dev-python/cryptography-2.6.1 compile failure

classic Classic list List threaded Threaded
11 messages Options
Reply | Threaded
Open this post in threaded view
|

dev-python/cryptography-2.6.1 compile failure

Walter Dnes
  I can't build dev-python/cryptography-2.6.1 on my 64-bit Gentoo
desktop.  Earlier on in the system update process the readline build
told me to...

revdep-rebuild --library '/lib64/libhistory.so.6' && rm '/lib64/libhistory.so.6'
revdep-rebuild --library '/lib64/libreadline.so.6' && rm '/lib64/libreadline.so.6'

  I've done that.  I've also tried cutting MAKEOPTS down to -j1.  Didn't
help.  This is 1 of 15 builds remaining in my system update.  I'm
attaching the log file.  The error appears to be...

ImportError: libffi.so.6: cannot open shared object file: No such file or directory

  I do have libffi.  The output of "emerge -pv libffi" is...
[ebuild   R    ] dev-libs/libffi-3.3_rc0:0/7::gentoo  USE="-debug -pax_kernel -static-libs -test" 0 KiB

  "find / -name libffi.so.*" turns up...

/usr/lib64/libffi.so.7
/usr/lib64/libffi.so.7.1.0

...so it looks like some sort of version mis-match, if that helps.

--
Walter Dnes <[hidden email]>
I don't run "desktop environments"; I run useful applications

log.txt.gz (1K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: dev-python/cryptography-2.6.1 compile failure

Hartmut Figge
Walter Dnes:

>  "find / -name libffi.so.*" turns up...
>
>/usr/lib64/libffi.so.7
>/usr/lib64/libffi.so.7.1.0
>
>...so it looks like some sort of version mis-match, if that helps.

I have the same libffi. "find" takes much time, I prefer "locate". And
"qlist".

hafi@i5-64 ~ $ qlist libffi | grep libffi.so
/usr/lib/libffi.so
/usr/lib/libffi.so.6
/usr/lib/libffi.so.6.0.4
/usr/lib/libffi.so.7
/usr/lib/libffi.so.7.1.0
/usr/lib64/libffi.so
/usr/lib64/libffi.so.6
/usr/lib64/libffi.so.6.0.4
/usr/lib64/libffi.so.7
/usr/lib64/libffi.so.7.1.0

Hartmut


Reply | Threaded
Open this post in threaded view
|

Re: dev-python/cryptography-2.6.1 compile failure

Walter Dnes
On Mon, Oct 28, 2019 at 03:16:07PM +0100, Hartmut Figge wrote
>
> I have the same libffi. "find" takes much time, I prefer "locate". And
> "qlist".

  I greatly sped up the search with "find /usr/ -name libffi.so*".  The
full result of the search is...

/usr/lib64/libffi.so
/usr/lib64/libffi.so.7
/usr/lib64/libffi.so.7.1.0

  That's all she wrote.  So how do I force "slot 6" or whatever it's
called?

--
Walter Dnes <[hidden email]>
I don't run "desktop environments"; I run useful applications

Reply | Threaded
Open this post in threaded view
|

Re: dev-python/cryptography-2.6.1 compile failure

Hartmut Figge
Walter Dnes:

>/usr/lib64/libffi.so
>/usr/lib64/libffi.so.7
>/usr/lib64/libffi.so.7.1.0
>
>  That's all she wrote.  So how do I force "slot 6" or whatever it's
>called?

Hm. I didn't force anything. But output of -pv is different:

hafi@i5-64 ~ $ emerge -pv libffi

These are the packages that would be merged, in order:

Calculating dependencies... done!
[ebuild   R    ] dev-libs/libffi-3.3_rc0:0/7::gentoo  USE="-debug
-pax_kernel -static-libs -test" ABI_X86="32 (64) (-x32)" 0 KiB

The whole ABI part is missing from your -pv. Hm.

Hartmut


Reply | Threaded
Open this post in threaded view
|

Re: dev-python/cryptography-2.6.1 compile failure

Walter Dnes
On Mon, Oct 28, 2019 at 04:14:15PM +0100, Hartmut Figge wrote

> Hm. I didn't force anything. But output of -pv is different:
>
> hafi@i5-64 ~ $ emerge -pv libffi
>
> These are the packages that would be merged, in order:
>
> Calculating dependencies... done!
> [ebuild   R    ] dev-libs/libffi-3.3_rc0:0/7::gentoo  USE="-debug
> -pax_kernel -static-libs -test" ABI_X86="32 (64) (-x32)" 0 KiB
>
> The whole ABI part is missing from your -pv. Hm.

  I use the regular no-multilib profile
[26]  default/linux/amd64/17.1/no-multilib (stable) *

  I tried keywording =dev-python/cryptography-2.7 in the hope that it
would work with the newer libffi, but no luck.  Could you please run

equery b /usr/lib64/libffi.so.6

...on your system, and post the output?  I may have to revert to an
older version, or hopefully slot it in.  Looking at my
/usr/portage/dev-libs/libffi directory...

i660 /usr/portage/dev-libs/libffi # grep -i slot *
grep: files: Is a directory
libffi-3.2.1-r1.ebuild:SLOT="0"
libffi-3.2.1-r2.ebuild:SLOT="0"
libffi-3.2.1-r3.ebuild:SLOT="0"
libffi-3.2.1.ebuild:SLOT="0"
libffi-3.3_rc0.ebuild:SLOT="0/7" # SONAME=libffi.so.7
libffi-3.3_rc1.ebuild:SLOT="0/7" # SONAME=libffi.so.7

--
Walter Dnes <[hidden email]>
I don't run "desktop environments"; I run useful applications

Reply | Threaded
Open this post in threaded view
|

Re: dev-python/cryptography-2.6.1 compile failure

Hartmut Figge
Walter Dnes:

>Could you please run
>
>equery b /usr/lib64/libffi.so.6
>
>...on your system, and post the output?

hafi@i5-64 ~ $ equery b /usr/lib64/libffi.so.6
 * Searching for /usr/lib64/libffi.so.6 ...
dev-libs/libffi-3.3_rc0 (/usr/lib64/libffi.so.6 -> libffi.so.6.0.4)
dev-libs/libffi-3.3_rc0 (/usr/lib64/libffi.so.6.0.4)

Hartmut


Reply | Threaded
Open this post in threaded view
|

Re: dev-python/cryptography-2.6.1 compile failure

Walter Dnes
On Mon, Oct 28, 2019 at 05:50:39PM +0100, Hartmut Figge wrote

> hafi@i5-64 ~ $ equery b /usr/lib64/libffi.so.6
>  * Searching for /usr/lib64/libffi.so.6 ...
> dev-libs/libffi-3.3_rc0 (/usr/lib64/libffi.so.6 -> libffi.so.6.0.4)
> dev-libs/libffi-3.3_rc0 (/usr/lib64/libffi.so.6.0.4)

  I think I'm officially screwed.  Now I can't even open a new browser
profile due to it not finding /usr/lib64/libffi.so.6.  I tried
re-emerging libffi and it ends up with...

===============================================================
 * Final size of build directory: 6700 KiB (6.5 MiB)
 * Final size of installed tree:   764 KiB

strip: x86_64-pc-linux-gnu-strip --strip-unneeded -N __gentoo_check_ldflags__ -R .comment -R .GCC.command.line -R .note.gnu.gold-version
   /usr/lib64/libffi.so.7.1.0

>>> Installing (1 of 1) dev-libs/libffi-3.3_rc0::gentoo
>>> Auto-cleaning packages...
===============================================================

  Now what?  Further checking.  libffi is a dependancy of
virtual/libffi, and the virtual/libffi-3.3_rc0.ebuild is...

===============================================================
# Copyright 1999-2019 Gentoo Authors
# Distributed under the terms of the GNU General Public License v2

# The purpose of this ebuild is to provide quick fallback if and when we need to
# switch back to internal copy of libffi from sys-devel/gcc.

EAPI=7

inherit multilib-build

DESCRIPTION="A virtual for the Foreign Function Interface implementation"
SLOT="0/7" # SONAME=libffi.so.7
KEYWORDS="alpha amd64 arm ~arm64 hppa ia64 ~m68k ~mips ppc ppc64 ~riscv s390 ~sh
 sparc x86 ~ppc-aix ~x64-cygwin ~amd64-linux ~x86-linux ~ppc-macos ~x64-macos ~x
86-macos ~m68k-mint ~sparc-solaris ~sparc64-solaris ~x64-solaris ~x86-solaris"

RDEPEND=">=dev-libs/libffi-3.3_rc0:0/7[${MULTILIB_USEDEP}]"
===============================================================


--
Walter Dnes <[hidden email]>
I don't run "desktop environments"; I run useful applications

Reply | Threaded
Open this post in threaded view
|

Re: dev-python/cryptography-2.6.1 compile failure

Mick-10
On Monday, 28 October 2019 17:54:02 GMT Walter Dnes wrote:

> On Mon, Oct 28, 2019 at 05:50:39PM +0100, Hartmut Figge wrote
>
> > hafi@i5-64 ~ $ equery b /usr/lib64/libffi.so.6
> >
> >  * Searching for /usr/lib64/libffi.so.6 ...
> >
> > dev-libs/libffi-3.3_rc0 (/usr/lib64/libffi.so.6 -> libffi.so.6.0.4)
> > dev-libs/libffi-3.3_rc0 (/usr/lib64/libffi.so.6.0.4)
>
>   I think I'm officially screwed.  Now I can't even open a new browser
> profile due to it not finding /usr/lib64/libffi.so.6.  I tried
> re-emerging libffi and it ends up with...
>
> ===============================================================
>  * Final size of build directory: 6700 KiB (6.5 MiB)
>  * Final size of installed tree:   764 KiB
>
> strip: x86_64-pc-linux-gnu-strip --strip-unneeded -N
> __gentoo_check_ldflags__ -R .comment -R .GCC.command.line -R
> .note.gnu.gold-version /usr/lib64/libffi.so.7.1.0
>
> >>> Installing (1 of 1) dev-libs/libffi-3.3_rc0::gentoo
> >>> Auto-cleaning packages...
>
> ===============================================================
>
>   Now what?  Further checking.  libffi is a dependancy of
> virtual/libffi, and the virtual/libffi-3.3_rc0.ebuild is...
>
> ===============================================================
> # Copyright 1999-2019 Gentoo Authors
> # Distributed under the terms of the GNU General Public License v2
>
> # The purpose of this ebuild is to provide quick fallback if and when we
> need to # switch back to internal copy of libffi from sys-devel/gcc.
>
> EAPI=7
>
> inherit multilib-build
>
> DESCRIPTION="A virtual for the Foreign Function Interface implementation"
> SLOT="0/7" # SONAME=libffi.so.7
> KEYWORDS="alpha amd64 arm ~arm64 hppa ia64 ~m68k ~mips ppc ppc64 ~riscv s390
> ~sh sparc x86 ~ppc-aix ~x64-cygwin ~amd64-linux ~x86-linux ~ppc-macos
> ~x64-macos ~x 86-macos ~m68k-mint ~sparc-solaris ~sparc64-solaris
> ~x64-solaris ~x86-solaris"
>
> RDEPEND=">=dev-libs/libffi-3.3_rc0:0/7[${MULTILIB_USEDEP}]"
> ===============================================================
If you have no back up you could probably download a LiveCD/DVD image and use
that.
--
Regards,

Mick

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: dev-python/cryptography-2.6.1 compile failure

Jack
In reply to this post by Walter Dnes
My guess is that it isn't libffi that needs re-emerging, but things  
that depend on it.  I have 3.30_rc0 installd (both dev_libs and  
virtual) and I have no libffi.so.6 anywhere to be found.  Have you  
tried revdep-rebuild?  Either that, or "emerge -pc virtual/libffi" and  
them emerge -1 everything it lists.

Jack

On 2019.10.28 13:54, Walter Dnes wrote:

> On Mon, Oct 28, 2019 at 05:50:39PM +0100, Hartmut Figge wrote
>
> > hafi@i5-64 ~ $ equery b /usr/lib64/libffi.so.6
> >  * Searching for /usr/lib64/libffi.so.6 ...
> > dev-libs/libffi-3.3_rc0 (/usr/lib64/libffi.so.6 -> libffi.so.6.0.4)
> > dev-libs/libffi-3.3_rc0 (/usr/lib64/libffi.so.6.0.4)
>
>   I think I'm officially screwed.  Now I can't even open a new browser
> profile due to it not finding /usr/lib64/libffi.so.6.  I tried
> re-emerging libffi and it ends up with...
>
> ===============================================================
>  * Final size of build directory: 6700 KiB (6.5 MiB)
>  * Final size of installed tree:   764 KiB
>
> strip: x86_64-pc-linux-gnu-strip --strip-unneeded -N  
> __gentoo_check_ldflags__ -R .comment -R .GCC.command.line -R  
> .note.gnu.gold-version
>    /usr/lib64/libffi.so.7.1.0
>
> >>> Installing (1 of 1) dev-libs/libffi-3.3_rc0::gentoo
> >>> Auto-cleaning packages...
> ===============================================================
>
>   Now what?  Further checking.  libffi is a dependancy of
> virtual/libffi, and the virtual/libffi-3.3_rc0.ebuild is...
>
> ===============================================================
> # Copyright 1999-2019 Gentoo Authors
> # Distributed under the terms of the GNU General Public License v2
>
> # The purpose of this ebuild is to provide quick fallback if and when  
> we need to
> # switch back to internal copy of libffi from sys-devel/gcc.
>
> EAPI=7
>
> inherit multilib-build
>
> DESCRIPTION="A virtual for the Foreign Function Interface  
> implementation"
> SLOT="0/7" # SONAME=libffi.so.7
> KEYWORDS="alpha amd64 arm ~arm64 hppa ia64 ~m68k ~mips ppc ppc64  
> ~riscv s390 ~sh
>  sparc x86 ~ppc-aix ~x64-cygwin ~amd64-linux ~x86-linux ~ppc-macos  
> ~x64-macos ~x
> 86-macos ~m68k-mint ~sparc-solaris ~sparc64-solaris ~x64-solaris  
> ~x86-solaris"
>
> RDEPEND=">=dev-libs/libffi-3.3_rc0:0/7[${MULTILIB_USEDEP}]"
> ===============================================================
>
>
> --
> Walter Dnes <[hidden email]>
> I don't run "desktop environments"; I run useful applications
Reply | Threaded
Open this post in threaded view
|

Re: dev-python/cryptography-2.6.1 compile failure

Walter Dnes
On Mon, Oct 28, 2019 at 02:34:41PM -0400, Jack wrote
> My guess is that it isn't libffi that needs re-emerging, but things  
> that depend on it.  I have 3.30_rc0 installd (both dev_libs and  
> virtual) and I have no libffi.so.6 anywhere to be found.  Have you  
> tried revdep-rebuild?  Either that, or "emerge -pc virtual/libffi" and  
> them emerge -1 everything it lists.

  Facepalm.  I did revdep-rebuild when the update first failed part way
through, but not at the most recent failure.  After another
revdep-rebuild, Pale Moon (web browser) now starts up.  The list of
pending updates is down from 15 to 9.  I'm now run the (hopefully) final
emerge update run.

--
Walter Dnes <[hidden email]>
I don't run "desktop environments"; I run useful applications

Reply | Threaded
Open this post in threaded view
|

[SOLVED] Re: dev-python/cryptography-2.6.1 compile failure

Walter Dnes
On Mon, Oct 28, 2019 at 03:17:30PM -0400, Walter Dnes wrote
>
>   Facepalm.  I did revdep-rebuild when the update first failed part way
> through, but not at the most recent failure.  After another
> revdep-rebuild, Pale Moon (web browser) now starts up.  The list of
> pending updates is down from 15 to 9.  I'm now run the (hopefully) final
> emerge update run.

  Success.  Thanks for everybody's help.  Lesson learned... run
revdep-rebuild after every emerge that installs even one ebuild.

--
Walter Dnes <[hidden email]>
I don't run "desktop environments"; I run useful applications