safe/jmp/linux-2.6
18 years ago[SCSI] convert sg to scsi_execute_async
Mike Christie [Tue, 8 Nov 2005 10:06:41 +0000 (04:06 -0600)]
[SCSI] convert sg to scsi_execute_async

Convert sg to always send scatterlists, and kill scsi_request usage.

TODO:

- move DIO code to common place or make block layers usable for ULDs.
- move buffer allocation code to common place for all ULDs to use. And
make buffer allocation code obey all queue limits so we can find
out about problems before calling scsi_execute_async. Currently, sg.c
could allocate a buffer that is too large, and send the request
to scsi_execute_async. scsi_execute_async will then check it against
all the queue limits and return a failure in this case. It would nicer
to know about the queue limit violation right away.
- move indirect (copy_to/from_user) paths commone place or make block
layers usable for ULDs.

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] add kmemcache for scsi_io_context
Mike Christie [Fri, 11 Nov 2005 11:31:41 +0000 (05:31 -0600)]
[SCSI] add kmemcache for scsi_io_context

Add kmemcache of scsi io contexts.

In the future when we finalize on where these functions will live
we can add a mempool for it and do a bioset for out REQ_BLOCK_PC
bios. This is needed becuase the dm-multipath handlers will
want to use the scsi_exectute* functions for failover and we cannot
have them and the bio device allocating from the same mempool.

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] complete the whole command when it is REQ_BLOCK_PC
Mike Christie [Fri, 11 Nov 2005 11:31:40 +0000 (05:31 -0600)]
[SCSI] complete the whole command when it is REQ_BLOCK_PC

sd does not allow scsi_io_completion to retry commands for
SG_IO requests, and it make sense that it should not happen for st
SG_IO commands too. If for st we hit the bottom of scsi_io_completion
we will probably screw things up pretty bad. This patch returns to the
block layer that the whole command completed and relies on the caller to check
the request errors field. For initialization commands like in sd, this adds
the previous behavior where scsi_io_completion did not process the error.

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] add retries field to request for REQ_BLOCK_PC use
Mike Christie [Fri, 11 Nov 2005 11:31:37 +0000 (05:31 -0600)]
[SCSI] add retries field to request for REQ_BLOCK_PC use

For tape we need to control the retries. This patch adds a retries
counter on the request for REQ_BLOCK_PC commands originating from
scsi_execute* to use. REQ_BLOCK_PC commands comming from the block
layer SG_IO path continue to use the retires set in the ULD init_command.
(scsi_execute* does not set the gendisk so we do not execute
the init_command in that path).

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] Convert SCSI mid-layer to scsi_execute_async
Mike Christie [Fri, 11 Nov 2005 11:30:27 +0000 (05:30 -0600)]
[SCSI] Convert SCSI mid-layer to scsi_execute_async

Add scsi helpers to create really-large-requests and convert
scsi-ml to scsi_execute_async().

Per Jens's previous comments, I placed this function in scsi_lib.c.
I made it follow all the queue's limits - I think I did at least :), so
I removed the warning on the function header.

I think the scsi_execute_* functions should eventually take a request_queue
and be placed some place where the dm-multipath hw_handler can use them
if that failover code is going to stay in the kernel. That conversion
patch will be sent in another mail though.

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] export blk layer functions needed for blk_execute_rq_nowait
Mike Christie [Fri, 11 Nov 2005 11:30:24 +0000 (05:30 -0600)]
[SCSI] export blk layer functions needed for blk_execute_rq_nowait

To send async requests we need these two functions exported.

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] handle scsi_add_host failure for aic7xxx and fix compiler warning
Jesper Juhl [Wed, 14 Dec 2005 18:27:20 +0000 (19:27 +0100)]
[SCSI] handle scsi_add_host failure for aic7xxx and fix compiler warning

Add scsi_add_host() failure handling for aic7xxx
Also silence a compiler warning :
 drivers/scsi/aic7xxx/aic7xxx_osm.c: In function `ahc_linux_register_host':
 drivers/scsi/aic7xxx/aic7xxx_osm.c:1100: warning: ignoring return value of `scsi_add_host', declared with attribute warn_unused_result

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] handle scsi_add_host failure for aic79xx and fix compiler warning
Jesper Juhl [Wed, 14 Dec 2005 18:27:28 +0000 (19:27 +0100)]
[SCSI] handle scsi_add_host failure for aic79xx and fix compiler warning

Add scsi_add_host() failure handling for aic79xx
Also silence a compiler warning :
 drivers/scsi/aic7xxx/aic79xx_osm.c: In function `ahd_linux_register_host':
 drivers/scsi/aic7xxx/aic79xx_osm.c:1099: warning: ignoring return value of `scsi_add_host', declared with attribute warn_unused_result

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] Make scsi_transport_spi.h includable by itself
Matthew Wilcox [Tue, 13 Dec 2005 17:29:31 +0000 (10:29 -0700)]
[SCSI] Make scsi_transport_spi.h includable by itself

Add forward declarations to allow scsi_transport_spi.h to be compiled by
itself.

Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] sd: Always do write-protect check
Alan Stern [Fri, 9 Dec 2005 16:34:45 +0000 (11:34 -0500)]
[SCSI] sd: Always do write-protect check

Since nobody has offered an explanation for why the sd driver makes a
write-protect check only for devices with removable media, I'm submitting
this patch to get rid of the removable-media test.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] pci_ids.h: add subclass code for SAS Controllers
Moore, Eric Dean [Thu, 1 Dec 2005 23:51:02 +0000 (16:51 -0700)]
[SCSI] pci_ids.h: add subclass code for SAS Controllers

Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] mptfusion - bump version
Moore, Eric Dean [Thu, 17 Nov 2005 01:54:27 +0000 (18:54 -0700)]
[SCSI] mptfusion - bump version

Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] mptfusion - mapping fixs required support for transport layers.
Moore, Eric Dean [Thu, 17 Nov 2005 01:54:25 +0000 (18:54 -0700)]
[SCSI] mptfusion - mapping fixs required support for transport layers.

This utilizes the hostdata area that is hung off of scsi_target and
scsi_device for saving unique firmware mapping. This will be required
for supporting new Fibre and SPI transport support.

This also fixs problems in error handling error code for SAS
controllers, in which the incorrect mapping was passed to the
firmware.

Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] mptfusion - prep for removing domain validation
Moore, Eric Dean [Thu, 17 Nov 2005 01:54:23 +0000 (18:54 -0700)]
[SCSI] mptfusion - prep for removing domain validation

This moves some functions around from within the #define
MPTSCSIH_ENABLE_DOMAIN_VALIDATION area, in preperation for generic
domain validation.

Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] mptfusion - bus_type, change SCSI to SPI
Moore, Eric Dean [Thu, 17 Nov 2005 01:54:20 +0000 (18:54 -0700)]
[SCSI] mptfusion - bus_type, change SCSI to SPI

This changes to SPI for the bus_type enumeration.

Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] mptfusion - cleaning up xxx_probe error handling
Moore, Eric Dean [Thu, 17 Nov 2005 01:54:17 +0000 (18:54 -0700)]
[SCSI] mptfusion - cleaning up xxx_probe error handling

This cleans the returning failure conditions of the
mptsas/mptfc/mptspi probe routines.

Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] mptfusion - adding = THIS_MODULE
Moore, Eric Dean [Thu, 17 Nov 2005 01:54:14 +0000 (18:54 -0700)]
[SCSI] mptfusion - adding = THIS_MODULE

Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] lpfc 8.1.1 : Change version number to 8.1.1
James.Smart@Emulex.Com [Mon, 28 Nov 2005 16:42:49 +0000 (11:42 -0500)]
[SCSI] lpfc 8.1.1 : Change version number to 8.1.1

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] lpfc 8.1.1 : kill use of pci_read_config_xxx
James.Smart@Emulex.Com [Mon, 28 Nov 2005 20:08:56 +0000 (15:08 -0500)]
[SCSI] lpfc 8.1.1 : kill use of pci_read_config_xxx

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] lpfc 8.1.1 : Added code to adjust lun queue depth to avoid target overloading
James.Smart@Emulex.Com [Mon, 28 Nov 2005 16:42:38 +0000 (11:42 -0500)]
[SCSI] lpfc 8.1.1 : Added code to adjust lun queue depth to avoid target overloading

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] lpfc 8.1.1 : Add polled-mode support
James.Smart@Emulex.Com [Tue, 29 Nov 2005 21:32:13 +0000 (16:32 -0500)]
[SCSI] lpfc 8.1.1 : Add polled-mode support

- Add functionality to run in polled mode only. Includes run time
  attribute to enable mode.
- Enable runtime writable hba settings for coallescing and delay parameters

Customers have requested a mode in the driver to run strictly polled.
This is generally to support an environment where the server is extremely
loaded and is looking to reclaim some cpu cycles from adapter interrupt
handling.

This patch adds a new "poll" attribute, and the following behavior:

if value is 0 (default):
  The driver uses the normal method for i/o completion. It uses the
  firmware feature of interrupt coalesing. The firmware allows a
  minimum number of i/o completions before an interrupt, or a maximum
  time delay between interrupts.  By default, the driver sets these
  to no delay (disabled) or 1 i/o - meaning coalescing is disabled.

  Attributes were provided to change the coalescing values, but it was
  a module-load time only and global across all adapters.
  This patch allows them to be writable on a per-adapter basis.

if value is 1 :
  Interrupts are left enabled, expecting that the user has tuned the
  interrupt coalescing values. When this setting is enabled, the driver
  will attempt to service completed i/o whenever new i/o is submitted
  to the adapter. If the coalescing values are large, and the i/o
  generation rate steady, an interrupt will be avoided by servicing
  completed i/o prior to the coalescing thresholds kicking in. However,
  if the i/o completion load is high enough or i/o generation slow, the
  coalescion values will ensure that completed i/o is serviced in a timely
  fashion.

if value is 3 :
  Turns off FCP i/o interrupts altogether. The coalescing values now have
  no effect. A new attribute "poll_tmo" (default 10ms) exists to set
  the polling interval for i/o completion. When this setting is enabled,
  the driver will attempt to service completed i/o and restart the
  interval timer whenever new i/o is submitted. This behavior allows for
  servicing of completed i/o sooner than the interval timer, but ensures
  that if no i/o is being issued, then the interval timer will kick in
  to service the outstanding i/o.

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] lpfc 8.1.1 : Bring model descriptions in sync with Emulex standard generic...
James.Smart@Emulex.Com [Mon, 28 Nov 2005 16:42:19 +0000 (11:42 -0500)]
[SCSI] lpfc 8.1.1 : Bring model descriptions in sync with Emulex standard generic names

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] lpfc 8.1.1 : Add support for more members of the Light Pulse 11xxx (4Gb) family
James.Smart@Emulex.Com [Mon, 28 Nov 2005 16:42:12 +0000 (11:42 -0500)]
[SCSI] lpfc 8.1.1 : Add support for more members of the Light Pulse 11xxx (4Gb) family

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] lpfc 8.1.1 : Fixes to error handlers
James.Smart@Emulex.Com [Mon, 28 Nov 2005 16:42:05 +0000 (11:42 -0500)]
[SCSI] lpfc 8.1.1 : Fixes to error handlers

- Release task management command before counting outstanding commands.
  TMF was being erroneously counted as an active outstanding command.
- Serialize EH calls and block requests when EH function is running.

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] lpfc 8.1.1 : Remove locking wrappers around error handlers
James.Smart@Emulex.Com [Mon, 28 Nov 2005 16:41:53 +0000 (11:41 -0500)]
[SCSI] lpfc 8.1.1 : Remove locking wrappers around error handlers

Remove locking wrappers around error handlers. Wrappers were added in
early 2.6.13 api change

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] lpfc 8.1.1 : Adjust use of scsi_block_requests and interaction w/ FC transport
James.Smart@Emulex.Com [Mon, 28 Nov 2005 16:41:44 +0000 (11:41 -0500)]
[SCSI] lpfc 8.1.1 : Adjust use of scsi_block_requests and interaction w/ FC transport

- Remove unnecessary scsi_block_requests calls on rport deletes.
  This was deadlocking the sdev removals as they wanted to flush commands.
- No longer block requests when adding the remote port (to block
  discovery). Instead, register, then change port role. Maps to Qlogic
  behavior, and closer to the register-node-upon-first-ELS behavior.

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] lpfc 8.1.1 : Fixes for short cable pulls
James.Smart@Emulex.Com [Mon, 28 Nov 2005 16:41:33 +0000 (11:41 -0500)]
[SCSI] lpfc 8.1.1 : Fixes for short cable pulls

Cause: Link bounces were causing discovery ELS's to be killed.
Driver was not properly flushing ELS commands upon the subsequent
link bounces. Thus, processing of ELS post link bounce erroneously
assumed discovery failure and device loss.

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] lpfc 8.1.1 : Correct some 8bit to 16bit field conversions/comparisons
James.Smart@Emulex.Com [Mon, 28 Nov 2005 16:41:24 +0000 (11:41 -0500)]
[SCSI] lpfc 8.1.1 : Correct some 8bit to 16bit field conversions/comparisons

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] lpfc 8.1.1: Miscellaneous Cleanups
James.Smart@Emulex.Com [Mon, 28 Nov 2005 16:41:15 +0000 (11:41 -0500)]
[SCSI] lpfc 8.1.1: Miscellaneous Cleanups

Miscellaneous Cleanups:
- Remove ProgType READ_REV mailbox command value check in lpfc_config_port_prep.
- Convert simple printk to an lpfc_printf_log in queuecommand.
- Modify lpfc_abort_handler message 0749 to display more accurate text and data.
- Minor style cleanup: fix 3 long lines in lpfc_hw.h

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] qla2xxx: fix compile error caused by pci_dev.owner move
James Bottomley [Thu, 1 Dec 2005 18:51:50 +0000 (12:51 -0600)]
[SCSI] qla2xxx: fix compile error caused by pci_dev.owner move

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] correct some dropped const compiler warnings
James Bottomley [Thu, 1 Dec 2005 18:50:13 +0000 (12:50 -0600)]
[SCSI] correct some dropped const compiler warnings

Make the vendor, model and rev fields in scsi_device pointers to const
and update a few prototypes of functions using them.

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] iscsi: check header digests for mgmt tasks
Mike Christie [Wed, 30 Nov 2005 08:27:19 +0000 (02:27 -0600)]
[SCSI] iscsi: check header digests for mgmt tasks

From Wang Zhenyu:

check header digest for cmd and mgmt tasks

Signed-off-by: Wang Zhenyu <zhenyu.z.wang@intel.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] iscsi: update version
Mike Christie [Wed, 30 Nov 2005 05:13:01 +0000 (23:13 -0600)]
[SCSI] iscsi: update version

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] iscsi: lower queue depth
Mike Christie [Wed, 30 Nov 2005 05:12:59 +0000 (23:12 -0600)]
[SCSI] iscsi: lower queue depth

From Wang Zhenyu:

High queue depth was a problem for some targets so make queue_depth adjustable

From Mike Christie

Make default queue_depth a little lower

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] iscsi: data digest calculation fix
Mike Christie [Wed, 30 Nov 2005 05:12:57 +0000 (23:12 -0600)]
[SCSI] iscsi: data digest calculation fix

From Wang Zhenyu:

data digest fix (the bug caused data corruption w/Wasabi StorageBuilder target)

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] iscsi: iscsi response fix
Mike Christie [Wed, 30 Nov 2005 05:12:54 +0000 (23:12 -0600)]
[SCSI] iscsi: iscsi response fix

from Wang Zhenyu:

Must check SCSI CMD and R2T response according to the spec

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] iscsi: redirect fix
Mike Christie [Wed, 30 Nov 2005 05:12:49 +0000 (23:12 -0600)]
[SCSI] iscsi: redirect fix

From tomof@acm.org:

There is one more issue about Equallogic systems. They send
re-direction info with FIN. I think that the kernel module needs to
let iscsid to read data from the socket before killing it.

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] iscsi: opcode check fix
Mike Christie [Wed, 30 Nov 2005 05:12:46 +0000 (23:12 -0600)]
[SCSI] iscsi: opcode check fix

Must check only valid opcode bits.

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] sym2: Version 2.2.2
Matthew Wilcox [Wed, 30 Nov 2005 04:08:48 +0000 (23:08 -0500)]
[SCSI] sym2: Version 2.2.2

Update version number to 2.2.2

Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] sym2: Report disabled devices and LUNs more attractively
Matthew Wilcox [Wed, 30 Nov 2005 04:08:46 +0000 (23:08 -0500)]
[SCSI] sym2: Report disabled devices and LUNs more attractively

Rather than print a list of targets at driver init time, print each
disabled target as we attempt to scan it.

Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] sym2: Allow NVRAM settings to limit speed and width
Matthew Wilcox [Wed, 30 Nov 2005 04:08:44 +0000 (23:08 -0500)]
[SCSI] sym2: Allow NVRAM settings to limit speed and width

The NVRAM for both Tekram and Symbios boards allows the user to set the
speed and width for individual targets.  I took that code out in March
2004 when we introduced Domain Validation, but it seems there's still
a legitimate need for it in some configurations.

Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] sym2: Use scsi_print_msg
Matthew Wilcox [Wed, 30 Nov 2005 04:08:42 +0000 (23:08 -0500)]
[SCSI] sym2: Use scsi_print_msg

sym_show_msg was almost a duplicate of scsi_print_msg, except not as
featureful.  So use the common code instead.

Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] sym2: Use DMA_40BIT_MASK constant
Matthew Wilcox [Wed, 30 Nov 2005 04:08:36 +0000 (23:08 -0500)]
[SCSI] sym2: Use DMA_40BIT_MASK constant

Now that this constant has been added to dma-mapping.h, we don't need our
own definition

Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] sym2: Remove code to handle DMA_BIDIRECTION requests
Matthew Wilcox [Wed, 30 Nov 2005 04:08:33 +0000 (23:08 -0500)]
[SCSI] sym2: Remove code to handle DMA_BIDIRECTION requests

The upper layer doesn't send these down since 2.4.x (or 2.6 in
practice), so no need to handle it.  Inline sym_setup_data_pointers
into its only caller so we can fail gracefully in the case we'd get
one neverless.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] sym2: Manage sym_lcb properly
Matthew Wilcox [Wed, 30 Nov 2005 04:08:31 +0000 (23:08 -0500)]
[SCSI] sym2: Manage sym_lcb properly

Allocate the lcb in slave_alloc and free it in slave_destroy.  This allows
us to remove all the code that checks to see if it's already been allocated.

From: Christoph Hellwig <hch@lst.de>
Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] sym2: Remove last vestiges of sym_sniff_inquiry
Matthew Wilcox [Wed, 30 Nov 2005 04:08:29 +0000 (23:08 -0500)]
[SCSI] sym2: Remove last vestiges of sym_sniff_inquiry

The SYM_OPT_SNIFF_INQUIRY define is never set any more, and the
sym_sniff_inquiry() function doesn't exist

From: Christoph Hellwig <hch@lst.de>
Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] sym2: Remove FreeBSD ifdefs
Matthew Wilcox [Wed, 30 Nov 2005 04:08:27 +0000 (23:08 -0500)]
[SCSI] sym2: Remove FreeBSD ifdefs

Remove FreeBSD ifdefs from sym2 driver

Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] Mention scsi_scan_host() in scsi_mid_low_api.txt
Matthew Wilcox [Thu, 10 Nov 2005 14:45:55 +0000 (07:45 -0700)]
[SCSI] Mention scsi_scan_host() in scsi_mid_low_api.txt

Update to mention scsi_scan_host()

Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] qla2xxx: Resync with latest released ISP24xx firmware -- 4.00.16.
Andrew Vasquez [Wed, 9 Nov 2005 23:49:33 +0000 (15:49 -0800)]
[SCSI] qla2xxx: Resync with latest released ISP24xx firmware -- 4.00.16.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] qla2xxx: Add support for embedded ISP24xx firmware.
Andrew Vasquez [Wed, 9 Nov 2005 23:49:19 +0000 (15:49 -0800)]
[SCSI] qla2xxx: Add support for embedded ISP24xx firmware.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] qla2xxx: Add full firmware(-request) hotplug support for all ISPs.
Andrew Vasquez [Wed, 9 Nov 2005 23:49:04 +0000 (15:49 -0800)]
[SCSI] qla2xxx: Add full firmware(-request) hotplug support for all ISPs.

Transition driver to exclusively use the request_firmware()
interfaces to retrieve firmware-blobs from user-space.  This
will be the default behaviour going forward until the
embedded firmware-binary images are removed from the
upstream kernel.

Upon request, the driver caches the firmware image until the
driver is unloaded.

NOTE: The option is present to allow the user to continue to
use the firmware-loader modules, but, should be considered
deprecated.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Rejections fixed up and
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] scsi_lib: stricter checks for clearing use_10_for_rw
Jens Axboe [Tue, 29 Nov 2005 20:03:34 +0000 (21:03 +0100)]
[SCSI] scsi_lib: stricter checks for clearing use_10_for_rw

Check the asc and ascq for being "invalid command opcode" as well.

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] Delete trailing full stop
Matthew Wilcox [Thu, 17 Nov 2005 18:13:43 +0000 (11:13 -0700)]
[SCSI] Delete trailing full stop

None of the other domain validation messages have a trailing full stop,
so I don't see why this one should.

Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] Mark some core scsi data structures const
Arjan van de Ven [Mon, 28 Nov 2005 15:22:25 +0000 (16:22 +0100)]
[SCSI] Mark some core scsi data structures const

patch below marks a few scsi core datastructures as const, so that they end up
in the .rodata section and don't cacheline share with things that get dirtied

Signed-off-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] ipr: Driver initialization fix for kexec/kdump
brking@us.ibm.com [Thu, 17 Nov 2005 15:35:12 +0000 (09:35 -0600)]
[SCSI] ipr: Driver initialization fix for kexec/kdump

When kexec booting a kernel when the previous kernel did not
call ipr's shutdown method, the ipr adapter does not get
properly initialized, which can result in the ipr adapter
completing commands issued by the previous kernel. Fix ipr
to detect this scenario by reading the adapter's interrupt
mask register and the microprocessor interrupt register.
If the interrupt mask register indicates that interrupts
are enabled or the reset alert bit is set when the card is
probed, this means the card is in an unknown state and we
hard reset the card.

Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] megaraid_{mbox,mm} : remove PCI Id overlaping between megaraid_legacy and...
Ju, Seokmann [Thu, 17 Nov 2005 18:17:25 +0000 (13:17 -0500)]
[SCSI] megaraid_{mbox,mm} : remove PCI Id overlaping between megaraid_legacy and megaraid_{mbox,mm}

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[SCSI] megaraid_legacy: removed PCI ID overlap from the driv er
Ju, Seokmann [Thu, 17 Nov 2005 18:13:31 +0000 (13:13 -0500)]
[SCSI] megaraid_legacy: removed PCI ID overlap from the driv er

This patch fixes
- PCI ID overlap issue
- node name changed to 'megaraid_legacy'
I hope this patch addresses concerns brought by Daniel Drake.

Signed-off by: Seokmann Ju <seokmann.ju@enginio.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[PATCH] fbdev: Fix incorrect unaligned access in little-endian machines
Antonino A. Daplas [Tue, 13 Dec 2005 06:17:21 +0000 (22:17 -0800)]
[PATCH] fbdev: Fix incorrect unaligned access in little-endian machines

The drawing function cfbfillrect does not work correctly when access is not
unsigned-long aligned.  It manifests as extra lines of pixels that are not
complete drawn.  Reversing the shift operator solves the problem, so I would
presume that this bug would manifest only on little endian machines.  The
function cfbcopyarea may also have this bug.

Aligned access should present no problems.

Signed-off-by: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fbdev: Shift pixel value before entering loop in cfbimageblit
Antonino A. Daplas [Tue, 13 Dec 2005 06:17:20 +0000 (22:17 -0800)]
[PATCH] fbdev: Shift pixel value before entering loop in cfbimageblit

In slow imageblit, the pixel value is shifted by a certain amount (dependent
on the bpp and endianness) for each iteration.  This is inefficient.  Better
do the shifting once before going into the loop.

Signed-off-by: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fbdev: fix switch to KD_TEXT, enhanced version
Knut Petersen [Tue, 13 Dec 2005 06:17:19 +0000 (22:17 -0800)]
[PATCH] fbdev: fix switch to KD_TEXT, enhanced version

Every framebuffer driver relies on the assumption that the set_par()
function of the driver is called before drawing functions and other
functions dependent on the hardware state are executed.

Whenever you switch from X to a framebuffer console for the very first
time, there is a chance that a broken X system has _not_ set the mode to
KD_GRAPHICS, thus the vt and framebuffer code executes a screen redraw and
several other functions before a set_par() is executed.  This is believed
to be not a bug of linux but a bug of X/xdm.  At least some X releases used
by SuSE and Debian show this behaviour.

There was a 2nd case, but that has been fixed by Antonino Daplas on
10-dec-2005.

This patch allows drivers to set a flag to inform fbcon_switch() that they
prefer a set_par() call on every console switch, working around the
problems caused by the broken X releases.

The flag will be used by the next release of cyblafb and might help other
drivers that assume a hardware state different to the one used by X.

As the default behaviour does not change, this patch should be acceptable
to everybody.

Signed-off-by: Knut Petersen <Knut_Petersen@t-online.de>
Acked-by: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fbcon: Avoid illegal display panning
Antonino A. Daplas [Tue, 13 Dec 2005 06:17:18 +0000 (22:17 -0800)]
[PATCH] fbcon: Avoid illegal display panning

Avoid calls to fb_pan_display when driver is suspended or not in text mode.

Signed-off-by: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fbdev: Pan display fixes
Antonino A. Daplas [Tue, 13 Dec 2005 06:17:17 +0000 (22:17 -0800)]
[PATCH] fbdev: Pan display fixes

- Fix fb_pan_display rejecting yoffsets that are valid if panning mode
  is ywrap.

- Add more robust error checking in fb_pan_display specially since this
  function is accessible by userland apps.

Signed-off-by: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fbcon: Add ability to save/restore graphics state
Antonino A. Daplas [Tue, 13 Dec 2005 06:17:16 +0000 (22:17 -0800)]
[PATCH] fbcon: Add ability to save/restore graphics state

Add hooks to save and restore the graphics state.  These hooks are called in
fbcon_blank() when entering/leaving KD_GRAPHICS mode.  This is needed by
savagefb at least so it can cooperate with savage_dri and by cyblafb.

State save/restoration can be full or partial.

Signed-off-by: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fbcon: fix complement_mask() with 512 character map
Antonino A. Daplas [Tue, 13 Dec 2005 06:17:15 +0000 (22:17 -0800)]
[PATCH] fbcon: fix complement_mask() with 512 character map

There is a bug in the complement_mask when you have a 512-character map.
Linux boots to a default 256-character map and most probably your login
profile is loading a 512-character map which results in a bad gpm cursor.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cciss: fix for deregister_disk
Mike Miller [Tue, 13 Dec 2005 06:17:14 +0000 (22:17 -0800)]
[PATCH] cciss: fix for deregister_disk

This patch adds setting our drv->queue = NULL back in deregister_disk.  The
drv->queue is part of our controller struct.  blk_cleanup_queue works only
on the queue in the gendisk struct.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: Jens Axboe <axboe@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] x86_64: Bug correction in populate_memnodemap()
Eric Dumazet [Tue, 13 Dec 2005 06:17:14 +0000 (22:17 -0800)]
[PATCH] x86_64: Bug correction in populate_memnodemap()

As reported by Keith Mannthey, there are problems in populate_memnodemap()

The bug was that the compute_hash_shift() was returning 31, with incorrect
initialization of memnodemap[]

To correct the bug, we must use (1UL << shift) instead of (1 << shift) to
avoid an integer overflow, and we must check that shift < 64 to avoid an
infinite loop.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] x86_64: Fix collision between pmtimer and pit/hpet
john stultz [Tue, 13 Dec 2005 06:17:13 +0000 (22:17 -0800)]
[PATCH] x86_64: Fix collision between pmtimer and pit/hpet

On systems that do not support the HPET legacy functions (basically the IBM
x460, but there could be others), in time_init() we accidentally fall into a
PM timer conditional and set the vxtime_hz value to the PM timer's frequency.
We then use this value with the HPET for timekeeping.

This patch (which mimics the behavior in time_init_gtod) corrects the
collision.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] x86_64: Fix 32bit thread coredumps
Andi Kleen [Tue, 13 Dec 2005 06:17:12 +0000 (22:17 -0800)]
[PATCH] x86_64: Fix 32bit thread coredumps

When a register set is passed in don't try to fix up the pointer.

Noticed by Al Viro

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] i386/x86-64 Correct for broken MCFG tables on K8 systems
Andi Kleen [Tue, 13 Dec 2005 06:17:11 +0000 (22:17 -0800)]
[PATCH] i386/x86-64 Correct for broken MCFG tables on K8 systems

They report all busses as MMCONFIG capable, but it never works for the
internal devices in the CPU's builtin northbridge.

It just probes all func 0 devices on bus 0 (the internal northbridge is
currently always on bus 0) and if they are not accessible using MCFG they are
put into a special fallback bitmap.

On systems where it isn't we assume the BIOS vendor supplied correct MCFG.

Requires the earlier patch for mmconfig type1 fallback

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] i386/x86-64 Fall back to type 1 access when no entry found
Andi Kleen [Tue, 13 Dec 2005 06:17:10 +0000 (22:17 -0800)]
[PATCH] i386/x86-64 Fall back to type 1 access when no entry found

When there is no entry for a bus in MCFG fall back to type1.  This is
especially important on K8 systems where always some devices can't be accessed
using mmconfig (in particular the builtin northbridge doesn't support it for
its own devices)

Cc: <gregkh@suse.de>
Cc: <jgarzik@pobox.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] i386/x86-64: Don't call change_page_attr with a spinlock held
Andi Kleen [Tue, 13 Dec 2005 06:17:09 +0000 (22:17 -0800)]
[PATCH] i386/x86-64: Don't call change_page_attr with a spinlock held

It's illegal because it can sleep.

Use a two step lookup scheme instead.  First look up the vm_struct, then
change the direct mapping, then finally unmap it.  That's ok because nobody
can change the particular virtual address range as long as the vm_struct is
still in the global list.

Also added some LinuxDoc documentation to iounmap.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] i386/x86-64 disable LAPIC completely for offline CPU
Shaohua Li [Tue, 13 Dec 2005 06:17:08 +0000 (22:17 -0800)]
[PATCH] i386/x86-64 disable LAPIC completely for offline CPU

Disabling LAPIC timer isn't sufficient.  In some situations, such as we
enabled NMI watchdog, there is still unexpected interrupt (such as NMI)
invoked in offline CPU.  This also avoids offline CPU receives spurious
interrupt and anything similar.

Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Acked-by: "Seth, Rohit" <rohit.seth@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] x86_64: Make sure hpet_address is 0 when any part of HPET initialization...
Andi Kleen [Tue, 13 Dec 2005 06:17:07 +0000 (22:17 -0800)]
[PATCH] x86_64: Make sure hpet_address is 0 when any part of HPET initialization fails

Otherwise TSC->HPET fallback could see incorrect state and crash later.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge branch 'upstream-fixes' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Tue, 13 Dec 2005 05:42:17 +0000 (21:42 -0800)]
Merge branch 'upstream-fixes' of /linux/kernel/git/jgarzik/netdev-2.6

18 years agoMerge branch 'upstream-fixes' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Tue, 13 Dec 2005 05:41:58 +0000 (21:41 -0800)]
Merge branch 'upstream-fixes' of /linux/kernel/git/jgarzik/libata-dev

18 years ago[drm] fix radeon aperture issue
Dave Airlie [Tue, 13 Dec 2005 04:18:41 +0000 (04:18 +0000)]
[drm] fix radeon aperture issue

Ben noticed that on certain cards we've landed the AGP space on top of
the second aperture instead of after it..  Which messes things up a lot
on those machines.

This just moves the gart further out, a more correct fix is in the works
from Ben for after 2.6.15.

Signed-off-by: Dave Airlie <airlied@linux.ie>
CC: Ben Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] ieee80211_crypt_tkip depends on NET_RADIO
Olaf Hering [Thu, 1 Dec 2005 20:22:37 +0000 (21:22 +0100)]
[PATCH] ieee80211_crypt_tkip depends on NET_RADIO

*** Warning: ".wireless_send_event" [net/ieee80211/ieee80211_crypt_tkip.ko] undefined!

Signed-off-by: Olaf Hering <olh@suse.de>
 net/ieee80211/Kconfig |    2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
18 years ago[PATCH] skge: get rid of warning on race
Stephen Hemminger [Tue, 6 Dec 2005 23:01:49 +0000 (15:01 -0800)]
[PATCH] skge: get rid of warning on race

Get rid of warning in case of race with ring full and lockless
tx on the skge driver. It is possible to be in the transmit
routine with no available slots and already stopped.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
18 years ago[PATCH] libata-core.c: fix parameter bug on kunmap_atomic() calls
Mark Lord [Tue, 13 Dec 2005 04:19:28 +0000 (23:19 -0500)]
[PATCH] libata-core.c:  fix parameter bug on kunmap_atomic() calls

Fix incorrect pointer usage on two calls to kunmap_atomic().
This seems to happen a lot, because kunmap() wants the struct page *,
whereas kunmap_atomic() instead wants the mapped virtual address.

Signed-off-by: Mark Lord <liml@rtr.ca>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
18 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6
Linus Torvalds [Tue, 13 Dec 2005 00:48:29 +0000 (16:48 -0800)]
Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

18 years agoMerge branch 'upstream-fixes' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Tue, 13 Dec 2005 00:41:37 +0000 (16:41 -0800)]
Merge branch 'upstream-fixes' of /linux/kernel/git/jgarzik/netdev-2.6

18 years agoget_user_pages: don't try to follow PFNMAP pages
Linus Torvalds [Tue, 13 Dec 2005 00:24:33 +0000 (16:24 -0800)]
get_user_pages: don't try to follow PFNMAP pages

Nick Piggin points out that a few drivers play games with VM_IO (why?
who knows..) and thus a pfn-remapped area may not have that bit set even
if remap_pfn_range() set it originally.

So make it explicit in get_user_pages() that we don't follow VM_PFNMAP
pages, since pretty much by definition they do not have a "struct page"
associated with them.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/nf-2.6
Linus Torvalds [Mon, 12 Dec 2005 23:49:56 +0000 (15:49 -0800)]
Merge /pub/scm/linux/kernel/git/davem/nf-2.6

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Mon, 12 Dec 2005 23:49:45 +0000 (15:49 -0800)]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Mon, 12 Dec 2005 23:49:27 +0000 (15:49 -0800)]
Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

18 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Mon, 12 Dec 2005 23:25:58 +0000 (15:25 -0800)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

18 years agoMerge master.kernel.org:/home/rmk/linux-2.6-mmc
Linus Torvalds [Mon, 12 Dec 2005 23:25:15 +0000 (15:25 -0800)]
Merge master.kernel.org:/home/rmk/linux-2.6-mmc

18 years agoMerge master.kernel.org:/home/rmk/linux-2.6-serial
Linus Torvalds [Mon, 12 Dec 2005 23:24:36 +0000 (15:24 -0800)]
Merge master.kernel.org:/home/rmk/linux-2.6-serial

18 years ago[NETFILTER]: ip_nat_tftp: Fix expectation NAT
Marcus Sundberg [Mon, 12 Dec 2005 23:02:48 +0000 (15:02 -0800)]
[NETFILTER]: ip_nat_tftp: Fix expectation NAT

When a TFTP client is SNATed so that the port is also changed, the
port is never changed back for the expected connection.

Signed-off-by: Marcus Sundberg <marcus@ingate.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC]: block/ needed in final image link
Paul Jackson [Mon, 12 Dec 2005 22:42:44 +0000 (14:42 -0800)]
[SPARC]: block/ needed in final image link

With this fix, sparc links vmlinuz again using crosstool.  Without this
fix, the final link fails missing several dozen dozen symbols, beginning
with:

    kernel/built-in.o(.text+0x6fd0): In function `do_exit':
    : undefined reference to `exit_io_context'

(exit_io_context is defined in block/ll_rw_blk.c).

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC]: atomic_clear_mask build fix
Paul Jackson [Mon, 12 Dec 2005 22:42:31 +0000 (14:42 -0800)]
[SPARC]: atomic_clear_mask build fix

This fixes one build error introduced in sparc with the patch of Oct 30,
resent Nov 4 "[patch 3/5] atomic: atomic_inc_not_zero" I still can't get
sparc to build, but at least it gets further after I remove this line.
Apparently, this change was agreed to by Andrew and Nick on Nov 14, but
everyone thought someone else was doing it.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SBUSFB] tcx: Use FB_BLANK_UNBLANK instead of magic constant.
Hareesh Nagarajan [Mon, 12 Dec 2005 22:42:07 +0000 (14:42 -0800)]
[SBUSFB] tcx: Use FB_BLANK_UNBLANK instead of magic constant.

From: Hareesh Nagarajan <hnagar2@gmail.com>

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SBUSFB]: Kill 'list' member from foo_par structs, totally unused.
David S. Miller [Mon, 12 Dec 2005 22:41:20 +0000 (14:41 -0800)]
[SBUSFB]: Kill 'list' member from foo_par structs, totally unused.

Based upon a patch from Hareesh Nagarajan.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TCPv6]: Fix skb leak
Arnaldo Carvalho de Melo [Mon, 12 Dec 2005 22:38:10 +0000 (14:38 -0800)]
[TCPv6]: Fix skb leak

Spotted by Francois Romieu, thanks!

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[PATCH] Fix SCSI scanning slab corruption
Brian King [Mon, 12 Dec 2005 19:05:08 +0000 (13:05 -0600)]
[PATCH] Fix SCSI scanning slab corruption

There is a double free in the scsi scan code if a LLDD's slave_alloc()
call fails.  There is a direct call to scsi_free_queue and then the
following put_device calls the release function, which also frees the
queue.

Remove the redundant scsi_free_queue.

Signed-off-by: Brian King <brking@us.ibm.com>
Tested-by: Nathan Lynch <ntl@pobox.com>
[ Also removed some strange whitespace artifacts in that area ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] pcnet32: use MAC address from prom also on powerpc
Olaf Hering [Fri, 9 Dec 2005 18:12:10 +0000 (19:12 +0100)]
[PATCH] pcnet32: use MAC address from prom also on powerpc

The CSR contains garbage after a coldboot on RS/6000.
One some systems (like my 44p 270) the MAC address is all FF,
on others (like my B50) it is ff:ff:ff:fd:ff:6b.

It can eventually be fixed by loading pcnet32, set the interface
into the UP state, rmmod pcnet32 and load it again. But this worked
only on the 270.

Only netbooting after a cold start provides the correct MAC address
via prom and CSR. This makes it very unreliable.
I dont know why the MAC is stored in two different places. Remove
the special case for powerpc, which was added in early 2.4 development.

Signed-off-by: Olaf Hering <olh@suse.de>
 drivers/net/pcnet32.c |    5 -----
 1 files changed, 5 deletions(-)
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
18 years agoMerge branch 'master'
Jeff Garzik [Mon, 12 Dec 2005 20:30:15 +0000 (15:30 -0500)]
Merge branch 'master'

18 years agoRevert revert of "[SCSI] fix usb storage oops"
Linus Torvalds [Mon, 12 Dec 2005 19:25:04 +0000 (11:25 -0800)]
Revert revert of "[SCSI] fix usb storage oops"

This reverts commit 1b0997f561bf46689cc6e0903f342e9bf2506bf1, which in
turn reverted 34ea80ec6a02ad02e6b9c75c478c18e5880d6713 (which is thus
re-instated).

Quoth James Bottomley:

  "All it's doing is deferring the device_put() from the
   scsi_put_command() to after the scsi_run_queue(), which doesn't fix
   the sleep while atomic problem of the device release method.  In both
   cases we still get the semaphore in atomic context problem which is
   caused by scsi_reap_target() doing a device_del(), which I assumed
   (wrongly) was valid from atomic context."

who also promised to fix scsi_reap_target().

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: use correct size of raid5 stripe cache when measuring how full it is
NeilBrown [Mon, 12 Dec 2005 10:39:17 +0000 (02:39 -0800)]
[PATCH] md: use correct size of raid5 stripe cache when measuring how full it is

The raid5 stripe cache was recently changed from fixed size (NR_STRIPES) to
variable size (conf->max_nr_stripes).  However there are two places that still
use the constant and as a result, reducing the size of the stripe cache can
result in a deadlock.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: fix a use-after-free bug in raid1
NeilBrown [Mon, 12 Dec 2005 10:39:16 +0000 (02:39 -0800)]
[PATCH] md: fix a use-after-free bug in raid1

Who would submit code with a FIXME like that in it !!!!

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>