[SCSI] qla2xxx: Addition of shutdown callback handler.

This patch adds a shutdown handler to qla2xxx driver to make sure that all
DMA and firmware activities are stopped, and any associated driver resources
are released. The need for this handler arose when executing kexec in specific
environments caused the data of the 2nd kernel to be corrupted, due to DMA
activities.

Signed-off-by: Madhuranath Iyengar <Madhu.Iyengar@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla2xxx: Initialize the vport_slock spinlock.

Commit feafb7b1714cf599a6d0fed45801ab3f66046cbd neglected to initialize
the spinlock.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: Madhuranath Iyengar <Madhu.Iyengar@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla2xxx: Remove scsi_cmnd->serial_number from debug traces

This patch cleans up any printk or debug tracing of the the
serial_number field in the qla2xxx driver.

Signed-off-by: Madhuranath Iyengar <Madhu.Iyengar@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla2xxx: Add flash read/update support using BSG interface

Signed-off-by: Harish Zunjarrao <harish.zunjarrao@qlogic.com>
Signed-off-by: Madhuranath Iyengar <Madhu.Iyengar@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla2xxx: Group CS_RESET return status with other link level event statuses.

Currently when we receive a CS_RESET as a response for a SCSI command the
driver will return DID_TRANSPORT_DISRUPTED back to the SCSI mid-layer.  There
are certain circumstances where this could cause the mid-layer to exhaust all of
its retries if the FC port goes away for a short time.  This will result in
commands being prematurly failed.  Moving the CS_RESET return code to be
grouped with other link level events will cause the FC transport layer to block
that target's queue thus preventing the premature exhaustion of retries.

Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: Madhuranath Iyengar <Madhu.Iyengar@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla2xxx: Change del_timer_sync() to del_timer() in qla2x00_ctx_sp_free().

Using del_timer_sync() in the qla2x00_ctx_sp_free() function may cause a kernel
panic as it is not interrupt context safe and qla2x00_ctx_sp_free() may be
called from a softirq context.  Changing the call from del_timer_sync() to
del_timer() will make the function interrupt context safe.

Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: Madhuranath Iyengar <Madhu.Iyengar@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla2xxx: Add module parameter to enable/disable GFF_ID device type check.

Add the module parameter ql2xgffidenable to disable/enable the use of the
GFF_ID name server command to prevent non FCP SCSI devices from being added to
the driver's internal fc_port database.

Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: Madhuranath Iyengar <Madhu.Iyengar@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla2xxx: Remove port down retry count.

This patch removes the use of the port down retry counter as a mechanism to
update a fcport state. The internal driver counter is a residual carry-over
from pre-FC-transport aware driver inteaction. The ql2xport_down_retry module
parameter and NVRAM set ha->port_down_retry_count remain in order to seed the
fc-host's default dev-loss-tmo.

Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: Madhuranath Iyengar <Madhu.Iyengar@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla2xxx: locking problem in qla2x00_init_rings()

IRQs are already disabled here so we don't need to disable them again.
But more importantly, the spin_lock_irqsave() overwrites "flags" and
that breaks things when we want to re-enable the IRQs when we call
spin_unlock_irqrestore(&ha->hardware_lock, flags);

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Madhuranath Iyengar <Madhu.Iyengar@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] sr: fix sr_drive_status handling when initialization required

An sr device that reports sense data with SK/ASC/ASCQ of 2/4/2 (Not ready,
Logical unit not ready, Initializing command required) will be handled
in sr_drive_status as (2/4/!1) and assumed to be a 'format in progress'
which returns CDS_DISC_OK.  The drive will not be made ready in this case.

Prior to 210ba1d1724f5c4ed87a2ab1a21ca861a915f734 sr_drive_status would
have returned CDS_TRAY_OPEN and this results in an START_STOP_UNIT to
close the tray, which resolves the initialization requirement.

This patch adds handling for SK/ASC/ASCQ of 2/4/2 where it will return
CDS_TRAY_OPEN as a means of triggering a START_STOP_UNIT.

This issue is seen on the IBM POWER platform when using a file-backed,
virtual optical device.  The device does not support media queries
through the Get Event Status Notification command which could otherwise
trigger a START_STOP_UNIT call to close an open tray.

Signed-off-by: Robert Jennings <rcj@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] fcoe: Fix broken NPIV with correction to MAC validation

A previous patch attempted to validate the destination
MAC address of a FCoE frame by checking that MAC
address against the received port's MAC address. The
implementation seems fine on the surface, but any
VN_Ports added using the NPIV feature will have their
own MAC addresses and these MACs were not being checked,
which prevented any NPIV VN_Ports from receiving frames.

In other words, the following patch has broken NPIV.

519e5135e2537c9dbc1cbcc0891b0a936ff5dcd2
 [SCSI] fcoe: adds src and dest mac address
              checking for fcoe frames

Part of the offending patch is correct, but the part
that broke NPIV was attempting to satisfy FC-BB-5
section D.5, 2.1-

(discard frames that) "contain a destination MAC
address/destination N_Port_ID pair that was not
assigned by an FCF to one of the VN_Ports on the ENode"

The language does _not_ say to compare the destination
FC-MAP/destination N_Port_ID, but instead to compare
the destination MAC address/destination N_Port_ID.

>From the FC-BB-5 specification,

"A properly formed FPMA is one in which the 24 most
significant bits equal the Fabric’s FC-MAP value and
the least significant 24 bits equal the N_Port_ID
assigned to the VN_Port by the FCF."

This means that we need to compare the FC Frame's
destination FCID against the embedded FCID in the
destination MAC address. This patch checks the lower
24 bits of the destination MAC address against
destination FCID in the Fibre Channel frame.

For MAC validation the first line of defense is the
hardware MAC filtering. Each VN_Port will have a
unicast MAC addresses added to the hardware's
filtering table. The Ethernet driver should drop any
MACs not destined for a programmed MAC. This patch
adds a second line of defense that very specfically
compares an element in the FC frame against an element
in the Ethernet header, which is appropriate for the
FCoE layer.

Many alternative approaches were considered, including
a LLD callback from libfc. The second most reasonable
approach seemed to be walking the list of NPIV ports
and check each of their MAC addresses against the
destination MAC address of the received frame. The
problem with this approach was that it is likely that
performance would suffer with the more NPIV ports added
to the system since every received frame would need to
walk this list, comparing each entry's MAC.

Signed-off-by: Robert Love <robert.w.love@intel.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] libfcoe: VN2VN connection setup causing stack memory corruption.

Fix: When FIP frame is received, function fcoe_ctlr_vn_recv calls function
fcoe_ctlr_vn_parse which does memset for addr (&buf.rdata) which leads to
memory corruption. Code was trying to treat "buf" as struct but it was defined
as union. Fix is to change from union to struct for "buf" in function fcoe_ctlr_vn_recv.

Technical Details: N/A

Signed-off-by: Kiran Patil <kiran.patil@intel.com>
Acked-by: Joe Eykholt <jeykholt@cisco.com>
Signed-off-by: Robert Love <robert.w.love@intel.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] libfc: Do not let disc work cancel itself

When number of NPIV ports created are greater than the xids
allocated per pool -- for eg., creating 255 NPIV ports on a
system with nr_cpu_ids of 32, with each pool containing 128
xids -- and then generating a link event - for eg.,
shutdown/no shutdown -- on the switch port causes the hang
with the following stack trace.

Call Trace:
schedule_timeout+0x19d/0x230
wait_for_common+0xc0/0x170
__cancel_work_timer+0xcf/0x1b0
fc_disc_stop+0x16/0x30 [libfc]
fc_lport_reset_locked+0x47/0x90 [libfc]
fc_lport_enter_reset+0x67/0xe0 [libfc]
fc_lport_disc_callback+0xbc/0xe0 [libfc]
fc_disc_done+0xa8/0xf0 [libfc]
fc_disc_timeout+0x29/0x40 [libfc]
run_workqueue+0xb8/0x140
worker_thread+0x96/0x110
kthread+0x96/0xa0
child_rip+0xa/0x20

Fix is to not cancel the disc_work if discovery is already
stopped, thus allowing lport state machine to restart and try
discovery again.

Signed-off-by: Bhanu Prakash Gollapudi <bprakash@broadcom.com>
Acked-by: Joe Eykholt <jeykholt@cisco.com>
Signed-off-by: Robert Love <robert.w.love@intel.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] libfc: possible race could panic system due to NULL fsp->cmd

It is unlikely but in case if it hits then it would cause panic
due to null cmd ptr, so far only one instance seen recently with
ESX though this was introduced long ago with this commit:-

commit c1ecb90a66c5afc7cc5c9349f9c3714eef4a5cfb
Author: Chris Leech <christopher.leech@intel.com>
Date:   Thu Dec 10 09:59:26 2009 -0800
[SCSI] libfc: reduce hold time on SCSI host lock

Currently fsp->cmd is set to NULL w/o scsi_queue_lock before
dequeuing from scsi_pkt_queue and that could cause NULL
fsp->cmd in fc_fcp_cleanup_each_cmd for cmd completing
with fsp->cmd = NULL after fc_fcp_cleanup_each_cmd taken
reference. No need to set fsp->cmd to NULL as this is also
protected by fc_fcp_lock_pkt(), for above race the
fc_fcp_lock_pkt() in fc_fcp_cleanup_each_cmd() will fail
as that cmd is  already done.

Mike mentioned same issue at
http://www.open-fcoe.org/pipermail/devel/2010-September/010533.html

Similarly moved sc_cmd->SCp.ptr = NULL under scsi_queue_lock so
that scsi abort error handler won't abort on completed cmds.

Signed-off-by: Vasu Dev <vasu.dev@intel.com>
Signed-off-by: Robert Love <robert.w.love@intel.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] fcoe: set default FIP mode as FIP_MODE_FABRIC

Since sometimes current FIP_MODE_AUTO mode falls back to non-FIP
mode while DCB link still getting ready in fabric mode with
its peer switch, it falls back after few libfc flogi retries
and that is not we want while working with FIP enabled
switches in FABRIC mode, therefore sets default as FIP_MODE_FABRIC
as discussed and agreed before in this mail thread
http://www.open-fcoe.org/pipermail/devel/2010-August/010511.html

Signed-off-by: Vasu Dev <vasu.dev@intel.com>
Signed-off-by: Robert Love <robert.w.love@intel.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] libfc: adds flogi retry in case DID is zero in RJT

Sometimes switch in NPV mode rejects flogi request with DID
zero and in that case flogi is not tried again and port
remains offline, so this patch validates DID for non zero
along with only ACC response to allow flogi retry
for RJT with DID=0 also succeed FLOGI in next try.

Signed-off-by: Vasu Dev <vasu.dev@intel.com>
Signed-off-by: Robert Love <robert.w.love@intel.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] libfc: use DID_TRANSPORT_DISRUPTED while lport not ready

This is per Mile Christie feedback since in this case IO
could get retried for tape devices and therefore DID_REQUEUE
cannot be used, more details in this thread.

http://marc.info/?l=linux-scsi&m=127970522630136&w=2

Signed-off-by: Vasu Dev <vasu.dev@intel.com>
Signed-off-by: Robert Love <robert.w.love@intel.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] libfc: fix setting of rport dev loss

There does not seem to be a reason why libfc adds a 5
second delay to the user requested value for the dev loss
tmo. There also does not seem to be a reason to allow
setting it to 0 (or really close).

This patch removes the extra 5 sec delay, and for 0 it
sets it to 1 like other fc drivers. We should actually
be able to set it to 0 since the queue_delayed_work API
will just call queue_work, but other drivers set it to 1 in
that case.

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Robert Love <robert.w.love@intel.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] gdth: integer overflow in ioctl

gdth_ioctl_alloc() takes the size variable as an int.
copy_from_user() takes the size variable as an unsigned long.
gen.data_len and gen.sense_len are unsigned longs.
On x86_64 longs are 64 bit and ints are 32 bit.

We could pass in a very large number and the allocation would truncate
the size to 32 bits and allocate a small buffer.  Then when we do the
copy_from_user(), it would result in a memory corruption.

CC: stable@kernel.org
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] Fix race when removing SCSI devices

Removing SCSI devices through
echo 1 > /sys/bus/scsi/devices/ ... /delete

while the FC transport class removes the SCSI target can lead to an
oops:

Unable to handle kernel pointer dereference at virtual kernel address 00000000b6815000
Oops: 0011 [#1] PREEMPT SMP DEBUG_PAGEALLOC
Modules linked in: sunrpc qeth_l3 binfmt_misc dm_multipath scsi_dh dm_mod ipv6 qeth ccwgroup [last unloaded: scsi_wait_scan]
CPU: 1 Not tainted 2.6.35.5-45.x.20100924-s390xdefault #1
Process fc_wq_0 (pid: 861, task: 00000000b7331240, ksp: 00000000b735bac0)
Krnl PSW : 0704200180000000 00000000003ff6e4 (__scsi_remove_device+0x24/0xd0)
           R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:0 CC:2 PM:0 EA:3
Krnl GPRS: 0000000000000001 0000000000000000 00000000b6815000 00000000bc24a8c0
           00000000003ff7c8 000000000056dbb8 0000000000000002 0000000000835d80
           ffffffff00000000 0000000000001000 00000000b6815000 00000000bc24a7f0
           00000000b68151a0 00000000b6815000 00000000b735bc20 00000000b735bbf8
Krnl Code: 00000000003ff6d6: a7840001            brc 8,3ff6d8
           00000000003ff6da: a7fbffd8            aghi %r15,-40
           00000000003ff6de: e3e0f0980024        stg %r14,152(%r15)
          >00000000003ff6e4: e31021200004        lg %r1,288(%r2)
           00000000003ff6ea: a71f0000            cghi    %r1,0
           00000000003ff6ee: a7a40011            brc 10,3ff710
           00000000003ff6f2: a7390003            lghi    %r3,3
           00000000003ff6f6: c0e5ffffc8b1        brasl %r14,3f8858
Call Trace:
([<0000000000001000>] 0x1000)
 [<00000000003ff7d2>] scsi_remove_device+0x42/0x54
 [<00000000003ff8ba>] __scsi_remove_target+0xca/0xfc
 [<00000000003ff99a>] __remove_child+0x3a/0x48
 [<00000000003e3246>] device_for_each_child+0x72/0xbc
 [<00000000003ff93a>] scsi_remove_target+0x4e/0x74
 [<0000000000406586>] fc_rport_final_delete+0xb2/0x23c
 [<000000000015d080>] worker_thread+0x200/0x344
 [<000000000016330c>] kthread+0xa0/0xa8
 [<0000000000106c1a>] kernel_thread_starter+0x6/0xc
 [<0000000000106c14>] kernel_thread_starter+0x0/0xc
INFO: lockdep is turned off.
Last Breaking-Event-Address:
 [<00000000003ff7cc>] scsi_remove_device+0x3c/0x54

The function __scsi_remove_target iterates through the SCSI devices on
the host, but it drops the host_lock before calling
scsi_remove_device. When the SCSI device is deleted from another
thread, the pointer to the SCSI device in scsi_remove_device can
become invalid. Fix this by getting a reference to the SCSI device
before dropping the host_lock to keep the SCSI device alive for the
call to scsi_remove_device.

Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Cc: Stable Tree <stable@kernel.org>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] sd: Export effective protection mode in sysfs

Create a sysfs entry that reports the negotiated DIX/DIF protection mode
for a SCSI disk. This depends on the protection type the disk is
formatted with as well as the protection capabilities advertised by the
controller.

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: Update driver version to 5.02.00-k4

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: grab hardware_lock in eh_abort before accessing srb

grab hardware_lock in eh_abort before accessing srb to avoid
race between command completion and get refcount on srb.

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: remove unwanted check for bad spd

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: update AER support for ISP82XX

* Cleanup qla4xxx_pci_mmio_enabled():
  don't want to return PCI_ERS_NEED_RESET if firmware hung.
  IDC will take care of it.
* Request irq after initialize_adapter() in qla82xx_error_recovery().
* Return all active commands from qla4xxx_pci_error_detected().
* Cleanup ql4_def.h

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: Clear the rom lock if the firmware died while holding it.

There is a possibility that the firmware dies while the rom
lock is held. The only way to recover from this condition is
to forcefully unlock.

Signed-off-by: Shyam Sundar <shyam.sundar@qlogic.com>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: use CRB Register for Request Queue in-pointer

Switching from doorbell mechanism to CRB register based

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Shyam Sundar <shyam.sundar@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: dump mailbox registers on System Error

Signed-off-by: Karen Higgins <karen.higgins@qlogic.com>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: Add support for 8130/8131 AENs.

AEN 8130 Corresponds to an event representing the insertion (detection)
of a transceiver. It also reports the type of the SFP+.
AEN 8131 corresponds to the removal of a transceiver.

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Shyam Sundar <shyam.sundar@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: Reset seconds_since_last_heartbeat correctly.

The seconds_since_last_heartbeat should be checked for consecutive
heartbeat checks. Currently it could happen that it gets set to
max (2 seconds) for non-consecutive heartbeat checks.

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Lalit Chandivade <lalit.chandivade@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: On firmware hang do not wait for the outstanding commands to complete

Signed-off-by: Nilesh Javali <nilesh.javali@qlogic.com>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: free_irqs on failed initialize_adapter

Since interrupts are registered in start_firmware(load_risc) for 82xx,
free them if init_firmware fails.

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Karen Higgins <karen.higgins@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: correct data type of sense_len in qla4xxx_status_cont_entry

change data type of sense_len from uint8_t to uint16_t

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: remove "ha->retry_reset_ha_cnt" from wait_for_hba_online

remove "ha->retry_reset_ha_cnt" from wait_for_hba_online as its
initialize to zero at driver init time so it could always return
QLA_ERROR from wait_for_hba_online() without waiting for hba to
come online.

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: honor return status of qla4xxx_hw_reset

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] qla4xxx: Trivial cleanup

* cleanup function qla4xxx_recovery_timeout
- No need to wakeup dpc thread from function
qla4xxx_recovery_timeout() as we are not doing anything
in do_dpc() thread when wakeup from
qla4xxx_recovery_timeout()

* cleanup function qla4xxx_wait_for_hba_online
- Remove hard coded value from qla4xxx_wait_for_hba_online().

* cleanup function qla4xxx_start_firmware_from_flash
- display seconds

* cleanup function  qla4_8xxx_load_risc
- Remove redundant code.

* cleanup function qla4xxx_get_firmware_status
- update debug statement

* cleanup function qla4_8xxx_try_start_fw
- update return status

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: Ravi Anand <ravi.anand@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] Fix regressions in scsi_internal_device_block

Deleting a SCSI device on a blocked fc_remote_port (before
fast_io_fail_tmo fires) results in a hanging thread:

  STACK:
  0 schedule+1108 [0x5cac48]
  1 schedule_timeout+528 [0x5cb7fc]
  2 wait_for_common+266 [0x5ca6be]
  3 blk_execute_rq+160 [0x354054]
  4 scsi_execute+324 [0x3b7ef4]
  5 scsi_execute_req+162 [0x3b80ca]
  6 sd_sync_cache+138 [0x3cf662]
  7 sd_shutdown+138 [0x3cf91a]
  8 sd_remove+112 [0x3cfe4c]
  9 __device_release_driver+124 [0x3a08b8]
10 device_release_driver+60 [0x3a0a5c]
11 bus_remove_device+266 [0x39fa76]
12 device_del+340 [0x39d818]
13 __scsi_remove_device+204 [0x3bcc48]
14 scsi_remove_device+66 [0x3bcc8e]
15 sysfs_schedule_callback_work+50 [0x260d66]
16 worker_thread+622 [0x162326]
17 kthread+160 [0x1680b0]
18 kernel_thread_starter+6 [0x10aaea]

During the delete, the SCSI device is in moved to SDEV_CANCEL.  When
the FC transport class later calls scsi_target_unblock, this has no
effect, since scsi_internal_device_unblock ignores SCSI devics in this
state.

It looks like all these are regressions caused by:
5c10e63c943b4c67561ddc6bf61e01d4141f881f
[SCSI] limit state transitions in scsi_internal_device_unblock

Fix by rejecting offline and cancel in the state transition.

Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
[jejb: Original patch by Christof Schmitt, modified by Mike Christie]
Cc: Stable Tree <stable@kernel.org>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] zfcp: Use correct length for FCP_RSP_INFO

Use the FCP_RSP_INFO length to correctly skip the FCP_RSP_INFO field.

Reviewed-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] zfcp: Call get_device on port before calling put_device

zfcp_unit_release calls put_device on the port. Ensure that get_device
has been called before possibly triggering the release function
through put_device or device_unregister.

Reviewed-by: Swen Schillig <swen@vnet.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

[SCSI] zfcp: Fix adapter activation on link down

If an exchange config is executed while the local link is down, the
request succeeds but the returned data is incomplete.  Proceeding with
the adapter activation is leading to an unpredictable behaviour (e.g.
kernel panic) caused by invalid values.  In such a scenario the
recommended ERP is to retry the action and wait for a link up event.
If the issue persists the activation has to fail.

Signed-off-by: Swen Schillig <swen@vnet.ibm.com>
Sigend-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>

Merge branch 'for-next' of git://git./linux/kernel/git/jikos/trivial

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/trivial: (39 commits)
  Update broken web addresses in arch directory.
  Update broken web addresses in the kernel.
  Revert "drivers/usb: Remove unnecessary return's from void functions" for musb gadget
  Revert "Fix typo: configuation => configuration" partially
  ida: document IDA_BITMAP_LONGS calculation
  ext2: fix a typo on comment in ext2/inode.c
  drivers/scsi: Remove unnecessary casts of private_data
  drivers/s390: Remove unnecessary casts of private_data
  net/sunrpc/rpc_pipe.c: Remove unnecessary casts of private_data
  drivers/infiniband: Remove unnecessary casts of private_data
  drivers/gpu/drm: Remove unnecessary casts of private_data
  kernel/pm_qos_params.c: Remove unnecessary casts of private_data
  fs/ecryptfs: Remove unnecessary casts of private_data
  fs/seq_file.c: Remove unnecessary casts of private_data
  arm: uengine.c: remove C99 comments
  arm: scoop.c: remove C99 comments
  Fix typo configue => configure in comments
  Fix typo: configuation => configuration
  Fix typo interrest[ing|ed] => interest[ing|ed]
  Fix various typos of valid in comments
  ...

Fix up trivial conflicts in:
drivers/char/ipmi/ipmi_si_intf.c
drivers/usb/gadget/rndis.c
net/irda/irnet/irnet_ppp.c

Merge branch 'devel' of git://git./linux/kernel/git/mchehab/edac

* 'devel' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/edac: (25 commits)
  i7300_edac: Properly initialize per-csrow memory size
  V4L/DVB: i7300_edac: better initialize page counts
  MAINTAINERS: Add maintainer for i7300-edac driver
  i7300-edac: CodingStyle cleanup
  i7300_edac: Improve comments
  i7300_edac: Cleanup: reorganize the file contents
  i7300_edac: Properly detect channel on CE errors
  i7300_edac: enrich FBD error info for corrected errors
  i7300_edac: enrich FBD error info for fatal errors
  i7300_edac: pre-allocate a buffer used to prepare err messages
  i7300_edac: Fix MTR x4/x8 detection logic
  i7300_edac: Make the debug messages coherent with the others
  i7300_edac: Cleanup: remove get_error_info logic
  i7300_edac: Add a code to cleanup error registers
  i7300_edac: Add support for reporting FBD errors
  i7300_edac: Properly detect the type of error correction
  i7300_edac: Detect if the device is on single mode
  i7300_edac: Adds detection for enhanced scrub mode on x8
  i7300_edac: Clear the error bit after reading
  i7300_edac: Add error detection code for global errors
  ...

Merge branch 'for-linus' of git://git./linux/kernel/git/penberg/slab-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/penberg/slab-2.6: (27 commits)
  SLUB: Fix memory hotplug with !NUMA
  slub: Move functions to reduce #ifdefs
  slub: Enable sysfs support for !CONFIG_SLUB_DEBUG
  SLUB: Optimize slab_free() debug check
  slub: Move NUMA-related functions under CONFIG_NUMA
  slub: Add lock release annotation
  slub: Fix signedness warnings
  slub: extract common code to remove objects from partial list without locking
  SLUB: Pass active and inactive redzone flags instead of boolean to debug functions
  slub: reduce differences between SMP and NUMA
  Revert "Slub: UP bandaid"
  percpu: clear memory allocated with the km allocator
  percpu: use percpu allocator on UP too
  percpu: reduce PCPU_MIN_UNIT_SIZE to 32k
  vmalloc: pcpu_get/free_vm_areas() aren't needed on UP
  SLUB: Fix merged slab cache names
  Slub: UP bandaid
  slub: fix SLUB_RESILIENCY_TEST for dynamic kmalloc caches
  slub: Fix up missing kmalloc_cache -> kmem_cache_node case for memoryhotplug
  slub: Add dummy functions for the !SLUB_DEBUG case
  ...

Merge branch 'kvm-updates/2.6.37' of git://git./virt/kvm/kvm

* 'kvm-updates/2.6.37' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (321 commits)
  KVM: Drop CONFIG_DMAR dependency around kvm_iommu_map_pages
  KVM: Fix signature of kvm_iommu_map_pages stub
  KVM: MCE: Send SRAR SIGBUS directly
  KVM: MCE: Add MCG_SER_P into KVM_MCE_CAP_SUPPORTED
  KVM: fix typo in copyright notice
  KVM: Disable interrupts around get_kernel_ns()
  KVM: MMU: Avoid sign extension in mmu_alloc_direct_roots() pae root address
  KVM: MMU: move access code parsing to FNAME(walk_addr) function
  KVM: MMU: audit: check whether have unsync sps after root sync
  KVM: MMU: audit: introduce audit_printk to cleanup audit code
  KVM: MMU: audit: unregister audit tracepoints before module unloaded
  KVM: MMU: audit: fix vcpu's spte walking
  KVM: MMU: set access bit for direct mapping
  KVM: MMU: cleanup for error mask set while walk guest page table
  KVM: MMU: update 'root_hpa' out of loop in PAE shadow path
  KVM: x86 emulator: Eliminate compilation warning in x86_decode_insn()
  KVM: x86: Fix constant type in kvm_get_time_scale
  KVM: VMX: Add AX to list of registers clobbered by guest switch
  KVM guest: Move a printk that's using the clock before it's ready
  KVM: x86: TSC catchup mode
  ...

Merge branch 'i2c-for-linus' of git://git./linux/kernel/git/jdelvare/staging

* 'i2c-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelvare/staging:
  i2c-viapro: Don't log nacks
  i2c/pca954x: Remove __devinit and __devexit from probe and remove functions
  MAINTAINERS: Add maintainer for PCA9541 I2C bus master selector driver
  i2c/mux: Driver for PCA9541 I2C Master Selector
  i2c: Optimize function i2c_detect()
  i2c: Discard warning message on device instantiation from user-space
  i2c-amd8111: Add proper error handling
  i2c: Change to new flag variable
  i2c: Remove unneeded inclusions of <linux/i2c-id.h>
  i2c: Let i2c_parent_is_i2c_adapter return the parent adapter
  i2c: Simplify i2c_parent_is_i2c_adapter
  i2c-pca-platform: Change device name of request_irq
  i2c: Fix Kconfig dependencies

Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/hid

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid: (47 commits)
  HID: fix mismerge in hid-lg
  HID: hidraw: fix window in hidraw_release
  HID: hid-sony: override usbhid_output_raw_report for Sixaxis
  HID: add absolute axis resolution calculation
  HID: force feedback support for Logitech RumblePad gamepad
  HID: support STmicroelectronics and Sitronix with hid-stantuml driver
  HID: magicmouse: Adjust major / minor axes to scale
  HID: Fix for problems with eGalax/DWAV multi-touch-screen
  HID: waltop: add support for Waltop Slim Tablet 12.1 inch
  HID: add NOGET quirk for AXIS 295 Video Surveillance Joystick
  HID: usbhid: remove unused hiddev_driver
  HID: magicmouse: Use hid-input parsing rather than bypassing it
  HID: trivial formatting fix
  HID: Add support for Logitech Speed Force Wireless gaming wheel
  HID: don't Send Feature Reports on Interrupt Endpoint
  HID: 3m: Adjust major / minor axes to scale
  HID: 3m: Correct touchscreen emulation
  HID: 3m: Convert to MT slots
  HID: 3m: Output proper orientation range
  HID: 3m: Adjust to sequential MT HID protocol
  ...

Merge git://git./linux/kernel/git/herbert/crypto-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: Makefile - replace the use of <module>-objs with <module>-y
  crypto: hifn_795x - use cancel_delayed_work_sync()
  crypto: talitos - sparse check endian fixes
  crypto: talitos - fix checkpatch warning
  crypto: talitos - fix warning: 'alg' may be used uninitialized in this function
  crypto: cryptd - Adding the AEAD interface type support to cryptd
  crypto: n2_crypto - Niagara2 driver needs to depend upon CRYPTO_DES
  crypto: Kconfig - update broken web addresses
  crypto: omap-sham - Adjust DMA parameters
  crypto: fips - FIPS requires algorithm self-tests
  crypto: omap-aes - OMAP2/3 AES hw accelerator driver
  crypto: updates to enable omap aes
  padata: add missing __percpu markup in include/linux/padata.h
  MAINTAINERS: Add maintainer entries for padata/pcrypt

Merge branch 'master' into for-linus

Conflicts:
include/linux/percpu.h
mm/percpu.c

i2c-viapro: Don't log nacks

Transactions not acked can happen every now and then, in particular
during device detection, and various transaction types can be used for
this purpose. So stop logging this event, except when debugging is
enabled. This is what other similar drivers (e.g. i2c-i801 or
i2c-piix4) do.

Signed-off-by: Jean Delvare <khali@linux-fr.org>

i2c/pca954x: Remove __devinit and __devexit from probe and remove functions

The underlying I2C adapter may or may not be present when this driver
gets initialized, and may disappear later, so there is no safe time at
which the probe and remove functions can be discarded.

Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>

MAINTAINERS: Add maintainer for PCA9541 I2C bus master selector driver

Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>

i2c/mux: Driver for PCA9541 I2C Master Selector

This patch adds support for PCA9541, an I2C Bus Master Selector.
The driver is modeled as single channel I2C Multiplexer to be able to utilize
the I2C multiplexer framework.

Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com>
Reviewed-by: Tom Grennan <tom.grennan@ericsson.com>
Acked-by: Jean Delvare <khali@linux-fr.org>

i2c: Optimize function i2c_detect()

Check the class flags before allocating the temporary i2c_client
structure, to avoid allocating it when we don't need it.

Also optimize the inner loop a bit.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Jacob Pan <jacob.jun.pan@linux.intel.com>

i2c: Discard warning message on device instantiation from user-space

The "new_device" sysfs interface has been there for quite some time
now, nobody complained about it so it must be good enough. Time to
remove the warning and call it stable.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Acked-by: Michael Lawnick <ml.lawnick@gmx.de>

i2c-amd8111: Add proper error handling

The functions the functions amd_ec_wait_write and amd_ec_wait_read have an
unsigned return type, but return a negative constant to indicate an error
condition.

A sematic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@exists@
identifier f;
constant C;
@@

 unsigned f(...)
 { <+...
*  return -C;
 ...+> }
// </smpl>

Fixing amd_ec_wait_write and amd_ec_wait_read leads to the need to adjust
the return type of the functions amd_ec_write and amd_ec_read, which are
the only functions that call amd_ec_wait_write and amd_ec_wait_read.
amd_ec_write and amd_ec_read, in turn, are only called from within the
function amd8111_access, which already returns a signed typed value.  Each
of the calls to amd_ec_write and amd_ec_read are updated using the
following semantic patch:

// <smpl>
@@
@@

+ status = amd_ec_write
- amd_ec_write
  (...);
+ if (status) return status;

@@
@@

+ status = amd_ec_read
- amd_ec_read
  (...);
+ if (status) return status;
// </smpl>

The patch also adds the declaration of the status variable.

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Jean Delvare <khali@linux-fr.org>

i2c: Change to new flag variable

Replace EXTRA_CFLAGS with ccflags-y.

Signed-off-by: matt mooney <mfm@muteddisk.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>

i2c: Remove unneeded inclusions of <linux/i2c-id.h>

These drivers don't use anything which is defined in <linux/i2c-id.h>.
This header file was never meant to be included directly anyway, and
will be deleted soon.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Acked-by: Ben Dooks <ben-linux@fluff.org>
Acked-by: Dave Airlie <airlied@linux.ie>
Cc: Hans Verkuil <hverkuil@xs4all.nl>

i2c: Let i2c_parent_is_i2c_adapter return the parent adapter

This makes the calling site's code clearer IMHO.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Acked-by: Michael Lawnick <ml.lawnick@gmx.de>

i2c: Simplify i2c_parent_is_i2c_adapter

Only i2c devices can have their type set to i2c_adapter_type, so
testing the bus type is redundant.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Michael Lawnick <ml.lawnick@gmx.de>

i2c-pca-platform: Change device name of request_irq

i2c->adap.name shouldn't be used in request_irq.
Instead the driver name "i2c-pca-platform" should be used.

Signed-off-by: Nobuhiro Iwamatsu <nobuhiro.iwamatsu.yj@renesas.com>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: stable@kernel.org
Signed-off-by: Jean Delvare <khali@linux-fr.org>

i2c: Fix Kconfig dependencies

drivers/i2c/algos/Kconfig makes all the algorithms dependent on
!I2C_HELPER_AUTO, which triggers a Kconfig warning about broken
dependencies when some driver selects one of the algorithms. Ideally
we would make only the prompts dependent on !I2C_HELPER_AUTO, however
Kconfig doesn't currently support that. So we have to redefine the
symbols separately for the I2C_HELPER_AUTO=y case.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Acked-by: Michal Marek <mmarek@suse.cz>

KVM: Drop CONFIG_DMAR dependency around kvm_iommu_map_pages

We also have to call kvm_iommu_map_pages for CONFIG_AMD_IOMMU. So drop
the dependency on Intel IOMMU, kvm_iommu_map_pages will be a nop anyway
if CONFIG_IOMMU_API is not defined.

KVM-Stable-Tag.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: Fix signature of kvm_iommu_map_pages stub

Breaks otherwise if CONFIG_IOMMU_API is not set.

KVM-Stable-Tag.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: MCE: Send SRAR SIGBUS directly

Originally, SRAR SIGBUS is sent to QEMU-KVM via touching the poisoned
page. But commit 96054569190bdec375fe824e48ca1f4e3b53dd36 prevents the
signal from being sent. So now the signal is sent via
force_sig_info_fault directly.

[marcelo: use send_sig_info instead]

Reported-by: Dean Nelson <dnelson@redhat.com>
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: MCE: Add MCG_SER_P into KVM_MCE_CAP_SUPPORTED

Now we have MCG_SER_P (and corresponding SRAO/SRAR MCE) support in
kernel and QEMU-KVM, the MCG_SER_P should be added into
KVM_MCE_CAP_SUPPORTED to make all these code really works.

Reported-by: Dean Nelson <dnelson@redhat.com>
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: fix typo in copyright notice

Fix typo in copyright notice.

Signed-off-by: Nicolas Kaiser <nikai@nikai.net>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: Disable interrupts around get_kernel_ns()

get_kernel_ns() wants preemption disabled.  It doesn't make a lot of sense
during the get/set ioctls (no way to make them non-racy) but the callee wants
it.

Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: MMU: Avoid sign extension in mmu_alloc_direct_roots() pae root address

Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: MMU: move access code parsing to FNAME(walk_addr) function

Move access code parsing from caller site to FNAME(walk_addr) function

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: MMU: audit: check whether have unsync sps after root sync

After root synced, all unsync sps are synced, this patch add a check to make
sure it's no unsync sps in VCPU's page table

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: MMU: audit: introduce audit_printk to cleanup audit code

Introduce audit_printk, and record audit point instead audit name

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: MMU: audit: unregister audit tracepoints before module unloaded

fix:

Call Trace:
 [<ffffffffa01e46ba>] ? kvm_mmu_pte_write+0x229/0x911 [kvm]
 [<ffffffffa01c6ba9>] ? gfn_to_memslot+0x39/0xa0 [kvm]
 [<ffffffffa01c6c26>] ? mark_page_dirty+0x16/0x2e [kvm]
 [<ffffffffa01c6d6f>] ? kvm_write_guest_page+0x67/0x7f [kvm]
 [<ffffffff81066fbd>] ? local_clock+0x2a/0x3b
 [<ffffffffa01d52ce>] emulator_write_phys+0x46/0x54 [kvm]
 ......
Code:  Bad RIP value.
RIP  [<ffffffffa0172056>] 0xffffffffa0172056
 RSP <ffff880134f69a70>
CR2: ffffffffa0172056

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: MMU: audit: fix vcpu's spte walking

After nested nested paging, it may using long mode to shadow 32/PAE paging
guest, so this patch fix it

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: MMU: set access bit for direct mapping

Set access bit while setup up direct page table if it's nonpaing or npt enabled,
it's good for CPU's speculate access

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: MMU: cleanup for error mask set while walk guest page table

Small cleanup for set page fault error code

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: MMU: update 'root_hpa' out of loop in PAE shadow path

The value of 'vcpu->arch.mmu.pae_root' is not modified, so we can update
'root_hpa' out of the loop.

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: x86 emulator: Eliminate compilation warning in x86_decode_insn()

Eliminate:
arch/x86/kvm/emulate.c:801: warning: ‘sv’ may be used uninitialized in this
function

on gcc 4.1.2

Signed-off-by: Sheng Yang <sheng@linux.intel.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: x86: Fix constant type in kvm_get_time_scale

Older gcc versions complain about the improper type (for x86-32), 4.5
seems to fix this silently. However, we should better use the right type
initially.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: VMX: Add AX to list of registers clobbered by guest switch

By chance this caused no harm so far. We overwrite AX during switch
to/from guest context, so we must declare this.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

KVM guest: Move a printk that's using the clock before it's ready

Fix a hang during SMP kernel boot on KVM that showed up
after commit 489fb490dbf8dab0249ad82b56688ae3842a79e8
(2.6.35) and 59aab522154a2f17b25335b63c1cf68a51fb6ae0
(2.6.34.1). The problem only occurs when
CONFIG_PRINTK_TIME is set.

KVM-Stable-Tag.
Signed-off-by: Arjan Koers <0h61vkll2ly8@xutrox.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: x86: TSC catchup mode

Negate the effects of AN TYM spell while kvm thread is preempted by tracking
conversion factor to the highest TSC rate and catching the TSC up when it has
fallen behind the kernel view of time.  Note that once triggered, we don't
turn off catchup mode.

A slightly more clever version of this is possible, which only does catchup
when TSC rate drops, and which specifically targets only CPUs with broken
TSC, but since these all are considered unstable_tsc(), this patch covers
all necessary cases.

Signed-off-by: Zachary Amsden <zamsden@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: x86: Rename timer function

This just changes some names to better reflect the usage they
will be given.  Separated out to keep confusion to a minimum.

Signed-off-by: Zachary Amsden <zamsden@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: x86: Make math work for other scales

The math in kvm_get_time_scale relies on the fact that
NSEC_PER_SEC < 2^32.  To use the same function to compute
arbitrary time scales, we must extend the first reduction
step to shrink the base rate to a 32-bit value, and
possibly reduce the scaled rate into a 32-bit as well.

Note we must take care to avoid an arithmetic overflow
when scaling up the tps32 value (this could not happen
with the fixed scaled value of NSEC_PER_SEC, but can
happen with scaled rates above 2^31.

Signed-off-by: Zachary Amsden <zamsden@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: cpu_relax() during spin waiting for reboot

It doesn't really matter, but if we spin, we should spin in a more relaxed
manner.  This way, if something goes wrong at least it won't contribute to
global warming.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: VMX: Respect interrupt window in big real mode

If an interrupt is pending, we need to stop emulation so we
can inject it.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: VMX: Emulated real mode interrupt injection

Replace the inject-as-software-interrupt hack we currently have with
emulated injection.

Signed-off-by: Mohammed Gamal <m.gamal005@gmail.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: Add kvm_inject_realmode_interrupt() wrapper

This adds a wrapper function kvm_inject_realmode_interrupt() around the
emulator function emulate_int_real() to allow real mode interrupt injection.

[avi: initialize operand and address sizes before emulating interrupts]
[avi: initialize rip for real mode interrupt injection]
[avi: clear interrupt pending flag after emulating interrupt injection]

Signed-off-by: Mohammed Gamal <m.gamal005@gmail.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: x86 emulator: Expose emulate_int_real()

Signed-off-by: Mohammed Gamal <m.gamal005@gmail.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: MMU: fix counting of rmap entries in rmap_add()

It seems that rmap entries are under counted.

Signed-off-by: Hillf Danton <dhillf@gmail.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: document 'kvm.mmu_audit' parameter

Document this parameter into Documentation/kernel-parameters.txt

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: fix the description of kvm-amd.nested in documentation

The default state of 'kvm-amd.nested' is enabled now, so fix the documentation

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: SVM: do not generate "external interrupt exit" if other exit is pending

Nested SVM checks for external interrupt after injecting nested exception.
In case there is external interrupt pending the code generates "external
interrupt exit" and overwrites previous exit info. If previously injected
exception already generated exit it will be lost.

Signed-off-by: Gleb Natapov <gleb@redhat.com>
Acked-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: Convert PIC lock from raw spinlock to ordinary spinlock

The PIC code used to be called from preempt_disable() context, which
wasn't very good for PREEMPT_RT.  That is no longer the case, so move
back from raw_spinlock_t to spinlock_t.

Signed-off-by: Avi Kivity <avi@redhat.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: x86: Fix kvmclock bug

If preempted after kvmclock values are updated, but before hardware
virtualization is entered, the last tsc time as read by the guest is
never set.  It underflows the next time kvmclock is updated if there
has not yet been a successful entry / exit into hardware virt.

Fix this by simply setting last_tsc to the newly read tsc value so
that any computed nsec advance of kvmclock is nulled.

Signed-off-by: Zachary Amsden <zamsden@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

KVM: MMU: Don't track nested fault info in error-code

This patch moves the detection whether a page-fault was
nested or not out of the error code and moves it into a
separate variable in the fault struct.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: VMX: Move fixup_rmode_irq() to avoid forward declaration

No code changes.

Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: Non-atomic interrupt injection

Change the interrupt injection code to work from preemptible, interrupts
enabled context.  This works by adding a ->cancel_injection() operation
that undoes an injection in case we were not able to actually enter the guest
(this condition could never happen with atomic injection).

Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: VMX: Parameterize vmx_complete_interrupts() for both exit and entry

Currently vmx_complete_interrupts() can decode event information from vmx
exit fields into the generic kvm event queues.  Make it able to decode
the information from the entry fields as well by parametrizing it.

Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: VMX: Move real-mode interrupt injection fixup to vmx_complete_interrupts()

This allows reuse of vmx_complete_interrupts() for cancelling injections.

Signed-off-by: Avi Kivity <avi@redhat.com>

KVM: VMX: Split up vmx_complete_interrupts()

vmx_complete_interrupts() does too much, split it up:
 - vmx_vcpu_run() gets the "cache important vmcs fields" part
 - a new vmx_complete_atomic_exit() gets the parts that must be done atomically
 - a new vmx_recover_nmi_blocking() does what its name says
 - vmx_complete_interrupts() retains the event injection recovery code

This helps in reducing the work done in atomic context.

Signed-off-by: Avi Kivity <avi@redhat.com>