1
0
Fork 0
alistair23-linux/net/mac80211
Linus Lüssing e980f9cb2f mac80211: mesh: fix discovery timer re-arming issue / crash
commit e2d4a80f93 upstream.

On a non-forwarding 802.11s link between two fairly busy
neighboring nodes (iperf with -P 16 at ~850MBit/s TCP;
1733.3 MBit/s VHT-MCS 9 80MHz short GI VHT-NSS 4), so with
frequent PREQ retries, usually after around 30-40 seconds the
following crash would occur:

[ 1110.822428] Unable to handle kernel read from unreadable memory at virtual address 00000000
[ 1110.830786] Mem abort info:
[ 1110.833573]   Exception class = IABT (current EL), IL = 32 bits
[ 1110.839494]   SET = 0, FnV = 0
[ 1110.842546]   EA = 0, S1PTW = 0
[ 1110.845678] user pgtable: 4k pages, 48-bit VAs, pgd = ffff800076386000
[ 1110.852204] [0000000000000000] *pgd=00000000f6322003, *pud=00000000f62de003, *pmd=0000000000000000
[ 1110.861167] Internal error: Oops: 86000004 [#1] PREEMPT SMP
[ 1110.866730] Modules linked in: pppoe ppp_async batman_adv ath10k_pci ath10k_core ath pppox ppp_generic nf_conntrack_ipv6 mac80211 iptable_nat ipt_REJECT ipt_MASQUERADE cfg80211 xt_time xt_tcpudp xt_state xt_nat xt_multiport xt_mark xt_mac xt_limit xt_conntrack xt_comment xt_TCPMSS xt_REDIRECT xt_LOG xt_FLOWOFFLOAD slhc nf_reject_ipv4 nf_nat_redirect nf_nat_masquerade_ipv4 nf_conntrack_ipv4 nf_nat_ipv4 nf_nat nf_log_ipv4 nf_flow_table_hw nf_flow_table nf_defrag_ipv6 nf_defrag_ipv4 nf_conntrack_rtcache nf_conntrack iptable_mangle iptable_filter ip_tables crc_ccitt compat nf_log_ipv6 nf_log_common ip6table_mangle ip6table_filter ip6_tables ip6t_REJECT x_tables nf_reject_ipv6 usb_storage xhci_plat_hcd xhci_pci xhci_hcd dwc3 usbcore usb_common
[ 1110.932190] Process swapper/3 (pid: 0, stack limit = 0xffff0000090c8000)
[ 1110.938884] CPU: 3 PID: 0 Comm: swapper/3 Not tainted 4.14.162 #0
[ 1110.944965] Hardware name: LS1043A RGW Board (DT)
[ 1110.949658] task: ffff8000787a81c0 task.stack: ffff0000090c8000
[ 1110.955568] PC is at 0x0
[ 1110.958097] LR is at call_timer_fn.isra.27+0x24/0x78
[ 1110.963055] pc : [<0000000000000000>] lr : [<ffff0000080ff29c>] pstate: 00400145
[ 1110.970440] sp : ffff00000801be10
[ 1110.973744] x29: ffff00000801be10 x28: ffff000008bf7018
[ 1110.979047] x27: ffff000008bf87c8 x26: ffff000008c160c0
[ 1110.984352] x25: 0000000000000000 x24: 0000000000000000
[ 1110.989657] x23: dead000000000200 x22: 0000000000000000
[ 1110.994959] x21: 0000000000000000 x20: 0000000000000101
[ 1111.000262] x19: ffff8000787a81c0 x18: 0000000000000000
[ 1111.005565] x17: ffff0000089167b0 x16: 0000000000000058
[ 1111.010868] x15: ffff0000089167b0 x14: 0000000000000000
[ 1111.016172] x13: ffff000008916788 x12: 0000000000000040
[ 1111.021475] x11: ffff80007fda9af0 x10: 0000000000000001
[ 1111.026777] x9 : ffff00000801bea0 x8 : 0000000000000004
[ 1111.032080] x7 : 0000000000000000 x6 : ffff80007fda9aa8
[ 1111.037383] x5 : ffff00000801bea0 x4 : 0000000000000010
[ 1111.042685] x3 : ffff00000801be98 x2 : 0000000000000614
[ 1111.047988] x1 : 0000000000000000 x0 : 0000000000000000
[ 1111.053290] Call trace:
[ 1111.055728] Exception stack(0xffff00000801bcd0 to 0xffff00000801be10)
[ 1111.062158] bcc0:                                   0000000000000000 0000000000000000
[ 1111.069978] bce0: 0000000000000614 ffff00000801be98 0000000000000010 ffff00000801bea0
[ 1111.077798] bd00: ffff80007fda9aa8 0000000000000000 0000000000000004 ffff00000801bea0
[ 1111.085618] bd20: 0000000000000001 ffff80007fda9af0 0000000000000040 ffff000008916788
[ 1111.093437] bd40: 0000000000000000 ffff0000089167b0 0000000000000058 ffff0000089167b0
[ 1111.101256] bd60: 0000000000000000 ffff8000787a81c0 0000000000000101 0000000000000000
[ 1111.109075] bd80: 0000000000000000 dead000000000200 0000000000000000 0000000000000000
[ 1111.116895] bda0: ffff000008c160c0 ffff000008bf87c8 ffff000008bf7018 ffff00000801be10
[ 1111.124715] bdc0: ffff0000080ff29c ffff00000801be10 0000000000000000 0000000000400145
[ 1111.132534] bde0: ffff8000787a81c0 ffff00000801bde8 0000ffffffffffff 000001029eb19be8
[ 1111.140353] be00: ffff00000801be10 0000000000000000
[ 1111.145220] [<          (null)>]           (null)
[ 1111.149917] [<ffff0000080ff77c>] run_timer_softirq+0x184/0x398
[ 1111.155741] [<ffff000008081938>] __do_softirq+0x100/0x1fc
[ 1111.161130] [<ffff0000080a2e28>] irq_exit+0x80/0xd8
[ 1111.166002] [<ffff0000080ea708>] __handle_domain_irq+0x88/0xb0
[ 1111.171825] [<ffff000008081678>] gic_handle_irq+0x68/0xb0
[ 1111.177213] Exception stack(0xffff0000090cbe30 to 0xffff0000090cbf70)
[ 1111.183642] be20:                                   0000000000000020 0000000000000000
[ 1111.191461] be40: 0000000000000001 0000000000000000 00008000771af000 0000000000000000
[ 1111.199281] be60: ffff000008c95180 0000000000000000 ffff000008c19360 ffff0000090cbef0
[ 1111.207101] be80: 0000000000000810 0000000000000400 0000000000000098 ffff000000000000
[ 1111.214920] bea0: 0000000000000001 ffff0000089167b0 0000000000000000 ffff0000089167b0
[ 1111.222740] bec0: 0000000000000000 ffff000008c198e8 ffff000008bf7018 ffff000008c19000
[ 1111.230559] bee0: 0000000000000000 0000000000000000 ffff8000787a81c0 ffff000008018000
[ 1111.238380] bf00: ffff00000801c000 ffff00000913ba34 ffff8000787a81c0 ffff0000090cbf70
[ 1111.246199] bf20: ffff0000080857cc ffff0000090cbf70 ffff0000080857d0 0000000000400145
[ 1111.254020] bf40: ffff000008018000 ffff00000801c000 ffffffffffffffff ffff0000080fa574
[ 1111.261838] bf60: ffff0000090cbf70 ffff0000080857d0
[ 1111.266706] [<ffff0000080832e8>] el1_irq+0xe8/0x18c
[ 1111.271576] [<ffff0000080857d0>] arch_cpu_idle+0x10/0x18
[ 1111.276880] [<ffff0000080d7de4>] do_idle+0xec/0x1b8
[ 1111.281748] [<ffff0000080d8020>] cpu_startup_entry+0x20/0x28
[ 1111.287399] [<ffff00000808f81c>] secondary_start_kernel+0x104/0x110
[ 1111.293662] Code: bad PC value
[ 1111.296710] ---[ end trace 555b6ca4363c3edd ]---
[ 1111.301318] Kernel panic - not syncing: Fatal exception in interrupt
[ 1111.307661] SMP: stopping secondary CPUs
[ 1111.311574] Kernel Offset: disabled
[ 1111.315053] CPU features: 0x0002000
[ 1111.318530] Memory Limit: none
[ 1111.321575] Rebooting in 3 seconds..

With some added debug output / delays we were able to push the crash from
the timer callback runner into the callback function and by that shedding
some light on which object holding the timer gets corrupted:

[  401.720899] Unable to handle kernel read from unreadable memory at virtual address 00000868
[...]
[  402.335836] [<ffff0000088fafa4>] _raw_spin_lock_bh+0x14/0x48
[  402.341548] [<ffff000000dbe684>] mesh_path_timer+0x10c/0x248 [mac80211]
[  402.348154] [<ffff0000080ff29c>] call_timer_fn.isra.27+0x24/0x78
[  402.354150] [<ffff0000080ff77c>] run_timer_softirq+0x184/0x398
[  402.359974] [<ffff000008081938>] __do_softirq+0x100/0x1fc
[  402.365362] [<ffff0000080a2e28>] irq_exit+0x80/0xd8
[  402.370231] [<ffff0000080ea708>] __handle_domain_irq+0x88/0xb0
[  402.376053] [<ffff000008081678>] gic_handle_irq+0x68/0xb0

The issue happens due to the following sequence of events:

1) mesh_path_start_discovery():
-> spin_unlock_bh(&mpath->state_lock) before mesh_path_sel_frame_tx()

2) mesh_path_free_rcu()
-> del_timer_sync(&mpath->timer)
   [...]
-> kfree_rcu(mpath)

3) mesh_path_start_discovery():
-> mod_timer(&mpath->timer, ...)
   [...]
-> rcu_read_unlock()

4) mesh_path_free_rcu()'s kfree_rcu():
-> kfree(mpath)

5) mesh_path_timer() starts after timeout, using freed mpath object

So a use-after-free issue due to a timer re-arming bug caused by an
early spin-unlocking.

This patch fixes this issue by re-checking if mpath is about to be
free'd and if so bails out of re-arming the timer.

Cc: stable@vger.kernel.org
Fixes: 050ac52cbe ("mac80211: code for on-demand Hybrid Wireless Mesh Protocol")
Cc: Simon Wunderlich <sw@simonwunderlich.de>
Signed-off-by: Linus Lüssing <ll@simonwunderlich.de>
Link: https://lore.kernel.org/r/20200522170413.14973-1-linus.luessing@c0d3.blue
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-06-03 08:21:30 +02:00
..
Kconfig Merge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6 2019-07-08 20:57:08 -07:00
Makefile mac80211: minstrel: merge with minstrel_ht, always enable VHT support 2018-10-11 16:01:01 +02:00
aead_api.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
aead_api.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
aes_ccm.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
aes_cmac.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
aes_cmac.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
aes_gcm.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
aes_gmac.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
aes_gmac.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
agg-rx.c mac80211: add missing null return check from call to ieee80211_get_sband 2019-07-31 10:51:17 +02:00
agg-tx.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
cfg.c cfg80211: Fix radar event during another phy CAC 2020-02-05 21:22:46 +00:00
chan.c treewide: Add SPDX license identifier for missed files 2019-05-21 10:50:45 +02:00
debug.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
debugfs.c mac80211: AMPDU handling for rekeys with Extended Key ID 2019-07-26 13:29:10 +02:00
debugfs.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
debugfs_key.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2019-06-22 08:59:24 -04:00
debugfs_key.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
debugfs_netdev.c mac80211: fix txq null pointer dereference 2019-10-01 17:56:19 +02:00
debugfs_netdev.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
debugfs_sta.c mac80211: drop data frames without key on encrypted links 2020-04-01 11:02:01 +02:00
debugfs_sta.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
driver-ops.c mac80211: don't warn about CW params when not using them 2019-07-20 21:40:32 +02:00
driver-ops.h mac80211: pass the vif to cancel_remain_on_channel 2019-07-26 13:08:28 +02:00
ethtool.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 432 2019-06-05 17:37:16 +02:00
fils_aead.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
fils_aead.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
he.c mac80211: fix possible NULL pointerderef in obss pd code 2019-08-21 10:58:32 +02:00
ht.c mac80211: add support for the ADDBA extension element 2019-07-29 16:40:22 +02:00
ibss.c mac80211: IBSS: send deauth when expiring inactive STAs 2019-09-11 09:33:29 +02:00
ieee80211_i.h mac80211: IBSS: send deauth when expiring inactive STAs 2019-09-11 09:33:29 +02:00
iface.c mac80211: allow drivers to set max MTU 2019-09-11 09:33:29 +02:00
key.c mac80211: drop data frames without key on encrypted links 2020-04-01 11:02:01 +02:00
key.h mac80211: clear crypto tx tailroom counter upon keys enable 2019-09-11 09:33:28 +02:00
led.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
led.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
main.c mac80211: populate debugfs only after cfg80211 init 2020-04-29 16:33:18 +02:00
mesh.c mac80211: fix channel switch trigger from unknown mesh peer 2020-05-02 08:48:58 +02:00
mesh.h mac80211: implement HE support for mesh 2019-07-26 16:14:12 +02:00
mesh_hwmp.c mac80211: mesh: fix discovery timer re-arming issue / crash 2020-06-03 08:21:30 +02:00
mesh_pathtbl.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
mesh_plink.c mac80211: implement HE support for mesh 2019-07-26 16:14:12 +02:00
mesh_ps.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
mesh_sync.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
michael.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
michael.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
mlme.c mac80211: add ieee80211_is_any_nullfunc() 2020-05-10 10:31:32 +02:00
ocb.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
offchannel.c mac80211: pass the vif to cancel_remain_on_channel 2019-07-26 13:08:28 +02:00
pm.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
rate.c mac80211: populate debugfs only after cfg80211 init 2020-04-29 16:33:18 +02:00
rate.h mac80211: populate debugfs only after cfg80211 init 2020-04-29 16:33:18 +02:00
rc80211_minstrel.c mac80211: call rate_control_send_low() internally 2019-06-14 14:17:37 +02:00
rc80211_minstrel.h mac80211: minstrel_ht: improve rate probing for devices with static fallback 2019-08-21 11:10:13 +02:00
rc80211_minstrel_debugfs.c mac80211: rc80211_minstrel: remove variance / stddev calculation 2018-10-11 16:01:05 +02:00
rc80211_minstrel_ht.c mac80211: populate debugfs only after cfg80211 init 2020-04-29 16:33:18 +02:00
rc80211_minstrel_ht.h mac80211: minstrel_ht: improve rate probing for devices with static fallback 2019-08-21 11:10:13 +02:00
rc80211_minstrel_ht_debugfs.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
rx.c mac80211: add ieee80211_is_any_nullfunc() 2020-05-10 10:31:32 +02:00
scan.c mac80211: fix scan when operating on DFS channels in ETSI domains 2019-10-07 22:10:50 +02:00
spectmgmt.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
sta_info.c mac80211: sta_info: Add lockdep condition for RCU list usage 2020-05-10 10:31:28 +02:00
sta_info.h mac80211: drop data frames without key on encrypted links 2020-04-01 11:02:01 +02:00
status.c mac80211: add ieee80211_is_any_nullfunc() 2020-05-10 10:31:32 +02:00
tdls.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2019-06-17 15:55:34 -07:00
tkip.c mac80211: Fix TKIP replay protection immediately after key setup 2020-02-05 21:22:46 +00:00
tkip.h Merge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6 2019-07-08 20:57:08 -07:00
trace.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
trace.h mac80211: pass the vif to cancel_remain_on_channel 2019-07-26 13:08:28 +02:00
trace_msg.h mac80211: Increase MAX_MSG_LEN 2019-03-29 11:20:36 +01:00
tx.c mac80211: add ieee80211_is_any_nullfunc() 2020-05-10 10:31:32 +02:00
util.c mac80211: fix wrong 160/80+80 MHz setting 2020-03-05 16:43:41 +01:00
vht.c mac80211: vht: add support VHT EXT NSS BW in parsing VHT 2019-09-11 09:13:03 +02:00
wep.c Merge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6 2019-07-08 20:57:08 -07:00
wep.h Merge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6 2019-07-08 20:57:08 -07:00
wme.c treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
wme.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00
wpa.c mac80211: add IEEE80211_KEY_FLAG_GENERATE_MMIE to ieee80211_key_flags 2019-07-26 16:14:12 +02:00
wpa.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 500 2019-06-19 17:09:55 +02:00