ISI crash when disconnecting N900 from USB

ofonod[6748]: Aborting (signal 11)
ofonod[6748]: ++++++++ backtrace ++++++++
ofonod[6748]: [0]: /lib64/libpthread.so.0() [0x38fa60f440]
ofonod[6748]: [1]: ./src/ofonod() [0x4151d8]
ofonod[6748]: [2]: ./src/ofonod() [0x41546a]
ofonod[6748]: [3]: ./src/ofonod() [0x4112b9]
ofonod[6748]: [4]: /lib64/libc.so.6(tdestroy+0x3d) [0x38f9edfc7d] ofonod[6748]: [5]: ./src/ofonod() [0x4116cb]
ofonod[6748]: [6]: ./src/ofonod() [0x416221]
ofonod[6748]: [7]: ./src/ofonod() [0x45db29]
ofonod[6748]: [8]: ./src/ofonod() [0x457268]
ofonod[6748]: [9]: ./src/ofonod(ofono_modem_set_powered+0x10a) [0x457f5a] ofonod[6748]: [10]: ./src/ofonod() [0x420dd1]
ofonod[6748]: [11]: ./src/ofonod() [0x4108e0]
ofonod[6748]: [12]: /lib64/libglib-2.0.so.0(g_main_context_dispatch+0x1f2) [0x38fba3bd02] ofonod[6748]: [13]: /lib64/libglib-2.0.so.0() [0x38fba3fae8]
ofonod[6748]: [14]: /lib64/libglib-2.0.so.0(g_main_loop_run+0x195) [0x38fba3fff5] ofonod[6748]: [15]: ./src/ofonod() [0x4561f0]
ofonod[6748]: [16]: /lib64/libc.so.6(__libc_start_main+0xfd) [0x38f9e1ec5d] ofonod[6748]: [17]: ./src/ofonod() [0x40b7e9]
ofonod[6748]: +++++++++++++++++++++++++++

-------- backtrace --------
[0]: isi_registration_status() [network-registration.c:271] [1]: set_auto_resp_cb() [network-registration.c:535] [2]: g_isi_cleanup_req() [client.c:241]
[3]: g_isi_client_destroy() [client.c:310]
[4]: isi_netreg_remove() [network-registration.c:974] [5]: netreg_remove() [network.c:1724]
[6]: flush_atoms() [modem.c:350]
[7]: ofono_modem_set_powered() [modem.c:794]
[8]: report_powered() [isigen.c:77]
[9]: g_pn_nl_link() [netlink.c:203]
[10]: main() [main.c:240]
[11]: _start() [audio-settings.c:0]
---------------------------

It looks like there was an ongoing registration attempt when you disconnected the USB cable. The isimodem driver cleans up that pending registration request, and in doing so, calls the atom's registration callback indicating a failure.

However, in that callback, the netreg atom makes a new query for current registration state, and that is what is crashing the isimodem driver. The driver shouldn't try to initiate any more ISI requests, as it is in the process of being removed.

I'll push a fix shortly.

Cheers,
Aki