Re: global-file-syms for clang

[Nick Clifton <nickc@redhat.com>]

Hi Grant,

> 1) Will the new annobin plugin/annocheck work with clang version16.0.6-1.el9)?

Yes.  Well assuming that you are using a version of the annobin clang
plugin that was compiled with clang 16.

If you are using RHEL-9 then there may be a slight issue in that annobin
version 12.46 has not been made available for CentOS-9/RHEL-9 yet.  You
are always free to compile it yourself, but if you want to use the official
CentOS-9/RHEL-9 annobin rpm then you will probably need to file a bug
report/enhancement request to get the new version.

As for annocheck, it is completely architecture and version agnostic.
It will run on any architecture and can check files built for any
architecture (not just the architecture type of the host on which it
is run) by any version of any compiler.


> Specifically, will the interaction between annocheck & the compile cause any issues/false positives or negatives?

No.  Or at least it should not, and if any issues or false positives
do arise then this is due to bug(s) in annocheck.  (In which case
please do report them...)


> 2) I noticed that annocheck has a considerable number of errors around gaps. This is after regenerating all the source (except standard libraries) with annobin plugins. Is this expected. For reference, LTO is enabled in these cases.

Err, maybe ?

It depends upon where the gaps are located.  If they are in those
standard libraries that have not been recompiled, then it would
make sense.  If the gaps are in code that has been recompiled and
therefore should have annobin data, then something has gone wrong.

LTO compilation does cause problems for annobin and annocheck, but
these days they should mostly be solved.  Having said that though,
I have mostly been working with LTO using GCC rather than Clang,
so there could still be some issues that need addressing.

If you have an example binary that you would care to upload I can
take a closer look and maybe find out some more.

Cheers
   Nick