asix: fix BUG in receive path when lowering MTU
Subsystems (3)
cover asix: fix BUG in receive path when lowering MTU
Hi Oliver,
there has been reported a bug in the asix usbnet driver, occuring at the time a packet arrives larger than MTU size (minus some overhead):
BUG: unable to handle kernel paging request at 0000004000000001 IP: [<ffffffff8126f65b>] skb_release_head_state+0x2d/0xd2 PGD 0 Oops: 0002 [#1] SMP Modules linked in: xt_connmark xt_tcpudp xt_set xt_multiport xt_addrtype ip_set_hash_ip nf_nat_pptp nf_nat_proto_gre nf_nat_irc nf_nat_ftp nf_conntrack_pptp nf_conntrack_proto_gre nf_conntrack_irc nf_conntrack_ftp af_packet ebtable_filter ebtables ip6table_ips ip6table_mangle ip6table_nat nf_nat_ipv6 iptable_ips iptable_mangle iptable_nat nf_nat_ipv4 nf_nat xt_NFLOG xt_condition(O) xt_logmark xt_confirmed xt_owner ip6t_REJECT ipt_REJECT xt_state ip_set red2 ip_scheduler red nfnetlink_log nf_conntrack_ipv6 nf_defrag_ipv6 ip6table_filter ip6table_raw nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack iptable_filter iptable_raw xt_CT nf_conntrack_netlink nfnetlink nf_conntrack ip6_tables ip_tables x_tables ipv6 loop asix libphy usbnet mii acpi_cpufreq mperf crc32c_intel coretemp ehci_pci e1000e(O ) ehci_hcd xhci_hcd sg sr_mod evdev rtc_cmos pcspkr i2c_i801 ppdev parport_pc e1000 parport microcode button cdrom fan thermal sd_mod processor thermal_sys hwmon pata_acpi ata_generic edd ata_piix libata scsi_mod CPU 0 Pid: 0, comm: swapper/0 Tainted: G O 3.8.6-1.gae5c790-smp64 #1 System manufacturer System Product Name/P8H61/USB3 RIP: 0010:[<ffffffff8126f65b>] [<ffffffff8126f65b>] skb_release_head_state+0x2d/0xd2 RSP: 0018:ffff88011ec03e20 EFLAGS: 00010202 RAX: 0000000000000000 RBX: ffff880117869000 RCX: 0000000000000086 RDX: ffff8801178690f4 RSI: 0000000000000046 RDI: 0000004000000001 RBP: 0000000000000000 R08: 0000000000000046 R09: ffffffff815515ec R10: 0000000000000000 R11: 0000000000000078 R12: ffff88011923e8d0 R13: ffff880118d5a600 R14: ffff88011923e780 R15: 000000000000000a FS: 0000000000000000(0000) GS:ffff88011ec00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000004000000001 CR3: 0000000001430000 CR4: 00000000000407f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process swapper/0 (pid: 0, threadinfo ffffffff81424000, task ffffffff81438410) Stack: ffff880117869000 ffffffff8126f86d ffff880117869000 ffffffff8126f8ad ffff8801181aa050 ffffffffa00b4200 ffff8801189584aa ffff880118f5f048 ffff88011923e780 ffff880118d5a600 ffff88011923e8d0 0000000000000001 Call Trace: <IRQ> [<ffffffff8126f86d>] ? skb_release_all+0x9/0x1e [<ffffffff8126f8ad>] ? __kfree_skb+0x9/0x6f [<ffffffffa00b4200>] ? asix_rx_fixup_internal+0xff/0x1ae [asix] [<ffffffffa00fb3dc>] ? usbnet_bh+0x4f/0x226 [usbnet] [<ffffffff812056fe>] ? add_interrupt_randomness+0x39/0x157 [<ffffffff81039ff6>] ? tasklet_action+0x74/0xbd [<ffffffff8103a742>] ? __do_softirq+0x9d/0x15f [<ffffffff8130099c>] ? call_softirq+0x1c/0x30 [<ffffffff81009fd7>] ? do_softirq+0x3f/0x79 [<ffffffff8103a538>] ? irq_exit+0x43/0xb0 [<ffffffff81009865>] ? do_IRQ+0x98/0xae [<ffffffff812536d2>] ? disable_cpuidle+0xb/0xb [<ffffffff812fa1ad>] ? common_interrupt+0x6d/0x6d <EOI> [<ffffffff8104f21b>] ? __hrtimer_start_range_ns+0x271/0x284 [<ffffffff811ad3f2>] ? __setup_broadcast_timer+0x2d/0x2d [<ffffffff8100dd4e>] ? sched_clock+0x5/0x8 [<ffffffff81253c47>] ? cpuidle_wrap_enter+0x39/0x71 [<ffffffff81253c40>] ? cpuidle_wrap_enter+0x32/0x71 [<ffffffff812536ee>] ? cpuidle_enter_state+0xa/0x33 [<ffffffff81253b70>] ? cpuidle_idle_call+0x9e/0xcc [<ffffffff8100ef8f>] ? cpu_idle+0x61/0xa9 [<ffffffff81499120>] ? early_idt_handlers+0x120/0x120 [<ffffffff81499c5f>] ? start_kernel+0x372/0x37e [<ffffffff81499705>] ? repair_env_string+0x5d/0x5d [<ffffffff814993e1>] ? x86_64_start_kernel+0x102/0x10f Code: 89 fb 48 8b 7f 58 48 85 ff 74 17 40 f6 c7 01 75 09 48 83 e7 fe e8 e5 ef 00 00 48 c7 43 58 00 00 00 00 48 8b 7b 60 48 85 ff 74 0f <f0> ff 0f 0f 94 c0 84 c0 74 05 e8 7b 16 07 00 48 83 bb 80 00 00 RIP [<ffffffff8126f65b>] skb_release_head_state+0x2d/0xd2 RSP <ffff88011ec03e20> CR2: 0000004000000001 ---[ end trace 9c13f42cc4b59d9c ]---
It is easily reproducable by setting an MTU of 512 e. g. and sending something like
$ ping -s 1472 -c 1 -M do $SELF
from another box.
I can see that the overlong packet case is detected correctly in asix_rx_fixup_internal(), but in this case the asix_rx_fix_info data is not reset properly, so that a subsequent call frees that skb again.
The attached patch fixes that for me by resetting rx->ax_skb and rx->size on error. Please check.
Thank you.
/Holger
PATCH 1/1 asix: fix BUG in receive path when lowering MTU
There is bug in the receive path of the asix driver at the time a packet is received larger than MTU size and DF bit set:
BUG: unable to handle kernel paging request at 0000004000000001 IP: [<ffffffff8126f65b>] skb_release_head_state+0x2d/0xd2 ... Call Trace: <IRQ> [<ffffffff8126f86d>] ? skb_release_all+0x9/0x1e [<ffffffff8126f8ad>] ? __kfree_skb+0x9/0x6f [<ffffffffa00b4200>] ? asix_rx_fixup_internal+0xff/0x1ae [asix] [<ffffffffa00fb3dc>] ? usbnet_bh+0x4f/0x226 [usbnet] ...
It is easily reproducable by setting an MTU of 512 e. g. and sending something like
ping -s 1472 -c 1 -M do $SELF
from another box.
And this is because the rx->ax_skb is freed on error, but rx->ax_skb is not reset, and the size is not reset to zero in this case.
And since the skb is added again to the usbnet->done skb queue it is accessing already freed memory, resulting in the BUG when freeing a 2nd time. I therefore think the value 0x0000004000000001 show in the trace is more or less random data.
drivers/net/usb/asix_common.c +3 -0
@@ -100,6 +100,9 @@netdev_err (dev->net,"asix_rx_fixup() Bad RX Length %d\n" , rx->size);kfree_skb (rx->ax_skb); + rx->ax_skb =NULL ; + rx->size =0U ; +return 0 ; }
From: [email protected]
Date: Fri, 03 May 2013 12:02:20 +0200
Quoted 24 lines
There is bug in the receive path of the asix driver at the time a packet is received larger than MTU size and DF bit set:
BUG: unable to handle kernel paging request at 0000004000000001 IP: [<ffffffff8126f65b>] skb_release_head_state+0x2d/0xd2 ... Call Trace: <IRQ> [<ffffffff8126f86d>] ? skb_release_all+0x9/0x1e [<ffffffff8126f8ad>] ? __kfree_skb+0x9/0x6f [<ffffffffa00b4200>] ? asix_rx_fixup_internal+0xff/0x1ae [asix] [<ffffffffa00fb3dc>] ? usbnet_bh+0x4f/0x226 [usbnet] ...It is easily reproducable by setting an MTU of 512 e. g. and sending something like
ping -s 1472 -c 1 -M do $SELF
from another box.
And this is because the
rx->ax_skbis freed on error, butrx->ax_skbis not reset, and the size is not reset to zero in this case.
Quoted 6 lines
And since the skb is added again to the
usbnet->doneskb queue it is accessing already freed memory, resulting in the BUG when freeing a 2nd time. I therefore think the value0x0000004000000001show in the trace is more or less random data.Signed-off-by Holger Eitzenberger <[email protected]>
Applied.