On Mon, Dec 28, 2015 at 04:42:03PM +0100, gregor herrmann wrote: > > I can get this to happen with an unset TERM (on both amd64 and i386). > > > > Core was generated by `perl -Iblib/lib -Iblib/arch t/05flags.t'. > > Program terminated with signal SIGSEGV, Segmentation fault. > > Same here. Another interesting data point:
> # TERM=dumb prove --blib --verbose t/05flags.t > t/05flags.t .. > 1..4 > ok 1 - Explicit UTF-8 flag preserved > ok 2 - Explicit RAW flag preserved > > And here it hangs since 5 minutes using 100% of one core ... Interrupting it in gdb gives this backtrace: #0 0x00007ffff6cc8006 in compress_trie (n=0xa3f2d0) at driver-ti.c:148 #1 0x00007ffff6cc8713 in new_driver (tk=<optimized out>, term=<optimized out>) at driver-ti.c:281 #2 0x00007ffff6cc5b1d in termkey_init (tk=tk@entry=0x9fe330, term=0x7ec445 "dumb") at termkey.c:326 #3 0x00007ffff6cc5c14 in termkey_new (fd=<optimized out>, flags=136, flags@entry=128) at termkey.c:404 #4 0x00007ffff6ed3389 in XS_Term__TermKey_new (my_perl=<optimized out>, cv=<optimized out>) at lib/Term/TermKey.xs:400 #5 0x00000000004bd6ea in Perl_pp_entersub (my_perl=0x7d2010) at pp_hot.c:3270 #6 0x00000000004b6586 in Perl_runops_standard (my_perl=0x7d2010) at run.c:41 #7 0x0000000000443999 in S_run_body (oldscope=1, my_perl=0x7d2010) at perl.c:2453 #8 perl_run (my_perl=0x7d2010) at perl.c:2376 #9 0x000000000041cbbb in main (argc=4, argv=0x7fffffffe8f8, env=0x7fffffffe920) at perlmain.c:116 > Not sure what to do here. Is there an underlying bug or do we need to > set TERM to something nice or should we skip the test? Setting TERM=vt100 seems to not trigger either of these problems, so I guess we could try doing that in debian/rules. Cc'ing James, the libtermkey and unibilium maintainer. Any thoughts? Would you like separate bugs about these 'unset TERM' SIGSEGV and TERM=dumb busy loop issues? -- Niko Tyni nt...@debian.org