aarch64 mock gcc/annobin errors

[Sam Varshavchik]

I'm running an F32 aarch64 guest qemu VM on an x86_64 F32 host, and, in that  
VM, I attempted to build something fairly simple via mock. This is my reward:

configure:3590: gcc -O2 -g -pipe -Wall -Werror=format-security -Wp,- 
D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fexceptions -fstack-protector- 
strong -grecord-gcc-switches -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 - 
specs=/usr/lib/rpm/redhat/redhat-annobin-cc1 -fasynchronous-unwind-tables - 
fstack-clash-protection  -Wl,-z,relro -Wl,--as-needed  -Wl,-z,now - 
specs=/usr/lib/rpm/redhat/redhat-hardened-ld conftest.c  >&5

annobin: conftest.c: AArch64: The annobin plugin is out of date with respect  
to  gcc
*** WARNING *** there are active plugins, do not report this as a bug unless  
you can reproduce it without enabling any plugins.
Event                            | Plugins
PLUGIN_FINISH_UNIT               | annobin: Generate final annotations
PLUGIN_START_UNIT                | annobin: Generate global annotations
PLUGIN_ALL_PASSES_START          | annobin: Generate per-function annotations
PLUGIN_ALL_PASSES_END            | annobin: Register per-function end symbol
conftest.c:1: internal compiler error: Segmentation fault

First time I've seen something like that, any idea what to do here, for this?
This if from a first build in mock, which presumably pulled in the most  
recent versions of all rpms into the chroot.

[Jerry James]

On Mon, Aug 3, 2020 at 4:25 PM Sam Varshavchik <mrsam(a)courier-mta.com&gt; wrote:
>  I'm running an F32 aarch64 guest qemu VM on an x86_64 F32 host,
> and, in that
>  VM, I attempted to build something fairly simple via mock. This is my reward:
>
>  configure:3590: gcc -O2 -g -pipe -Wall -Werror=format-security -Wp,-
>  D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fexceptions -fstack-protector-
>  strong -grecord-gcc-switches -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -
>  specs=/usr/lib/rpm/redhat/redhat-annobin-cc1 -fasynchronous-unwind-tables -
>  fstack-clash-protection  -Wl,-z,relro -Wl,--as-needed  -Wl,-z,now -
>  specs=/usr/lib/rpm/redhat/redhat-hardened-ld conftest.c  >&5
>
>  annobin: conftest.c: AArch64: The annobin plugin is out of date with respect
>  to  gcc
>  *** WARNING *** there are active plugins, do not report this as a bug unless
>  you can reproduce it without enabling any plugins.
>  Event                            | Plugins
>  PLUGIN_FINISH_UNIT               | annobin: Generate final annotations
>  PLUGIN_START_UNIT                | annobin: Generate global annotations
>  PLUGIN_ALL_PASSES_START          | annobin: Generate per-function annotations
>  PLUGIN_ALL_PASSES_END            | annobin: Register per-function end symbol
>  conftest.c:1: internal compiler error: Segmentation fault
>
>  First time I've seen something like that, any idea what to do here, for this?
>  This if from a first build in mock, which presumably pulled in the most
>  recent versions of all rpms into the chroot. 
Are you building for a rawhide target?  If so, mock is pulling in rpms
from the most recent rawhide compose.  Add "--enablerepo=local" to get
the very latest rpms from koji.

There have been a fewl instances recently of the annobin plugin not
matching gcc.  As far as I know, the issue has been resolved, so
hopefully the extra mock flag will work for you.
-- 
Jerry James
http://www.jamezone.org/

[Todd Zullinger]

Jerry James wrote:
>  On Mon, Aug 3, 2020 at 4:25 PM Sam Varshavchik
> <mrsam(a)courier-mta.com&gt; wrote:
> > I'm running an F32 aarch64 guest qemu VM on an x86_64 F32 host, and, in that
> > VM, I attempted to build something fairly simple via mock. This is my reward:
> >
> > configure:3590: gcc -O2 -g -pipe -Wall -Werror=format-security -Wp,-
> > D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fexceptions -fstack-protector-
> > strong -grecord-gcc-switches -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -
> > specs=/usr/lib/rpm/redhat/redhat-annobin-cc1 -fasynchronous-unwind-tables -
> > fstack-clash-protection  -Wl,-z,relro -Wl,--as-needed  -Wl,-z,now -
> > specs=/usr/lib/rpm/redhat/redhat-hardened-ld conftest.c  >&5
> >
> > annobin: conftest.c: AArch64: The annobin plugin is out of date with respect
> > to  gcc
> > *** WARNING *** there are active plugins, do not report this as a bug unless
> > you can reproduce it without enabling any plugins.
> > Event                            | Plugins
> > PLUGIN_FINISH_UNIT               | annobin: Generate final annotations
> > PLUGIN_START_UNIT                | annobin: Generate global annotations
> > PLUGIN_ALL_PASSES_START          | annobin: Generate per-function annotations
> > PLUGIN_ALL_PASSES_END            | annobin: Register per-function end symbol
> > conftest.c:1: internal compiler error: Segmentation fault
> >
> > First time I've seen something like that, any idea what to do here, for this?
> > This if from a first build in mock, which presumably pulled in the most
> > recent versions of all rpms into the chroot.
>  
>  Are you building for a rawhide target?  If so, mock is pulling in rpms
>  from the most recent rawhide compose.  Add "--enablerepo=local" to get
>  the very latest rpms from koji.
>  
>  There have been a fewl instances recently of the annobin plugin not
>  matching gcc.  As far as I know, the issue has been resolved, so
>  hopefully the extra mock flag will work for you. 
I think the same gcc version from rawhide was pushed to f32,
while the corresponding annonbin release (9.27-1.fc32) is
still in updates-testing¹.  So it's possible this comes up
even for an f32 target.  In that case, enabling either the
updates-testing or the 'local' repo for the mock build
target would probably help.

Karma for the annobin release would likely be appreciated by
the maintainer(s) and anyone else who's running into this
problem, if you are able to test and confirm it fixes the
issue(s). :)

¹ https://bodhi.fedoraproject.org/updates/FEDORA-2020-db7719f642
  https://bugzilla.redhat.com/1862297

-- 
Todd

[Sam Varshavchik]

Todd Zullinger writes:

>  Jerry James wrote:
>  > On Mon, Aug 3, 2020 at 4:25 PM Sam Varshavchik <mrsam(a)courier-mta.com&gt;  
>  wrote:
>  >> I'm running an F32 aarch64 guest qemu VM on an x86_64 F32 host, and, in  
>  that
>  >> VM, I attempted to build something fairly simple via mock. This is my  
>  reward:
>  >>
>  >> configure:3590: gcc -O2 -g -pipe -Wall -Werror=format-security -Wp,-
>  >> D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fexceptions -fstack- 
>  protector-
>  >> strong -grecord-gcc-switches -specs=/usr/lib/rpm/redhat/redhat-hardened- 
>  cc1 -
>  >> specs=/usr/lib/rpm/redhat/redhat-annobin-cc1 -fasynchronous-unwind-tables  
>  -
>  >> fstack-clash-protection  -Wl,-z,relro -Wl,--as-needed  -Wl,-z,now -
>  >> specs=/usr/lib/rpm/redhat/redhat-hardened-ld conftest.c  >&5
>  >>
>  >> annobin: conftest.c: AArch64: The annobin plugin is out of date with  
>  respect
>  >> to  gcc
>  >> *** WARNING *** there are active plugins, do not report this as a bug  
>  unless
>  >> you can reproduce it without enabling any plugins.
>  >> Event                            | Plugins
>  >> PLUGIN_FINISH_UNIT               | annobin: Generate final annotations
>  >> PLUGIN_START_UNIT                | annobin: Generate global annotations
>  >> PLUGIN_ALL_PASSES_START          | annobin: Generate per-function  
>  annotations
>  >> PLUGIN_ALL_PASSES_END            | annobin: Register per-function end  
>  symbol
>  >> conftest.c:1: internal compiler error: Segmentation fault
>  >>
>  >> First time I've seen something like that, any idea what to do here, for  
>  this?
>  >> This if from a first build in mock, which presumably pulled in the most
>  >> recent versions of all rpms into the chroot.
>  >
>  > Are you building for a rawhide target?  If so, mock is pulling in rpms
>  > from the most recent rawhide compose.  Add "--enablerepo=local" to get
>  > the very latest rpms from koji.
>  >
>  > There have been a fewl instances recently of the annobin plugin not
>  > matching gcc.  As far as I know, the issue has been resolved, so
>  > hopefully the extra mock flag will work for you.
>
>  I think the same gcc version from rawhide was pushed to f32,
>  while the corresponding annonbin release (9.27-1.fc32) is
>  still in updates-testing¹.  So it's possible this comes up
>  even for an f32 target. 
Yup, this is in stable:

root@arm mock]# ls -al default.cfg
lrwxrwxrwx. 1 root root 21 Jul 28 13:51 default.cfg → fedora-32-aarch64.cfg

>   In that case, enabling either the
>  updates-testing or the 'local' repo for the mock build
>  target would probably help.
>
>  Karma for the annobin release would likely be appreciated by
>  the maintainer(s) and anyone else who's running into this
>  problem, if you are able to test and confirm it fixes the
>  issue(s). :)
>
>  ¹ https://bodhi.fedoraproject.org/updates/FEDORA-2020-db7719f642
>    https://bugzilla.redhat.com/1862297 
Yup, that update fixes my mock builds, and is on its way to stable.

[Todd Zullinger]

Sam Varshavchik wrote:
>  Todd Zullinger writes:
> > Karma for the annobin release would likely be appreciated by
> > the maintainer(s) and anyone else who's running into this
> > problem, if you are able to test and confirm it fixes the
> > issue(s). :)
> > 
> > ¹ https://bodhi.fedoraproject.org/updates/FEDORA-2020-db7719f642
> >   https://bugzilla.redhat.com/1862297
>  
>  Yup, that update fixes my mock builds, and is on its way to stable. 
Excellent.  I'd love to think that's the last time we'll see
this problem, but it seems like it's just too easy for
annobin to get out of sync.

Who knows, maybe this will be the time that leads to a
better technical solution to prevent this problem (at the
very least, breakage like this shouldn't hit the stable
releases _ever_).

-- 
Todd

[Sam Varshavchik]

Todd Zullinger writes:

>  Sam Varshavchik wrote:
>  > Todd Zullinger writes:
>  >> Karma for the annobin release would likely be appreciated by
>  >> the maintainer(s) and anyone else who's running into this
>  >> problem, if you are able to test and confirm it fixes the
>  >> issue(s). :)
>  >>
>  >> ¹ https://bodhi.fedoraproject.org/updates/FEDORA-2020-db7719f642
>  >>   https://bugzilla.redhat.com/1862297
>  >
>  > Yup, that update fixes my mock builds, and is on its way to stable.
>
>  Excellent.  I'd love to think that's the last time we'll see
>  this problem, but it seems like it's just too easy for
>  annobin to get out of sync.
>
>  Who knows, maybe this will be the time that leads to a
>  better technical solution to prevent this problem (at the
>  very least, breakage like this shouldn't hit the stable
>  releases _ever_). 
Isn't this what RPM dependencies are for? This is the exact problems RPM  
dependencies are designed to solve.

If gcc has a dependency on a specific version of annobin, it should require  
it.

In case of weak dependencies, a declared conflict would, at least, uninstall  
the incompatible package, and give you an early alert.