Bug 330085 – net-wireless/wpa_supplicant-0.7.2-r2 constant segfault

• Status: RESOLVED UPSTREAM
• Product: Gentoo Linux

[Description Daniel Gryniewicz (RETIRED) 2010-07-27 14:51:12 UTC]

wpa-supplicnat has a constant segfault on (apparently) multilib systems.  It happens every 5 minutes or so.  Patch attached that fixes it for me and others.

[Comment 1 Daniel Gryniewicz (RETIRED) 2010-07-27 14:53:03 UTC]

Created attachment 240329 [details]
Empty lists when deleting containers

The problem is that containers for a list are deleted without emptying the lists.  Later, the list elements try to access their containers and crash.

[Comment 2 Ihar Hrachyshka 2010-07-27 15:14:15 UTC]

Have you sent it upstream?

[Comment 3 Jeroen Roovers (RETIRED) 2010-07-27 17:22:29 UTC]

Why don't you assign your own bugs?

[Comment 4 Daniel Gryniewicz (RETIRED) 2010-07-27 20:05:24 UTC]

I haven't reported it, no.  I don't need Yet Another Bugzilla Account.  Unfortunately, I don't have my backtraces anymore.  I'll generate a new one.

[Comment 5 Daniel Gryniewicz (RETIRED) 2010-07-27 20:49:45 UTC]

Here's a backtrace:

Program received signal SIGSEGV, Segmentation fault.
0x0805ae27 in pmksa_candidate_add (sm=0x82dc260, bssid=0x82dbf6c "", prio=1000, preauth=1)
    at ../src/rsn_supp/preauth.c:374
374			if (os_memcmp(pos->bssid, bssid, ETH_ALEN) == 0) {
(gdb) bt
#0  0x0805ae27 in pmksa_candidate_add (sm=0x82dc260, bssid=0x82dbf6c "", prio=1000, preauth=1)
    at ../src/rsn_supp/preauth.c:374
#1  0x0805b062 in rsn_preauth_scan_result (sm=0x82dc260, bssid=0x82dbf6c "", 
    ssid=0x82dbf9c "", rsn=0x82dbfe0 "0\024\001") at ../src/rsn_supp/preauth.c:465
#2  0x08085d12 in wpa_supplicant_rsn_preauth_scan_results (wpa_s=<value optimized out>, 
    data=<value optimized out>) at events.c:801
#3  wpa_supplicant_event_scan_results (wpa_s=<value optimized out>, 
    data=<value optimized out>) at events.c:941
#4  0x08087369 in wpa_supplicant_event (ctx=0x82dc018, event=EVENT_SCAN_RESULTS, data=0x0)
    at events.c:1627
#5  0x08090ce8 in wpa_driver_wext_event_wireless (drv=<value optimized out>, 
    data=<value optimized out>, len=<value optimized out>) at ../src/drivers/driver_wext.c:509
#6  0x080914ea in wpa_driver_wext_event_rtm_newlink (ctx=0x82db8a0, ifi=0xff8bf4f0, 
    buf=0xff8bf500 "\n", len=32) at ../src/drivers/driver_wext.c:655
#7  0x0809b452 in netlink_receive_link (sock=6, eloop_ctx=0x82db940, sock_ctx=0x0)
    at ../src/drivers/netlink.c:36
#8  netlink_receive (sock=6, eloop_ctx=0x82db940, sock_ctx=0x0) at ../src/drivers/netlink.c:71
#9  0x08053ab4 in eloop_sock_table_dispatch (table=<value optimized out>, 
    fds=<value optimized out>) at ../src/utils/eloop.c:216
#10 0x08053fb3 in eloop_run () at ../src/utils/eloop.c:548
#11 0x0808155e in wpa_supplicant_run (global=0x82d9460) at wpa_supplicant.c:2353
#12 0x0808a2e6 in main (argc=2, argv=0xff8c1744) at main.c:274

[Comment 6 Daniel Gryniewicz (RETIRED) 2010-07-27 20:50:35 UTC]

I don't maintain wpa_supplicant.  This bug should not be assigned to me.

[Comment 7 Michael Weber (RETIRED) 2010-07-28 20:24:08 UTC]

(In reply to comment #3)
> Why don't you assign your own bugs?

(In reply to comment #6)
> I don't maintain wpa_supplicant. This bug should not be assigned to me.

Jer was speaking of assigning it to the maintainers of wpa_supplicant, not cluttering bugwranglers queue (since you have the power to do so).

[Comment 8 Bjarke Istrup Pedersen (RETIRED) 2010-08-14 10:18:07 UTC]

I have reported it upstream (see the URL for this bug).

Also, I have added the patch and bumped wpa_supplicant to 0.7.2-r3 .

[Comment 9 Bjarke Istrup Pedersen (RETIRED) 2010-08-14 20:29:50 UTC]

Also, this has just been included upstream, so 0.7.3 will be fixed :)