06 Feb, 2008
40 commits
-
I got the following oops during interface ifup. Unfortunately its not
easily reproducable so I cant say for sure that my fix fixes this
problem, but I am confident and I think its correct anyway:kernel BUG at /space/kvm/drivers/virtio/virtio_ring.c:234!
illegal operation: 0001 [#1] PREEMPT SMP
Modules linked in:
CPU: 0 Not tainted 2.6.24zlive-guest-07293-gf1ca151-dirty #91
Process swapper (pid: 0, task: 0000000000800938, ksp: 000000000084ddb8)
Krnl PSW : 0404300180000000 0000000000466374 (vring_disable_cb+0x30/0x34)
R:0 T:1 IO:0 EX:0 Key:0 M:1 W:0 P:0 AS:0 CC:3 PM:0 EA:3
Krnl GPRS: 0000000000000001 0000000000000001 0000000010003800 0000000000466344
000000000e980900 00000000008848b0 000000000084e748 0000000000000000
000000000087b300 0000000000001237 0000000000001237 000000000f85bdd8
000000000e980920 00000000001137c0 0000000000464754 000000000f85bdd8
Krnl Code: 0000000000466368: e3b0b0700004 lg %r11,112(%r11)
000000000046636e: 07fe bcr 15,%r14
0000000000466370: a7f40001 brc 15,466372
>0000000000466374: a7f4fff6 brc 15,466360
0000000000466378: eb7ff0500024 stmg %r7,%r15,80(%r15)
000000000046637e: a7f13e00 tmll %r15,15872
0000000000466382: b90400ef lgr %r14,%r15
0000000000466386: a7840001 brc 8,466388
Call Trace:
([] 0x201500f85c000)
[] vring_interrupt+0x72/0x88
[] kvm_extint_handler+0x34/0x44
[] do_extint+0xbc/0xf8
[] ext_no_vtime+0x16/0x1a
[] cpu_idle+0x216/0x238
([] cpu_idle+0x1f6/0x238)
[] rest_init+0xaa/0xb8
[] start_kernel+0x3fc/0x490
[] _stext+0x20/0x80
Kernel panic - not syncing: Fatal exception in interrupt
After looking at the code and the dump I think the following scenario
happened: Ifup was running on cpu2 and the interrupt arrived on cpu0.
Now virtnet_open on cpu 2 managed to execute napi_enable and disable_cb
but did not execute rx_schedule. Meanwhile on cpu 0 skb_recv_done was
called by vring_interrupt, executed netif_rx_schedule_prep, which
succeeded and therefore called disable_cb. This triggered the BUG_ON,
as interrupts were already disabled by cpu 2.I think the proper solution is to make the call to disable_cb depend on
the atomic update of NAPI_STATE_SCHED by using netif_rx_schedule_prep
in the same way as skb_recv_done.Signed-off-by: Christian Borntraeger
Acked-by: Rusty Russell
Signed-off-by: Jeff Garzik -
The PHY Lib now uses mutexes instead of spin_locks. ucc_geth
and gianfar both grab the locks in their mdio_reset functions,
so they need to use mutex_(un)lock instead. This was not caught
until someone tested it on an SMP system.Signed-off-by: Andy Fleming
Signed-off-by: Jeff Garzik -
Various registers need to be preserved before resetting the device.
Signed-off-by: Ayaz Abdulla
Signed-off-by: Jeff Garzik -
The driver needs to ack only the phy status bits that it is currently
handling and preserve the other bits for the other handlers. For
example, when reading/writing from the phy, it should not clear the link
change interrupt bit. This will cause a missing link change interrupt.Signed-off-by: Ayaz Abdulla
Signed-off-by: Jeff Garzik -
This patch fixes the issue where the transmitter and receiver must be
restarted when applying new changes to certain registers.Signed-off-by: Ayaz Abdulla
Signed-off-by: Jeff Garzik -
When CONFIG_PROC_FS is not set and CONFIG_PPPOL2TP is set,
we have the following warning in build:
drivers/net/pppol2tp.c: In function 'pppol2tp_init':
drivers/net/pppol2tp.c:2472: warning: label
'out_unregister_pppox_proto' defined but not usedThis patches fixes this warning by adding appropriate #ifdef.
Signed-off-by: Rami Rosen
Acked-by: James Chapman
Signed-off-by: David S. Miller -
This checks if the DMA address is bigger than what the controller can manage.
It will reallocate the buffers in the GFP_DMA zone in that case.Signed-off-by: Michael Buesch
Signed-off-by: John W. Linville -
This patch fixes two off-by-one errors resulting in array overflows
spotted by the Coverity checker.Signed-off-by: Adrian Bunk
Signed-off-by: John W. Linville -
This fixes four resource leakages.
In any error path we must deallocate the DMA frame slots we
previously allocated by request_slot().
This is done by storing the ring pointers before doing any ring
allocation and restoring the old pointers in case of an error.This patch by Michael Buesch has been ported to b43legacy.
Cc: Michael Buesch
Signed-off-by: Stefano Brivio
Signed-off-by: John W. Linville -
We must drop any packets we are not able to encrypt.
We must not send them unencrypted or with an all-zero-key (which
basically is the same as unencrypted, from a security point of view).This might only trigger shortly after resume before mac80211 reassociated
and reconfigured the keys.It is safe to drop these packets, as the association they belong to
is not guaranteed anymore anyway.
This is a security fix in the sense that it prevents information leakage.This patch by Michael Buesch has been ported to b43legacy.
Cc: Michael Buesch
Signed-off-by: Stefano Brivio
Signed-off-by: John W. Linville -
This patch makes suspend/resume work with the b43legacy driver.
We must not overwrite the MAC addresses in the init function, as this
would also overwrite the MAC on resume. With an all-zero MAC the device
firmware is not able to ACK any received packets anymore.
Fix this by moving the initializion stuff that must be done on init but
not on resume to the start function.
Also zero out filter_flags to make sure we don't have some flags
from a previous instance for a tiny timeframe until mac80211 reconfigures
them.This patch by Michael Buesch has been ported to b43legacy.
Cc: Michael Buesch
Signed-off-by: Stefano Brivio
Signed-off-by: John W. Linville -
Fix the crash reported below, which seems to happen on bcm4306 rev. 2 devices
only while using PIO:Oops: 0000 [#1] PREEMPT
Modules linked in: b43(F) rfkill(F) led_class(F) input_polldev(F) arc4 b43legacy mac80211 cfg80211 i915 drm snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device ohci1394 ieee1394 ssb pcmcia snd_intel8x0m ehci_hcd uhci_hcd evdevPid: 0, comm: swapper Tainted: GF (2.6.24st3 #2)
EIP: 0060:[] EFLAGS: 00010002 CPU: 0
EIP is at b43legacy_pio_handle_txstatus+0xbb/0x210 [b43legacy]
EAX: 0000049b EBX: f11f8044 ECX: 00000001 EDX: 00000000
ESI: f1ff8000 EDI: 00000000 EBP: f11f8040 ESP: c04f4ef4
DS: 007b ES: 007b FS: 0000 GS: 0000 SS: 0068
Process swapper (pid: 0, ti=c04f4000 task=c0488300 task.ti=c04b8000)
Stack: f90f2788 c05009f0 c0500900 000010f7 f1053823 c04f4f24 dfb8e800 00000003
f1368000 00000007 00000296 f90f1975 00001000 010c0800 01000000 00000007
f90f6391 f11f8000 00000082 c04f4f4a 00000000 00004fd0 10f70000 8c061000
Call Trace:
[] b43legacy_debugfs_log_txstat+0x48/0xb0 [b43legacy]
[] b43legacy_handle_hwtxstatus+0x75/0x80 [b43legacy]
[] b43legacy_pio_rx+0x201/0x280 [b43legacy]
[] b43legacy_interrupt_tasklet+0x2e3/0x870 [b43legacy]
[] tasklet_action+0x27/0x60
[] __do_softirq+0x54/0xb0
[] do_softirq+0x7b/0xe0
[] handle_level_irq+0x0/0x110
[] handle_level_irq+0x0/0x110
[] irq_exit+0x38/0x40
[] do_IRQ+0x83/0xd0
[] __update_rq_clock+0x4f/0x180
[] common_interrupt+0x23/0x28
[] wakeup_code+0x7b/0xde
[] acpi_processor_idle+0x24a/0x3c9
[] cpu_idle+0x47/0x80
[] start_kernel+0x205/0x290
[] unknown_bootoption+0x0/0x1f0
=======================
Code: 0f 00 00 81 fb ff 00 00 00 0f 87 36 01 00 00 8d 04 db 85 ff 8d 6c c6 40 8d 5d 04 0f 85 ef 00 00 00 fe 4e 0e 0f b7 46 0c 8b 53 04 4a 50 29 c8 83 e8 52 66 89 46 0c 8b 54 24 14 80 7a 0b 00 74
EIP: [] b43legacy_pio_handle_txstatus+0xbb/0x210 [b43legacy] SS:ESP 0068:c04f4ef4
Kernel panic - not syncing: Fatal exception in interruptSigned-off-by: Stefano Brivio
Signed-off-by: John W. Linville -
Generic HDLC now uses random_ether_addr() for generating MAC addresse
for Ethernet-alike interfaces.Signed-off-by: Krzysztof Halasa
Signed-off-by: Jeff Garzik -
Removes now unneeded struct hdlc_device_desc
Signed-off-by: Krzysztof Halasa
Signed-off-by: Jeff Garzik -
Fixes kernel panic in Frame Relay mode
Signed-off-by: Krzysztof Halasa
Signed-off-by: Jeff Garzik -
Just like our other drivers before we can switch ixgbe to
provide real-time packet/byte counters to the stack easily.Signed-off-by: Ayyappan Veeraiyan
Signed-off-by: Auke Kok
Signed-off-by: Jeff Garzik -
Several counters behave differently on 82598 causing them to display
incorrect values. Adjust the accounting so the reported numbers
make sense and do not double count or represent the wrong item.Signed-off-by: Ayyappan Veeraiyan
Signed-off-by: Auke Kok
Signed-off-by: Jeff Garzik -
We were not returning CHECKSUM_NONE in a lot of cases which is
wrong. Move common exit points in this function and error code
up before the actual work in this function.Signed-off-by: Ayyappan Veeraiyan
Signed-off-by: Auke Kok
Signed-off-by: Jeff Garzik -
A gap was left in the FW release/grab code in up/down path. Fix
it by making the release/grab code a function and calling it in
appropriate locations.Signed-off-by: Ayyappan Veeraiyan
Signed-off-by: Auke Kok
Signed-off-by: Jeff Garzik -
Signed-off-by: Ayyappan Veeraiyan
Signed-off-by: Auke Kok
Signed-off-by: Jeff Garzik -
The i82598 can support various media types but this ethtool
code only was coded for fiber just yet.Signed-off-by: Ayyappan Veeraiyan
Signed-off-by: Auke Kok
Signed-off-by: Jeff Garzik -
This ports Herbert Xu's "maybe_stop_tx" code and removes the tx_lock
which is not needed.Signed-off-by: Ayyappan Veeraiyan
Signed-off-by: Auke Kok
Signed-off-by: Jeff Garzik -
After testing we confirmed that the irq_sem can safely be
removed from ixgbe.Add strict state checking code to various ethtool parts to
properly protect against races between various driver reset
paths.Signed-off-by: Ayyappan Veeraiyan
Signed-off-by: Auke Kok
Signed-off-by: Jeff Garzik -
Changes in other networking paths uncovered a bug in the xircom_cb
driver which made the kernel spew lots of the following error messages:BUG eth1 code -5 qlen 0
It turned out that the driver returned -EIO when there was no
descriptor available for sending packets. It should return
NETDEV_TX_BUSY instead. This was discussed on the netdev list before,
see http://thread.gmane.org/gmane.linux.network/84603 .Signed-off-by: Erik Mouw
Signed-off-by: Jeff Garzik -
Add a check for the pci_register_driver() return value. Removed unused
variable pad_allocated.The aim of this patch is to remove the following warning messages:
drivers/net/tlan.c: In function 'tlan_probe':
drivers/net/tlan.c:486: warning: ignoring return value of 'pci_register_driver', declared with attribute warn_unused_resultSigned-off-by: Leonardo Potenza
Signed-off-by: Andrew Morton
Signed-off-by: Jeff Garzik -
The Marvell Orion system on chips have an integrated mv643xx MAC. On these
little endian ARM devices mv643xx will oops when checksum offload is
enabled. Swapping the byte order of the protocol and checksum solves this
problem.Signed-off-by: Byron Bradley
Cc: Dale Farnsworth
Cc: Manish Lachwani
Cc: Jeff Garzik
Cc: Al Viro
Signed-off-by: Andrew Morton
Signed-off-by: Jeff Garzik -
The Yukon FE chip has a ram buffer therefore it needs the alignment
restriction and hang check workarounds.Therefore:
* Autodetect the prescence/absence of ram buffer
* Rename the flag value to reflect this
* Use it consistently (ie don't reread register)Signed-off-by: Stephen Hemminger
Signed-off-by: Jeff Garzik -
WARNING: vmlinux.o(.text+0x25dca0): Section mismatch in reference from the function .veth_probe() to the function .init.text:.veth_probe_one()
Signed-off-by: Stephen Rothwell
Signed-off-by: Jeff Garzik -
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (21 commits)
[PKT_SCHED]: vlan tag match
[NET]: Add if_addrlabel.h to sanitized headers.
[NET] rtnetlink.c: remove no longer used functions
[ICMP]: Restore pskb_pull calls in receive function
[INET]: Fix accidentally broken inet(6)_hash_connect's port offset calculations.
[NET]: Remove further references to net-modules.txt
bluetooth rfcomm tty: destroy before tty_close()
bluetooth: blacklist another Broadcom BCM2035 device
drivers/bluetooth/btsdio.c: fix double-free
drivers/bluetooth/bpa10x.c: fix memleak
bluetooth: uninlining
bluetooth: hidp_process_hid_control remove unnecessary parameter dealing
tun: impossible to deassert IFF_ONE_QUEUE or IFF_NO_PI
hamradio: fix dmascc section mismatch
[SCTP]: Fix kernel panic while received AUTH chunk with BAD shared key identifier
[SCTP]: Fix kernel panic while received AUTH chunk while enabled auth
[IPV4]: Formatting fix for /proc/net/fib_trie.
[IPV6]: Fix sysctl compilation error.
[NET_SCHED]: Add #ifdef CONFIG_NET_EMATCH in net/sched/cls_flow.c (latest git broken build)
[IPV4]: Fix compile error building without CONFIG_FS_PROC
... -
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
RDMA/nes: Add a driver for NetEffect RNICs
IB/mthca: Return proper error codes from mthca_fmr_alloc()
IB: Avoid marking __devinitdata as const
IB/mlx4: Actually print out the driver version
IB/ib_mthca: Pre-link receive WQEs in Tavor mode
IB/mthca: Remove checks for srq->first_free < 0
IB/fmr_pool: Allocate page list for pool FMRs only when caching enabled
IB/srp: Retry stale connections
mlx4_core: Don't read reserved fields in mlx4_QUERY_ADAPTER()
IB/mthca: Don't read reserved fields in mthca_QUERY_ADAPTER()
IPoIB: Remove a misleading debug print
IPoIB: Handle bonding failover race for connected neighbours too
IB/mthca: Fix and simplify page size calculation in mthca_reg_phys_mr()
IB/ehca: Add PMA support
IB/ehca: Update sma_attr also in case of disruptive config change
IB/ehca: Prevent sending UD packets to QP0
IB/cm: Add interim support for routed paths
mlx4_core: Fix more section mismatches -
Modify the b43 driver to avoid deadlocking suspend and resume, which happens
as a result of attempting to unregister device objects locked by the PM core
during suspend/resume cycles. Also, make it use a suspend-safe method of
unregistering device object in the resume error path.Signed-off-by: Rafael J. Wysocki
Acked-by: Michael Buesch
Cc: Pavel Machek
Cc: "John W. Linville"
Cc: Alan Stern
Cc: Len Brown
Cc: Greg KH
Cc: Kay Sievers
Cc: Richard Purdie
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Replace latency.c use with pm_qos_params use.
Signed-off-by: mark gross
Cc: "John W. Linville"
Cc: Len Brown
Cc: Jaroslav Kysela
Cc: Takashi Iwai
Cc: Arjan van de Ven
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Checking if an address is a vmalloc address is done in a couple of places.
Define a common version in mm.h and replace the other checks.Again the include structures suck. The definition of VMALLOC_START and
VMALLOC_END is not available in vmalloc.h since highmem.c cannot be included
there.Signed-off-by: Christoph Lameter
Cc: Nick Piggin
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Fixing:
CHECK drivers/net/pcmcia/pcnet_cs.c
drivers/net/pcmcia/pcnet_cs.c:523:15: warning: symbol 'hw_info' shadows an earlier one
drivers/net/pcmcia/pcnet_cs.c:148:18: originally declared hereSigned-off-by: Richard Knutsson
Acked-by: Alan Cox
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Fixing:
CHECK drivers/net/pcmcia/fmvj18x_cs.c
drivers/net/pcmcia/fmvj18x_cs.c:1205:6: warning: symbol 'i' shadows an earlier one
drivers/net/pcmcia/fmvj18x_cs.c:1179:9: originally declared hereSigned-off-by: Richard Knutsson
Acked-by: Alan Cox
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Use 'max(x,y)' instead of 'x < y ? y : x'.
Signed-off-by: Richard Knutsson
Acked-by: Alan Cox
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Fixing:
CHECK drivers/net/pcmcia/axnet_cs.c
drivers/net/pcmcia/axnet_cs.c:994:5: warning: symbol 'ax_close' was not declared. Should it be static?
drivers/net/pcmcia/axnet_cs.c:1017:6: warning: symbol 'ei_tx_timeout' was not declared. Should it be static?Signed-off-by: Richard Knutsson
Acked-by: Alan Cox
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Fixing:
CHECK drivers/net/pcmcia/3c574_cs.c
drivers/net/pcmcia/3c574_cs.c:695:7: warning: symbol 'i' shadows an earlier one
drivers/net/pcmcia/3c574_cs.c:636:6: originally declared hereSigned-off-by: Richard Knutson
Acked-by: Alan Cox
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds
