Bug 5284
| Summary: | kernel oops and crash with rtl8110/8169 + vlan | ||
|---|---|---|---|
| Product: | Drivers | Reporter: | Bgs (bgs) |
| Component: | Network | Assignee: | Francois Romieu (romieu) |
| Status: | CLOSED CODE_FIX | ||
| Severity: | blocking | ||
| Priority: | P2 | ||
| Hardware: | i386 | ||
| OS: | Linux | ||
| Kernel Version: | 2.6.12.6/2.6.13.1/2.6.13.2 | Subsystem: | |
| Regression: | --- | Bisected commit-id: | |
Patch by Tommy Christensen <tommy.christensen@tpack.net> has been merged in mainline on 22/09/2005 and will appear in 2.6.14-rc3. -- Ueimor |
Most recent kernel where this bug did not occur: 2.6.13.2 Distribution: vanilla Hardware Environment: AMD Duron/VIA KT133/Realtek 8110/8169 Software Environment: Problem Description: Kernel oops and kernel panic occurs regularly. Steps to reproduce: Moderate LAN traffic through 8169 NIC with vlan. Kernel oops: Badness in __kfree_skb at net/core/skbuff.c:291 [<c03e2ff1>] __kfree_skb+0xb1/0x100 [<c03e658d>] sk_stream_kill_queues+0x4d/0x180 [<c042e14c>] tcp_v4_destroy_sock+0x1c/0x180 [<c041d4d7>] tcp_destroy_sock+0x47/0x120 [<c042561a>] tcp_rcv_state_process+0x51a/0x9c0 [<c042d1b9>] tcp_v4_do_rcv+0x99/0x120 [<c042d7d0>] tcp_v4_rcv+0x590/0x770 [<c0413c7f>] ip_local_deliver+0xbf/0x1c0 [<c0414260>] ip_local_deliver_finish+0x0/0x130 [<c04140d6>] ip_rcv+0x356/0x4e0 [<c0414390>] ip_rcv_finish+0x0/0x270 [<c01b1165>] pathrelse_and_restore+0x45/0x50 [<c03e88c7>] netif_receive_skb+0x127/0x190 [<c036eeac>] rtl8169_rx_vlan_skb+0x15c/0x1a0 [<c037134a>] rtl8169_rx_interrupt+0x18a/0x400 [<c0371a50>] pci_dma_sync_single_for_device+0x0/0x1 [<c03716cb>] rtl8169_interrupt+0x10b/0x150 [<c01385a0>] handle_IRQ_event+0x30/0x70 [<c0138682>] __do_IRQ+0xa2/0xe0 [<c0448056>] tcp_in_window+0x496/0x4a0 [<c0104d79>] do_IRQ+0x19/0x30 [<c0103292>] common_interrupt+0x1a/0x20 [<c0139ad6>] find_get_page+0x26/0x30 [<c0158373>] __find_get_block_slow+0x53/0x170 [<c044a1c1>] ipt_do_table+0x251/0x330 [<c0159190>] __find_get_block+0x80/0xe0 [<c0159227>] __getblk+0x37/0x70 [<c01b15f7>] search_by_key+0x97/0xd80 [<c03edbfc>] neigh_resolve_output+0xec/0x190 [<c0416c6e>] ip_finish_output+0xfe/0x220 [<c041730a>] ip_queue_xmit+0x36a/0x510 [<c0419300>] dst_output+0x0/0x30 [<c019aab9>] make_cpu_key+0x59/0x70 [<c019cfad>] inode2sd+0xfd/0x160 [<c012cc20>] autoremove_wake_function+0x0/0x60 [<c019aab9>] make_cpu_key+0x59/0x70 [<c019d326>] reiserfs_update_sd_size+0xa6/0x220 [<c0139b3f>] find_lock_page+0x1f/0x90 [<c019aab9>] make_cpu_key+0x59/0x70 [<c01a8424>] reiserfs_dirty_inode+0x74/0x90 [<c01a2365>] reiserfs_submit_file_region_for_write+0x215/0x280 [<c01a3342>] reiserfs_file_write+0x4f2/0x690 [<c0414260>] ip_local_deliver_finish+0x0/0x130 [<c04140d6>] ip_rcv+0x356/0x4e0 [<c0414390>] ip_rcv_finish+0x0/0x270 [<c03e88c7>] netif_receive_skb+0x127/0x190 [<c036eeac>] rtl8169_rx_vlan_skb+0x15c/0x1a0 [<c03713a9>] rtl8169_rx_interrupt+0x1e9/0x400 [<c03716b5>] rtl8169_interrupt+0xf5/0x150 [<c0156ad7>] vfs_write+0xa7/0x180 [<c0156c81>] sys_write+0x51/0x80 [<c01030d5>] syscall_call+0x7/0xb The rtl8169 is eth2 here and has vlans eth2.2 through eth2.10 VLAN setup is done by the following basic script: ============================= #!/bin/zsh VC=/sbin/vconfig ETH=eth2 for i in {2..10}; do ${VC} add ${ETH} $i ip ad add 192.168.$i.254 dev eth2.$i ip li set eth2.$i up ip ro add 192.168.$i.0/24 dev eth2.$i done ============================ I also experience dead loop/fix it messages from time to time. (Only ono one specific vlan, not all!)