Henning Meier-Geinitz <[EMAIL PROTECTED]> wrote: Hi Henning,
> At least on first sight it looks like stack corruption for me. I would Same here. There's something fishy that needs to be ironed out :) > Running saned -d may lead to different behaviour because the original > error may not happen (e.g. because saned -d runs as root while from > (x)inetd its ran as user "saned" or "scanner" or whatever). Indeed; there's room for improvement in saned/net protocol :) > server (for whatever reason) and tries to free the device list. #8 > tries to free one of the strings e.g. sane_device.name. Now > *w->codec.w_string should be called (sanei_wire.c, 350). However this > happens instead: > > #5 0x40746d61 in free () from /lib/libc.so.6 > #6 0x4115f0bd in sanei_w_array () from /usr/lib/sane/libsane-net.so.1 > #7 0x4115e5f2 in sanei_config_read () from /usr/lib/sane/libsane-net.so.1 > > Either gdb is confused somehow, or sanei_config_read is called > mistakenly (or I miss something). Looks like the stack is in a weird state and gdb doesn't detect it. > I remember similar problems when the sanei_net/sanei_codec* code Hmm, last time it only produced a deadlock between the client and the server (both reading at the same time, waiting for data from the other side). > between xscanimage and sane-backends ran out of sync. However xsane > doesn't seem to use sanei at all. Well, it doesn't have to use it anyway. I'm still unsure as to why does scanimage use sanei :) JB. -- Julien BLACHE - Debian & GNU/Linux Developer - <[EMAIL PROTECTED]> Public key available on <http://www.jblache.org> - KeyID: F5D6 5169 GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]