openwrt/staging/blogic.git
14 years ago[SCSI] qla2xxx: Add APEX support.
Sarang Radke [Sat, 20 Mar 2010 00:03:59 +0000 (17:03 -0700)]
[SCSI] qla2xxx: Add APEX support.

Allows priority setting for FCP_CMNDs.

Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] qla2xxx: Re-organized BSG interface specific code.
Giridhar Malavali [Sat, 20 Mar 2010 00:03:58 +0000 (17:03 -0700)]
[SCSI] qla2xxx: Re-organized BSG interface specific code.

1. Segregate BSG interface specific code to new files.
2. Handle multiple vendor specific commands indepedently.
3. Reorganised support for reset, management and update FCoE firmware commands.
4. Fixed memory leak issue in Loopback.
5. Added new vendor command to support iiDMA using BSG interface.
6. Proper cleanup of dma mapped and dma allocated buffers for BSG request.

[jejb: fix up conflict and merge in Jiri Slaby lock imbalance patch]
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Harish Zunjarrao <harish.zunjarrao@qlogic.com>
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] bfa: sg addr big endian fix
Jing Huang [Fri, 19 Mar 2010 18:07:36 +0000 (11:07 -0700)]
[SCSI] bfa: sg addr big endian fix

sg address in IO request is not set up correctly for big endian platform.
add new macros to properly swap the address.

Signed-off-by: Jing Huang <huangj@brocade.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] bfa: protect idr using bfad_mutex
Jing Huang [Fri, 19 Mar 2010 18:07:09 +0000 (11:07 -0700)]
[SCSI] bfa: protect idr using bfad_mutex

idr is a global resource, protect it with global bfad_mutex.

Signed-off-by: Jing Huang <huangj@brocade.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] bfa: use pci_iomap() and pci_iounmap()
Jing Huang [Fri, 19 Mar 2010 18:06:44 +0000 (11:06 -0700)]
[SCSI] bfa: use pci_iomap() and pci_iounmap()

Use pci_iomap() and pci_iounmap() to simplify the code. Remove uncessary #ifdef
check for ia64 (it was added as a workaound for some RHEL 5.x release which
doesn't export iounmap function)

Signed-off-by: Jing Huang <huangj@brocade.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] bfa: fix the issue of not handling scsi_cmnd sg chaining case
Jing Huang [Fri, 19 Mar 2010 18:06:05 +0000 (11:06 -0700)]
[SCSI] bfa: fix the issue of not handling scsi_cmnd sg chaining case

Currently the driver doesn't take into consideraion of possible sg chaining
when it walks through the sg list. This is fixed by using the sg_next()
which automatically handles the chaining case. Obosolete code is removed
as a result of this change.

Signed-off-by: Jing Huang <huangj@brocade.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] bfa: add fc transport class based vport create/delete
Jing Huang [Fri, 19 Mar 2010 18:05:39 +0000 (11:05 -0700)]
[SCSI] bfa: add fc transport class based vport create/delete

Use duplicate fc transport template for physical and vitual port. Add
vport create/delete/disalbe functions in the transport template of physical
port. Changes to make the vport create/delete function to work under this
framework.

Signed-off-by: Jing Huang <huangj@brocade.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] don't kfree an initialized struct device
Alan Stern [Thu, 18 Mar 2010 19:41:22 +0000 (15:41 -0400)]
[SCSI] don't kfree an initialized struct device

This patch (as1359) fixes a bug in scsi_alloc_target().  After a
device structure has been initialized (and especially after its name
has been set), it must not be freed directly.  One has to call
put_device() instead.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] don't reap targets upon device_add failure
Alan Stern [Thu, 18 Mar 2010 19:41:11 +0000 (15:41 -0400)]
[SCSI] don't reap targets upon device_add failure

This patch (as1358) fixes a bug in the error pathway of
scsi_target_add().  If registration fails, the target should not be
reaped.  The reaping occurs later, when scanning is finished and all
the child devices are removed.  The current code leaves an unbalanced
value in starget->reap_ref.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mptfusion: Bump version 03.04.15
Kashyap, Desai [Thu, 18 Mar 2010 13:56:02 +0000 (19:26 +0530)]
[SCSI] mptfusion: Bump version 03.04.15

Upgrade version from 3.04.14 to 3.04.15

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mptfusion: Proper error handling is added after mpt_config timeout
Kashyap, Desai [Thu, 18 Mar 2010 13:54:57 +0000 (19:24 +0530)]
[SCSI] mptfusion: Proper error handling is added after mpt_config timeout

Added proper error handling after mpt_config.
Now check of MPI_IOCSTATUS_CONFIG_INVALID_PAGE is added.
If error is MPI_IOCSTATUS_CONFIG_INVALID_PAGE, driver will return -ENODEV.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mptfusion: Event data alignment with 4 byte.
Kashyap, Desai [Thu, 18 Mar 2010 13:54:20 +0000 (19:24 +0530)]
[SCSI] mptfusion: Event data alignment with 4 byte.

event_data needs to be 4 byte aligned to makes sure there is no unaligned
memory access take place.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mptfusion: Check for command status is added after completion.
Kashyap, Desai [Thu, 18 Mar 2010 13:53:50 +0000 (19:23 +0530)]
[SCSI] mptfusion: Check for command status is added after completion.

1) Corrected name string as "MPT SAS HOST"
2) Added proper check conditions for MPT_MGMT_STATUS_COMMAND_GOOD.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mptfusion: Task abort is not supported for Volumes
Kashyap, Desai [Thu, 18 Mar 2010 13:53:19 +0000 (19:23 +0530)]
[SCSI] mptfusion: Task abort is not supported for Volumes

1) corrected return value as SUCCESS instead of 0.
2) Added check in mptscsih_abort.
mptfusion do not support task abort for Volumes.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mptfusion: sanity check for vdevice pointer is added
Kashyap, Desai [Thu, 18 Mar 2010 13:52:45 +0000 (19:22 +0530)]
[SCSI] mptfusion: sanity check for vdevice pointer is added

Added sanity checks before accessing vdevice and added vdevice->deleted
setting for mptfc.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mptfusion: Setting period,offset and width for SPI driver
Kashyap, Desai [Thu, 18 Mar 2010 13:51:34 +0000 (19:21 +0530)]
[SCSI] mptfusion: Setting period,offset and width for SPI driver

Set factor, offset and width while target negotiation.

Added config timeout 60 seconds. It was missing for only
mptspi_read_spi_device_pg0

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mptfusion: Proper bus_type check is added
Kashyap, Desai [Thu, 18 Mar 2010 13:50:38 +0000 (19:20 +0530)]
[SCSI] mptfusion: Proper bus_type check is added

Added proper bus_type check before processing event/ reset handler.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mptfusion: mpt_detach is called properly at the time of rmmod
Kashyap, Desai [Thu, 18 Mar 2010 13:48:30 +0000 (19:18 +0530)]
[SCSI] mptfusion: mpt_detach is called properly at the time of rmmod

Current design of mptsas is as follow.
MPTSAS will do probe() if pci id matches for available card in
system, irrespective of mode of controller. If controller is I/T mode
or I mode, things are fine. If controller is only in T mode, mptsas is
not doing complete process of mptsas_probe(). It will only make
sure IOC structure is created and IOC reference is available for
mptstm driver. Now While removing module we should take care
case of Target mode only mptsas. If we are removing IOC which is
only in Target mode, We should only detach IOC instead of
following rest of the cleanup process which is only required for T
mode controller. Now For T mode controller, only part clean up is
done instead of complete cleanup. mpt_detach will call early in case
of Target mode only controller.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mptfusion: mpt config will do Hard Reset based upon retry counts
Kashyap, Desai [Thu, 18 Mar 2010 13:44:51 +0000 (19:14 +0530)]
[SCSI] mptfusion: mpt config will do Hard Reset based upon retry counts

mpt_config would only attempt a MUR before retrying the command. The
driver will now retry a second time with a hard reset before leaving
the function.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mptfusion: Updated SCSI IO IOCTL error handling.
Kashyap, Desai [Thu, 18 Mar 2010 13:43:10 +0000 (19:13 +0530)]
[SCSI] mptfusion: Updated SCSI IO IOCTL error handling.

Behavior changes only for IOCTLs that time out.  Current behavior of
Bus Reset remains the same for RAID Passthru Timeouts Current behavior
of Diagnostic reset for any other type of IOCTL remains the same

CHANGE: For IOCTL SCSI IOs that timeout, a Target Reset TM is sent,
instead of Bus Reset. All error handing from that point is the same as
what the driver currently does, which is to say that if the Device
Reset TM fails it escalates do diagnostic reset.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mptfusion: Added new less expensive RESET (Message Unit Reset)
Kashyap, Desai [Thu, 18 Mar 2010 13:42:17 +0000 (19:12 +0530)]
[SCSI] mptfusion: Added new less expensive RESET (Message Unit Reset)

Message Unit Reset - instructs the IOC to reset the Reply Post and
Free FIFO's. All the Message Frames on Reply Free FIFO are
discarded. All posted buffers are freed, and event notification is
turned off.  IOC doesnt reply to any outstanding request. This will
transfer IOC to READY state.  Message unit ready is less expensive
operations than Hard Reset.  soft reset will not force Firmware to
reload again, it only do clean up of Message units.

mpt_Soft_Hard_ResetHandler will first try for Soft Reset,if
it fails then go for big hammer reset which is Hard Reset.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: Bump version 05.100.00.00
Kashyap, Desai [Wed, 17 Mar 2010 10:58:34 +0000 (16:28 +0530)]
[SCSI] mpt2sas: Bump version 05.100.00.00

Upgraded version string.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: Copyright 2010.
Kashyap, Desai [Wed, 17 Mar 2010 10:58:04 +0000 (16:28 +0530)]
[SCSI] mpt2sas: Copyright 2010.

Copyright changes for year 2010.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: Default descriptor for RAID Passthru command.
Kashyap, Desai [Wed, 17 Mar 2010 10:57:25 +0000 (16:27 +0530)]
[SCSI] mpt2sas: Default descriptor for RAID Passthru command.

RAID_SCSI_IO_PASSTHROUGH: Driver needs to be send the default
descriptor for RAID Passthru, currently its sending SCSI_IO descriptor.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: return -ENOMEM if memory allocation failed.
Kashyap, Desai [Wed, 17 Mar 2010 10:56:48 +0000 (16:26 +0530)]
[SCSI] mpt2sas: return -ENOMEM if memory allocation failed.

Added proper return type values in case memory allocation failed.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: removed use of tm_cmds.mutex in IOCTL branch.
Kashyap, Desai [Wed, 17 Mar 2010 10:55:59 +0000 (16:25 +0530)]
[SCSI] mpt2sas: removed use of tm_cmds.mutex in IOCTL branch.

Removed all the mutex's for ioc->tm_cmds.mutex, then created one
single mutex inside the function mpt2sas_scsih_issue_tm. This is the
single function used when sending task management. Also the sanity
checks required for scsi mid layer escalation were moved to inside the
same function because these checks need to be done while the mutex is
held. The ioc->tm_cmds.mutex inside the IOCTL branch is really not
required since there is another mutex in this code called for ctl_cmds
handling this sync.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: Fix for little endian
Kashyap, Desai [Wed, 17 Mar 2010 10:54:52 +0000 (16:24 +0530)]
[SCSI] mpt2sas: Fix for little endian

1. Fixes for little endian issues.
2. Now Debug info for Discovery event is more readable.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: Driver will not treat NEEDS_INIT as failure.
Kashyap, Desai [Wed, 17 Mar 2010 10:54:14 +0000 (16:24 +0530)]
[SCSI] mpt2sas: Driver will not treat NEEDS_INIT as failure.

Now Driver will not treat NEEDS_INIT as failure. In addition to this,
the driver will now display message to describe the the access flags
when bits are set, so the end user can better understand failures.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: Early return from function if shost is in recovery.
Kashyap, Desai [Wed, 17 Mar 2010 10:53:36 +0000 (16:23 +0530)]
[SCSI] mpt2sas: Early return from function if shost is in recovery.

Aded checks for shost_recovery flag for early return from function.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: Use of get_free_pages for huge memorary allocation.
Kashyap, Desai [Wed, 17 Mar 2010 10:52:52 +0000 (16:22 +0530)]
[SCSI] mpt2sas: Use of get_free_pages for huge memorary allocation.

use the get_free_pages API for larger contigious physical memory chunk.
Also, the ioc->chain_depth need to be changed from
a 16bit to 32bit variable because the number of chains will exceed 64k
when the queue depth is large.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: Corrected conditional checks for Internal device Reset
Kashyap, Desai [Wed, 17 Mar 2010 10:52:21 +0000 (16:22 +0530)]
[SCSI] mpt2sas: Corrected conditional checks for Internal device Reset

bug fix in the handling of the internal device reset event
The reason code check in scsih_sas_device_status_change_event never
evaluates as true for internal device reset, hence driver never quiesce s IO
when firmware is sending a device reset. The fix is to change the
evaluate to:
if (event_data->ReasonCode !=
    MPI2_EVENT_SAS_DEV_STAT_RC_INTERNAL_DEVICE_RESET &&
   event_data->ReasonCode !=
    MPI2_EVENT_SAS_DEV_STAT_RC_CMP_INTERNAL_DEV_RESET)
        return;

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: Corrected time stamp
Kashyap, Desai [Wed, 17 Mar 2010 10:51:33 +0000 (16:21 +0530)]
[SCSI] mpt2sas: Corrected time stamp

incorrect timestamp on 32 bit platforms: The upper 32 bit of
the timestamp was getting truncated when converting seconds to
milliseconds, which was due to the variable being long. To fix the problem,
the variable needs to be u64. Also the microseconds conversion to
milliseconds was incorrect; it should be divide by 1000 instead of divide by
8.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas : Do not reset handle before calling _scsih_remove_device in RESCAN...
Kashyap, Desai [Wed, 17 Mar 2010 09:50:37 +0000 (15:20 +0530)]
[SCSI] mpt2sas : Do not reset handle before calling _scsih_remove_device in RESCAN task after HBA RESET

Setting handle to zero is not required before _scsih_remove_device.
Driver uses sas_device->handle reference in _scsih_remove_device.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] lpfc 8.3.11: Update Driver version to 8.3.11
James Smart [Mon, 15 Mar 2010 15:25:58 +0000 (11:25 -0400)]
[SCSI] lpfc 8.3.11: Update Driver version to 8.3.11

Signed-off-by: Alex Iannicelli <alex.iannicelli@emulex.com>
Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] lpfc 8.3.11: NPIV changes
James Smart [Mon, 15 Mar 2010 15:25:44 +0000 (11:25 -0400)]
[SCSI] lpfc 8.3.11: NPIV changes

- Enable NPIV by default.
- Added code to handle unsolicited LOGO on physical port.

Signed-off-by: Alex Iannicelli <alex.iannicelli@emulex.com>
Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] lpfc 8.3.11: Fix AER uncorrectable non-fatal error handling
James Smart [Mon, 15 Mar 2010 15:25:32 +0000 (11:25 -0400)]
[SCSI] lpfc 8.3.11: Fix AER uncorrectable non-fatal error handling

Only abort outstanding I/O to force the OS to retry failed I/Os for AER
uncorrectable non-fatal errors instead of reseting the adapter.

Signed-off-by: Alex Iannicelli <alex.iannicelli@emulex.com>
Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] lpfc 8.3.11: Driver management improvements via BSG
James Smart [Mon, 15 Mar 2010 15:25:20 +0000 (11:25 -0400)]
[SCSI] lpfc 8.3.11: Driver management improvements via BSG

- Add BSG support for PCI loopback testing.
- Add BSG support for extended mailbox commands.

Signed-off-by: Alex Iannicelli <alex.iannicelli@emulex.com>
Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] lpfc 8.3.11: SLI4 Improvements
James Smart [Mon, 15 Mar 2010 15:25:07 +0000 (11:25 -0400)]
[SCSI] lpfc 8.3.11: SLI4 Improvements

- Correct all SLI4 code to work on big endian systems.
- Move read of sli4 params earlier so returned values are used correctly.

Signed-off-by: Alex Iannicelli <alex.iannicelli@emulex.com>
Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] lpfc 8.3.11: FCF failover improvements
James Smart [Mon, 15 Mar 2010 15:24:56 +0000 (11:24 -0400)]
[SCSI] lpfc 8.3.11: FCF failover improvements

FCF failover improvements

- Add random FCF failover when there are multiple FCFs available.
- Prevent FCF log messages from being displayed for FC adapters.
- Separate the New FCF and Modified FCF log messages.

Signed-off-by: Alex Iannicelli <alex.iannicelli@emulex.com>
Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt: modify mptctl_exit() to call proper deregister functions
Kei Tokunaga [Mon, 15 Mar 2010 05:48:43 +0000 (14:48 +0900)]
[SCSI] mpt: modify mptctl_exit() to call proper deregister functions

This patch fixes some issues of mptctl_exit().

  1) It doesn't call mpt_deregister() for mptctl_taskmgmt_id
     => Insmoding/rmmoding mptctl.ko repeadtedly (up to
        MPT_MAX_PROTOCOL_DRIVERS-1 at most) can eat up all cb_idx,
        and that would cause a lack of MptCallbacks[], MptDriverClass[],
        and MptEvHandlers[].

  2) It doesn't call mpt_event_deregister() for mptctl_id
     => Need to call it.

  3) It calls mpt_reset_deregister() for mptctl_taskmgmt_id
     => This could accidentally deregister an innocent reset handler
        that you don't want to.

This patch also adds a check for mptctl_taskmgmt_id.

Signed-off-by: Kei Tokunaga <tokunaga.keiich@jp.fujitsu.com>
Acked-by: "Desai, Kashyap" <Kashyap.Desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] libfc, libfcoe, fcoe: use smp_processor_id() only when preempt disabled
Joe Eykholt [Sat, 13 Mar 2010 00:08:55 +0000 (16:08 -0800)]
[SCSI] libfc, libfcoe, fcoe: use smp_processor_id() only when preempt disabled

When the kernel is configured for preemption, using smp_processor_id()
when preemption is enabled causes a warning backtrace and is wrong
since we could move off of that CPU as soon as we get the ID,
and we would be referencing the wrong CPU, and possibly an invalid one
if it could be hotswapped out.

Remove the fc_lport_get_stats() function and explicitly use per_cpu_ptr()
to get the statistics.  Where preemption has been disabled by holding
a _bh lock continue to use smp_processor_id(), but otherwise use
get_cpu()/put_cpu().

In fcoe_recv_frame() also changed the cases where we return in the
middle to do a goto to the code which bumps ErrorFrames and does
a put_cpu().  Two of these cases didn't bump ErrorFrames before, but
doing so is harmless because they "can't happen", due to prior length
checks.

Also rearranged code in fcoe_recv_frame() to have only one call to
fc_exch_recv().  It's just as efficient and saves a call to put_cpu().

In fc_fcp.c, adjusted a FIXME comment for code which doesn't need fixing.

Signed-off-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>
14 years ago[SCSI] libfc: Add debug statements when fc_fcp returns DID_ERROR to scsi-ml
Robert Love [Sat, 13 Mar 2010 00:08:50 +0000 (16:08 -0800)]
[SCSI] libfc: Add debug statements when fc_fcp returns DID_ERROR to scsi-ml

DID_ERROR cases can be ambigouos. Debugging FCP error cases
will be much easier if we have debug statements when we hit
these error conditions.

This patch simply adds debug messages using the FC_FCP_DBG
macro when we return DID_ERROR to SCSI. This way if a DID_ERROR
is reproducible turning on debug_logging will give a clue
to developers as to what the problem might be.

Signed-off-by: Robert Love <robert.w.love@intel.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] libfc: fix fcp pkt recovery in fc_fcp_recv_data
Vasu Dev [Sat, 13 Mar 2010 00:08:44 +0000 (16:08 -0800)]
[SCSI] libfc: fix fcp pkt recovery in fc_fcp_recv_data

Currently fc_fcp_recv_data calls fc_fcp_retry_cmd to
retry failed IO but in this case tgt is still sending
data frames, therefore exchange needs to be aborted
first before initiating retry. So this patch fixes
this by aborting exchange first then have retry.

Renames fc_timeout_error to fc_fcp_recovery since
fc_timeout_error is already called from several other
places beside from fcp timeout handler and then
used fc_fcp_recovery for abort & retry from
fc_fcp_recv_data, this rename also required renaming
FC_CMD_TIME_OUT status to FC_CMD_RECOVERY to be
consistent with new fc_fcp_recovery.

Data frames are not expected for an DDPed exchange and
potentially it could be tampered data frame, so does
recovery in this case by calling fc_fcp_recovery.

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>
14 years ago[SCSI] libfc: use offload EM instance again instead jumping to next EM
Vasu Dev [Sat, 13 Mar 2010 00:08:39 +0000 (16:08 -0800)]
[SCSI] libfc: use offload EM instance again instead jumping to next EM

Since use of offloads is more efficient than switching
to non-offload EM. However kept logic same to call em_match
if it is provided in the list of EMs.

Converted fc_exch_alloc to inline being now tiny a function
and already not an exported libfc API any more.

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>
14 years ago[SCSI] libfc: fixes unnecessary seq id jump
Vasu Dev [Sat, 13 Mar 2010 00:08:34 +0000 (16:08 -0800)]
[SCSI] libfc: fixes unnecessary seq id jump

In some cases seq is incremented twice causing unnecessary
seq jump, for instance fc_exch_recv_seq_resp increments
seq id when fc_sof_is_init is true and that is true for
each incoming xfer ready but then fc_fcp_send_data does
another seq increment to send data for xfer ready.

This patch removes all such seq id jumps, at least it
eliminates few calls to fc_seq_start_next using ex_lock.

Also removes seq id update with incoming frame's seq id
as this is not needed since each end (I or T) just need
to send incremented their own seq id on each TSI from
other end & before sending new sequence within a
exchange.

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>
14 years ago[SCSI] libfc: fix sequence-initiative WARN in fc_seq_start_next
Joe Eykholt [Sat, 13 Mar 2010 00:08:29 +0000 (16:08 -0800)]
[SCSI] libfc: fix sequence-initiative WARN in fc_seq_start_next

When starting a new response sequence in a multi-sequence
exchange, a warning was issued that sequence initiative
wasn't held.

The bug was that sequence initiative was cleared by the previous
sequence due to the END_SEQ flag being on.  The intent may have
been to check LAST_SEQ.  Change just to check SEQ_INIT.

Signed-off-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>
14 years ago[SCSI] libfcoe: eliminate unused link and last_link fields
Joe Eykholt [Sat, 13 Mar 2010 00:08:23 +0000 (16:08 -0800)]
[SCSI] libfcoe: eliminate unused link and last_link fields

The link and last_link fields in the fcoe_ctlr struct are no
longer useful, since they are always set to the same value,
and FIP always calls libfc to pass link information to the lport.

Eliminate those fields and rename link_work to timer_work, since
it no longer has any link change work to do.

Thanks to Brian Uchino for discovering this issue.

Signed-off-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>
14 years ago[SCSI] fcoe: remove an unused variable in fcoe_recv_frame()
Joe Eykholt [Sat, 13 Mar 2010 00:08:18 +0000 (16:08 -0800)]
[SCSI] fcoe: remove an unused variable in fcoe_recv_frame()

Remove an unused variable, mac, in fcoe_recv_frame().

Signed-off-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>
14 years ago[SCSI] libfc: send point-to-poin FLOGI LS_ACC to assigned D_DID
Joe Eykholt [Sat, 13 Mar 2010 00:08:12 +0000 (16:08 -0800)]
[SCSI] libfc: send point-to-poin FLOGI LS_ACC to assigned D_DID

The method we've been using for point-to-point mode requires
that the LS_ACC for the FLOGI uses the D_ID and S_ID assigned
to the remote port and local port, not those in the exchange.

This is not the correct method, but for now, it's what works
with the old target, as well as with new targets based on libfc.

This patch changes the addresses used accordingly.

Signed-off-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>
14 years ago[SCSI] libfc: recognize incoming FLOGI for point-to-point mode
Joe Eykholt [Sat, 13 Mar 2010 00:08:07 +0000 (16:08 -0800)]
[SCSI] libfc: recognize incoming FLOGI for point-to-point mode

When receiving a FLOGI request from a point-to-point peer,
the D_ID of 0xfffffe was not recognized as belonging to one
of the lports, so it was dropped.

Change fc_vport_id_lookup() to treat d_id 0xfffffe as a match.

Signed-off-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>
14 years ago[SCSI] fcoe: save gateway address when receiving FLOGI request
Joe Eykholt [Sat, 13 Mar 2010 00:08:02 +0000 (16:08 -0800)]
[SCSI] fcoe: save gateway address when receiving FLOGI request

In point-to-point mode, we need to save the source MAC
from received FLOGI requests to use as the destination MAC
for all outgoing frames.  We stopped doing that at some point.

Use the lport_set_port_id method to catch incoming FLOGI frames
and pass them to fcoe_ctlr_recv_flogi() so it can save the source MAC.

Signed-off-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>
14 years ago[SCSI] libfcoe: fix debug message entering non-FIP mode
Joe Eykholt [Sat, 13 Mar 2010 00:07:57 +0000 (16:07 -0800)]
[SCSI] libfcoe: fix debug message entering non-FIP mode

The debug message that indicated we are using non-FIP mode was
being printed only if we were already in non-FIP mode.
Also changed the message text to make it more clear the mode
is being set, not that the message is indicating how FLOGI
was received.

Signed-off-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>
14 years ago[SCSI] fcoe: call fcoe_ctlr_els_send even for ELS responses
Joe Eykholt [Sat, 13 Mar 2010 00:07:52 +0000 (16:07 -0800)]
[SCSI] fcoe: call fcoe_ctlr_els_send even for ELS responses

In point-to-point mode, the destination MAC address for
the FLOGI response was zero because the LS_ACC for the FLOGI
wasn't getting intercepted by FIP.

Change to call fcoe_ctlr_els_send when sending any ELS,
not just requests.

Signed-off-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>
14 years ago[SCSI] libfc: fix oops in point-to-point mode
Joe Eykholt [Sat, 13 Mar 2010 00:07:46 +0000 (16:07 -0800)]
[SCSI] libfc: fix oops in point-to-point mode

In point-to-point mode, if the PLOGI to the remote port times
out, it can get deleted by the remote port module.  Since there's
no reference by the local port, lport->ptp_data points to a freed
rport, and when the local port is reset and tries to logout again,
an oops occurs in mutex_lock_nested().

Hold a reference count on the point-to-point rdata.

Signed-off-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>
14 years ago[SCSI] libfc: add definition for task attribute mask
Joe Eykholt [Sat, 13 Mar 2010 00:07:41 +0000 (16:07 -0800)]
[SCSI] libfc: add definition for task attribute mask

The FCP command header definition should define a mask for
the task attribute field.  This adds that #define.

Signed-off-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>
14 years ago[SCSI] libfc: recode incoming PRLI handling
Joe Eykholt [Sat, 13 Mar 2010 00:07:36 +0000 (16:07 -0800)]
[SCSI] libfc: recode incoming PRLI handling

Reduce indentation in fc_rport_recv_prli_req() using gotos.
Also add payload length checks.

Signed-off-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>
14 years ago[SCSI] mpt2sas: use correct pci_resource_flag for comparison
Richard A Lary [Fri, 12 Mar 2010 23:27:06 +0000 (15:27 -0800)]
[SCSI] mpt2sas: use correct pci_resource_flag for comparison

This patch replaces incorrect base address space flag with correct IO
resource flag.  Also, performs check of memory resource to validate
resource before using.

Signed-off-by: Richard A Lary <rlary@linux.vnet.ibm.com>
Acked-by: "Desai, Kashyap" <Kashyap.Desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] bfa: eliminate useless code
Julia Lawall [Thu, 11 Mar 2010 22:09:55 +0000 (14:09 -0800)]
[SCSI] bfa: eliminate useless code

The variable bfa_itnim is initialized twice to the same (side effect-free)
expression.  Drop one initialization.

A simplified version of the semantic match that finds this problem is:
(http://coccinelle.lip6.fr/)

// <smpl>
@forall@
idexpression *x;
identifier f!=ERR_PTR;
@@

x = f(...)
... when != x
(
x = f(...,<+...x...+>,...)
|
* x = f(...)
)
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Acked-by: Jing Huang <huangj@brocade.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] pmcraid: redundant check in pmcraid_check_ioctl_buffer()
Roel Kluin [Thu, 11 Mar 2010 22:09:44 +0000 (14:09 -0800)]
[SCSI] pmcraid: redundant check in pmcraid_check_ioctl_buffer()

struct pmcraid_ioctl_header member buffer_length is unsigned, so this
check appears redundant.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Acked-by: Anil Ravindranath <anil_ravindranath@pmc-sierra.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] gdth: fix buffer overflow
Roel Kluin [Thu, 11 Mar 2010 22:09:35 +0000 (14:09 -0800)]
[SCSI] gdth: fix buffer overflow

This allows i == MAXHA, which is out of range

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] pm8001: drop redundant memset
Julia Lawall [Tue, 9 Mar 2010 21:14:34 +0000 (22:14 +0100)]
[SCSI] pm8001: drop redundant memset

The region set by the call to memset is immediately overwritten by the
subsequent call to memcpy.

The semantic patch that makes this change is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
expression e1,e2,e3,e4;
@@

- memset(e1,e2,e3);
  memcpy(e1,e4,e3);
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Acked-by: Jack Wang <jack_wang@usish.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas : Device removal algorithm in interrupt context only
Kashyap, Desai [Tue, 9 Mar 2010 12:04:13 +0000 (17:34 +0530)]
[SCSI] mpt2sas : Device removal algorithm in interrupt context only

external host not connecting after controller reboot: The
problem is : devices are not coming back after having the cable
disconnected then reconnected. The problem is because the
driver/firmware device removal handshake is failing. Due to this failure,
the controller firmware is not sending out device add events when the target
is reconnected. This is root caused to a race in the driver/firmware device
removal algorithm. There is duplicate code in both interrupt and user
context; where target reset is being issue from user context path while
sas_iounit_control(OP_REMOVE) is being sent from interrupt context. An
active target_reset will fail the OP_REMOVE. To fix this problem, the
duplicate code has been removed from user context path.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: Upgrading version to 04.100.01.02
Kashyap, Desai [Tue, 9 Mar 2010 11:02:47 +0000 (16:32 +0530)]
[SCSI] mpt2sas: Upgrading version to 04.100.01.02

Upgraded version string.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Reviewed-by: Eric Moore <eric.moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: modified _scsih_sas_device_find_by_handle/sas_address
Kashyap, Desai [Tue, 9 Mar 2010 11:02:17 +0000 (16:32 +0530)]
[SCSI] mpt2sas: modified _scsih_sas_device_find_by_handle/sas_address

modified _scsih_sas_device_find_by_handle
so to handle the search on both list(device list and device_init_list)
Also, we moved the priority of the
search so the ioc->sas_device_list is done first.  The
"sas_device_init_list" is only used during the 1st port enable, so its
unlikely there’s devices on it.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Reviewed-by: Eric Moore <eric.moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: RESCAN Barrier work is added in case of HBA reset.
Kashyap, Desai [Tue, 9 Mar 2010 11:01:43 +0000 (16:31 +0530)]
[SCSI] mpt2sas: RESCAN Barrier work is added in case of HBA reset.

Add the cancel_pending_work flag from the fw_event_work structure, and then to
set the flag during host reset, check the flag later from work threads
context and if cancel_pending_work_flag is set ingore those events.

Now Rescan after host reset is changed.
Added special task MPT2SAS_RESCAN_AFTER_HOST_RESET. This task will be queued
at the time of HBA reset. this task is treated as barrier. All work after
MPT2SAS_RESCAN_AFTER_HOST_RESET will be treated as new work and will be
server by callback handle. If host_recovery is going on while running RESCAN
task, it will wait for shos_recovery_done completion which will be called
from HBA reset DONE context.

Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Reviewed-by: Eric Moore <eric.moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] scsi_transport_fc: Protect against overflow in dev_loss_tmo
Hannes Reinecke [Tue, 9 Mar 2010 09:18:48 +0000 (10:18 +0100)]
[SCSI] scsi_transport_fc: Protect against overflow in dev_loss_tmo

The rport structure defines dev_loss_tmo as u32, which is
later multiplied with HZ to get the actual timeout value.
This might overflow for large dev_loss_tmo values. So we
should be better using u64 as intermediate variables here
to protect against overflow.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Acked-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] mpt2sas: fix the incorrect scsi_dma_map error checking
FUJITA Tomonori [Tue, 9 Mar 2010 02:09:50 +0000 (11:09 +0900)]
[SCSI] mpt2sas: fix the incorrect scsi_dma_map error checking

scsi_dma_map() returns -1 if an error occurred (zero means that the
command has no data). So the following current code can't catch an
error:

sges_left = scsi_dma_map(scmd);
if (!sges_left) {
sdev_printk(KERN_ERR, scmd->device, "pci_map_sg"
" failed: request for %d bytes!\n", scsi_bufflen(scmd));
return -ENOMEM;
}

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: "Desai, Kashyap" <Kashyap.Desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] 3ware maintainers update
adam radford [Mon, 8 Mar 2010 20:40:36 +0000 (12:40 -0800)]
[SCSI] 3ware maintainers update

This patch updates the 3ware maintainers in the MAINTAINERS file.

Signed-off-by: Adam Radford <aradford@gmail.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] 3w-xxxx, 3w-9xxx: force 60 second timeout
adam radford [Mon, 8 Mar 2010 20:37:46 +0000 (12:37 -0800)]
[SCSI] 3w-xxxx, 3w-9xxx: force 60 second timeout

This small patch forces 60 second timeouts for the older 3w-xxxx &
3w-9xxx drivers for systems that don't contain the udev rule for
setting scsi timeouts to 60 seconds.

Signed-off-by: Adam Radford <aradford@gmail.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] qla2xxx: fix lock imbalance
Jiri Slaby [Tue, 16 Mar 2010 15:23:59 +0000 (16:23 +0100)]
[SCSI] qla2xxx: fix lock imbalance

Stanse found that one error path in qla24xx_bsg_timeout omits to
unlock ha->hardware_lock. Fix that.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Acked-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] lpfc: fix lock imbalances
Jiri Slaby [Tue, 16 Mar 2010 15:23:58 +0000 (16:23 +0100)]
[SCSI] lpfc: fix lock imbalances

Stanse found that two error paths in lpfc_bsg_rport_els_cmp and
lpfc_issue_ct_rsp_cmp omits to unlock phba->ct_ev_lock. It is
because they wrongly unlock phba->hbalock instead. Fix that.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Acked-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] be2iscsi: fix lock imbalance
Jiri Slaby [Tue, 16 Mar 2010 15:23:57 +0000 (16:23 +0100)]
[SCSI] be2iscsi: fix lock imbalance

Stanse found that one error path in mgmt_invalidate_icds omits to unlock
ctrl->mbox_lock. Fix that.

Added in 756d29c8c7ed8887ed7d752371ce2f (Enable async mode for mcc rings)
where the spinlock was moved.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Acked-by: Jayamohan Kallickal <jayamohank@serverengines.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years ago[SCSI] dpt_i2o: several use after free issues
Dan Carpenter [Mon, 15 Mar 2010 08:26:56 +0000 (11:26 +0300)]
[SCSI] dpt_i2o: several use after free issues

adpt_i2o_delete_hba() calls kfree() so we have to save "pHba->next"
before calling it.  Also inside adpt_i2o_delete_hba() itself, there
was another use after free bug which I fixed by moving the kfree()
down a line.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
14 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzi...
Linus Torvalds [Thu, 8 Apr 2010 17:02:02 +0000 (10:02 -0700)]
Merge branch 'upstream-linus' of git://git./linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  libata: Fix accesses at LBA28 boundary (old bug, but nasty) (v2)

14 years agolibata: Fix accesses at LBA28 boundary (old bug, but nasty) (v2)
Mark Lord [Wed, 7 Apr 2010 17:52:08 +0000 (13:52 -0400)]
libata: Fix accesses at LBA28 boundary (old bug, but nasty) (v2)

Most drives from Seagate, Hitachi, and possibly other brands,
do not allow LBA28 access to sector number 0x0fffffff (2^28 - 1).
So instead use LBA48 for such accesses.

This bug could bite a lot of systems, especially when the user has
taken care to align partitions to 4KB boundaries. On misaligned systems,
it is less likely to be encountered, since a 4KB read would end at
0x10000000 rather than at 0x0fffffff.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
14 years agoMerge branch 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 8 Apr 2010 15:37:05 +0000 (08:37 -0700)]
Merge branch 'sched-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  sched: Fix sched_getaffinity()

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/ide-2.6
Linus Torvalds [Thu, 8 Apr 2010 14:45:36 +0000 (07:45 -0700)]
Merge git://git./linux/kernel/git/davem/ide-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/ide-2.6:
  ide: Fix IDE taskfile with cfq scheduler
  ide: Must hold queue lock when requeueing
  ide: Requeue request after DMA timeout

14 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Thu, 8 Apr 2010 14:44:53 +0000 (07:44 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6:
  ACPI / PM: Move ACPI video resume to a PM notifier
  ACPI: Reduce ACPI resource conflict message to KERN_WARNING, printk cleanup
  ACPI: battery drivers should call power_supply_changed()
  ACPI: battery: Fix CONFIG_ACPI_SYSFS_POWER=n
  PNPACPI: truncate _CRS windows with _LEN > _MAX - _MIN + 1
  ACPI: Don't send KEY_UNKNOWN for random video notifications
  ACPI: NUMA: map pxms to low node ids
  ACPI: use _HID when supplied by root-level devices
  ACPI / ACPICA: Do not check reference counters in acpi_ev_enable_gpe()
  ACPI: fixes a false alarm from lockdep
  ACPI dock: support multiple ACPI dock devices
  ACPI: EC: Allow multibyte access to EC

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus
Linus Torvalds [Thu, 8 Apr 2010 01:49:20 +0000 (18:49 -0700)]
Merge git://git./linux/kernel/git/rusty/linux-2.6-for-linus

* git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus:
  hvc_console: Fix race between hvc_close and hvc_remove
  virtio: disable multiport console support.
  virtio: console makes incorrect assumption about virtio API
  virtio: console: Fix early_put_chars usage
  MAINTAINERS: Put the virtio-console entry in correct alphabetical order

14 years agohvc_console: Fix race between hvc_close and hvc_remove
Anton Blanchard [Tue, 6 Apr 2010 11:42:38 +0000 (21:42 +1000)]
hvc_console: Fix race between hvc_close and hvc_remove

I don't claim to understand the tty layer, but it seems like hvc_open and
hvc_close should be balanced in their kref reference counting.

Right now we get a kref every call to hvc_open:

        if (hp->count++ > 0) {
                tty_kref_get(tty); <----- here
                spin_unlock_irqrestore(&hp->lock, flags);
                hvc_kick();
                return 0;
        } /* else count == 0 */

        tty->driver_data = hp;

        hp->tty = tty_kref_get(tty); <------ or here if hp->count was 0

But hvc_close has:

        tty_kref_get(tty);

        if (--hp->count == 0) {
...
                /* Put the ref obtained in hvc_open() */
                tty_kref_put(tty);
...
        }

        tty_kref_put(tty);

Since the outside kref get/put balance we only do a single kref_put when
count reaches 0.

The patch below changes things to call tty_kref_put once for every
hvc_close call, and with that my machine boots fine.

Signed-off-by: Anton Blanchard <anton@samba.org>
Acked-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
14 years agovirtio: disable multiport console support.
Michael S. Tsirkin [Wed, 31 Mar 2010 18:56:42 +0000 (21:56 +0300)]
virtio: disable multiport console support.

Move MULTIPORT feature and related config changes
out of exported headers, and disable the feature
at runtime.

At this point, it seems less risky to keep code around
until we can enable it than rip it out completely.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
14 years agovirtio: console makes incorrect assumption about virtio API
Rusty Russell [Thu, 8 Apr 2010 15:46:16 +0000 (09:46 -0600)]
virtio: console makes incorrect assumption about virtio API

The get_buf() API sets the second arg to the number of bytes *written*
by the other side; in this case it should be zero as these are output buffers.

lguest gets this right (obviously kvm's console doesn't), resulting in
continual buildup of console writes.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Acked-by: Amit Shah <amit.shah@redhat.com>
14 years agovirtio: console: Fix early_put_chars usage
François Diakhaté [Tue, 23 Mar 2010 12:53:15 +0000 (18:23 +0530)]
virtio: console: Fix early_put_chars usage

Currently early_put_chars is not used by virtio_console because it can
only be used once a port has been found, at which point it's too late
because it is no longer needed. This patch should fix it.

Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
14 years agoMAINTAINERS: Put the virtio-console entry in correct alphabetical order
Amit Shah [Tue, 23 Mar 2010 12:53:09 +0000 (18:23 +0530)]
MAINTAINERS: Put the virtio-console entry in correct alphabetical order

Move around the entry for virtio-console to keep the file sorted.

Signed-off-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
14 years agorwsem generic spinlock: use IRQ save/restore spinlocks
Kevin Hilman [Wed, 7 Apr 2010 18:52:46 +0000 (11:52 -0700)]
rwsem generic spinlock: use IRQ save/restore spinlocks

rwsems can be used with IRQs disabled, particularily in early boot
before IRQs are enabled.  Currently the spin_unlock_irq() usage in the
slow-patch will unconditionally enable interrupts and cause problems
since interrupts are not yet initialized or enabled.

This patch uses save/restore versions of IRQ spinlocks in the slowpath
to ensure interrupts are not unintentionally disabled.

Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoHave nfs ->d_revalidate() report errors properly
Al Viro [Wed, 7 Apr 2010 23:06:07 +0000 (00:06 +0100)]
Have nfs ->d_revalidate() report errors properly

If nfs atomic open implementation ends up doing open request from
->d_revalidate() codepath and gets an error from server, return that error
to caller explicitly and don't bother with lookup_instantiate_filp() at all.
->d_revalidate() can return an error itself just fine...

See
http://bugzilla.kernel.org/show_bug.cgi?id=15674
http://marc.info/?l=linux-kernel&m=126988782722711&w=2

for original report.

Reported-by: Daniel J Blueman <daniel.blueman@gmail.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMerge branch 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Wed, 7 Apr 2010 21:01:51 +0000 (14:01 -0700)]
Merge branch 'perf-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  perf, x86: Enable Nehalem-EX support
  perf kmem: Fix breakage introduced by 5a0e3ad slab.h script

14 years agoMerge branch 'davinci-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Wed, 7 Apr 2010 18:03:06 +0000 (11:03 -0700)]
Merge branch 'davinci-fixes-for-linus' of git://git./linux/kernel/git/khilman/linux-davinci

* 'davinci-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/khilman/linux-davinci:
  davinci: fix compile warning: <mach/da8xx.h>: #include <linux/platform_device.h>
  davinci: DM365: fix duplicate default IRQ priorities
  davinci: edma: clear events in edma_start()
  davinci: da8xx/omap-l1: fix build error when CONFIG_DAVINCI_MUX is undefined
  davinci: timers: don't enable timer until clocksource is initialized

14 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 7 Apr 2010 18:02:23 +0000 (11:02 -0700)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/x86/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-tip:
  x86: Fix double enable_IR_x2apic() call on SMP kernel on !SMP boards
  x86: Increase CONFIG_NODES_SHIFT max to 10
  ibft, x86: Change reserve_ibft_region() to find_ibft_region()
  x86, hpet: Fix bug in RTC emulation
  x86, hpet: Erratum workaround for read after write of HPET comparator
  bootmem, x86: Fix 32bit numa system without RAM on node 0
  nobootmem, x86: Fix 32bit numa system without RAM on node 0
  x86: Handle overlapping mptables
  x86: Make e820_remove_range to handle all covered case
  x86-32, resume: do a global tlb flush in S4 resume

14 years agodavinci: fix compile warning: <mach/da8xx.h>: #include <linux/platform_device.h>
Sergei Shtylyov [Fri, 26 Mar 2010 14:56:58 +0000 (17:56 +0300)]
davinci: fix compile warning: <mach/da8xx.h>: #include <linux/platform_device.h>

This hushes the following warning:

arch/arm/mach-davinci/include/mach/da8xx.h:104: warning: ‘struct platform_device’
declared inside parameter list
arch/arm/mach-davinci/include/mach/da8xx.h:104: warning: its scope is only this
definition or declaration, which is probably not what you want

Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
14 years agoMerge branch 'for-linus' of git://git.monstr.eu/linux-2.6-microblaze
Linus Torvalds [Wed, 7 Apr 2010 15:48:39 +0000 (08:48 -0700)]
Merge branch 'for-linus' of git://git.monstr.eu/linux-2.6-microblaze

* 'for-linus' of git://git.monstr.eu/linux-2.6-microblaze:
  microblaze: Remove unused variable from ptrace
  microblaze: io.h: Add io big-endian function
  microblaze: Enable memory leak detector
  microblaze: Fix futex code
  microblaze: Fix ftrace_update_ftrace_func panic

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Wed, 7 Apr 2010 15:42:25 +0000 (08:42 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  ALSA: mixart: range checking proc file
  ALSA: hda - Fix a wrong array range check in patch_realtek.c
  ALSA: ASoC: move dma_data from snd_soc_dai to snd_soc_pcm_stream
  ALSA: hda - Enable amplifiers on Acer Inspire 6530G
  ASoC: Only do WM8994 bias off transition from standby
  ASoC: Don't use DCS_DATAPATH_BUSY for WM hubs devices
  ASoC: Don't do runtime wm_hubs DC servo updates if using offset correction
  ASoC: Support second DC servo readback method for wm_hubs
  ASoC: Avoid wraparound in wm_hubs DC servo correction
  ALSA: echoaudio - Eliminate use after free
  ALSA: i2c: cleanup: change parameter to pointer
  ALSA: hda - Add MSI blacklist for Aopen MZ915-M
  ASoC: OMAP: Fix capture pointer handling for OMAP1510 to work correctly with recent ALSA PCM code
  ALSA: hda - Update document about MSI and interrupts
  ALSA: hda: Fix 0 dB offset for Lenovo Thinkpad models using AD1981
  ALSA: hda - Add missing printk argument in previous patch
  ASoC: Fix passing platform_data to ac97 bus users and fix a leak
  ALSA: hda - Fix ADC/MUX assignment of ALC269 codec
  ALSA: hda - Fix invalid bit values passed to snd_hda_codec_amp_stereo()
  ASoC: wm8994: playback => capture

14 years agoMerge branch 'slabh' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/misc
Linus Torvalds [Wed, 7 Apr 2010 15:38:27 +0000 (08:38 -0700)]
Merge branch 'slabh' of git://git./linux/kernel/git/tj/misc

* 'slabh' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/misc:
  nodemask: include slab.h from drivers/base/node.c

14 years agofs-cache: order the debugfs stats correctly
David Howells [Tue, 6 Apr 2010 21:35:09 +0000 (14:35 -0700)]
fs-cache: order the debugfs stats correctly

Order the debugfs statistics correctly.  The values displayed through a
seq_printf() statement should be in the same order as the names in the
format string.

In the 'Lookups' line, objects created ('crt=') and lookups timed out
('tmo=') have their values transposed.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agofrv: fix kernel/user segment handling in NOMMU mode
David Howells [Tue, 6 Apr 2010 21:35:09 +0000 (14:35 -0700)]
frv: fix kernel/user segment handling in NOMMU mode

In NOMMU mode, the FRV segment handling is broken because KERNEL_DS ==
USER_DS.  This causes tests of the following sort:

/* don't pin down non-user-based iovecs */
if (segment_eq(get_fs(), KERNEL_DS))
return NULL;

to malfunction.

To fix this, make USER_DS the top of RAM instead of the top of the non-IO
address space, and make KERNEL_DS one more than the top of the non-IO
address space.

Also get rid of FRV's __addr_ok() as nothing uses it.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agofrv: hide uncached_access() when pgprot_noncached is not #defined
David Howells [Tue, 6 Apr 2010 21:35:08 +0000 (14:35 -0700)]
frv: hide uncached_access() when pgprot_noncached is not #defined

Hide uncached_access() when pgprot_noncached is not #defined.  This prevents
the following warning:

  CC      drivers/char/mem.o
drivers/char/mem.c:229: warning: 'uncached_access' defined but not used

Repairs d7d4d849b4e3acc405ec222884936800ffb26d48 ("drivers/char/mem.c:
cleanups").

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agortc-mxc: multiple fixes in rtc-mxc probe method
Vladimir Zapolskiy [Tue, 6 Apr 2010 21:35:07 +0000 (14:35 -0700)]
rtc-mxc: multiple fixes in rtc-mxc probe method

On exit paths in mxc_rtc_probe() method some resources are not freed
correctly.

This patch fixes:
* unrequested memory region containing imx RTC registers
* iounmap() isn't called on exit_free_pdata branch
* clock get rate is called for freed clock source
* clock isn't disabled on exit_put_clk branch

To simplify the fix managed device resources are used.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Vladimir Zapolskiy <vzapolskiy@gmail.com>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agomemcg: fix race in file_mapped accounting
KAMEZAWA Hiroyuki [Tue, 6 Apr 2010 21:35:05 +0000 (14:35 -0700)]
memcg: fix race in file_mapped accounting

Presently, memcg's FILE_MAPPED accounting has following race with
move_account (happens at rmdir()).

    increment page->mapcount (rmap.c)
    mem_cgroup_update_file_mapped()           move_account()
      lock_page_cgroup()
      check page_mapped() if
      page_mapped(page)>1 {
FILE_MAPPED -1 from old memcg
FILE_MAPPED +1 to old memcg
      }
      .....
      overwrite pc->mem_cgroup
      unlock_page_cgroup()
    lock_page_cgroup()
    FILE_MAPPED + 1 to pc->mem_cgroup
    unlock_page_cgroup()

Then,
old memcg (-1 file mapped)
new memcg (+2 file mapped)

This happens because move_account see page_mapped() which is not guarded
by lock_page_cgroup().  This patch adds FILE_MAPPED flag to page_cgroup
and move account information based on it.  Now, all checks are synchronous
with lock_page_cgroup().

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Balbir Singh <balbir@in.ibm.com>
Reviewed-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Andrea Righi <arighi@develer.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopagemap: fix pfn calculation for hugepage
Naoya Horiguchi [Tue, 6 Apr 2010 21:35:04 +0000 (14:35 -0700)]
pagemap: fix pfn calculation for hugepage

When we look into pagemap using page-types with option -p, the value of
pfn for hugepages looks wrong (see below.) This is because pte was
evaluated only once for one vma although it should be updated for each
hugepage.  This patch fixes it.

  $ page-types -p 3277 -Nl -b huge
  voffset   offset  len     flags
  7f21e8a00 11e400  1       ___U___________H_G________________
  7f21e8a01 11e401  1ff     ________________TG________________
               ^^^
  7f21e8c00 11e400  1       ___U___________H_G________________
  7f21e8c01 11e401  1ff     ________________TG________________
               ^^^

One hugepage contains 1 head page and 511 tail pages in x86_64 and each
two lines represent each hugepage.  Voffset and offset mean virtual
address and physical address in the page unit, respectively.  The
different hugepages should not have the same offset value.

With this patch applied:

  $ page-types -p 3386 -Nl -b huge
  voffset   offset   len    flags
  7fec7a600 112c00   1      ___UD__________H_G________________
  7fec7a601 112c01   1ff    ________________TG________________
               ^^^
  7fec7a800 113200   1      ___UD__________H_G________________
  7fec7a801 113201   1ff    ________________TG________________
               ^^^
               OK

More info:

- This patch modifies walk_page_range()'s hugepage walker.  But the
  change only affects pagemap_read(), which is the only caller of hugepage
  callback.

- Without this patch, hugetlb_entry() callback is called per vma, that
  doesn't match the natural expectation from its name.

- With this patch, hugetlb_entry() is called per hugepte entry and the
  callback can become much simpler.

Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Matt Mackall <mpm@selenic.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoratelimit: fix the return value when __ratelimit() fails to acquire the lock
Yong Zhang [Tue, 6 Apr 2010 21:35:03 +0000 (14:35 -0700)]
ratelimit: fix the return value when __ratelimit() fails to acquire the lock

The log of commit edaac8e3167501cda336231d00611bf59c164346 ("ratelimit:
Fix/allow use in atomic contexts"), indicates that we want to suppress the
callback when the trylock fails.

Signed-off-by: Yong Zhang <yong.zhang@windriver.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>