Following up on this:

On Sun, May 26, 2024 at 11:42:56AM +0200, Florian Ernst wrote:
> On Sun, May 26, 2024 at 10:54:21AM +0200, Christoph Biedl wrote:
> > Étienne Mollier wrote...
> > > If that helps, the symbols strlcat and strlcpy have been added
> > > to the glibc 2.38, so I believe it should be safe to remove the
> > > following symbols without soname bump:
> > 
> > Just came here to say the same (and why does the BTS¹ not show Étienne's
> > message?). There might be a small chance this breaks packages due to
> > slight implementation differences - but that's what testing and
> > autopkgtests are for.
> 
> Thanks for your investigation. I suspected as much, but didn't yet get
> around to actually check on this any further, so thanks for the pointer.
> 
> I plan to deal with this on next Sunday, and I'll make sure to check
> building all reverse build deps against the updated package as well.

Only one of the reverse build depends actually made use of strlcat as
provided by libdumbnet:

- farpd
  The last build of 0.2-11.3+b1_amd64 from 2024-03-26 still shows this,
  but with the upload of 0.2-11.4 on 2024-05-30 it starts to reference
  the glibc version, so this seems fine now.

Two of the reverse build deps provide and use their own copy of
strlcat/strlcpy unless those are already available, but don't reference
libdumbnet's:

- scanssh
  The last build of 2.0-4.3+b1_amd64 from 2024-03-26 still shows the
  internal copies, whereas the build of 2.1.3.1-0.1_amd64 from
  2024-05-27 shows the glibc versions, so this seems fine.
- tcpreplay
  The last build of 4.4.4-1+b1_amd64 from 2024-03-29 still shows the
  internal copies, whereas a local rebuild references the glibc
  versions. This change is unrelated to libdumbnet dropping those
  symbols, though, and should happen seamlessly some time in the future
  with the next rebuild. There seems to be no need for such a rebuild as
  of now.

Five of the reverse build deps don't use strlcat/strlcpy at all, so they
are all fine:
- arpon
- daemonlogger
- daq
- firewalk
- libnet-libdnet-perl

One of the reverse build deps FTBFS and is overall in a rather sorry
state, so I didn't really investigate here:
- snort

All in all, to me it seems no further action is required here: Bastian's
upload of libdumbnet 1.18.0-1.1 solved this bug#1071321 here, and his
farpd 0.2-11.4 resolved the followup issue of the then-missing strlcat
for that package; all other reverse build deps are not affected or, as
is the case with snort, too broken anyway.

Cheers,
Flo

Attachment: signature.asc
Description: PGP signature

Reply via email to