u...@debian.org (Aaron M. Ucko) writes: > Ferenc Wágner <wf...@niif.hu> writes: > >> I wanted to, but hit some problem... Maybe failed to access a >> porterbox. Anyway, that error does not make any sense, knet_handle_new >> is the first thing the tests call, so I assumed it was a toolchain or >> binutils issue which would resolve itself eventually... > > Ah, ok. Sounds like whatever setup this test uses to check symbol > availability managed to misfire, then.
The test looks for the exported API function names in the text (T) section, but the ppc64 nm reports them in the data (D) section: $ nm -B -D libknet.so.1.0.0 0000000000000000 A LIBKNET [...] 000000000003ec40 D knet_handle_new [...] Is this a known ppc64 peculiarity or a binutils bug? -- Thanks, Feri
