pandora-kernel.git
17 years ago[SCSI] fusion : mpi header update
Eric Moore [Tue, 27 Jun 2006 20:39:06 +0000 (14:39 -0600)]
[SCSI] fusion : mpi header update

MPI Header Update

Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] scsi: Device scanning oops for offlined devices (resend)
Brian King [Tue, 27 Jun 2006 16:10:43 +0000 (11:10 -0500)]
[SCSI] scsi: Device scanning oops for offlined devices (resend)

If a device gets offlined as a result of the Inquiry sent
during scanning, the following oops can occur. After the
disk gets put into the SDEV_OFFLINE state, the error handler
sends back the failed inquiry, which wakes the thread doing
the scan. This starts a race between the scanning thread
freeing the scsi device and the error handler calling
scsi_run_host_queues to restart the host. Since the disk
is in the SDEV_OFFLINE state, scsi_device_get will still
work, which results in __scsi_iterate_devices getting
a reference to the scsi disk when it shouldn't.

The following execution thread causes the oops:

CPU 0 (scan) CPU 1 (eh)

---------------------------------------------------------
scsi_probe_and_add_lun
                        ....
                                        scsi_eh_offline_sdevs
                                        scsi_eh_flush_done_q
scsi_destroy_sdev
scsi_device_dev_release
                                        scsi_restart_operations
                                         scsi_run_host_queues
                                          __scsi_iterate_devices
                                           get_device
scsi_device_dev_release_usercontext
                                          scsi_run_queue
                                            <---OOPS--->

The patch fixes this by changing the state of the sdev to SDEV_DEL
before doing the final put_device, which should prevent the race
from occurring.

Original oops follows:

Badness in kref_get at lib/kref.c:32
Call Trace:
[C00000002F4476D0] [C00000000000EE20] .show_stack+0x68/0x1b0 (unreliable)
[C00000002F447770] [C00000000037515C] .program_check_exception+0x1cc/0x5a8
[C00000002F447840] [C00000000000446C] program_check_common+0xec/0x100
 Exception: 700 at .kref_get+0x10/0x28
    LR = .kobject_get+0x20/0x3c
[C00000002F447B30] [C00000002F447BC0] 0xc00000002f447bc0 (unreliable)
[C00000002F447BB0] [C000000000254BDC] .get_device+0x20/0x3c
[C00000002F447C30] [D000000000063188] .scsi_device_get+0x34/0xdc [scsi_mod]
[C00000002F447CC0] [D0000000000633EC] .__scsi_iterate_devices+0x50/0xbc [scsi_mod]
[C00000002F447D60] [D00000000006A910] .scsi_run_host_queues+0x34/0x5c [scsi_mod]
[C00000002F447DF0] [D000000000069054] .scsi_error_handler+0xdb4/0xe44 [scsi_mod]
[C00000002F447EE0] [C00000000007B4E0] .kthread+0x128/0x178
[C00000002F447F90] [C000000000025E84] .kernel_thread+0x4c/0x68
Unable to handle kernel paging request for <7>PCI: Enabling device: (0002:41:01.1), cmd 143
data at address 0x000001b8
Faulting instruction address: 0xd0000000000698e4
sym1: <1010-66> rev 0x1 at pci 0002:41:01.1 irq 216
sym1: No NVRAM, ID 7, Fast-80, LVD, parity checking
sym1: SCSI BUS has been reset.
scsi2 : sym-2.2.2
cpu 0x0: Vector: 300 (Data Access) at [c00000002f447a30]
    pc: d0000000000698e4: .scsi_run_queue+0x2c/0x218 [scsi_mod]
    lr: d00000000006a904: .scsi_run_host_queues+0x28/0x5c [scsi_mod]
    sp: c00000002f447cb0
   msr: 9000000000009032
   dar: 1b8
 dsisr: 40000000
  current = 0xc0000000045fecd0
  paca    = 0xc00000000048ee80
    pid   = 1123, comm = scsi_eh_1
enter ? for help
[c00000002f447d60d00000000006a904 .scsi_run_host_queues+0x28/0x5c [scsi_mod]
[c00000002f447df0d000000000069054 .scsi_error_handler+0xdb4/0xe44 [scsi_mod]
[c00000002f447ee0c00000000007b4e0 .kthread+0x128/0x178
[c00000002f447f90c000000000025e84 .kernel_thread+0x4c/0x68

Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] scsi: Add allow_restart sysfs class attribute
Brian King [Tue, 27 Jun 2006 16:10:31 +0000 (11:10 -0500)]
[SCSI] scsi: Add allow_restart sysfs class attribute

This is a resend of a patch I generated in response to an email sent
by Ruben Faelens <parasietje@gmail.com>. His original email to
linux-scsi requested a method in which he could spin down a scsi disk
when not in use and have the kernel automatically spin it back up when
an I/O was generated to the disk. The infrastructure to automatically
spin a disk up has been in the scsi error handler for some time now,
but it is not enabled by default. This patch adds an sd sysfs attribute
which allows userspace to enable this behavior.

Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] fc transport: bug fix: correct references
James Smart [Mon, 26 Jun 2006 18:19:59 +0000 (14:19 -0400)]
[SCSI] fc transport: bug fix: correct references

Original post was incorrect as it didn't realize that we already had
a self-referenc due to device_initialize(), and we were really only
missing the put on our own reference. This was hidden by the other bug
which had the midlayer reusing stargets after they were already free,
which was doing too many puts on our rport.

Updating FC transport for:
- Add put in fc_rport_final_delete(), to release the rport.
  Prior, we were leaving the rport with a reference, thus the shost
  with references, etc. If the driver was unloaded, shosts and rports
  remained, along with work threads, etc
- Fix fc_rport_create failure path - too many put's on parent
- Add commenting to easily track ref taking.

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] update max sdev block limit
James Smart [Tue, 16 May 2006 20:13:36 +0000 (16:13 -0400)]
[SCSI] update max sdev block limit

Updated patch to address comments from Pat Mansfield and Michael Reed:
Bumped max to 600 (10mins). Set default dev_loss_tmo to a value other
than the max (30s).

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] fc transport: resolve scan vs delete deadlocks
James Smart [Thu, 11 May 2006 17:27:09 +0000 (13:27 -0400)]
[SCSI] fc transport: resolve scan vs delete deadlocks

In a prior posting to linux-scsi on the fc transport and workq
deadlocks, we noted a second error that did not have a patch:
  http://marc.theaimsgroup.com/?l=linux-scsi&m=114467847711383&w=2
  - There's a deadlock where scsi_remove_target() has to sit behind
    scsi_scan_target() due to contention over the scan_lock().

Subsequently we posted a request for comments about the deadlock:
  http://marc.theaimsgroup.com/?l=linux-scsi&m=114469358829500&w=2

This posting resolves the second error. Here's what we now understand,
and are implementing:

  If the lldd deletes the rport while a scan is active, the sdev's queue
  is blocked which stops the issuing of commands associated with the scan.
  At this point, the scan stalls, and does so with the shost->scan_mutex held.
  If, at this point, if any scan or delete request is made on the host, it
  will stall waiting for the scan_mutex.

  For the FC transport, we queue all delete work to a single workq.
  So, things worked fine when competing with the scan, as long as the
  target blocking the scan was the same target at the top of our delete
  workq, as the delete workq routine always unblocked just prior to
  requesting the delete.  Unfortunately, if the top of our delete workq
  was for a different target, we deadlock.  Additionally, if the target
  blocking scan returned, we were unblocking it in the scan workq routine,
  which really won't execute until the existing stalled scan workq
  completes (e.g. we're re-scheduling it while it is in the midst of its
  execution).

  This patch moves the unblock out of the workq routines and moves it to
  the context that is scheduling the work. This ensures that at some point,
  we will unblock the target that is blocking scan.  Please note, however,
  that the deadlock condition may still occur while it waits for the
  transport to timeout an unblock on a target.  Worst case, this is bounded
  by the transport dev_loss_tmo (default: 30 seconds).

Finally, Michael Reed deserves the credit for the bulk of this patch,
analysis, and it's testing. Thank you for your help.

Note: The request for comments statements about the gross-ness of the
  scan_mutex still stand.

Signed-off-by: Michael Reed <mdr@sgi.com>
Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] Block I/O while SG reset operation in progress - lpfc portion
James Smart [Wed, 8 Mar 2006 19:50:18 +0000 (14:50 -0500)]
[SCSI] Block I/O while SG reset operation in progress - lpfc portion

This removes the duplicate functionality which had been added to
the lpfc driver.

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] Block I/O while SG reset operation in progress - the midlayer patch
James Smart [Wed, 8 Mar 2006 19:50:12 +0000 (14:50 -0500)]
[SCSI] Block I/O while SG reset operation in progress - the midlayer patch

The scsi midlayer portion of the patch

Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] aacraid: remove x86_64 IOMMU dependent code
Salyzyn, Mark [Mon, 26 Jun 2006 12:37:39 +0000 (08:37 -0400)]
[SCSI] aacraid: remove x86_64 IOMMU dependent code

This may seem like a DILLIGAF, but after chatting with the F/W folks,
there is no harm in dropping the page calculation as denoted in the
enclosed patch for these older adapters in this new age of 4GB+ memory
sticks. Any resource optimization within the old-old-old adapters for
systems with less than 4G of memory is of little consequence. The
existing AAC_QUIRK_31BIT flag in linit.c should look after the rest of
the legacy hardware DMA limitations.

Signed-off-by: Mark Salyzyn <aacraid@adaptec.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] random: remove redundant SA_SAMPLE_RANDOM from NinjaSCSI
Matt Mackall [Sun, 25 Jun 2006 08:58:54 +0000 (01:58 -0700)]
[SCSI] random: remove redundant SA_SAMPLE_RANDOM from NinjaSCSI

The scsi layer is already calling add_disk_randomness in scsi_end_request.

Signed-off-by: Matt Mackall <mpm@selenic.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] Bogus disk geometry on large disks
Alan Cox [Sun, 25 Jun 2006 08:58:58 +0000 (01:58 -0700)]
[SCSI] Bogus disk geometry on large disks

We currently stuff a truncated size into the geometry logic and return the
result which can produce bizarre reports for a 4Tb array.  Since that
mapping logic isn't useful for disks that big don't try and map this way at
all.

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] Add scsi_add_host() failure handling for nsp32
GOTO Masanori [Sun, 25 Jun 2006 08:58:56 +0000 (01:58 -0700)]
[SCSI] Add scsi_add_host() failure handling for nsp32

Add scsi_add_host() failure handling for nsp32
and silence warning.
  drivers/scsi/nsp32.c:2888: 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: GOTO Masanori <gotom@sanori.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] lpfc: sparse NULL warnings
Randy Dunlap [Sun, 25 Jun 2006 08:58:51 +0000 (01:58 -0700)]
[SCSI] lpfc: sparse NULL warnings

From: Randy Dunlap <rdunlap@xenotime.net>

Fix sparse warnings: use NULL instead of 0 for pointers:
drivers/scsi/lpfc/lpfc_els.c:827:56: warning: Using plain integer as NULL pointer
drivers/scsi/lpfc/lpfc_els.c:2781:18: warning: Using plain integer as NULL pointer
drivers/scsi/lpfc/lpfc_els.c:2782:18: warning: Using plain integer as NULL pointer
drivers/scsi/lpfc/lpfc_init.c:951:21: warning: Using plain integer as NULL pointer
drivers/scsi/lpfc/lpfc_init.c:956:20: warning: Using plain integer as NULL pointer

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Acked-by: James Smart <James.Smart@Emulex.Com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Update version number to 8.01.05-k3.
Andrew Vasquez [Fri, 23 Jun 2006 23:11:32 +0000 (16:11 -0700)]
[SCSI] qla2xxx: Update version number to 8.01.05-k3.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Correctly set the firmware NOS/OLS timeout during initialization.
Andrew Vasquez [Fri, 23 Jun 2006 23:11:27 +0000 (16:11 -0700)]
[SCSI] qla2xxx: Correctly set the firmware NOS/OLS timeout during initialization.

Original code incorrectly assigned it to the driver's
link-down-timeout value (a value in seconds).

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Convert from pci_module_init() to pci_register_driver().
Andrew Vasquez [Fri, 23 Jun 2006 23:11:22 +0000 (16:11 -0700)]
[SCSI] qla2xxx: Convert from pci_module_init() to pci_register_driver().

Also remove qla2xxx_probe_one/qla2xxx_remove_one stubs previously
used with external firmware module loaders.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Correct 'loop-down' determination logic in qla2x00_fw_ready().
Andrew Vasquez [Fri, 23 Jun 2006 23:11:17 +0000 (16:11 -0700)]
[SCSI] qla2xxx: Correct 'loop-down' determination logic in qla2x00_fw_ready().

As there is no point in failing the initialization process when
firmware informs the host software that it could not transition
beyond a CONFIG_WAIT nor WAIT_FOR_LOGIN state.  Previous logic
would mark such conditions as a general *failure* and subsequently
tear-down the scsi-host during initialization.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Add support for extended error logging.
Andrew Vasquez [Fri, 23 Jun 2006 23:11:10 +0000 (16:11 -0700)]
[SCSI] qla2xxx: Add support for extended error logging.

Similar in form to QLogic's standard offering -- via
the 'extended_error_logging' module parameter.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Cleanup DEBUG macro usage.
Andrew Vasquez [Fri, 23 Jun 2006 23:11:05 +0000 (16:11 -0700)]
[SCSI] qla2xxx: Cleanup DEBUG macro usage.

- macro usage statements should terminate with a ';'
- remove unused macros.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Remove no-op IOCTL codes and macros.
Andrew Vasquez [Fri, 23 Jun 2006 23:11:00 +0000 (16:11 -0700)]
[SCSI] qla2xxx: Remove no-op IOCTL codes and macros.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Create an VPD sysfs entry for supported ISPs only.
Andrew Vasquez [Fri, 23 Jun 2006 23:10:55 +0000 (16:10 -0700)]
[SCSI] qla2xxx: Create an VPD sysfs entry for supported ISPs only.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Add DMI (Diagnostics Monitoring Interface) support.
Andrew Vasquez [Fri, 23 Jun 2006 23:10:50 +0000 (16:10 -0700)]
[SCSI] qla2xxx: Add DMI (Diagnostics Monitoring Interface) support.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Honour 'skip process-login' option during fabric-login IOCB.
Andrew Vasquez [Fri, 23 Jun 2006 23:10:44 +0000 (16:10 -0700)]
[SCSI] qla2xxx: Honour 'skip process-login' option during fabric-login IOCB.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Add NVRAM 'Disable Serdes' bit support.
Andrew Vasquez [Fri, 23 Jun 2006 23:10:39 +0000 (16:10 -0700)]
[SCSI] qla2xxx: Add NVRAM 'Disable Serdes' bit support.

The host section of ISP24xx NVRAMs contain a new bit which
allows a user to selectively disable ports of an HBA.  These
ports (hosts) will not be presented to the midlayer.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Resync with latest HBA SSID specification -- 2.2j.
Andrew Vasquez [Fri, 23 Jun 2006 23:10:34 +0000 (16:10 -0700)]
[SCSI] qla2xxx: Resync with latest HBA SSID specification -- 2.2j.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] qla2xxx: Rework firmware-trace facilities.
Andrew Vasquez [Fri, 23 Jun 2006 23:10:29 +0000 (16:10 -0700)]
[SCSI] qla2xxx: Rework firmware-trace facilities.

- Defer firmware dump-data raw-to-textual conversion to
  user-space.
- Add module parameter (ql2xallocfwdump) to allow for per-HBA
  allocations of firmware dump memory.
- Dump request and response queue data as per firmware group
  request.
- Add extended firmware trace support for ISP24XX/ISP54XX chips.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] SCSI core: Allow QUIESCE -> CANCEL sdev transition
Alan Stern [Fri, 23 Jun 2006 18:25:34 +0000 (14:25 -0400)]
[SCSI] SCSI core: Allow QUIESCE -> CANCEL sdev transition

We have to be able to remove SCSI devices even when they are suspended, so
QUIESCE -> CANCEL must be a legal state transition.  This patch (as727)
adds the transition to the state machine.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] sd/scsi_lib simplify sd_rw_intr and scsi_io_completion
Luben Tuikov [Fri, 23 Jun 2006 16:39:09 +0000 (09:39 -0700)]
[SCSI] sd/scsi_lib simplify sd_rw_intr and scsi_io_completion

This patch simplifies "good_bytes" computation in sd_rw_intr().
sd: "good_bytes" computation is always done in terms of the resolution
of the device's medium, since after that it is the number of good bytes
we pass around and other layers/contexts (as opposed ot sd) can translate
that to their own resolution (block layer:512).  It also makes
scsi_io_completion() processing more straightforward, eliminating the
3rd argument to the function.

It also fixes a couple of bugs like not checking return value,
using "break" instead of "return;", etc.

I've been running with this patch for some time now on a
test (do-it-all) system.

Signed-off-by: Luben Tuikov <ltuikov@yahoo.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] aic79xx: remove slave_destroy
Hannes Reinecke [Thu, 22 Jun 2006 09:45:00 +0000 (11:45 +0200)]
[SCSI] aic79xx: remove slave_destroy

Even with the latest fixes aic79xx still occasionally triggers the
BUG_ON in slave_destroy. Rather than trying to figure out the various
levels of interaction here I've decided to remove the callback altogether.

The primary reason for the slave_alloc / slave_destroy is to keep an
index of pointers to the sdevs associated with a given target.
However, by changing the arguments to the affected functions slightly
it's possible to avoid the use of that index entirely.
The only performance penalty we'll incur is in writing the
information for /proc/scsi/XXX, as we'll have to recurse over all
available sdevs to find the correct ones. But I doubt that reading
from /proc is in any way time-critical.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] HP XP devinfo update
Hannes Reinecke [Thu, 22 Jun 2006 07:19:51 +0000 (09:19 +0200)]
[SCSI] HP XP devinfo update

According to Anthony Cheung all HP XP arrays with "OPEN-"
types support REPORT_LUN. So there is no reason why we
shouldn't use it.

Signed-off-by: Anthony Cheung <anthony.cheung@hp.com>
Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] megaraid_sas: zcr with fix
Sumant Patro [Tue, 20 Jun 2006 22:32:37 +0000 (15:32 -0700)]
[SCSI] megaraid_sas: zcr with fix

The patch adds support for a ZCR controller (Device ID : 0x413).

It also has a critical bug fix :

Disable controller interrupt before firing INIT cmd to FW.  Interrupt
is enabled after required initialization is over. This is done to
ensure that driver is ready to handle interrupts when it is generated
by the controller.

Signed-off-by: Sumant Patro <Sumant.Patro@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] ibmvscsi: treat busy and error conditions separately
Dave C Boutcher [Tue, 13 Jun 2006 02:22:51 +0000 (21:22 -0500)]
[SCSI] ibmvscsi: treat busy and error conditions separately

This patch fixes a condition where ibmvscsi treats a transport error as a
"busy" condition, so no errors were returned to the scsi mid-layer.
In a RAID environment this means that I/O hung rather than failing
over.

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] scsi_debug version 1.79
Douglas Gilbert [Tue, 6 Jun 2006 04:11:24 +0000 (00:11 -0400)]
[SCSI] scsi_debug version 1.79

  - add 'virtual_gb' parameter to simulate large storage
    (by wrapping in dev_size_mb megabytes of actual ram)
  - add 'no_lun_0' parameter to skip lun 0 on each target
    (but still respond as required to INQUIRY + REPORT LUNS)
  - add well know lu support
  - add MODE SELECT commands support [pages: 0xa and 0x1c]
  - add LOG SENSE command support [pages: 0xd and 0x2f]
  - add READ CAPACITY (16) support
  - increase number of mode pages supported (to read),
    mainly transport specific (SAS) mode (sub)pages
  - add more VPD pages and extend others, including
    ATA information VPD page
  - START STOP UNIT now maintains a state machine
  - READ (16) and WRITE (16) cope with lbas larger
    than 32 bits (needed for the 'virtual_gb' parameter)
  - allow single command transfers up to 32 MB
  - more precise error (sense data) messages

Signed-off-by: Douglas Gilbert <dougg@torque.net>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] kmalloc argument switcheroo in recent 53c700 change.
Dave Jones [Thu, 22 Jun 2006 19:46:15 +0000 (15:46 -0400)]
[SCSI] kmalloc argument switcheroo in recent 53c700 change.

On Wed, Jun 21, 2006 at 07:00:34PM +0000, Linux Kernel wrote:
 > commit 67d59dfdeb21df2c16dcd478b66177e91178ecd0
 > tree ae85703651d81740f4a6cd398f9dd4d6aabe6a2f
 > parent 6db874fbdbedba5e15e76cc03b42f52ea70338c0
 > author James Bottomley <James.Bottomley@steeleye.com> Wed, 14 Jun 2006 07:31:19 -0500
 > committer James Bottomley <jejb@mulgrave.il.steeleye.com> Tue, 20 Jun 2006 05:34:01 -0500
 >
 > [SCSI] 53c700: remove reliance on deprecated cmnd fields
 >  ...
 >
 > + SDp->hostdata = kmalloc(GFP_KERNEL, sizeof(struct NCR_700_sense));
 > +
 > + if (!SDp->hostdata)
 > + return -ENOMEM;

"I'll take reversed arguments for $100 please Alex".

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[PATCH] uclinux: use PER_LINUX_32BIT in binfmt_flat
Malcolm Parsons [Mon, 26 Jun 2006 01:49:41 +0000 (11:49 +1000)]
[PATCH] uclinux: use PER_LINUX_32BIT in binfmt_flat

binfmt_flat.c calls set_personality with PER_LINUX as the personality.
On the arm architecture this results in the program running in 26bit
usermode.  PER_LINUX_32BIT should be used instead.  This doesn't affect
other architectures that use binfmt_flat.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: make enable_irq() macro statement
Greg Ungerer [Mon, 26 Jun 2006 01:47:35 +0000 (11:47 +1000)]
[PATCH] m68knommu: make enable_irq() macro statement

Change enable_irq() macro to be a statement, not expression.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: fix ColdFire 5249 PLL set
Greg Ungerer [Mon, 26 Jun 2006 01:43:35 +0000 (11:43 +1000)]
[PATCH] m68knommu: fix ColdFire 5249 PLL set

Fix PLL setting for the Coldfire 5249 CPU. This brings it into line with
the new style frequency configuration of m68knommu parts.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: ColdFire cache flush fixups
Greg Ungerer [Mon, 26 Jun 2006 01:40:44 +0000 (11:40 +1000)]
[PATCH] m68knommu: ColdFire cache flush fixups

Fix flush code for the ColdFire 5206/5206e/5272 cases.
Add support for the new ColdFire 532x CPU family

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: improve syscall entry and fix strace
Philippe De Muyter [Mon, 26 Jun 2006 01:37:57 +0000 (11:37 +1000)]
[PATCH] m68knommu: improve syscall entry and fix strace

Here is a patch to the system call handling for 5307/5272/etc to:

- fix the strace support (one tested the wrong bit)
- make all system calls a little bit faster by inlining set_esp0 and
  supporting ENOSYS out of the critical path.
- remove extraneous spaces

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: force stack alignment on ColdFire
Greg Ungerer [Mon, 26 Jun 2006 01:01:32 +0000 (11:01 +1000)]
[PATCH] m68knommu: force stack alignment on ColdFire

This patch solve a bug triggered by execvp (this function use calloc to
store the argument list and gcc 3.4.x align the stack to word, not to dword).
This situation aren't related to signal handling and all 2.6.x have the bug.
On ColdFire targets we must force the stack to be aligned.

Original patch from Andrea Tarani <andrea.tarani@gilbarco.com>,

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: configurable frequency selection header
Greg Ungerer [Mon, 26 Jun 2006 00:58:09 +0000 (10:58 +1000)]
[PATCH] m68knommu: configurable frequency selection header

Remove list of fixed clock frequency options used for configuring master
clock, and make field an int. Much more flexible this way, no need to add
more options for every new used freqency.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: configurable frequency selection
Greg Ungerer [Mon, 26 Jun 2006 00:55:36 +0000 (10:55 +1000)]
[PATCH] m68knommu: configurable frequency selection

Remove list of fixed clock frequency options used for configuring master
clock, and make field an int. Much more flexible this way, no need to add
more options for every new used freqency.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: configure support for Avnet5282 board
Greg Ungerer [Mon, 26 Jun 2006 00:47:13 +0000 (10:47 +1000)]
[PATCH] m68knommu: configure support for Avnet5282 board

Add support for the AVNET 5282 board.
Patch submitted by Daniel Alomar <dalomar@serrasold.com>.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: ColdFire 532x CPU configure support
Greg Ungerer [Mon, 26 Jun 2006 00:45:45 +0000 (10:45 +1000)]
[PATCH] m68knommu: ColdFire 532x CPU configure support

Add configure support for the new Freescale 532x family of CPUs.
Patch submitted by Matt Waddel <Matt.Waddel@freescale.com>.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years agoRevert "swsusp special saveable pages support" commits
Linus Torvalds [Mon, 26 Jun 2006 01:41:00 +0000 (18:41 -0700)]
Revert "swsusp special saveable pages support" commits

This reverts commits

  3e3318dee0878d42ed62a19c292a2ac284135db3 [PATCH] swsusp: x86_64 mark special saveable/unsaveable pages
  b6370d96e09944c6e3ae8d5743ca8a8ab1f79f6c [PATCH] swsusp: i386 mark special saveable/unsaveable pages
  ce4ab0012b32c1a4a1d6e934aeb73bf3151c48d9 [PATCH] swsusp: add architecture special saveable pages support

because not only do they apparently cause page faults on x86, the
infrastructure doesn't compile on powerpc.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: ColdFire 532x CPU platform build support
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: ColdFire 532x CPU platform build support

Add build support for the new Freescale 532x CPU platforms.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: add ColdFire 532x UART address support
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: add ColdFire 532x UART address support

Add support for the UART addressing on the new Freescale M532x CPU family.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: add ColdFire 532x timer build support
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: add ColdFire 532x timer build support

Add build support for new Freescale M532x CPU family timer.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: cleanup setup.c
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: cleanup setup.c

A cleanup of m68knommu/kernel/setup.c :

- No need to initialize global pointers to NULL, they will have that value
  automatically, and they eat up space in my data segment image in FLASH.
- Remove get_cpuinfo.  It has been replaced by show_cpuinfo.

Signed-off-by: Philippe De Muyter <phdm@macqel.be>
Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: check DEBUG defined arch code
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: check DEBUG defined arch code

Don't rely on DEBUG having a value, check for it being defined.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: include ColdFire 532x register definitions
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: include ColdFire 532x register definitions

Include the ColdFire 532x support when including ColdFire peripharp
support definitions.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: add ColdFire 532x cache init
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: add ColdFire 532x cache init

Add cache init support for the new ColdFire 532x CPU family.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: 532x UART support
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: 532x UART support

ColdFire serial driver support for the new 532x CPU family UARTs.

Patch submitted by Matt Waddel <Matt.Waddel@freescale.com>.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: ColdFire 532x register definitions
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: ColdFire 532x register definitions

Register definitions for the new Freescale 532x Coldfire CPU family.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: ColdFire 532x CPU startup code
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: ColdFire 532x CPU startup code

Add kernel startup code for the new Freescale 532x CPU family.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: read/write register access for PIT timer
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: read/write register access for PIT timer

Modify the m68knommu/ColdFire PIT timer code to use register offsets
with raw_read/raw_write access, instead of a mapped struct.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: remove use of rom_length
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: remove use of rom_length

Remove use of rom_length. It serves not real purpose, and when we move
to new stye RAM configuration it will not be needed.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: switch arch config name to CONFIG_M68K
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: switch arch config name to CONFIG_M68K

Switch to naming the architecture config options for the m68knommu branch
as "M68K", dropping "M68KNOMMU".  The CONFIG_MMU separates the 2 now, and
the m68knommu branch is still strictly speaking an M68K (including the
ColdFire parts).

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: read/write register access for ColdFire core timer
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: read/write register access for ColdFire core timer

Modify the m68knommu/ColdFire core timer code to use register offsets
with raw_read/raw_write access, instead of a mapped struct.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] m68knommu: 532x FEC eth struct map
Greg Ungerer [Mon, 26 Jun 2006 00:33:10 +0000 (10:33 +1000)]
[PATCH] m68knommu: 532x FEC eth struct map

Specify the struct mapping used on the new Freescale 532x CPU family's
FEC ethernet core.

Patch submitted by Matt Waddel <Matt.Waddel@freescale.com>.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] spufs: update ->flush method proto
Al Viro [Sun, 25 Jun 2006 23:42:33 +0000 (16:42 -0700)]
[PATCH] spufs: update ->flush method proto

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] xfs: update ->flush method proto
Alexey Dobriyan [Sun, 25 Jun 2006 23:42:33 +0000 (16:42 -0700)]
[PATCH] xfs: update ->flush method proto

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Nathan Scott <nathans@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years agoMerge branch 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/roland/infiniband
Linus Torvalds [Sun, 25 Jun 2006 23:07:58 +0000 (16:07 -0700)]
Merge branch 'for-linus' of /linux/kernel/git/roland/infiniband

* 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/roland/infiniband:
  IB/iser: iSER Kconfig and Makefile
  IB/iser: iSER handling of memory for RDMA
  IB/iser: iSER RDMA CM (CMA) and IB verbs interaction
  IB/iser: iSER initiator iSCSI PDU and TX/RX
  IB/iser: iSCSI iSER transport provider high level code
  IB/iser: iSCSI iSER transport provider header file
  IB/uverbs: Remove unnecessary list_del()s
  IB/uverbs: Don't free wr list when it's known to be empty

17 years agoFix NFS2 compile error
Linus Torvalds [Sun, 25 Jun 2006 19:30:33 +0000 (12:30 -0700)]
Fix NFS2 compile error

Trond had apparently merged the same patch twice, causing a duplicate
include of the "internal.h" file, with resulting obvious confusion.

Tssk.  I'm the only one allowed to send out trees that don't even
compile! Who does this Trond guy think he is?

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] i386: Fix softirq accounting with 4K stacks
Björn Steinbrink [Sun, 25 Jun 2006 14:24:40 +0000 (16:24 +0200)]
[PATCH] i386: Fix softirq accounting with 4K stacks

Copy the softirq bits in preempt_count from the current context into the
hardirq context when using 4K stacks to make the softirq_count macro work
correctly and thereby fix softirq cpu time accounting.

Signed-off-by: Björn Steinbrink <B.Steinbrink@gmx.de>
Acked-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years agoMerge git://git.linux-nfs.org/pub/linux/nfs-2.6
Linus Torvalds [Sun, 25 Jun 2006 17:54:14 +0000 (10:54 -0700)]
Merge git://git.linux-nfs.org/pub/linux/nfs-2.6

* git://git.linux-nfs.org/pub/linux/nfs-2.6: (51 commits)
  nfs: remove nfs_put_link()
  nfs-build-fix-99
  git-nfs-build-fixes
  Merge branch 'odirect'
  NFS: alloc nfs_read/write_data as direct I/O is scheduled
  NFS: Eliminate nfs_get_user_pages()
  NFS: refactor nfs_direct_free_user_pages
  NFS: remove user_addr, user_count, and pos from nfs_direct_req
  NFS: "open code" the NFS direct write rescheduler
  NFS: Separate functions for counting outstanding NFS direct I/Os
  NLM: Fix reclaim races
  NLM: sem to mutex conversion
  locks.c: add the fl_owner to nlm_compare_locks
  NFS: Display the chosen RPCSEC_GSS security flavour in /proc/mounts
  NFS: Split fs/nfs/inode.c
  NFS: Fix typo in nfs_do_clone_mount()
  NFS: Fix compile errors introduced by referrals patches
  NFSv4: Ensure that referral mounts bind to a reserved port
  NFSv4: A root pathname is sent as a zero component4
  NFSv4: Follow a referral
  ...

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/mchehab/v4l-dvb
Linus Torvalds [Sun, 25 Jun 2006 17:09:31 +0000 (10:09 -0700)]
Merge /pub/scm/linux/kernel/git/mchehab/v4l-dvb

* master.kernel.org:/pub/scm/linux/kernel/git/mchehab/v4l-dvb: (244 commits)
  V4L/DVB (4210b): git-dvb: tea575x-tuner build fix
  V4L/DVB (4210a): git-dvb versus matroxfb
  V4L/DVB (4209): Added some BTTV PCI IDs for newer boards
  Fixes some sync issues between V4L/DVB development and GIT
  V4L/DVB (4206): Cx88-blackbird: always set encoder height based on tvnorm->id
  V4L/DVB (4205): Merge tda9887 module into tuner.
  V4L/DVB (4203): Explicitly set the enum values.
  V4L/DVB (4202): allow selecting CX2341x port mode
  V4L/DVB (4200): Disable bitrate_mode when encoding mpeg-1.
  V4L/DVB (4199): Add cx2341x-specific control array to cx2341x.c
  V4L/DVB (4198): Avoid newer usages of obsoleted experimental MPEGCOMP API
  V4L/DVB (4197): Port new MPEG API to saa7134-empress with saa6752hs
  V4L/DVB (4196): Port cx88-blackbird to the new MPEG API.
  V4L/DVB (4193): Update cx2341x fw encoding API doc.
  V4L/DVB (4192): Use control helpers for saa7115, cx25840, msp3400.
  V4L/DVB (4191): Add CX2341X MPEG encoder module.
  V4L/DVB (4190): Add helper functions for control processing to v4l2-common.
  V4L/DVB (4189): Add videodev support for VIDIOC_S/G/TRY_EXT_CTRLS.
  V4L/DVB (4188): Add new MPEG control/ioctl definitions to videodev2.h
  V4L/DVB (4186): Add support for the DNTV Live! mini DVB-T card.
  ...

17 years agoFix PM_TRACE dependency: works only on 32-bit x86 for now
Linus Torvalds [Sun, 25 Jun 2006 17:04:15 +0000 (10:04 -0700)]
Fix PM_TRACE dependency: works only on 32-bit x86 for now

Not that x86-64 and other architecture support should be difficult to
add (trivial fixups to the data format and add the proper linker script
entry).

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] fix race in idr code
Sonny Rao [Sun, 25 Jun 2006 12:49:34 +0000 (05:49 -0700)]
[PATCH] fix race in idr code

I ran into a bug where the kernel died in the idr code:

cpu 0x1d: Vector: 300 (Data Access) at [c000000b7096f710]
    pc: c0000000001f8984: .idr_get_new_above_int+0x140/0x330
    lr: c0000000001f89b4: .idr_get_new_above_int+0x170/0x330
    sp: c000000b7096f990
   msr: 800000000000b032
   dar: 0
 dsisr: 40010000
  current = 0xc000000b70d43830
  paca    = 0xc000000000556900
    pid   = 2022, comm = hwup
1d:mon> t
[c000000b7096f990c0000000000d2ad8 .expand_files+0x2e8/0x364 (unreliable)
[c000000b7096faa0c0000000001f8bf8 .idr_get_new_above+0x18/0x68
[c000000b7096fb20c00000000002a054 .init_new_context+0x5c/0xf0
[c000000b7096fbc0c000000000049dc8 .copy_process+0x91c/0x1404
[c000000b7096fcd0c00000000004a988 .do_fork+0xd8/0x224
[c000000b7096fdc0c00000000000ebdc .sys_clone+0x5c/0x74
[c000000b7096fe30c000000000008950 .ppc_clone+0x8/0xc

17 years ago[PATCH] update-devices.txt
Jan Engelhardt [Sun, 25 Jun 2006 12:49:33 +0000 (05:49 -0700)]
[PATCH] update-devices.txt

Update Documentation/devices.txt with a new version from the LANANA site
http://www.lanana.org/docs/device-list/devices-2.6+.txt

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] remove unlikely(sb) in prune_dcache
Hua Zhong [Sun, 25 Jun 2006 12:49:32 +0000 (05:49 -0700)]
[PATCH] remove unlikely(sb) in prune_dcache

likely profiling shows that the following is a miss.

After boot:
[+- ] Type | # True | # False | Function:Filename@Line
+unlikely |     1074|        0  prune_dcache()@:fs/dcache.c@409

After a bonnie++ run:
+unlikely |    66716|    19584  prune_dcache()@:fs/dcache.c@409

So remove it.

Signed-off-by: Hua Zhong <hzhong@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] Fix "biovec-(256)" in /proc/slabinfo
Alexey Dobriyan [Sun, 25 Jun 2006 12:49:32 +0000 (05:49 -0700)]
[PATCH] Fix "biovec-(256)" in /proc/slabinfo

Stringify does what it was told to do.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] Fix kdump Crash Kernel boot memory reservation for NUMA machines
Amul Shah [Sun, 25 Jun 2006 12:49:31 +0000 (05:49 -0700)]
[PATCH] Fix kdump Crash Kernel boot memory reservation for NUMA machines

This patch will fix a boot memory reservation bug that trashes memory on
the ES7000 when loading the kdump crash kernel.

The code in arch/x86_64/kernel/setup.c to reserve boot memory for the crash
kernel uses the non-numa aware "reserve_bootmem" function instead of the
NUMA aware "reserve_bootmem_generic".  I checked to make sure that no other
function was using "reserve_bootmem" and found none, except the ones that
had NUMA ifdef'ed out.

I have tested this patch only on an ES7000 with NUMA on and off (numa=off)
in a single (non-NUMA) and multi-cell (NUMA) configurations.

Signed-off-by: Amul Shah <amul.shah@unisys.com>
Looks-good-to: Vivek Goyal <vgoyal@in.ibm.com>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] s390: setup.c cleanup + build fix
Heiko Carstens [Sun, 25 Jun 2006 12:49:30 +0000 (05:49 -0700)]
[PATCH] s390: setup.c cleanup + build fix

Cleanup & fix 31 bit compilation:

  CC      arch/s390/kernel/setup.o
arch/s390/kernel/setup.c:83: error: initializer element is not computable at
                                    load time
arch/s390/kernel/setup.c:83: error: (near initialization for
                                    'code_resource.start')
Not sure which patch in the -mm tree breaks this, but since this can be
considered a cleanup it can be merged anyway.

Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] au1xxx: oss sound support for au1200
Domen Puncer [Sun, 25 Jun 2006 12:49:29 +0000 (05:49 -0700)]
[PATCH] au1xxx: oss sound support for au1200

au1550 ac97 driver works fine on au1200 too.

Comments at the top of file state this code is GPL, so lets
mark it as GPL too.

Signed-off-by: Domen Puncer <domen.puncer@ultra.si>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] parport: add to kernel-doc
Randy Dunlap [Sun, 25 Jun 2006 12:49:29 +0000 (05:49 -0700)]
[PATCH] parport: add to kernel-doc

Add parport interfaces to kernel-doc template.  Small doc.  cleanups in 2
parport source files.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] au1550_ac97: spin_unlock in error path
Domen Puncer [Sun, 25 Jun 2006 12:49:27 +0000 (05:49 -0700)]
[PATCH] au1550_ac97: spin_unlock in error path

Error paths didn't spin_unlock.

Signed-off-by: Domen Puncer <domen.puncer@ultra.si>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] ext2: cleanup: put_page and comment fix
Evgeniy Dushistov [Sun, 25 Jun 2006 12:49:27 +0000 (05:49 -0700)]
[PATCH] ext2: cleanup: put_page and comment fix

Things which force me think a little: why so?

Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] pacct: none-delayed process accounting accumulation
KaiGai Kohei [Sun, 25 Jun 2006 12:49:26 +0000 (05:49 -0700)]
[PATCH] pacct: none-delayed process accounting accumulation

In current 2.6.17 implementation, signal_struct refered from task_struct is
used for per-process data structure.  The pacct facility also uses it as a
per-process data structure to store stime, utime, minflt, majflt.  But those
members are saved in __exit_signal().  It's too late.

For example, if some threads exits at same time, pacct facility has a
possibility to drop accountings for a part of those threads.  (see, the
following 'The results of original 2.6.17 kernel') I think accounting
information should be completely collected into the per-process data structure
before writing out an accounting record.

This patch fixes this matter.  Accumulation of stime, utime, minflt and majflt
are done before generating accounting record.

[mingo@elte.hu: fix acct_collect() siglock bug found by lockdep]
Signed-off-by: KaiGai Kohei <kaigai@ak.jp.nec.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] pacct: avoidance to refer the last thread as a representation of the process
KaiGai Kohei [Sun, 25 Jun 2006 12:49:25 +0000 (05:49 -0700)]
[PATCH] pacct: avoidance to refer the last thread as a representation of the process

When pacct facility generate an 'ac_flag' field in accounting record, it
refers a task_struct of the thread which died last in the process.  But any
other task_structs are ignored.

Therefore, pacct facility drops ASU flag even if root-privilege operations are
used by any other threads except the last one.  In addition, AFORK flag is
always set when the thread of group-leader didn't die last, although this
process has called execve() after fork().

We have a same matter in ac_exitcode.  The recorded ac_exitcode is an exit
code of the last thread in the process.  There is a possibility this exitcode
is not the group leader's one.

17 years ago[PATCH] pacct: add pacct_struct to fix some pacct bugs.
KaiGai Kohei [Sun, 25 Jun 2006 12:49:24 +0000 (05:49 -0700)]
[PATCH] pacct: add pacct_struct to fix some pacct bugs.

The pacct facility need an i/o operation when an accounting record is
generated.  There is a possibility to wake OOM killer up.  If OOM killer is
activated, it kills some processes to make them release process memory
regions.

But acct_process() is called in the killed processes context before calling
exit_mm(), so those processes cannot release own memory.  In the results, any
processes stop in this point and it finally cause a system stall.

17 years ago[PATCH] Corrections to memory barrier doc
David Howells [Sun, 25 Jun 2006 12:49:22 +0000 (05:49 -0700)]
[PATCH] Corrections to memory barrier doc

Apply some small corrections to the memory barrier document, as contributed by:

Christoph Lameter <clameter@sgi.com>
Kirill Smelkov <kirr@mns.spb.ru>
Randy Dunlap <rdunlap@xenotime.net>

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] synclink_gt: add GT2 adapter support
Paul Fulghum [Sun, 25 Jun 2006 12:49:22 +0000 (05:49 -0700)]
[PATCH] synclink_gt: add GT2 adapter support

Add support for SyncLink GT2 adapter to driver.

Signed-off-by: Paul Fulghum <paulkf@microgate.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] fix synclink_gt diagnostics error reporting
Paul Fulghum [Sun, 25 Jun 2006 12:49:21 +0000 (05:49 -0700)]
[PATCH] fix synclink_gt diagnostics error reporting

Fix diagnostics error reporting that was being overwritten by incorrect use
of return codes from individual diagnostic functions.

Signed-off-by: Paul Fulghum <paulkf@microgate.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] add synclink_gt crc return feature
Paul Fulghum [Sun, 25 Jun 2006 12:49:21 +0000 (05:49 -0700)]
[PATCH] add synclink_gt crc return feature

Add ability to return HDLC CRC to user application.

Signed-off-by: Paul Fulghum <paulkf@microgate.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] add synclink_gt custom hdlc idle
Paul Fulghum [Sun, 25 Jun 2006 12:49:20 +0000 (05:49 -0700)]
[PATCH] add synclink_gt custom hdlc idle

Add custom HDLC idle pattern feature.

It allows the user to specify an arbitrary 8 or 16 bit repeating pattern on
the transmit data pin between HDLC frames.

In most cases the idle pattern is continuous ones or flags as supported by off
the shelf synchronous controllers and defined in the ISO3309 standard.  Some
applications (radio/satellite modems, connections to legacy military hardware)
require non-standard patterns.

Signed-off-by: Paul Fulghum <paulkf@microgate.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] irda-usb printk fix
Andrew Morton [Sun, 25 Jun 2006 12:49:19 +0000 (05:49 -0700)]
[PATCH] irda-usb printk fix

drivers/net/irda/irda-usb.c: In function 'stir421x_patch_device':
drivers/net/irda/irda-usb.c:1108: warning: format '%u' expects type 'unsigned int', but argument 4 has type 'size_t'

Cc: Greg KH <greg@kroah.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] kthread: move kernel-doc and put it into DocBook
Randy Dunlap [Sun, 25 Jun 2006 12:49:19 +0000 (05:49 -0700)]
[PATCH] kthread: move kernel-doc and put it into DocBook

Move kthread API kernel-doc from kthread.h to kthread.c & fix it.
Add kthread API to kernel-api DocBook.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] DMI: cleanup kernel-doc, add to DocBook
Randy Dunlap [Sun, 25 Jun 2006 12:49:18 +0000 (05:49 -0700)]
[PATCH] DMI: cleanup kernel-doc, add to DocBook

Add DMI interface functions to a new Firmware Interfaces chapter in the
kernel-api DocBook.  Clean up kernel-doc in drivers/firmware/dmi_scan.c.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] Implement kasprintf
Jeremy Fitzhardinge [Sun, 25 Jun 2006 12:49:17 +0000 (05:49 -0700)]
[PATCH] Implement kasprintf

Implement kasprintf, a kernel version of asprintf.  This allocates the
memory required for the formatted string, including the trailing '\0'.
Returns NULL on allocation failure.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] Fix bounds check in vsnprintf, to allow for a 0 size and NULL buffer
Jeremy Fitzhardinge [Sun, 25 Jun 2006 12:49:17 +0000 (05:49 -0700)]
[PATCH] Fix bounds check in vsnprintf, to allow for a 0 size and NULL buffer

This change allows callers to use a 0-byte buffer and a NULL buffer pointer
with vsnprintf, so it can be used to determine how large the resulting
formatted string will be.

Previously the code effectively treated a size of 0 as a size of 4G (on
32-bit systems), with other checks preventing it from actually trying to
emit the string - but the terminal \0 would still be written, which would
crash if the buffer is NULL.

This change changes the boundary check so that 'end' points to the putative
location of the terminal '\0', which is only written if size > 0.

vsnprintf still allows the buffer size to be set very large, to allow
unbounded buffer sizes (to implement sprintf, etc).

[akpm@osdl.org: fix long-vs-longlong confusion]
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] LED: add LED heartbeat trigger
Atsushi Nemoto [Sun, 25 Jun 2006 12:49:15 +0000 (05:49 -0700)]
[PATCH] LED: add LED heartbeat trigger

Add an LED trigger acts like a heart beat.  This can be used as a
replacement of CONFIG_HEARTBEAT code exists in some arch's timer code.

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Acked-by: Richard Purdie <rpurdie@rpsys.net>
Cc: "Nish Aravamudan" <nish.aravamudan@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] ktime/hrtimer: fix kernel-doc comments
Randy Dunlap [Sun, 25 Jun 2006 12:49:15 +0000 (05:49 -0700)]
[PATCH] ktime/hrtimer: fix kernel-doc comments

Fix kernel-doc formatting in ktime.h and hrtimer.[ch] files.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] reed-solomon: fix kernel-doc comments
Randy Dunlap [Sun, 25 Jun 2006 12:49:14 +0000 (05:49 -0700)]
[PATCH] reed-solomon: fix kernel-doc comments

Fix kernel-doc formatting in Reed-Solomon code.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] kernel-doc: use Members for struct fields consistently
Randy Dunlap [Sun, 25 Jun 2006 12:49:14 +0000 (05:49 -0700)]
[PATCH] kernel-doc: use Members for struct fields consistently

kernel-doc struct fields should be consistently called "Members", not
"Arguments", so switch man-mode output to use "Members" like all of the
other formats do.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] kernel-doc: don't use XML escapes in text or man output mode
Randy Dunlap [Sun, 25 Jun 2006 12:49:13 +0000 (05:49 -0700)]
[PATCH] kernel-doc: don't use XML escapes in text or man output mode

For kernel-doc output modes of text and man, do not use XML escapes for
less-than, greater-than, and ampersand characters.  I.e., leave the text
and man output clean and readable.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] fix memory leak in rocketport rp_do_receive
Paul Fulghum [Sun, 25 Jun 2006 12:49:12 +0000 (05:49 -0700)]
[PATCH] fix memory leak in rocketport rp_do_receive

Fix memory leak caused by incorrect use of tty buffer facility.  tty
buffers are allocated but never processed by call to tty_flip_buffer_push
so they accumulate on the full buffer list.  Current code uses the buffers
as a temporary storage for data before passing it directly to the line
discipline.

Signed-off-by: Paul Fulghum <paulkf@microgate.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] Implement AT_SYMLINK_FOLLOW flag for linkat
Ulrich Drepper [Sun, 25 Jun 2006 12:49:11 +0000 (05:49 -0700)]
[PATCH] Implement AT_SYMLINK_FOLLOW flag for linkat

When the linkat() syscall was added the flag parameter was added in the
last minute but it wasn't used so far.  The following patch should change
that.  My tests show that this is all that's needed.

If OLDNAME is a symlink setting the flag causes linkat to follow the
symlink and create a hardlink with the target.  This is actually the
behavior POSIX demands for link() as well but Linux wisely does not do
this.  With this flag (which will most likely be in the next POSIX
revision) the programmer can choose the behavior, defaulting to the safe
variant.  As a side effect it is now possible to implement a
POSIX-compliant link(2) function for those who are interested.

  touch file
  ln -s file symlink

  linkat(fd, "symlink", fd, "newlink", 0)
    -> newlink is hardlink of symlink

  linkat(fd, "symlink", fd, "newlink", AT_SYMLINK_FOLLOW)
    -> newlink is hardlink of file

The value of AT_SYMLINK_FOLLOW is determined by the definition we already
use in glibc.

Signed-off-by: Ulrich Drepper <drepper@redhat.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] add "select GPIO_VR41XX" for TANBAC_TB0229
Yoichi Yuasa [Sun, 25 Jun 2006 12:49:11 +0000 (05:49 -0700)]
[PATCH] add "select GPIO_VR41XX" for TANBAC_TB0229

TANBAC_TB0229 requires GPIO_VR41XX.  This patch adds "select GPIO_VR41XX"
for TANBAC_TB0229.

Signed-off-by: Yoichi Yuasa <yoichi_yuasa@tripeaks.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] cpu hotplug: fix CPU_UP_CANCEL handling
Heiko Carstens [Sun, 25 Jun 2006 12:49:10 +0000 (05:49 -0700)]
[PATCH] cpu hotplug: fix CPU_UP_CANCEL handling

If a cpu hotplug callback fails on CPU_UP_PREPARE, all callbacks will be
called with CPU_UP_CANCELED.  A few of these callbacks assume that on
CPU_UP_PREPARE a pointer to task has been stored in a percpu array.  This
assumption is not true if CPU_UP_PREPARE fails and the following calls to
kthread_bind() in CPU_UP_CANCELED will cause an addressing exception
because of passing a NULL pointer.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>