <kzak@redhat.com> wrote:
> > 
> > The util-linux release v2.30-rc2 is available at
> > 
> >  http://www.kernel.org/pub/linux/utils/util-linux/v2.30/
> 
> Few results from various system:
> 
> 1.
> On several systems which do not have ncurses development files
> installed, the "configure" script does not detect it,
> and completion later fails.
> Example with Debian 8.1:
> 
> ===
>   CC       disk-utils/cfdisk-cfdisk.o
> disk-utils/cfdisk.c: In function ‘ui_get_string’:
> disk-utils/cfdisk.c:1757:3: warning: implicit declaration of function ‘get_wch’ [-Wimplicit-function-declaration]
>    if (get_wch(&c) == ERR) {
>    ^
> disk-utils/cfdisk.c:1757:3: warning: nested extern declaration of ‘get_wch’ [-Wnested-externs]
>   CCLD     cfdisk
> /usr/bin/ld: cannot find -lncursesw
> collect2: error: ld returned 1 exit status
> make[2]: *** [cfdisk] Error 1
> ===

It would be nice to have configure output. It seems you have
pkg-config (or ncursesw5-config or ncursesw6-config), but no header
files :-)

For example Fedora after "dnf remove ncurses-devel"

./configure
...
checking for NCURSESW... no
checking for ncursesw6-config... no
checking for ncursesw5-config... no
checking for initscr in -lncursesw... no
checking for NCURSES... no
checking for ncurses6-config... no
checking for ncurses5-config... no
checking for initscr in -lncurses... no


> On a X86-64 QEMU virtual machine running Fedora 25, Fedora 24, the following test failed:
>    fincore: count file contents in core    ... FAILED (fincore/count)
> Log here:
>    https://pretest.housegordon.org/g/4824/logs/make-check.log?inlined=1
> Build details here:
>    https://pretest.housegordon.org/d/4824

I have seen this test failed too. It seems like the test is unstable...

> Building with TinyCC compiler (tcc) fails with:
> ===
> CC       sys-utils/hwclock-cmos.o
> sys-utils/hwclock-cmos.c:139: warning: implicit declaration of function 'outb'
> sys-utils/hwclock-cmos.c:140: warning: implicit declaration of function 'inb'
> CC       sys-utils/hwclock-rtc.o
> CCLD     hwclock
> tcc: error: undefined symbol 'outb'
> tcc: error: undefined symbol 'inb'
> make[2]: *** [hwclock] Error 1
> ===
> (this is not a problem in util-linux, but perhaps can be detected and avoided?)

How I can force util-linux build system to accept tcc? 

    CC=tcc ./configure 
    
does not work as well as

    make CC=tcc


> 5.
> On Alpine-Linux 3.5.2 (musl-libc instead of glibc), compilation fails with:
> ===
> CC       lib/libcommon_la-loopdev.lo
> lib/loopdev.c:35:31: fatal error: linux/posix_types.h: No such file or directory
> #include <linux/posix_types.h>
> ^
> compilation terminated.
> make[2]: *** [Makefile:8466: lib/libcommon_la-loopdev.lo] Error 1
> ===
> The problem is that the alpine-linux package "linux-headers" [1]
> is not installed. It might be helpful to detect it in 'configure'.
> [1] https://pkgs.alpinelinux.org/package/v3.5/main/x86_64/linux-headers

Fixed, the header file is unnecessary.


> Still on Alpine-Linux, after installing "linux-headers" package,
> one of the binaries in "make check" fails to build due to another missing
> header:
> ===
> CC       misc-utils/test_uuidd-test_uuidd.o
> misc-utils/test_uuidd.c:26:19: fatal error: error.h: No such file or directory
> #include <error.h>
> ^
> compilation terminated.
> make[3]: *** [Makefile:11056: misc-utils/test_uuidd-test_uuidd.o] Error 1
> ===
> I guess "error.h" is glibc's header for error(3) - but alpine uses musl-libc
> and does not have it.

Fixed, err() and warn() should be enough in all package.

> 6.
> Building on GNU Hurd 0.7 (not linux), the following test fail:
>    fdisk: invalid input tests            ... FAILED (fdisk/oddinput)

need more information, for example tests/ts/diff/fdisk/oddinput

>    ipcs: headers                        ... FAILED (ipcs/headers)

not sure, what returns ipcs when executed on command line?

>    mountpoint                  ... FAILED (libmount/utils-mountpoint)
>    mountpoint-subdir           ... FAILED (libmount/utils-mountpoint-subdir)

should be fixed

>    misc: flock                          ...ps: unrecognized option '--ppid'
>      Try `ps --help' or `ps --usage' for more information.
>      ps: unrecognized option '--ppid'
>    FAILED (misc/flock)

fixed

> For the truly curious, more build reports
> from non-linux systems (e.g. OpenBSD, FreeBSD, SunOS, Darwin):
>   https://pretest.housegordon.org/q/util-linux-2.30-rc2

Thanks!

    Karel


-- 
 Karel Zak  <kzak@redhat.com>
 http://karelzak.blogspot.com

Re: [ANNOUNCE] util-linux v2.30-rc2

[Karel Zak]

On Tue, May 30, 2017 at 01:03:53AM -0400, Assaf Gordon wrote:
> Hello
> 
> > On May 23, 2017, at 06:50, Karel Zak <kzak@redhat.com> wrote:
> > 
> > The util-linux release v2.30-rc2 is available at
> > 
> >  http://www.kernel.org/pub/linux/utils/util-linux/v2.30/
> 
> Few results from various system:
> 
> 1.
> On several systems which do not have ncurses development files
> installed, the "configure" script does not detect it,
> and completion later fails.
> Example with Debian 8.1:

I see where is the problem. We assume that successful ncurses-config
call means that devel stuff is installed. Unfortunately, on Debian 8
this script is installed by ncurses-bin where is nothing like header
files etc. 

IMHO it's packaging bug, but I'll improve our build system to be more
paranoid...

   Karel

-- 
 Karel Zak  <kzak@redhat.com>
 http://karelzak.blogspot.com

Re: [ANNOUNCE] util-linux v2.30-rc2

[Karel Zak]

On Tue, May 30, 2017 at 01:03:53AM -0400, Assaf Gordon wrote:
> Hello
> 
> > On May 23, 2017, at 06:50, Karel Zak <kzak@redhat.com> wrote:
> > 
> > The util-linux release v2.30-rc2 is available at
> > 
> >  http://www.kernel.org/pub/linux/utils/util-linux/v2.30/
> 
> Few results from various system:
> 
> 1.
> On several systems which do not have ncurses development files
> installed, the "configure" script does not detect it,
> and completion later fails.
> Example with Debian 8.1:
> 
> ===
>   CC       disk-utils/cfdisk-cfdisk.o
> disk-utils/cfdisk.c: In function ‘ui_get_string’:
> disk-utils/cfdisk.c:1757:3: warning: implicit declaration of function ‘get_wch’ [-Wimplicit-function-declaration]
>    if (get_wch(&c) == ERR) {
>    ^
> disk-utils/cfdisk.c:1757:3: warning: nested extern declaration of ‘get_wch’ [-Wnested-externs]
>   CCLD     cfdisk
> /usr/bin/ld: cannot find -lncursesw
> collect2: error: ld returned 1 exit status
> make[2]: *** [cfdisk] Error 1
> ===

OK, I had nice afternoon with Debian (after 10 years;-)

All should be fixed. Please, test it. I'd like to release v2.30 ASAP.

    Karel


-- 
 Karel Zak  <kzak@redhat.com>
 http://karelzak.blogspot.com

Re: [ANNOUNCE] util-linux v2.30-rc2

[Assaf Gordon]

Hello,

> On May 30, 2017, at 11:18, Karel Zak <kzak@redhat.com> wrote:
>=20
> On Tue, May 30, 2017 at 01:03:53AM -0400, Assaf Gordon wrote:
>> /usr/bin/ld: cannot find -lncursesw
>> collect2: error: ld returned 1 exit status
>> make[2]: *** [cfdisk] Error 1
>> =3D=3D=3D
>=20
> OK, I had nice afternoon with Debian (after 10 years;-)


I'm still seeing this error (tested with util-linux-2.30-rc2-16-2ac1b on
same Debian 8.1 system).

"config.log" is here: http://sprunge.us/XXfX

This might be a problem with some interplay between 'libncurses' and =
'libncursesw' (the wide-character variant).

The installed packages on the system are:
=3D=3D=3D=3D
$ dpkg -l '*ncurs*'
Desired=3DUnknown/Install/Remove/Purge/Hold
| =
Status=3DNot/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-=
pend
|/ Err?=3D(none)/Reinst-required (Status,Err: uppercase=3Dbad)
||/ Name                    Version          Architecture     =
Description
=
+++-=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D-=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D-=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D-=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D
un  libncurses-dev          <none>           <none>           (no =
description available)
ii  libncurses5:amd64       5.9+20140913-1+b amd64            shared =
libraries for terminal handling
ii  libncurses5-dev:amd64   5.9+20140913-1+b amd64            =
developer's libraries for ncurses
ii  libncursesw5:amd64      5.9+20140913-1+b amd64            shared =
libraries for terminal handling (wide charact
un  ncurses                 <none>           <none>           (no =
description available)
ii  ncurses-base            5.9+20140913-1   all              basic =
terminal type definitions
ii  ncurses-bin             5.9+20140913-1+b amd64            =
terminal-related programs and man pages
un  ncurses-dev             <none>           <none>           (no =
description available)
un  ncurses-doc             <none>           <none>           (no =
description available)
un  ncurses-runtime         <none>           <none>           (no =
description available)
ii  ncurses-term            5.9+20140913-1   all              additional =
terminal type definitions
=3D=3D=3D=3D

I see that both the uni-byte and the wide-character versions are =
installed,
but only the uni-byte '-dev' version is installed,
And the error is about linking with the wide-char variant:
  /usr/bin/ld: cannot find -lncursesw

It could be that this is not an optimal system to build util-linux on
(e.g. messed up packages combination? yet all done with valid 'apt-get' =
commands).
However it would be nice to detect it during "./configure" if possible.

----

Also,
On alpine-linux, the compilation now fails with:
=3D=3D=3D
CC       sys-utils/dmesg-dmesg.o
sys-utils/dmesg.c:9:26: fatal error: linux/unistd.h: No such file or =
directory
compilation terminated.
Makefile:9274: recipe for target 'sys-utils/dmesg-dmesg.o' failed
make[2]: *** [sys-utils/dmesg-dmesg.o] Error 1
=3D=3D=3D
Again, this could be due to missing "linux-headers" package - but is it =
really the kernel's
"unistd.h" that's needed, or could this work with the standard one?



Thanks for all your quick fixes,
 - assaf

Re: [ANNOUNCE] util-linux v2.30-rc2

[Karel Zak]

On Tue, May 30, 2017 at 11:48:46PM -0400, Assaf Gordon wrote:
> Hello,
> 
> > On May 30, 2017, at 11:18, Karel Zak <kzak@redhat.com> wrote:
> > 
> > On Tue, May 30, 2017 at 01:03:53AM -0400, Assaf Gordon wrote:
> >> /usr/bin/ld: cannot find -lncursesw
> >> collect2: error: ld returned 1 exit status
> >> make[2]: *** [cfdisk] Error 1
> >> ===
> > 
> > OK, I had nice afternoon with Debian (after 10 years;-)
> 
> 
> I'm still seeing this error (tested with util-linux-2.30-rc2-16-2ac1b on
> same Debian 8.1 system).

Ah, I had another issues with ncurses on debian, so I forget to very
this one. Fixed now (I hope).

> +++-=======================-================-================-====================================================
> un  libncurses-dev          <none>           <none>           (no description available)
> ii  libncurses5:amd64       5.9+20140913-1+b amd64            shared libraries for terminal handling
> ii  libncurses5-dev:amd64   5.9+20140913-1+b amd64            developer's libraries for ncurses
> ii  libncursesw5:amd64      5.9+20140913-1+b amd64            shared libraries for terminal handling (wide charact

The issue is that we get -lncursesw from pkg-config (probably due to
installed ncursesw-config), but the rest of the build system assumes
non-widechar version. It seems better to use pkg-config as fallback
solution only.

> Also,
> On alpine-linux, the compilation now fails with:
> ===
> CC       sys-utils/dmesg-dmesg.o
> sys-utils/dmesg.c:9:26: fatal error: linux/unistd.h: No such file or directory
> compilation terminated.
> Makefile:9274: recipe for target 'sys-utils/dmesg-dmesg.o' failed
> make[2]: *** [sys-utils/dmesg-dmesg.o] Error 1
> ===
> Again, this could be due to missing "linux-headers" package - but is it really the kernel's
> "unistd.h" that's needed, or could this work with the standard one?

The file linux/unistd.h is unnecessary for dmesg (fixed), but we have
many another places where kernel header files are required.

So, now the header file <linux/version.h> is required by ./configure
on Linux. I guess it's not overkill.

    Karel

-- 
 Karel Zak  <kzak@redhat.com>
 http://karelzak.blogspot.com

Re: [ANNOUNCE] util-linux v2.30-rc2

[Assaf Gordon]

Hello,

On Wed, May 31, 2017 at 11:16:48AM +0200, Karel Zak wrote:
>On Tue, May 30, 2017 at 11:48:46PM -0400, Assaf Gordon wrote:
>> I'm still seeing this error (tested with util-linux-2.30-rc2-16-2ac1b on
>> same Debian 8.1 system).
>
>Ah, I had another issues with ncurses on debian, so I forget to very
>this one. Fixed now (I hope).

Thanks, works now (tested with Debian 8.1, 7.6, Ubuntu 16.04, 15.04,
14.04).


>> On alpine-linux, the compilation now fails with:
>
>So, now the header file <linux/version.h> is required by ./configure
>on Linux. I guess it's not overkill.

Yes, works well and the error message is clear.
Adding linux-headers with `sudo apk add linux-headers` solves the issue.

Now that it builds on alpine-linux, two other issues (not necessarily 
show-stoppers for the release):

1.
The test infrastrucutre uses some GNU extensions which aren't
available by default on Alpine-Linux (it has busybox instead of GNU 
coreuils). Specifically: ./tests/run.sh uses `#!/bin/bash`, `xargs -P` 
and `grep --after-context=1` .
The `grep` at least is easy to fix as busybox's grep supports the 
short-option equivalent of `-A1`.

As a work-around, the GNU programs can be installed on alpine with:
    sudo apk add findutils bash grep


2.
After installing the gnu programs `make check` works, but few tests 
fail:
============
   col: multibyte input                ... FAILED (col/multibyte)
colcrt: regressions                    ...
      : crash1                      ... FAILED (colcrt/regressions-crash1)
      : crash2                      ... FAILED (colcrt/regressions-crash2)
      : hang1                       ... FAILED (colcrt/regressions-hang1)
      ... FAILED (3 from 3 sub-tests)
column: invalid multibyte              ... FAILED (column/invalid-multibyte)
 fdisk: invalid input tests            ... FAILED (fdisk/oddinput)
getopt: options                        ...
      : alternative_option_clash    ... FAILED (getopt/options-alternative_option_clash)
      : posix_correctly             ... FAILED (getopt/options-posix_correctly)
      ... FAILED (2 from 29 sub-tests)
  misc: swaplabel                      ... FAILED (misc/swaplabel)
rename: exit codes                     ... FAILED (rename/exit_codes)
  utmp: subsecond                      ... FAILED (utmp/utmpdump-subsecond)
==============

Regarding the 'multibyte' failures - these could stem from musl-libc's differences
(compared to glibc).
'utmp' failure might be because musl-libc does not implement utmp support (only stubs).

I'm happy to test further if you have specific suggestions of what to test.

-----


Lastly,
I've hit a snag when trying to run 'make dist' on the latest version.
make failed with:
===
make[5]: Entering directory `/localdata2/sources/util-linux/po'
File da.new.po does not exist. If you are a translator, you can create it through 'msginit'.
make[5]: *** [da.new.po-create] Error 1
make[5]: Leaving directory `/localdata2/sources/util-linux/po'
===

I have no clue when this is coming from (it didn't fail on in previous builds).
Could be a problem on my system, or something else. Didn't investigate further
(just manually removed 'da.new.*' from po/Makefile).


regards,
 - assaf

Re: [ANNOUNCE] util-linux v2.30-rc2

[Ruediger Meier]

On Wednesday 31 May 2017, Assaf Gordon wrote:
> Hello,
>
> On Wed, May 31, 2017 at 11:16:48AM +0200, Karel Zak wrote:
> >On Tue, May 30, 2017 at 11:48:46PM -0400, Assaf Gordon wrote:
> >> I'm still seeing this error (tested with
> >> util-linux-2.30-rc2-16-2ac1b on same Debian 8.1 system).
> >
> >Ah, I had another issues with ncurses on debian, so I forget to very
> >this one. Fixed now (I hope).
>
> Thanks, works now (tested with Debian 8.1, 7.6, Ubuntu 16.04, 15.04,
> 14.04).
>
> >> On alpine-linux, the compilation now fails with:
> >
> >So, now the header file <linux/version.h> is required by ./configure
> >on Linux. I guess it's not overkill.
>
> Yes, works well and the error message is clear.
> Adding linux-headers with `sudo apk add linux-headers` solves the
> issue.
>
> Now that it builds on alpine-linux, two other issues (not necessarily
> show-stoppers for the release):
>
> 1.
> The test infrastrucutre uses some GNU extensions which aren't
> available by default on Alpine-Linux (it has busybox instead of GNU
> coreuils). Specifically: ./tests/run.sh uses `#!/bin/bash`, `xargs
> -P` and `grep --after-context=1` .
> The `grep` at least is easy to fix as busybox's grep supports the
> short-option equivalent of `-A1`.
>
> As a work-around, the GNU programs can be installed on alpine with:
>     sudo apk add findutils bash grep

This is a known issue in the test-suite only. It is pseudo-documented 
in .travis-functions.sh, osx_prepare_check(). We require bash and 
non-POSIX features of readlink, seq, truncate, find, xargs, tar and 
sed.

>
> 2.
> After installing the gnu programs `make check` works, but few tests
> fail:

> ============
>    col: multibyte input                ... FAILED (col/multibyte)
> colcrt: regressions                    ...
>
>       : crash1                      ... FAILED
>       : (colcrt/regressions-crash1) crash2                      ...
>       : FAILED (colcrt/regressions-crash2) hang1                     
>       :  ... FAILED (colcrt/regressions-hang1)
>
>       ... FAILED (3 from 3 sub-tests)
> column: invalid multibyte              ... FAILED
> (column/invalid-multibyte) fdisk: invalid input tests            ...
> FAILED (fdisk/oddinput) getopt: options                        ...
>
>       : alternative_option_clash    ... FAILED
>       : (getopt/options-alternative_option_clash) posix_correctly    
>       :         ... FAILED (getopt/options-posix_correctly)

>       ... FAILED (2 from 29 sub-tests)
>   misc: swaplabel                      ... FAILED (misc/swaplabel)
> rename: exit codes                     ... FAILED (rename/exit_codes)
>   utmp: subsecond                      ... FAILED
> (utmp/utmpdump-subsecond) ==============
>
> Regarding the 'multibyte' failures - these could stem from
> musl-libc's differences (compared to glibc).


> 'utmp' failure might be because musl-libc does not implement utmp
> support (only stubs).

musl supports both utmp and utmpx. We are using utmpx. Could you show us 
all your test diffs?:

  find tests/diff -type f  | xargs tail -n +1

> I'm happy to test further if you have specific suggestions of what to
> test.
>
> -----
>
>
> Lastly,
> I've hit a snag when trying to run 'make dist' on the latest version.
> make failed with:

"make dist" is something we do not really care to make it fully 
portable. It should run on "usual" maintainer machines to provide 
portable tarballs for other machines. "make distcheck" will be even 
less portable.

> ===
> make[5]: Entering directory `/localdata2/sources/util-linux/po'
> File da.new.po does not exist. If you are a translator, you can
> create it through 'msginit'. make[5]: *** [da.new.po-create] Error 1
> make[5]: Leaving directory `/localdata2/sources/util-linux/po'
> ===
>
> I have no clue when this is coming from (it didn't fail on in
> previous builds). Could be a problem on my system, or something else.
> Didn't investigate further (just manually removed 'da.new.*' from
> po/Makefile).
>
>
> regards,
>  - assaf
> --
> To unsubscribe from this list: send the line "unsubscribe util-linux"
> in the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [ANNOUNCE] util-linux v2.30-rc2

[Assaf Gordon]

[-- Attachment #1: Type: text/plain, Size: 243 bytes --]

On Wed, May 31, 2017 at 08:36:38PM +0200, Ruediger Meier wrote:

>musl supports both utmp and utmpx. We are using utmpx. Could you show us
>all your test diffs?:
>
>  find tests/diff -type f  | xargs tail -n +1

Attached.

regards,
 - assaf



[-- Attachment #2: err.log --]
[-- Type: text/plain, Size: 6606 bytes --]

==> tests/diff/column/invalid-multibyte <==
--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/expected/column/invalid-multibyte
+++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/column/invalid-multibyte
@@ -1 +1 @@
-column: read failed: Invalid or incomplete multibyte or wide character
+

==> tests/diff/getopt/options-posix_correctly <==
--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/expected/getopt/options-posix_correctly
+++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/getopt/options-posix_correctly
@@ -1,4 +1,4 @@
  -a -b '1' -c '2' --a-long --b-long '3' --c-long '' -- 'foo' 'bar' 'xyxxy' '--a-long'
 exit value: 0
- -a -b '1' -c '2' --a-long --b-long '3' --c-long '' -- 'foo' 'bar' 'xyxxy' '--a-long'
+ -a -b '1' -c '2' --a-long --b-long '3' --c-long '' --a-long -- 'foo' 'bar' 'xyxxy'
 exit value: 0

==> tests/diff/getopt/options-alternative_option_clash <==
--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/expected/getopt/options-alternative_option_clash
+++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/getopt/options-alternative_option_clash
@@ -1,2 +1,2 @@
- -a --abcde --abcde -a -c --
+ --abcde --abcde --abcde -a -c --
 exit value: 0

==> tests/diff/misc/swaplabel <==
--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/expected/misc/swaplabel
+++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/misc/swaplabel
@@ -1,7 +1,7 @@
 mkswap: error: swap area needs to be at least 10 pages
 mkswap: <swapfile>: insecure permissions <perm>, 0600 suggested.
-mkswap: Label was truncated.
 Setting up swapspace version 1, size = 9 pages (9xPGSZ bytes)
+mkswap: Label was truncated.
 LABEL=1234567890abcde, UUID=12345678-abcd-abcd-abcd-1234567890ab
 LABEL: 1234567890abcde
 UUID:  12345678-abcd-abcd-abcd-1234567890ab

==> tests/diff/fdisk/oddinput <==
--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/expected/fdisk/oddinput
+++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/fdisk/oddinput
@@ -8,4 +8,4 @@
 Nonexistent file
 fdisk: cannot open _a_file_that_does_not_exist_: No such file or directory
 Too small file
-fdisk: cannot open oddinput.toosmall: Inappropriate ioctl for device
+fdisk: cannot open oddinput.toosmall: Invalid argument

==> tests/diff/utmp/utmpdump-subsecond <==
--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/expected/utmp/utmpdump-subsecond
+++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/utmp/utmpdump-subsecond
@@ -1,5 +1,8 @@
 last 9 is expected to disappear in conversion
-3c3
-< [0] [00000] [ts/0] [nonvalid] [foo         ] [zero                ] [0.0.0.0        ] [2013-08-28T12:00:00,123456789+0000]
----
-> [0] [00000] [ts/0] [nonvalid] [foo         ] [zero                ] [0.0.0.0        ] [2013-08-28T12:00:00,12345678+0000]
+--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/ts/utmp/subsec
++++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/utmp/utmpdump-subsecond.txt
+@@ -1,3 +1,3 @@
+ [7] [00010] [ipv6] [IPv6    ] [root        ] [dns-server          ] [2001:503:ba3e::2:30] [2013-08-28T20:30:40,123456+0000]
+ [8] [00011] [ipv6] [IPv6    ] [root        ] [dns-server          ] [2001:503:ba3e::2:30] [2013-08-28T20:40:50,999999+0000]
+-[0] [00000] [ts/0] [nonvalid] [foo         ] [zero                ] [0.0.0.0        ] [2013-08-28T12:00:00,123456789+0000]
++[0] [00000] [ts/0] [nonvalid] [foo         ] [zero                ] [0.0.0.0        ] [2013-08-28T12:00:00,12345678+0000]

==> tests/diff/rename/exit_codes <==
--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/expected/rename/exit_codes
+++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/rename/exit_codes
@@ -2,6 +2,6 @@
 `rename_exit_codes.1' -> `rename_exit_values.1'
 `rename_exit_codes.2' -> `rename_exit_values.2'
 EXIT_SUCCESS: 0
-rename: rename_exit_values.2: rename to rename_exit_codes.2 failed: Is a directory
 `rename_exit_values.1' -> `rename_exit_codes.1'
+rename: rename_exit_values.2: rename to rename_exit_codes.2 failed: Is a directory
 RENAME_EXIT_SOMEOK: 2

==> tests/diff/colcrt/regressions-crash2 <==
--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/expected/colcrt/regressions-crash2
+++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/colcrt/regressions-crash2
@@ -1,10 +1,2 @@
-
-FGHIKIJKN\
-MN9|
-XYZRnT RnTUV|   NXP:w
-MN
-'QRnTUVWXYZQRnTU
-MN9|
-XYZ nT RnTUV|   NXP:w
-   -
-MNOP.return value: 0
+timeout: can't execute '2': No such file or directory
+return value: 127

==> tests/diff/colcrt/regressions-crash1 <==
--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/expected/colcrt/regressions-crash1
+++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/colcrt/regressions-crash1
@@ -1,2 +1,2 @@
- W U;    U;  D  f U;  D  f    D  f    >     @W  ]~   K- -  ----  -- -- -  -- -- ---- -- ---- --  -  --  --- --
-return value: 0
+timeout: can't execute '2': No such file or directory
+return value: 127

==> tests/diff/colcrt/regressions-hang1 <==
--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/expected/colcrt/regressions-hang1
+++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/colcrt/regressions-hang1
@@ -1,3 +1,2 @@
-789:;<=>=>?IABUVNXYZ[ `abcdefgg !"#$%&'()*+,-./01234)*:,-./0123456789:;[=>?1234)*:,-./0123456789:;[=>?4456789:;<=>?IABUVN`abcdefg !"
-                     -
-return value: 0
+timeout: can't execute '2': No such file or directory
+return value: 127

==> tests/diff/col/multibyte <==
--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/expected/col/multibyte
+++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/col/multibyte
@@ -1 +0,0 @@
-col: Invalid or incomplete multibyte or wide character

==> tests/diff/schedutils/chrt-non-root-batch-vs-nice <==
--- /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/expected/schedutils/chrt-non-root-batch-vs-nice
+++ /tmp/util-linux-2.30-rc2-22-3947.MJfaDm/util-linux-2.30-rc2-22-3947/tests/output/schedutils/chrt-non-root-batch-vs-nice
@@ -1,3 +1 @@
 <removed> (process ID) old priority 0, new priority 5
-SCHED_BATCH
-0

Re: [ANNOUNCE] util-linux v2.30-rc2

[Ruediger Meier]

On Wednesday 31 May 2017, Assaf Gordon wrote:
> On Wed, May 31, 2017 at 08:36:38PM +0200, Ruediger Meier wrote:
> >musl supports both utmp and utmpx. We are using utmpx. Could you
> > show us all your test diffs?:
> >
> >  find tests/diff -type f  | xargs tail -n +1
>
> Attached.

Thanks. The utmp problem is just another gnu-diff incompatibility. I'll 
fix a few of your issues and send patches later. 

cu,
Rudi

Re: [ANNOUNCE] util-linux v2.30-rc2

[Ruediger Meier]

On Wednesday 31 May 2017, Assaf Gordon wrote:

> rename: exit codes                     ... FAILED (rename/exit_codes)

This is because stdout and stderr are mixed:

musl seems to behave like the patch below, while glibc always prints 
the "wrong" order.

--------------
diff --git a/misc-utils/rename.c b/misc-utils/rename.c
index 502e888..edfc4e0 100644
--- a/misc-utils/rename.c
+++ b/misc-utils/rename.c
@@ -103,10 +103,13 @@ static int do_file(char *from, char *to, char *s, 
int verbose, int noact)
                return 0;
        else if (!noact && rename(s, newname) != 0) {
                warn(_("%s: rename to %s failed"), s, newname);
+               fflush(stderr);
                ret = 2;
        }
-       if (verbose && (noact || ret == 1))
+       if (verbose && (noact || ret == 1)) {
                printf("`%s' -> `%s'\n", s, newname);
+               fflush(stdout);
+       }
        free(newname);
        return ret;
 }
-------------

Not sure whether we should really use fflush() to define the order of 
error and non-error messages. We could also simply "fix" the the test 
script by not using "--verbose". or by using "stdbuf -o0 -e0 
rename ..."


cu,
Rudi