On 2008-08-10 at 01:03:37 [+0200], Andreas Färber <andreas.faerber@xxxxxx> wrote: > Am 09.08.2008 um 18:08 schrieb Andreas Färber: > > > Initializing BFS appears to work, i.e. `jam run ":<build>bfs_shell" > > --initialize my.image Haiku` writes something into an empty file and > > does not return an error. > > > > `jam run ":<build>bfs_shell" my.image` however crashes, this is what > > gdb made of it: > > > > (gdb) bt > > #0 0xfecd2e40 in countbytes () from /lib/libc.so.1 > > #1 0xfed14476 in _ndoprnt () from /lib/libc.so.1 > > #2 0xfed167ca in vprintf () from /lib/libc.so.1 > > #3 0x0808b642 in fssh_dprintf () > > #4 0x0808289b in bfs_mount () > > #5 0x08092398 in FSShell::fs_mount () > > #6 0x08092c73 in FSShell::_kern_mount () > > #7 0x0809661f in FSShell::standard_session () > > #8 0x08096a1d in main () > > With Rene's help I tracked this does to an INFORM macro in the BFS > kernel add-on. %Ld seems to be the offending part, %lld as suggested > by Rene makes it work. A patch is attached but untested on other > platforms. It should work. "%lld" is standard, "%Ld" is not. Applied in r26923. CU, Ingo