On Sun, Jan 14, 2018 at 11:20:49PM +0200, Adrian Bunk wrote:
> Source: libgc
> Version: 1:7.4.2-8.1
> Severity: serious
>
> https://buildd.debian.org/status/fetch.php?pkg=libgc=armel=1%3A7.4.2-8.1=1515764936=0
>
> ...
>dh_makeshlibs -a
> dh_makeshlibs: Compatibility levels before 9 are deprecated (level 7 in use)
> dpkg-gensymbols: warning: some symbols or patterns disappeared in the symbols
> file: see diff output below
> dpkg-gensymbols: warning: no debian/symbols file used as basis for generating
> debian/libgc1c2/DEBIAN/symbols
> --- new_symbol_file (libgc1c2_1:7.4.2-8.1_armel)
> +++ dpkg-gensymbolsCP1Zv9 2018-01-12 13:48:50.806274887 +
> @@ -1,7 +1,7 @@
> libgc.so.1 libgc1c2 #MINVER#
> (arch=armel)AO_compare_double_and_swap_double_emulation@Base 1:7.4.2
> (arch=armel)AO_fetch_compare_and_swap_emulation@Base 1:7.4.2
> - (arch=armel)AO_locks@Base 1:7.4.2
> +#MISSING: 1:7.4.2-8.1# (arch=armel)AO_locks@Base 1:7.4.2
> (arch=armel)AO_pause@Base 1:7.4.2
> (arch=armel)AO_pt_lock@Base 1:7.4.2
> (arch=armel)AO_store_full_emulation@Base 1:7.4.2
> dh_makeshlibs: failing due to earlier errors
> debian/rules:11: recipe for target 'binary-arch' failed
> make: *** [binary-arch] Error 2
>
>
> This might be caused by the recent armv4t -> armv5te armel
> baseline change.
>
> I am completely lost regarding what is going on here,
> starting with the fact that this looks a Libatomic-ops symbol?
This one turned out to be pretty easy; libatomic-ops 7.6.0 made AO_locks
static and therefore a local symbol; from the upstream changelog:
> == [7.6.0] 2017-05-19 ==
> [...]
> * Hide AO_locks symbol
Thus the symbol should be dropped from the symbols file in libgc, which
shouldn't cause any problems, since it was an implementation detail that
was only ever exported on armel so nobody should be using it (but of
course, when has that stopped anyone...).
Regards,
James
