firefly-linux-kernel-4.4.55.git
13 years ago[SCSI] libfc: improve flogi retries to avoid lport stuck
Vasu Dev [Fri, 28 Oct 2011 18:34:23 +0000 (11:34 -0700)]
[SCSI] libfc: improve flogi retries to avoid lport stuck

Adds more cases to do flogi retry, now also retry
on getting bad response due to either no ELS response
or flogi response payload length not large enough.
In those cases flogi was not retried and that
was leaving lport offline.

Signed-off-by: Vasu Dev <vasu.dev@intel.com>
Tested-by: Bhanu Prakash Gollapudi <bprakash@broadcom.com>
Signed-off-by: Yi Zou <yi.zou@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] libfc: avoid exchanges collision during lport reset
Vasu Dev [Fri, 28 Oct 2011 18:34:17 +0000 (11:34 -0700)]
[SCSI] libfc: avoid exchanges collision during lport reset

Currently timer delay is large and is using msleep to avoid
avoid exchanges collision across lport reset, so instead
do this by initializing exches pool indexes during
reset also.

Signed-off-by: Vasu Dev <vasu.dev@intel.com>
Tested-by: Bhanu Prakash Gollapudi <bprakash@broadcom.com>
Signed-off-by: Yi Zou <yi.zou@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] libfc: fix checking FC_TYPE_BLS
Vasu Dev [Fri, 28 Oct 2011 18:34:12 +0000 (11:34 -0700)]
[SCSI] libfc: fix checking FC_TYPE_BLS

Its checked after skb freed, so instead have fh_type
cached and then check FC_TYPE_BLS against cached
fh_type value.

This wrong check was causing double exch locking as
reported by Bhanu at
https://lists.open-fcoe.org/pipermail/devel/2011-October/011793.html

Signed-off-by: Vasu Dev <vasu.dev@intel.com>
Tested-by: Bhanu Prakash Gollapudi <bprakash@broadcom.com>
Signed-off-by: Yi Zou <yi.zou@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] edd: Treat "XPRS" host bus type the same as "PCI"
Michael Chan [Fri, 28 Oct 2011 18:34:07 +0000 (11:34 -0700)]
[SCSI] edd: Treat "XPRS" host bus type the same as "PCI"

PCI Express devices will return "XPRS" host bus type during BIOS EDD
call.  "XPRS" should be treated just like "PCI" so that the proper
pci_dev symlink will be created.  Scripts such as fcoe_edd.sh will
then work correctly.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Matt Domsch <Matt_Domsch@dell.com>
Signed-off-by: Yi Zou <yi.zou@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] isci: overriding max_concurr_spinup oem parameter by max(oem, user)
Andrzej Jakowski [Thu, 27 Oct 2011 22:05:42 +0000 (15:05 -0700)]
[SCSI] isci: overriding max_concurr_spinup oem parameter by max(oem, user)

Fixes bug where max_concurr_spinup oem parameter should be
overriden by max_concurr_spinup user parameter. Override should
happen only when max_concurr_spinup user parameter is specified
in command line (greater than 0). Also this fix shortens variables
representing max_conxurr_spinup for oem and user parameters.

Signed-off-by: Andrzej Jakowski <andrzej.jakowski@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] isci: revert bcn filtering
Dan Williams [Thu, 27 Oct 2011 22:05:37 +0000 (15:05 -0700)]
[SCSI] isci: revert bcn filtering

The initial bcn filtering implementation was validated on a kernel
baseline that predated the switch to new libata error handling.  Also,
prior to that conversion we borrowed the mvsas MVS_DEV_EH approach to
prevent the unwanted extra ap->ops->phy_reset(ap) that occurred in the
ata_bus_probe() path.

After the conversion to new libata eh resets at discovery are more
frequent and get filtered prematurely by IDEV_EH.  The result is that
our bcn filtering has been blocked from running and at discovery and it
appears to stall discovery completion to the point of triggering hung
task timeouts.  So, revert the implementation for now.  When it returns
it will go into libsas proper.

The domain rediscovery that takes place due to ->lldd_I_T_nexus_reset()
events should now be properly waited for by the ata_port_wait_eh() call
in ata_port_probe().  So the hard coded delay in the isci
->lldd_I_T_nexus_reset() and other libsas drivers should help debounce
the libsas thread from seeing temporary device removals.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] isci: Fix hard reset timeout conditions.
Jeff Skirvin [Thu, 27 Oct 2011 22:05:32 +0000 (15:05 -0700)]
[SCSI] isci: Fix hard reset timeout conditions.

A hard reset can timeout before or after the last phy in the
port goes away.  If after, then notify the OS that the last
phy has failed.

The recovery for the failed hard reset has been removed.
This recovery code was unecessary in that the link would
recover from the failure normally by a new link reset sequence
or hotplug of the remote device.

Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] isci: No need to manage the pending reset bit on pending requests.
Jeff Skirvin [Thu, 27 Oct 2011 22:05:27 +0000 (15:05 -0700)]
[SCSI] isci: No need to manage the pending reset bit on pending requests.

The lldd does not need to look at or manage the pending device
reset bit in pending sas_tasks.

Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] isci: Remove redundant isci_request.ttype field.
Jeff Skirvin [Thu, 27 Oct 2011 22:05:22 +0000 (15:05 -0700)]
[SCSI] isci: Remove redundant isci_request.ttype field.

Use the existing IREQ_TMF flag as a request type indicator.

Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] isci: Fix task management for SMP, SATA and on dev remove.
Jeff Skirvin [Thu, 27 Oct 2011 22:05:16 +0000 (15:05 -0700)]
[SCSI] isci: Fix task management for SMP, SATA and on dev remove.

libsas uses the LLDD abort task interface to handle I/O timeouts
in the SATA/STP and SMP discovery paths, so this change will terminate
STP/SMP requests. Also, if the device is gone, the lldd will prevent
libsas from further escalations in the error handler.

Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] isci: No task_done callbacks in error handler paths.
Jeff Skirvin [Thu, 27 Oct 2011 22:05:11 +0000 (15:05 -0700)]
[SCSI] isci: No task_done callbacks in error handler paths.

libsas will cleanup pending sas_tasks after error handler
path functions are called; do not call task_done callbacks.

Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] isci: Handle task request timeouts correctly.
Jeff Skirvin [Thu, 27 Oct 2011 22:05:06 +0000 (15:05 -0700)]
[SCSI] isci: Handle task request timeouts correctly.

In the case where "task" requests timeout (note that this class of
requests can also include SATA/STP soft reset FIS transmissions),
handle the case where the task was being managed by some call to
terminate the task request by completing both the tmf and the aborting
process.

Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] isci: Fix tag leak in tasks and terminated requests.
Jeff Skirvin [Thu, 27 Oct 2011 22:05:01 +0000 (15:05 -0700)]
[SCSI] isci: Fix tag leak in tasks and terminated requests.

Make sure terminated requests and completed task tags are freed.

Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] isci: Immediately fail I/O to removed devices.
Jeff Skirvin [Thu, 27 Oct 2011 22:04:56 +0000 (15:04 -0700)]
[SCSI] isci: Immediately fail I/O to removed devices.

In the case where an I/O fails to start in isci_request_execute,
only allow retries if the device is not already gone.

Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] isci: Lookup device references through requests in completions.
Jeff Skirvin [Thu, 27 Oct 2011 22:04:50 +0000 (15:04 -0700)]
[SCSI] isci: Lookup device references through requests in completions.

The LLDD needs to obtain a reference to the device through the request
itself and not through the domain_device, because the
domain_device.lldd_dev is set to NULL early in the lldd_dev_gone call.
This relies on the fact that the isci_remote_device object is keeping a
seperate reference count of outstanding requests.  TODO: unify the
request count tracking with the isci_remote_device kref.

The failure signature of this condition looks like the following
log, where the important bits are the call to lldd_dev_gone followed
by a crash in isci_terminate_request_core:

[  229.151541] isci 0000:0b:00.0: isci_remote_device_gone: domain_device = ffff8801492d4800, isci_device = ffff880143c657d0, isci_port = ffff880143c63658
[  229.166007] isci 0000:0b:00.0: isci_remote_device_stop: isci_device = ffff880143c657d0
[  229.175317] isci 0000:0b:00.0: isci_terminate_pending_requests: idev=ffff880143c657d0 request=ffff88014741f000; task=ffff8801470f46c0 old_state=2
[  229.189702] isci 0000:0b:00.0: isci_terminate_request_core: device = ffff880143c657d0; request = ffff88014741f000
[  229.201339] isci 0000:0b:00.0: isci_terminate_request_core: before completion wait (ffff88014741f000/ffff880149715ad0)
[  229.213414] isci 0000:0b:00.0: sci_controller_process_completions: completion queue entry:0x8000a0e9
[  229.214401] BUG: unable to handle kernel NULL pointer dereference at 0000000000000228
[  229.214401] IP:jdskirvi-testlbo [<ffffffffa00a58be>] sci_request_completed_state_enter+0x50/0xafb [isci]
[  229.214401] PGD 13d19e067 PUD 13d104067 PMD 0
[  229.214401] Oops: 0000 [#1] SMP
[  229.214401] CPU 0 x kernel: [  226
[  229.214401] Modules linked in: ipv6 dm_multipath uinput nouveau snd_hda_codec_realtek snd_hda_intel ttm drm_kms_helper drm snd_hda_codec snd_hwdep snd_pcm snd_timer i2c_algo_bit isci snd libsas ioatdma mxm_wmi iTCO_wdt soundcore snd_page_alloc scsi_transport_sas iTCO_vendor_support wmi dca video i2c_i801 i2c_core [last unloaded: speedstep_lib]
[  229.214401]
[  229.214401] Pid: 5, comm: kworker/u:0 Not tainted 3.0.0-isci-11.7.29+ #30.353196] Buffer  Intel Corporation Stoakley/Pearlcity Workstation
[  229.214401] RIP: 0010:[<ffffffffa00a58be>] I/O error on dev [<ffffffffa00a58be>] sci_request_completed_state_enter+0x50/0xafb [isci]
[  229.214401] RSP: 0018:ffff88014fc03d20  EFLAGS: 00010046
[  229.214401] RAX: 0000000000000000 RBX: ffff88014741f000 RCX: 0000000000000000
[  229.214401] RDX: ffffffffa00b2c90 RSI: 0000000000000017 RDI: ffff88014741f0a0
[  229.214401] RBP: ffff88014fc03d90 R08: 0000000000000018 R09: 0000000000000000
[  229.214401] R10: 0000000000000000 R11: ffffffff81a17d98 R12: 000000000000001d
[  229.214401] R13: ffff8801470f46c0 R14: 0000000000000000 R15: 0000000000008000
[  229.214401] FS:  0000000000000000(0000) GS:ffff88014fc00000(0000) knlGS:0000000000000000
[  229.214401] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[  229.214401] CR2: 0000000000000228 CR3: 000000013ceaa000 CR4: 00000000000406f0
[  229.214401] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  229.214401] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[  229.214401] Process kworker/u:0 (pid: 5, threadinfo ffff880149714000, task ffff880149718000)
[  229.214401] Call Trace:
[  229.214401]  <IRQ>
[  229.214401]  [<ffffffffa00aa6ce>] sci_change_state+0x4a/0x4f [isci]
[  229.214401]  [<ffffffffa00a4ca6>] sci_io_request_tc_completion+0x79c/0x7a0 [isci]
[  229.214401]  [<ffffffffa00acf35>] sci_controller_process_completions+0x14f/0x396 [isci]
[  229.214401]  [<ffffffffa00abbda>] ? spin_lock_irq+0xe/0x10 [isci]
[  229.214401]  [<ffffffffa00ad2cf>] isci_host_completion_routine+0x71/0x2be [isci]
[  229.214401]  [<ffffffff8107c6b3>] ? mark_held_locks+0x52/0x70
[  229.214401]  [<ffffffff810538e8>] tasklet_action+0x90/0xf1
[  229.214401]  [<ffffffff81054050>] __do_softirq+0xe5/0x1bf
[  229.214401]  [<ffffffff8106d9d1>] ? hrtimer_interrupt+0x129/0x1bb
[  229.214401]  [<ffffffff814ff69c>] call_softirq+0x1c/0x30
[  229.214401]  [<ffffffff8100bb67>] do_softirq+0x4b/0xa3
[  229.214401]  [<ffffffff81053d84>] irq_exit+0x53/0xb4
[  229.214401]  [<ffffffff814fffe7>] smp_apic_timer_interrupt+0x83/0x91
[  229.214401]  [<ffffffff814fee53>] apic_timer_interrupt+0x13/0x20
[  229.214401]  <EOI>
[  229.214401]  [<ffffffff814f7ad4>] ? retint_restore_args+0x13/0x13
[  229.214401]  [<ffffffff8107af29>] ? trace_hardirqs_off+0xd/0xf
[  229.214401]  [<ffffffff8104ea71>] ? vprintk+0x40b/0x452
[  229.214401]  [<ffffffff814f4b5a>] printk+0x41/0x47
[  229.214401]  [<ffffffff81314484>] __dev_printk+0x78/0x7a
[  229.214401]  [<ffffffff8131471e>] dev_printk+0x45/0x47
[  229.214401]  [<ffffffffa00ae2a3>] isci_terminate_request_core+0x15d/0x317 [isci]
[  229.214401]  [<ffffffffa00af1ad>] isci_terminate_pending_requests+0x1a4/0x204 [isci]
[  229.214401]  [<ffffffffa00229f6>] ? sas_phye_oob_error+0xc3/0xc3 [libsas]
[  229.214401]  [<ffffffffa00a7d9e>] isci_remote_device_nuke_requests+0xa6/0xff [isci]
[  229.214401]  [<ffffffffa00a811a>] isci_remote_device_stop+0x7c/0x166 [isci]
[  229.214401]  [<ffffffffa00229f6>] ? sas_phye_oob_error+0xc3/0xc3 [libsas]
[  229.214401]  [<ffffffffa00a827a>] isci_remote_device_gone+0x76/0x7e [isci]
[  229.214401]  [<ffffffffa002363e>] sas_notify_lldd_dev_gone+0x34/0x36 [libsas]
[  229.214401]  [<ffffffffa0023945>] sas_unregister_dev+0x57/0x9c [libsas]
[  229.214401]  [<ffffffffa00239c0>] sas_unregister_domain_devices+0x36/0x65 [libsas]
[  229.214401]  [<ffffffffa0022cb8>] sas_deform_port+0x72/0x1ac [libsas]
[  229.214401]  [<ffffffffa00229f6>] ? sas_phye_oob_error+0xc3/0xc3 [libsas]
[  229.214401]  [<ffffffffa0022a34>] sas_phye_loss_of_signal+0x3e/0x42 [libsas]

Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] ipr: add definitions for additional adapter
Wayne Boyer [Thu, 27 Oct 2011 18:58:21 +0000 (11:58 -0700)]
[SCSI] ipr: add definitions for additional adapter

Add the appropriate definition and table entry for an additional adapter.

Signed-off-by: Wayne Boyer <wayneb@linux.vnet.ibm.com>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] scsi_dh: check queuedata pointer before proceeding further
Moger, Babu [Wed, 26 Oct 2011 18:29:38 +0000 (14:29 -0400)]
[SCSI] scsi_dh: check queuedata pointer before proceeding further

This patch validates sdev pointer in scsi_dh_activate before proceeding further.

Without this check we might see the panic as below. I have seen this
panic multiple times..

Call trace:

 #0 [ffff88007d647b50] machine_kexec at ffffffff81020902
 #1 [ffff88007d647ba0] crash_kexec at ffffffff810875b0
 #2 [ffff88007d647c70] oops_end at ffffffff8139c650
 #3 [ffff88007d647c90] __bad_area_nosemaphore at ffffffff8102dd15
 #4 [ffff88007d647d50] page_fault at ffffffff8139b8cf
    [exception RIP: scsi_dh_activate+0x82]
    RIP: ffffffffa0041922  RSP: ffff88007d647e00  RFLAGS: 00010046
    RAX: 0000000000000000  RBX: 0000000000000000  RCX: 00000000000093c5
    RDX: 00000000000093c5  RSI: ffffffffa02e6640  RDI: ffff88007cc88988
    RBP: 000000000000000f   R8: ffff88007d646000   R9: 0000000000000000
    R10: ffff880082293790  R11: 00000000ffffffff  R12: ffff88007cc88988
    R13: 0000000000000000  R14: 0000000000000286  R15: ffff880037b845e0
    ORIG_RAX: ffffffffffffffff  CS: 0010  SS: 0000
 #5 [ffff88007d647e38] run_workqueue at ffffffff81060268
 #6 [ffff88007d647e78] worker_thread at ffffffff81060386
 #7 [ffff88007d647ee8] kthread at ffffffff81064436
 #8 [ffff88007d647f48] kernel_thread at ffffffff81003fba

Signed-off-by: Babu Moger <babu.moger@netapp.com>
Cc: stable@kernel.org
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] hpsa: detect controller lockup
Stephen M. Cameron [Wed, 26 Oct 2011 21:22:04 +0000 (16:22 -0500)]
[SCSI] hpsa: detect controller lockup

When controller lockup condition is detected,
we should fail all outstanding commands and disable
the controller.  This will enable multipath solutions
to recover gracefully.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] hpsa: fix flush cache transfer length
Stephen M. Cameron [Wed, 26 Oct 2011 21:21:17 +0000 (16:21 -0500)]
[SCSI] hpsa: fix flush cache transfer length

We weren't filling in the transfer length of the
flush cache command (it transfers 4 bytes of zeroes).
Firmware didn't seem to be bothered by this, but it
should be fixed.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] hpsa: fix potential array overflow in hpsa_update_scsi_devices
Scott Teel [Wed, 26 Oct 2011 21:21:12 +0000 (16:21 -0500)]
[SCSI] hpsa: fix potential array overflow in hpsa_update_scsi_devices

The currentsd[] array in hpsa_update_scsi_devices had room for
256 devices.  The code was iterating over however many physical
and logical devices plus an additional number of possible external
MSA2XXX controllers, which together could potentially exceed 256.

We increased the size of the currentsd array to 1024 + 1024 + 32 + 1
elements to reflect a reasonable maximum possible number of devices
which might be encountered.  We also don't just walk off the end
of the array if the array controller reports more devices than we
are prepared to handle, we just ignore the excessive devices.

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] hpsa: rename HPSA_MAX_SCSI_DEVS_PER_HBA
Scott Teel [Wed, 26 Oct 2011 21:21:07 +0000 (16:21 -0500)]
[SCSI] hpsa: rename HPSA_MAX_SCSI_DEVS_PER_HBA

Rename HPSA_MAX_SCSI_DEVS_PER_HBA to HPSA_MAX_DEVICES

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] hpsa: remove unused busy_initializing and busy_scanning
Stephen M. Cameron [Wed, 26 Oct 2011 21:20:58 +0000 (16:20 -0500)]
[SCSI] hpsa: remove unused busy_initializing and busy_scanning

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] hpsa: set max sectors instead of taking the default
Stephen M. Cameron [Wed, 26 Oct 2011 21:20:53 +0000 (16:20 -0500)]
[SCSI] hpsa: set max sectors instead of taking the default

Set the max hardware sectors in the SCSI host template to 8192
to allow for larger i/o's (8192 is the same limit the cciss
driver currently has.)

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] qla2xxx: Correct inadvertent clearing of RISC_INTR status.
Andrew Vasquez [Wed, 26 Oct 2011 21:20:13 +0000 (14:20 -0700)]
[SCSI] qla2xxx: Correct inadvertent clearing of RISC_INTR status.

During heavy I/O (CPU-affinity mode enabled) and CLI/Agent
interactions, the driver would report periodic mailbox command
timeout statuses.  Within the CPU-affinity ISR handler, the
driver should check the 'disable-msix-handshake' flag in deciding
whether or not to clear HCCRX_CLR_RISC_INT.  The mode is not
specific to a dedicated queue, instead, applies to the current
'ha' context.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] pmcraid: pmcraid_chr_ioctl uses incorrect argument order to kmalloc()
Dave Jones [Mon, 24 Oct 2011 22:10:26 +0000 (18:10 -0400)]
[SCSI] pmcraid: pmcraid_chr_ioctl uses incorrect argument order to kmalloc()

Size is 1st arg, not second.

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] bnx2fc: Bumped version to 1.0.9
Bhanu Prakash Gollapudi [Mon, 24 Oct 2011 06:23:58 +0000 (23:23 -0700)]
[SCSI] bnx2fc: Bumped version to 1.0.9

Signed-off-by: Bhanu Prakash Gollapudi <bprakash@broadcom.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] bnx2fc: Handle SRR LS_ACC drop scenario
Bhanu Prakash Gollapudi [Mon, 24 Oct 2011 06:23:57 +0000 (23:23 -0700)]
[SCSI] bnx2fc: Handle SRR LS_ACC drop scenario

When SRR LS_ACC is dropped, the driver was not issuing ABTS for SRR when it
times out. Since the target received SRR, it was able to send the XFER_RDY and
the the original IO request completed successfully. In this condition ABTS was
not sent during bnx2fc_srr_compl(). Fix this by first checking for ELS timeout
and issue ABTS before checking if original IO request is complete.

Signed-off-by: Bhanu Prakash Gollapudi <bprakash@broadcom.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] bnx2fc: Handle ABTS timeout during ulp timeout
Bhanu Prakash Gollapudi [Mon, 24 Oct 2011 06:23:56 +0000 (23:23 -0700)]
[SCSI] bnx2fc: Handle ABTS timeout during ulp timeout

If the IO and the corresponding ABTS are not responded by a target, cleanup the
IO and issue explicit logout when ulp timer expires while waiting for ABTS to
complete. Wait for the session to be ready before returning to the SCSI layer.
If the session is not ready let the SCSI-ml escalate the error recovery.

Signed-off-by: Bhanu Prakash Gollapudi <bprakash@broadcom.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] st: fix race in st_scsi_execute_end
Petr Uzel [Fri, 21 Oct 2011 11:31:09 +0000 (13:31 +0200)]
[SCSI] st: fix race in st_scsi_execute_end

The call to complete() in st_scsi_execute_end() wakes up sleeping thread
in write_behind_check(), which frees the st_request, thus invalidating
the pointer to the associated bio structure, which is then passed to the
blk_rq_unmap_user(). Fix by storing pointer to bio structure into
temporary local variable.

This bug is present since at least linux-2.6.32.

CC: stable@kernel.org
Signed-off-by: Petr Uzel <petr.uzel@suse.cz>
Reported-by: Juergen Groß <juergen.gross@ts.fujitsu.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Kai Mäkisara <kai.makisara@kolumbus.fi>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] Make scsi_free_queue() kill pending SCSI commands
Bart Van Assche [Fri, 23 Sep 2011 17:48:18 +0000 (19:48 +0200)]
[SCSI] Make scsi_free_queue() kill pending SCSI commands

Make sure that SCSI device removal via scsi_remove_host() does finish
all pending SCSI commands. Currently that's not the case and hence
removal of a SCSI host during I/O can cause a deadlock. See also
"blkdev_issue_discard() hangs forever if underlying storage device is
removed" (http://bugzilla.kernel.org/show_bug.cgi?id=40472). See also
http://lkml.org/lkml/2011/8/27/6.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Cc: <stable@kernel.org>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] sd: remove arbitrary SD_MAX_DISKS namespace limit
Dave Kleikamp [Wed, 19 Oct 2011 16:49:04 +0000 (11:49 -0500)]
[SCSI] sd: remove arbitrary SD_MAX_DISKS namespace limit

There is no reason to limit the SCSI disk namespace to sdXXX.

Add new error messages to sd_probe() in the unlikely event that either
ida_get_new() or sd_format_disk_name() fail.

Signed-off-by: Dave Kleikamp <dave.kleikamp@oracle.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] mpt2sas: Bump driver version to 10.100.00.00
nagalakshmi.nandigama@lsi.com [Wed, 19 Oct 2011 10:07:54 +0000 (15:37 +0530)]
[SCSI] mpt2sas: Bump driver version to 10.100.00.00

Bump driver vesion to 10.100.00.00

Signed-off-by: Nagalakshmi Nandigama <nagalakshmi.nandigama@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] mpt2sas: Fix for Panic when inactive volume is tried deleting
nagalakshmi.nandigama@lsi.com [Fri, 21 Oct 2011 04:38:07 +0000 (10:08 +0530)]
[SCSI] mpt2sas: Fix for Panic when inactive volume is tried deleting

The driver was setting the action to MPI2_CONFIG_ACTION_PAGE_READ_CURRENT,
which only returns active volumes. In order to get info on inactive volumes,
the driver needs to change the action to
MPI2_RAID_PGAD_FORM_GET_NEXT_CONFIGNUM, and traverse each config till the
iocstatus is MPI2_IOCSTATUS_CONFIG_INVALID_PAGE returned.
Added a change in the driver to remove the instance of
sas_device object when the driver returns "1" from the slave_configure callback.
Also fixed code to report the hot spares to the operating system with a /dev/sg
assigned.

Signed-off-by: Nagalakshmi Nandigama <nagalakshmi.nandigama@lsi.com>
Cc: stable@kernel.org
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] mpt2sas: Fix for issue Port Reset taking long time(around 5 mins) to complete...
nagalakshmi.nandigama@lsi.com [Wed, 19 Oct 2011 10:07:37 +0000 (15:37 +0530)]
[SCSI] mpt2sas: Fix for issue Port Reset taking long time(around 5 mins) to complete while issued during creating a volume

This is due to the slave_configuration routine is getting called when
host reset is active, and config page reads are failing, and driver
attempts to added device with stale config data.

To fix the issue, added error checking in slave_configure to check
for configuration pages failing, and return "1" so the device  is
not configured.  The config pages are failing if raid volume is
configured while issuing a host reset, thus driver is reading stale
data and proceeding to attempt to add.  The fix is to return error
so the volume is not configured.

Signed-off-by: Nagalakshmi Nandigama <nagalakshmi.nandigama@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] mpt2sas: Fix for deadlock between hot plug worker threads and host reset context
nagalakshmi.nandigama@lsi.com [Wed, 19 Oct 2011 10:07:24 +0000 (15:37 +0530)]
[SCSI] mpt2sas: Fix for deadlock between hot plug worker threads and host reset context

This is due to driver reporting a device missing to the OS then the OS sending
a SYNC_CACHE request to driver while the IO queues are locked due to host reset.

To fix the issue, the driver will be waking up the port enable context
immediately when the driver receives the reply message, instead of waiting
on the hot plug worker threads.

Signed-off-by: Nagalakshmi Nandigama <nagalakshmi.nandigama@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] mpt2sas: Fix for dead lock occurring between host_lock and sas_device_lock
nagalakshmi.nandigama@lsi.com [Wed, 19 Oct 2011 10:07:14 +0000 (15:37 +0530)]
[SCSI] mpt2sas: Fix for dead lock occurring between host_lock and sas_device_lock

Fix for dead lock occurring between host_lock and sas_device_lock.

The deadlock is between two spin locks, between the shost->host_lock
and driver ioc->sas_device_lock.

The fix is to rearrange the code in the  FW/Driver device removal
handshake so the ioc->sas_device_lock is not occurring when the
shost->host_lock is taken.

[jejb: zero initialise sas_address to fix spurious compiler warning]
Signed-off-by: Nagalakshmi Nandigama <nagalakshmi.nandigama@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] mpt2sas: Fix drives not getting properly deleted if sas cable is removed while...
nagalakshmi.nandigama@lsi.com [Wed, 19 Oct 2011 10:07:00 +0000 (15:37 +0530)]
[SCSI] mpt2sas: Fix drives not getting properly deleted if sas cable is removed while host reset is active

The fix is in the driver-firmware handshake device removal code. We
need to read the controller ioc_state to see if controller is OPERATIONAL
prior to sending target reset and OP_REMOVE. Previously it was checking
the flag ioc->shost_recovery flag, which is always set when host reset is
active, thus preventing drives from getting properly deleted.

Signed-off-by: Nagalakshmi Nandigama <nagalakshmi.nandigama@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] mpt2sas: Fix failure message displayed during diag reset
nagalakshmi.nandigama@lsi.com [Wed, 19 Oct 2011 10:06:47 +0000 (15:36 +0530)]
[SCSI] mpt2sas: Fix failure message displayed during diag reset

The fix is to inhibit the warning message in _scsih_get_sas_address
when the MPI2_IOCSTATUS_CONFIG_INVALID_PAGE ioc status is returned.

Signed-off-by: Nagalakshmi Nandigama <nagalakshmi.nandigama@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] mpt2sas: Fix for system hang when discovery in progress
nagalakshmi.nandigama@lsi.com [Fri, 21 Oct 2011 04:36:33 +0000 (10:06 +0530)]
[SCSI] mpt2sas: Fix for system hang when discovery in progress

Fix for issue : While discovery is in progress, hot unplug and hot plug of
enclosure connected to the controller card is causing system to hang.

When a device is in the process of being detected at driver load time then
if it is removed, the device that is no longer present will not be added
to the list. So the code in _scsih_probe_sas() is rearranged as such so
the devices that failed to be detected are not added to the list.

Signed-off-by: Nagalakshmi Nandigama <nagalakshmi.nandigama@lsi.com>
Cc: stable@kernel.org
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] mpt2sas: New feature - Fast Load Support
nagalakshmi.nandigama@lsi.com [Wed, 19 Oct 2011 10:06:26 +0000 (15:36 +0530)]
[SCSI] mpt2sas: New feature - Fast Load Support

New feature Fast Load Support.

(1)Asynchronous SCSI scanning: This will allow the drivers to scan
for devices in parallel while other device drivers are loading at
the same time. This will improve the amount of time it takes for the
OS to load.

(2) Reporting Devices while port enable is active: This feature will
allow devices to be reported to OS immediately while port enable is
active. The previous implementation waits for port enable to complete,
and then report devices. This feature is only enabled on IT firmware
configurations when there are no boot device configured in BIOS Configuration
Utility, else the driver will wait till port enable completes reporting
devices. For IR firmware, this feature is turned off. This feature is to
address large SAS topologies (>100 drives) when the boot OS is using onboard
SATA device, in other words, the boot devices is not
connected to our controller.

(3) Scanning for devices after diagnostic reset completes: A new routine
_scsih_scan_start is added. This will scan the expander pages, IR pages,
and sas device pages, then reporting new devices to SCSI Mid layer. It
seems the driver is not supporting adding devices while diagnostic reset
is active. Apparently this is due to the sanity checks on
ioc->shost_recovery flag throughout the context of kernel work thread FIFO,
and the mpt2sas_fw_work.

Signed-off-by: Nagalakshmi Nandigama <nagalakshmi.nandigama@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years ago[SCSI] mpt2sas: MPI next revision header update
nagalakshmi.nandigama@lsi.com [Wed, 19 Oct 2011 10:06:05 +0000 (15:36 +0530)]
[SCSI] mpt2sas: MPI next revision header update

1)Added ProxyVF_ID field to Configuration Request message.
2)Added IO Unit Page 8, IO Unit Page 9,and IO Unit Page 10.
3)Added SASNotifyPrimitiveMasks field to IOC Page 7.
4)Added SAS NOTIFY Primitive event.
5)Added Temperature Threshold Event.
6)Added Host Message Event.
7)Added Send Host Message request and reply.

Signed-off-by: Nagalakshmi Nandigama <nagalakshmi.nandigama@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6
Linus Torvalds [Fri, 28 Oct 2011 23:44:18 +0000 (16:44 -0700)]
Merge git://git./linux/kernel/git/jejb/scsi-misc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6: (204 commits)
  [SCSI] qla4xxx: export address/port of connection (fix udev disk names)
  [SCSI] ipr: Fix BUG on adapter dump timeout
  [SCSI] megaraid_sas: Fix instance access in megasas_reset_timer
  [SCSI] hpsa: change confusing message to be more clear
  [SCSI] iscsi class: fix vlan configuration
  [SCSI] qla4xxx: fix data alignment and use nl helpers
  [SCSI] iscsi class: fix link local mispelling
  [SCSI] iscsi class: Replace iscsi_get_next_target_id with IDA
  [SCSI] aacraid: use lower snprintf() limit
  [SCSI] lpfc 8.3.27: Change driver version to 8.3.27
  [SCSI] lpfc 8.3.27: T10 additions for SLI4
  [SCSI] lpfc 8.3.27: Fix queue allocation failure recovery
  [SCSI] lpfc 8.3.27: Change algorithm for getting physical port name
  [SCSI] lpfc 8.3.27: Changed worst case mailbox timeout
  [SCSI] lpfc 8.3.27: Miscellanous logic and interface fixes
  [SCSI] megaraid_sas: Changelog and version update
  [SCSI] megaraid_sas: Add driver workaround for PERC5/1068 kdump kernel panic
  [SCSI] megaraid_sas: Add multiple MSI-X vector/multiple reply queue support
  [SCSI] megaraid_sas: Add support for MegaRAID 9360/9380 12GB/s controllers
  [SCSI] megaraid_sas: Clear FUSION_IN_RESET before enabling interrupts
  ...

13 years agoMerge branch 'for-linus' of git://ceph.newdream.net/git/ceph-client
Linus Torvalds [Fri, 28 Oct 2011 23:42:18 +0000 (16:42 -0700)]
Merge branch 'for-linus' of git://ceph.newdream.net/git/ceph-client

* 'for-linus' of git://ceph.newdream.net/git/ceph-client:
  libceph: fix double-free of page vector
  ceph: fix 32-bit ino numbers
  libceph: force resend of osd requests if we skip an osdmap
  ceph: use kernel DNS resolver
  ceph: fix ceph_monc_init memory leak
  ceph: let the set_layout ioctl set single traits
  Revert "ceph: don't truncate dirty pages in invalidate work thread"
  ceph: replace leading spaces with tabs
  libceph: warn on msg allocation failures
  libceph: don't complain on msgpool alloc failures
  libceph: always preallocate mon connection
  libceph: create messenger with client
  ceph: document ioctls
  ceph: implement (optional) max read size
  ceph: rename rsize -> rasize
  ceph: make readpages fully async

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Linus Torvalds [Fri, 28 Oct 2011 21:25:01 +0000 (14:25 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound: (549 commits)
  ALSA: hda - Fix ADC input-amp handling for Cx20549 codec
  ALSA: hda - Keep EAPD turned on for old Conexant chips
  ALSA: hda/realtek - Fix missing volume controls with ALC260
  ASoC: wm8940: Properly set codec->dapm.bias_level
  ALSA: hda - Fix pin-config for ASUS W90V
  ALSA: hda - Fix surround/CLFE headphone and speaker pins order
  ALSA: hda - Fix typo
  ALSA: Update the sound git tree URL
  ALSA: HDA: Add new revision for ALC662
  ASoC: max98095: Convert codec->hw_write to snd_soc_write
  ASoC: keep pointer to resource so it can be freed
  ASoC: sgtl5000: Fix wrong mask in some snd_soc_update_bits calls
  ASoC: wm8996: Fix wrong mask for setting WM8996_AIF_CLOCKING_2
  ASoC: da7210: Add support for line out and DAC
  ASoC: da7210: Add support for DAPM
  ALSA: hda/realtek - Fix DAC assignments of multiple speakers
  ASoC: Use SGTL5000_LINREG_VDDD_MASK instead of hardcoded mask value
  ASoC: Set sgtl5000->ldo in ldo_regulator_register
  ASoC: wm8996: Use SND_SOC_DAPM_AIF_OUT for AIF2 Capture
  ASoC: wm8994: Use SND_SOC_DAPM_AIF_OUT for AIF3 Capture
  ...

13 years agoMerge branch 'next-rebase' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci
Linus Torvalds [Fri, 28 Oct 2011 21:20:44 +0000 (14:20 -0700)]
Merge branch 'next-rebase' of git://git./linux/kernel/git/jbarnes/pci

* 'next-rebase' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci:
  PCI: Clean-up MPS debug output
  pci: Clamp pcie_set_readrq() when using "performance" settings
  PCI: enable MPS "performance" setting to properly handle bridge MPS
  PCI: Workaround for Intel MPS errata
  PCI: Add support for PASID capability
  PCI: Add implementation for PRI capability
  PCI: Export ATS functions to modules
  PCI: Move ATS implementation into own file
  PCI / PM: Remove unnecessary error variable from acpi_dev_run_wake()
  PCI hotplug: acpiphp: Prevent deadlock on PCI-to-PCI bridge remove
  PCI / PM: Extend PME polling to all PCI devices
  PCI quirk: mmc: Always check for lower base frequency quirk for Ricoh 1180:e823
  PCI: Make pci_setup_bridge() non-static for use by arch code
  x86: constify PCI raw ops structures
  PCI: Add quirk for known incorrect MPSS
  PCI: Add Solarflare vendor ID and SFC4000 device IDs

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cjb/mmc
Linus Torvalds [Fri, 28 Oct 2011 21:16:11 +0000 (14:16 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/cjb/mmc

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cjb/mmc: (83 commits)
  mmc: fix compile error when CONFIG_BLOCK is not enabled
  mmc: core: Cleanup eMMC4.5 conditionals
  mmc: omap_hsmmc: if multiblock reads are broken, disable them
  mmc: core: add workaround for controllers with broken multiblock reads
  mmc: core: Prevent too long response times for suspend
  mmc: recognise SDIO cards with SDIO_CCCR_REV 3.00
  mmc: sd: Handle SD3.0 cards not supporting UHS-I bus speed mode
  mmc: core: support HPI send command
  mmc: core: Add cache control for eMMC4.5 device
  mmc: core: Modify the timeout value for writing power class
  mmc: core: new discard feature support at eMMC v4.5
  mmc: core: mmc sanitize feature support for v4.5
  mmc: dw_mmc: modify DATA register offset
  mmc: sdhci-pci: add flag for devices that can support runtime PM
  mmc: omap_hsmmc: ensure pbias configuration is always done
  mmc: core: Add Power Off Notify Feature eMMC 4.5
  mmc: sdhci-s3c: fix potential NULL dereference
  mmc: replace printk with appropriate display macro
  mmc: core: Add default timeout value for CMD6
  mmc: sdhci-pci: add runtime pm support
  ...

13 years agoMerge branch 'devel-stable' of http://ftp.arm.linux.org.uk/pub/linux/arm/kernel/git...
Linus Torvalds [Fri, 28 Oct 2011 19:02:27 +0000 (12:02 -0700)]
Merge branch 'devel-stable' of ftp.arm.linux.org.uk/pub/linux/arm/kernel/git-cur/linux-2.6-arm

* 'devel-stable' of http://ftp.arm.linux.org.uk/pub/linux/arm/kernel/git-cur/linux-2.6-arm: (178 commits)
  ARM: 7139/1: fix compilation with CONFIG_ARM_ATAG_DTB_COMPAT and large TEXT_OFFSET
  ARM: gic, local timers: use the request_percpu_irq() interface
  ARM: gic: consolidate PPI handling
  ARM: switch from NO_MACH_MEMORY_H to NEED_MACH_MEMORY_H
  ARM: mach-s5p64x0: remove mach/memory.h
  ARM: mach-s3c64xx: remove mach/memory.h
  ARM: plat-mxc: remove mach/memory.h
  ARM: mach-prima2: remove mach/memory.h
  ARM: mach-zynq: remove mach/memory.h
  ARM: mach-bcmring: remove mach/memory.h
  ARM: mach-davinci: remove mach/memory.h
  ARM: mach-pxa: remove mach/memory.h
  ARM: mach-ixp4xx: remove mach/memory.h
  ARM: mach-h720x: remove mach/memory.h
  ARM: mach-vt8500: remove mach/memory.h
  ARM: mach-s5pc100: remove mach/memory.h
  ARM: mach-tegra: remove mach/memory.h
  ARM: plat-tcc: remove mach/memory.h
  ARM: mach-mmp: remove mach/memory.h
  ARM: mach-cns3xxx: remove mach/memory.h
  ...

Fix up mostly pretty trivial conflicts in:
 - arch/arm/Kconfig
 - arch/arm/include/asm/localtimer.h
 - arch/arm/kernel/Makefile
 - arch/arm/mach-shmobile/board-ap4evb.c
 - arch/arm/mach-u300/core.c
 - arch/arm/mm/dma-mapping.c
 - arch/arm/mm/proc-v7.S
 - arch/arm/plat-omap/Kconfig
largely due to some CONFIG option renaming (ie CONFIG_PM_SLEEP ->
CONFIG_ARM_CPU_SUSPEND for the arm-specific suspend code etc) and
addition of NEED_MACH_MEMORY_H next to HAVE_IDE.

13 years agoMerge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/hch/vfs...
Linus Torvalds [Fri, 28 Oct 2011 17:49:34 +0000 (10:49 -0700)]
Merge branch 'for-next' of git://git./linux/kernel/git/hch/vfs-queue

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/hch/vfs-queue: (21 commits)
  leases: fix write-open/read-lease race
  nfs: drop unnecessary locking in llseek
  ext4: replace cut'n'pasted llseek code with generic_file_llseek_size
  vfs: add generic_file_llseek_size
  vfs: do (nearly) lockless generic_file_llseek
  direct-io: merge direct_io_walker into __blockdev_direct_IO
  direct-io: inline the complete submission path
  direct-io: separate map_bh from dio
  direct-io: use a slab cache for struct dio
  direct-io: rearrange fields in dio/dio_submit to avoid holes
  direct-io: fix a wrong comment
  direct-io: separate fields only used in the submission path from struct dio
  vfs: fix spinning prevention in prune_icache_sb
  vfs: add a comment to inode_permission()
  vfs: pass all mask flags check_acl and posix_acl_permission
  vfs: add hex format for MAY_* flag values
  vfs: indicate that the permission functions take all the MAY_* flags
  compat: sync compat_stats with statfs.
  vfs: add "device" tag to /proc/self/mountstats
  cleanup: vfs: small comment fix for block_invalidatepage
  ...

Fix up trivial conflict in fs/gfs2/file.c (llseek changes)

13 years agoMerge http://sucs.org/~rohan/git/gfs2-3.0-nmw
Linus Torvalds [Fri, 28 Oct 2011 17:44:50 +0000 (10:44 -0700)]
Merge http://sucs.org/~rohan/git/gfs2-3.0-nmw

* http://sucs.org/~rohan/git/gfs2-3.0-nmw: (24 commits)
  GFS2: Move readahead of metadata during deallocation into its own function
  GFS2: Remove two unused variables
  GFS2: Misc fixes
  GFS2: rewrite fallocate code to write blocks directly
  GFS2: speed up delete/unlink performance for large files
  GFS2: Fix off-by-one in gfs2_blk2rgrpd
  GFS2: Clean up ->page_mkwrite
  GFS2: Correctly set goal block after allocation
  GFS2: Fix AIL flush issue during fsync
  GFS2: Use cached rgrp in gfs2_rlist_add()
  GFS2: Call do_strip() directly from recursive_scan()
  GFS2: Remove obsolete assert
  GFS2: Cache the most recently used resource group in the inode
  GFS2: Make resource groups "append only" during life of fs
  GFS2: Use rbtree for resource groups and clean up bitmap buffer ref count scheme
  GFS2: Fix lseek after SEEK_DATA, SEEK_HOLE have been added
  GFS2: Clean up gfs2_create
  GFS2: Use ->dirty_inode()
  GFS2: Fix bug trap and journaled data fsync
  GFS2: Fix inode allocation error path
  ...

13 years agoMerge branch '3.2-without-smb2' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Fri, 28 Oct 2011 17:43:32 +0000 (10:43 -0700)]
Merge branch '3.2-without-smb2' of git://git.samba.org/sfrench/cifs-2.6

* '3.2-without-smb2' of git://git.samba.org/sfrench/cifs-2.6: (52 commits)
  Fix build break when freezer not configured
  Add definition for share encryption
  CIFS: Make cifs_push_locks send as many locks at once as possible
  CIFS: Send as many mandatory unlock ranges at once as possible
  CIFS: Implement caching mechanism for posix brlocks
  CIFS: Implement caching mechanism for mandatory brlocks
  CIFS: Fix DFS handling in cifs_get_file_info
  CIFS: Fix error handling in cifs_readv_complete
  [CIFS] Fixup trivial checkpatch warning
  [CIFS] Show nostrictsync and noperm mount options in /proc/mounts
  cifs, freezer: add wait_event_freezekillable and have cifs use it
  cifs: allow cifs_max_pending to be readable under /sys/module/cifs/parameters
  cifs: tune bdi.ra_pages in accordance with the rsize
  cifs: allow for larger rsize= options and change defaults
  cifs: convert cifs_readpages to use async reads
  cifs: add cifs_async_readv
  cifs: fix protocol definition for READ_RSP
  cifs: add a callback function to receive the rest of the frame
  cifs: break out 3rd receive phase into separate function
  cifs: find mid earlier in receive codepath
  ...

13 years agoMerge branch 'for-linus' of git://oss.sgi.com/xfs/xfs
Linus Torvalds [Fri, 28 Oct 2011 17:31:42 +0000 (10:31 -0700)]
Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfs

* 'for-linus' of git://oss.sgi.com/xfs/xfs: (69 commits)
  xfs: add AIL pushing tracepoints
  xfs: put in missed fix for merge problem
  xfs: do not flush data workqueues in xfs_flush_buftarg
  xfs: remove XFS_bflush
  xfs: remove xfs_buf_target_name
  xfs: use xfs_ioerror_alert in xfs_buf_iodone_callbacks
  xfs: clean up xfs_ioerror_alert
  xfs: clean up buffer allocation
  xfs: remove buffers from the delwri list in xfs_buf_stale
  xfs: remove XFS_BUF_STALE and XFS_BUF_SUPER_STALE
  xfs: remove XFS_BUF_SET_VTYPE and XFS_BUF_SET_VTYPE_REF
  xfs: remove XFS_BUF_FINISH_IOWAIT
  xfs: remove xfs_get_buftarg_list
  xfs: fix buffer flushing during unmount
  xfs: optimize fsync on directories
  xfs: reduce the number of log forces from tail pushing
  xfs: Don't allocate new buffers on every call to _xfs_buf_find
  xfs: simplify xfs_trans_ijoin* again
  xfs: unlock the inode before log force in xfs_change_file_space
  xfs: unlock the inode before log force in xfs_fs_nfs_commit_metadata
  ...

13 years agoleases: fix write-open/read-lease race
J. Bruce Fields [Wed, 21 Sep 2011 14:58:13 +0000 (10:58 -0400)]
leases: fix write-open/read-lease race

In setlease, we use i_writecount to decide whether we can give out a
read lease.

In open, we break leases before incrementing i_writecount.

There is therefore a window between the break lease and the i_writecount
increment when setlease could add a new read lease.

This would leave us with a simultaneous write open and read lease, which
shouldn't happen.

Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agonfs: drop unnecessary locking in llseek
Andi Kleen [Thu, 15 Sep 2011 23:06:52 +0000 (16:06 -0700)]
nfs: drop unnecessary locking in llseek

This makes NFS follow the standard generic_file_llseek locking scheme.

Cc: Trond.Myklebust@netapp.com
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agoext4: replace cut'n'pasted llseek code with generic_file_llseek_size
Andi Kleen [Thu, 15 Sep 2011 23:06:51 +0000 (16:06 -0700)]
ext4: replace cut'n'pasted llseek code with generic_file_llseek_size

This gives ext4 the benefits of unlocked llseek.

Cc: tytso@mit.edu
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agovfs: add generic_file_llseek_size
Andi Kleen [Thu, 15 Sep 2011 23:06:50 +0000 (16:06 -0700)]
vfs: add generic_file_llseek_size

Add a generic_file_llseek variant to the VFS that allows passing in
the maximum file size of the file system, instead of always
using maxbytes from the superblock.

This can be used to eliminate some cut'n'paste seek code in ext4.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agovfs: do (nearly) lockless generic_file_llseek
Andi Kleen [Thu, 15 Sep 2011 23:06:48 +0000 (16:06 -0700)]
vfs: do (nearly) lockless generic_file_llseek

The i_mutex lock use of generic _file_llseek hurts.  Independent processes
accessing the same file synchronize over a single lock, even though
they have no need for synchronization at all.

Under high utilization this can cause llseek to scale very poorly on larger
systems.

This patch does some rethinking of the llseek locking model:

First the 64bit f_pos is not necessarily atomic without locks
on 32bit systems. This can already cause races with read() today.
This was discussed on linux-kernel in the past and deemed acceptable.
The patch does not change that.

Let's look at the different seek variants:

SEEK_SET: Doesn't really need any locking.
If there's a race one writer wins, the other loses.

For 32bit the non atomic update races against read()
stay the same. Without a lock they can also happen
against write() now.  The read() race was deemed
acceptable in past discussions, and I think if it's
ok for read it's ok for write too.

=> Don't need a lock.

SEEK_END: This behaves like SEEK_SET plus it reads
the maximum size too. Reading the maximum size would have the
32bit atomic problem. But luckily we already have a way to read
the maximum size without locking (i_size_read), so we
can just use that instead.

Without i_mutex there is no synchronization with write() anymore,
however since the write() update is atomic on 64bit it just behaves
like another racy SEEK_SET.  On non atomic 32bit it's the same
as SEEK_SET.

=> Don't need a lock, but need to use i_size_read()

SEEK_CUR: This has a read-modify-write race window
on the same file. One could argue that any application
doing unsynchronized seeks on the same file is already broken.
But for the sake of not adding a regression here I'm
using the file->f_lock to synchronize this. Using this
lock is much better than the inode mutex because it doesn't
synchronize between processes.

=> So still need a lock, but can use a f_lock.

This patch implements this new scheme in generic_file_llseek.
I dropped generic_file_llseek_unlocked and changed all callers.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agodirect-io: merge direct_io_walker into __blockdev_direct_IO
Andi Kleen [Tue, 2 Aug 2011 04:38:09 +0000 (21:38 -0700)]
direct-io: merge direct_io_walker into __blockdev_direct_IO

This doesn't change anything for the compiler, but hch thought it would
make the code clearer.

I moved the reference counting into its own little inline.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agodirect-io: inline the complete submission path
Andi Kleen [Tue, 2 Aug 2011 04:38:08 +0000 (21:38 -0700)]
direct-io: inline the complete submission path

Add inlines to all the submission path functions. While this increases
code size it also gives gcc a lot of optimization opportunities
in this critical hotpath.

In particular -- together with some other changes -- this
allows gcc to get rid of the unnecessary clearing of
sdio at the beginning and optimize the messy parameter passing.
Any non inlining of a function which takes a sdio parameter
would break this optimization because they cannot be done if the
address of a structure is taken.

Note that benefits are only seen with CONFIG_OPTIMIZE_INLINING
and CONFIG_CC_OPTIMIZE_FOR_SIZE both set to off.

This gives about 2.2% improvement on a large database benchmark
with a high IOPS rate.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agodirect-io: separate map_bh from dio
Andi Kleen [Tue, 2 Aug 2011 04:38:07 +0000 (21:38 -0700)]
direct-io: separate map_bh from dio

Only a single b_private field in the map_bh buffer head is needed after
the submission path. Move map_bh separately to avoid storing
this information in the long term slab.

This avoids the weird 104 byte hole in struct dio_submit which also needed
to be memseted early.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agodirect-io: use a slab cache for struct dio
Andi Kleen [Tue, 2 Aug 2011 04:38:06 +0000 (21:38 -0700)]
direct-io: use a slab cache for struct dio

A direct slab call is slightly faster than kmalloc and can be better cached
per CPU. It also avoids rounding to the next kmalloc slab.

In addition this enforces cache line alignment for struct dio to avoid
any false sharing.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agodirect-io: rearrange fields in dio/dio_submit to avoid holes
Andi Kleen [Tue, 2 Aug 2011 04:38:05 +0000 (21:38 -0700)]
direct-io: rearrange fields in dio/dio_submit to avoid holes

Fix most problems reported by pahole.

There is still a weird 104 byte hole after map_bh. I'm not sure what
causes this.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agodirect-io: fix a wrong comment
Andi Kleen [Tue, 2 Aug 2011 04:38:04 +0000 (21:38 -0700)]
direct-io: fix a wrong comment

There's nothing on the stack, even before my changes.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agodirect-io: separate fields only used in the submission path from struct dio
Andi Kleen [Tue, 2 Aug 2011 04:38:03 +0000 (21:38 -0700)]
direct-io: separate fields only used in the submission path from struct dio

This large, but largely mechanic, patch moves all fields in struct dio
that are only used in the submission path into a separate on stack
data structure. This has the advantage that the memory is very likely
cache hot, which is not guaranteed for memory fresh out of kmalloc.

This also gives gcc more optimization potential because it can easier
determine that there are no external aliases for these variables.

The sdio initialization is a initialization now instead of memset.
This allows gcc to break sdio into individual fields and optimize
away unnecessary zeroing (after all the functions are inlined)

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agovfs: fix spinning prevention in prune_icache_sb
Christoph Hellwig [Fri, 28 Oct 2011 08:03:41 +0000 (10:03 +0200)]
vfs: fix spinning prevention in prune_icache_sb

We need to move the inode to the end of the list to actually make the
spinning prevention explained in the comment above it work.  With a
plain list_move it will simply stay in place as we're always reclaiming
from the head of the list.

Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agovfs: add a comment to inode_permission()
Andreas Gruenbacher [Sun, 23 Oct 2011 17:43:33 +0000 (23:13 +0530)]
vfs: add a comment to inode_permission()

Acked-by: J. Bruce Fields <bfields@redhat.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruen@kernel.org>
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agovfs: pass all mask flags check_acl and posix_acl_permission
Andreas Gruenbacher [Sun, 23 Oct 2011 17:43:32 +0000 (23:13 +0530)]
vfs: pass all mask flags check_acl and posix_acl_permission

Acked-by: J. Bruce Fields <bfields@redhat.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruen@kernel.org>
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agovfs: add hex format for MAY_* flag values
Aneesh Kumar K.V [Sun, 23 Oct 2011 17:43:31 +0000 (23:13 +0530)]
vfs: add hex format for MAY_* flag values

We are going to add more flags and having them in hex format
make it simpler

Acked-by: J. Bruce Fields <bfields@redhat.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agovfs: indicate that the permission functions take all the MAY_* flags
Andreas Gruenbacher [Sun, 23 Oct 2011 17:43:30 +0000 (23:13 +0530)]
vfs: indicate that the permission functions take all the MAY_* flags

Acked-by: J. Bruce Fields <bfields@redhat.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruen@kernel.org>
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agocompat: sync compat_stats with statfs.
Eric W. Biederman [Mon, 17 Oct 2011 20:40:02 +0000 (13:40 -0700)]
compat: sync compat_stats with statfs.

This was found by inspection while tracking a similar
bug in compat_statfs64, that has been fixed in mainline
since decemeber.

- This fixes a bug where not all of the f_spare fields
  were cleared on mips and s390.
- Add the f_flags field to struct compat_statfs
- Copy f_flags to userspace in case someone cares.
- Use __clear_user to copy the f_spare field to userspace
  to ensure that all of the elements of f_spare are cleared.
  On some architectures f_spare is has 5 ints and on some
  architectures f_spare only has 4 ints.  Which makes
  the previous technique of clearing each int individually
  broken.

I don't expect anyone actually uses the old statfs system
call anymore but if they do let them benefit from having
the compat and the native version working the same.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agoMerge branch 'drm-core-next' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Fri, 28 Oct 2011 12:54:23 +0000 (05:54 -0700)]
Merge branch 'drm-core-next' of git://people.freedesktop.org/~airlied/linux

* 'drm-core-next' of git://people.freedesktop.org/~airlied/linux: (290 commits)
  Revert "drm/ttm: add a way to bo_wait for either the last read or last write"
  Revert "drm/radeon/kms: add a new gem_wait ioctl with read/write flags"
  vmwgfx: Don't pass unused arguments to do_dirty functions
  vmwgfx: Emulate depth 32 framebuffers
  drm/radeon: Lower the severity of the radeon lockup messages.
  drm/i915/dp: Fix eDP on PCH DP on CPT/PPT
  drm/i915/dp: Introduce is_cpu_edp()
  drm/i915: use correct SPD type value
  drm/i915: fix ILK+ infoframe support
  drm/i915: add DP test request handling
  drm/i915: read full receiver capability field during DP hot plug
  drm/i915/dp: Remove eDP special cases from bandwidth checks
  drm/i915/dp: Fix the math in intel_dp_link_required
  drm/i915/panel: Always record the backlight level again (but cleverly)
  i915: Move i915_read/write out of line
  drm/i915: remove transcoder PLL mashing from mode_set per specs
  drm/i915: if transcoder disable fails, say which
  drm/i915: set watermarks for third pipe on IVB
  drm/i915: export a CPT mode set verification function
  drm/i915: fix transcoder PLL select masking
  ...

13 years agoMerge branch 'x86-vdso-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 28 Oct 2011 12:46:02 +0000 (05:46 -0700)]
Merge branch 'x86-vdso-for-linus' of git://git./linux/kernel/git/tip/tip

* 'x86-vdso-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86-64, doc: Remove int 0xcc from entry_64.S documentation
  x86, vsyscall: Add missing <asm/fixmap.h> to arch/x86/mm/fault.c

Fix up trivial conflicts in arch/x86/mm/fault.c (asm/fixmap.h vs
asm/vsyscall.h: both work, which to use? Whatever..)

13 years agoMerge branch 'x86-uv-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 28 Oct 2011 12:43:56 +0000 (05:43 -0700)]
Merge branch 'x86-uv-for-linus' of git://git./linux/kernel/git/tip/tip

* 'x86-uv-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86: uv2: Workaround for UV2 Hub bug (system global address format)

13 years agoMerge branch 'x86-spinlocks-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 28 Oct 2011 12:43:29 +0000 (05:43 -0700)]
Merge branch 'x86-spinlocks-for-linus' of git://git./linux/kernel/git/tip/tip

* 'x86-spinlocks-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, ticketlock: remove obsolete comment
  x86, cmpxchg: Use __compiletime_error() to make usage messages a bit nicer
  x86, ticketlock: Make __ticket_spin_trylock common
  x86, ticketlock: Convert __ticket_spin_lock to use xadd()
  x86, ticketlock: Convert spin loop to C
  x86, ticketlock: Clean up types and accessors
  x86: Use xadd helper more widely
  x86: Add xadd helper macro
  x86, cmpxchg: Unify cmpxchg into cmpxchg.h
  x86, cmpxchg: Move 64-bit set64_bit() to match 32-bit
  x86, cmpxchg: Move 32-bit __cmpxchg_wrong_size to match 64 bit.
  x86, cmpxchg: <linux/alternative.h> has LOCK_PREFIX

13 years agoMerge branch 'x86-rdrand-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 28 Oct 2011 12:29:07 +0000 (05:29 -0700)]
Merge branch 'x86-rdrand-for-linus' of git://git./linux/kernel/git/tip/tip

* 'x86-rdrand-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, random: Verify RDRAND functionality and allow it to be disabled
  x86, random: Architectural inlines to get random integers with RDRAND
  random: Add support for architectural random hooks

Fix up trivial conflicts in drivers/char/random.c: the architectural
random hooks touched "get_random_int()" that was simplified to use MD5
and not do the keyptr thing any more (see commit 6e5714eaf77d: "net:
Compute protocol sequence numbers and fragment IDs using MD5").

13 years agoMerge branch 'x86-microcode-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 28 Oct 2011 12:14:48 +0000 (05:14 -0700)]
Merge branch 'x86-microcode-for-linus' of git://git./linux/kernel/git/tip/tip

* 'x86-microcode-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, microcode, AMD: Add microcode revision to /proc/cpuinfo
  x86, microcode: Correct microcode revision format
  coretemp: Get microcode revision from cpu_data
  x86, intel: Use c->microcode for Atom errata check
  x86, intel: Output microcode revision in /proc/cpuinfo
  x86, microcode: Don't request microcode from userspace unnecessarily

Fix up trivial conflicts in arch/x86/kernel/cpu/amd.c (conflict between
moving AMD BSP code to cpu_dev helper function and adding AMD microcode
revision to /proc/cpuinfo code)

13 years agoMerge branch 'x86-hyperv-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 28 Oct 2011 12:08:40 +0000 (05:08 -0700)]
Merge branch 'x86-hyperv-for-linus' of git://git./linux/kernel/git/tip/tip

* 'x86-hyperv-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86: Hyper-V: Integrate the clocksource with Hyper-V detection code

Fix up conflicts in drivers/staging/hv/Makefile manually (some of the hv
code has moved out of staging to drivers/hv/)

13 years agoMerge branch 'x86-geode-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 28 Oct 2011 12:04:26 +0000 (05:04 -0700)]
Merge branch 'x86-geode-for-linus' of git://git./linux/kernel/git/tip/tip

* 'x86-geode-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86: geode: New PCEngines Alix system driver

13 years agoMerge branch 'x86-cpufeature-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 28 Oct 2011 12:04:04 +0000 (05:04 -0700)]
Merge branch 'x86-cpufeature-for-linus' of git://git./linux/kernel/git/tip/tip

* 'x86-cpufeature-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, cpu: Add cpufeature flag for PCIDs

13 years agoMerge branch 'x86-cpu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 28 Oct 2011 12:03:12 +0000 (05:03 -0700)]
Merge branch 'x86-cpu-for-linus' of git://git./linux/kernel/git/tip/tip

* 'x86-cpu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, amd: Include linux/elf.h since we use stuff from asm/elf.h
  x86: cache_info: Update calculation of AMD L3 cache indices
  x86: cache_info: Kill the atomic allocation in amd_init_l3_cache()
  x86: cache_info: Kill the moronic shadow struct
  x86: cache_info: Remove bogus free of amd_l3_cache data
  x86, amd: Include elf.h explicitly, prepare the code for the module.h split
  x86-32, amd: Move va_align definition to unbreak 32-bit build
  x86, amd: Move BSP code to cpu_dev helper
  x86: Add a BSP cpu_dev helper
  x86, amd: Avoid cache aliasing penalties on AMD family 15h

13 years agovfs: add "device" tag to /proc/self/mountstats
Bryan Schumaker [Fri, 7 Oct 2011 17:41:15 +0000 (13:41 -0400)]
vfs: add "device" tag to /proc/self/mountstats

nfsiostat was failing to find mounted filesystems on kernels after
2.6.38 because of changes to show_vfsstat() by commit
c7f404b40a3665d9f4e9a927cc5c1ee0479ed8f9.  This patch adds back the
"device" tag before the nfs server entry so scripts can parse the
mountstats file correctly.

Signed-off-by: Bryan Schumaker <bjschuma@netapp.com>
CC: stable@kernel.org [>=2.6.39]
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agocleanup: vfs: small comment fix for block_invalidatepage
Wang Sheng-Hui [Thu, 1 Sep 2011 00:22:57 +0000 (08:22 +0800)]
cleanup: vfs: small comment fix for block_invalidatepage

The patch is aganist 3.1-rc3.

Signed-off-by: Wang Sheng-Hui <shhuiw@gmail.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agovfs: iov_iter: have iov_iter_advance decrement nr_segs appropriately
Jeff Layton [Thu, 27 Oct 2011 21:53:08 +0000 (23:53 +0200)]
vfs: iov_iter: have iov_iter_advance decrement nr_segs appropriately

Currently, when you call iov_iter_advance, then the pointer to the iovec
array can be incremented, but it does not decrement the nr_segs value in
the iov_iter struct. The result is a iov_iter struct with a nr_segs
value that goes beyond the end of the array.

While I'm not aware of anything that's specifically broken by this, it
seems odd and a bit dangerous not to decrement that value. If someone
were to trust the nr_segs value to be correct, then they could end up
walking off the end of the array.

Changing this might also provide some micro-optimization when dealing
with the last iovec in an array. Many of the other routines that deal
with iov_iter have optimized codepaths when nr_segs == 1.

Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
13 years agoFix build break when freezer not configured
Steve French [Tue, 25 Oct 2011 15:02:53 +0000 (10:02 -0500)]
Fix build break when freezer not configured

fs/cifs/transport.c: In function 'wait_for_response':
fs/cifs/transport.c:328: error: implicit declaration of function 'wait_event_freezekillable'

Caused by commit f06ac72e9291 ("cifs, freezer: add
wait_event_freezekillable and have cifs use it").  In this config,
CONFIG_FREEZER is not set.

Reviewed-by: Shirish Pargaonkar <shirishp@us.ibm.com>
CC: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
13 years agoAdd definition for share encryption
Steve French [Tue, 25 Oct 2011 01:46:50 +0000 (20:46 -0500)]
Add definition for share encryption

Samba supports a setfs info level to negotiate encrypted
shares.  This patch adds the defines so we recognize
this info level.  Later patches will add the enablement
for it.

Acked-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Steve French <smfrench@gmail.com>
13 years agoPCI: Clean-up MPS debug output
Jon Mason [Fri, 14 Oct 2011 19:56:16 +0000 (14:56 -0500)]
PCI: Clean-up MPS debug output

Clean-up MPS debug output to make it a single line and aligned, thus
making it more readable for a large number of buses and devices in a
single system.

Suggested by Benjamin Herrenschmidt <benh@kernel.crashing.org>

Signed-off-by: Jon Mason <mason@myri.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
13 years agopci: Clamp pcie_set_readrq() when using "performance" settings
Benjamin Herrenschmidt [Fri, 14 Oct 2011 19:56:15 +0000 (14:56 -0500)]
pci: Clamp pcie_set_readrq() when using "performance" settings

When configuring the PCIe settings for "performance", we allow parents
to have a larger Max Payload Size than children and rely on children
Max Read Request Size to not be larger than their own MPS to avoid
having the host bridge generate responses they can't cope with.

However, various drivers in Linux call pci_set_readrq() with arbitrary
values, assuming this to be a simple performance tweak. This breaks
under our "performance" configuration.

Fix that by making sure the value programmed by pcie_set_readrq() is
never larger than the configured MPS for that device.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jon Mason <mason@myri.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
13 years agoPCI: enable MPS "performance" setting to properly handle bridge MPS
Jon Mason [Fri, 14 Oct 2011 19:56:14 +0000 (14:56 -0500)]
PCI: enable MPS "performance" setting to properly handle bridge MPS

Rework the "performance" MPS option to configure the device MPS with the
smaller of the device MPSS or the bridge MPS (which is assumed to be
properly configured at this point to the largest allowable MPS based on
its parent bus).

Also, rework the MRRS setting to report an inability to set the MRRS to
a valid setting.

Signed-off-by: Jon Mason <mason@myri.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
13 years agoPCI: Workaround for Intel MPS errata
Jon Mason [Fri, 14 Oct 2011 19:56:13 +0000 (14:56 -0500)]
PCI: Workaround for Intel MPS errata

Intel 5000 and 5100 series memory controllers have a known issue if read
completion coalescing is enabled and the PCI-E Maximum Payload Size is
set to 256B.  To work around this issue, disable read completion
coalescing in the memory controller and root complexes.  Unfortunately,
it must always be disabled, even if no 256B MPS devices are present, due
to the possibility of one being hotplugged.

Links to erratas:
http://www.intel.com/content/dam/doc/specification-update/5000-chipset-memory-controller-hub-specification-update.pdf
http://www.intel.com/content/dam/doc/specification-update/5100-memory-controller-hub-chipset-specification-update.pdf

Thanks to Jesse Brandeburg and Ben Hutchings for providing insight into
the problem.

Tested-and-Reported-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Jon Mason <mason@myri.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
13 years agoRevert "drm/ttm: add a way to bo_wait for either the last read or last write"
Dave Airlie [Thu, 27 Oct 2011 16:28:37 +0000 (18:28 +0200)]
Revert "drm/ttm: add a way to bo_wait for either the last read or last write"

This reverts commit dfadbbdb57b3f2bb33e14f129a43047c6f0caefa.

Further upstream discussion between Marek and Thomas decided this wasn't
fully baked and needed further work, so revert it before it hits mainline.

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agoRevert "drm/radeon/kms: add a new gem_wait ioctl with read/write flags"
Dave Airlie [Thu, 27 Oct 2011 16:15:10 +0000 (18:15 +0200)]
Revert "drm/radeon/kms: add a new gem_wait ioctl with read/write flags"

This reverts commit d3ed74027f1dd197b7e08247a40d3bf9be1852b0.

Further upstream discussion between Thomas and Marek decided this needed
more work and driver specifics. So revert before it goes upstream.

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agoMerge branch 'topic/hda' into for-linus
Takashi Iwai [Thu, 27 Oct 2011 15:06:43 +0000 (17:06 +0200)]
Merge branch 'topic/hda' into for-linus

13 years agoALSA: hda - Fix ADC input-amp handling for Cx20549 codec
Takashi Iwai [Fri, 14 Oct 2011 13:26:20 +0000 (15:26 +0200)]
ALSA: hda - Fix ADC input-amp handling for Cx20549 codec

It seems that Conexant CX20549 chip handle only a single input-amp even
though the audio-input widget has multiple sources.  This has been never
clear, and I implemented in the current way based on the debug information
I got at the early time -- the device reacts individual input-amp values
for different sources.  This is true for another Conexant codec, but it's
not applied to CX20549 actually.

This patch changes the auto-parser code to handle a single input-amp
per audio-in widget for CX20549.  After applying this, you'll see only a
single "Capture" volume control instead of separate "Mic" or "Line"
captures when the device is set up to use a single ADC.

We haven't tested 20551 and 20561 codecs yet.  If these show the similar
behavior like 20549, they need to set spec->single_adc_amp=1, too.

Cc: <stable@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
13 years agoALSA: hda - Keep EAPD turned on for old Conexant chips
Takashi Iwai [Fri, 14 Oct 2011 13:22:34 +0000 (15:22 +0200)]
ALSA: hda - Keep EAPD turned on for old Conexant chips

In the old Conexant chips (5045, 5047, 5051 and 5066), a single EAPD
may handle both headphone and speaker outputs while it's assigned only
to one of them.  Turning off dynamically leads to the unexpected silent
output in such a configuration with the auto-mute function.

Since it's difficult to know how the EAPD is handled in the actual h/w
implementation, better to keep EAPD on while running for such codecs.

Cc: <stable@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
13 years agoALSA: hda/realtek - Fix missing volume controls with ALC260
Takashi Iwai [Thu, 27 Oct 2011 14:33:27 +0000 (16:33 +0200)]
ALSA: hda/realtek - Fix missing volume controls with ALC260

ALC260 has multiple mixer widgets connected to the shared DAC, but the
driver currently doesn't check this possibility and ignores when the DAC
is shared with others.  This resulted in the silent output from some
routes because of lack of the amp setup.

This patch adds the workaround for it by checking the route even with the
shared DAC, but also checking the conflict with the existing control for
the very same widget NID.

Reference: https://bugzilla.novell.com/show_bug.cgi?id=726812

Signed-off-by: Takashi Iwai <tiwai@suse.de>
13 years agommc: fix compile error when CONFIG_BLOCK is not enabled
Namjae Jeon [Wed, 12 Oct 2011 15:43:14 +0000 (00:43 +0900)]
mmc: fix compile error when CONFIG_BLOCK is not enabled

'DISK_NAME_LEN' is undeclared when CONFIG_BLOCK is disabled; its use was
introduced via genhd.h by the general purpose partition patch.

To fix, we just add our own MAX_MMC_PART_NAME_LEN macro instead of using
DISK_NAME_LEN.

Reported-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Namjae Jeon <linkinjeon@gmail.com>
Acked-by: Randy Dunlap <rdunlap@xenotime.net>
Acked-by: Andrei Warkentin <andreiw@vmware.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
13 years agommc: core: Cleanup eMMC4.5 conditionals
Seungwon Jeon [Tue, 25 Oct 2011 00:43:12 +0000 (09:43 +0900)]
mmc: core: Cleanup eMMC4.5 conditionals

Code cleanup, putting all eMMC 4.5 detection cases together.
This patch removes one if-statement and assembles all. And it also
removes variable initialization below else-statement -- all members
of card structure are already set to zero at card-init.

Signed-off-by: Seungwon Jeon <tgih.jun@samsung.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
13 years agommc: omap_hsmmc: if multiblock reads are broken, disable them
Paul Walmsley [Thu, 6 Oct 2011 20:50:35 +0000 (14:50 -0600)]
mmc: omap_hsmmc: if multiblock reads are broken, disable them

When device data indicates that multiple block reads are not supported
on a given HSMMC controller instance, log a message to the console, and
pass the appropriate MMC capability flag to the MMC core.

Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Dave Hylands <dhylands@gmail.com>
Tested-by: Steve Sakoman <sakoman@gmail.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
13 years agommc: core: add workaround for controllers with broken multiblock reads
Paul Walmsley [Thu, 6 Oct 2011 20:50:33 +0000 (14:50 -0600)]
mmc: core: add workaround for controllers with broken multiblock reads

Due to hardware bugs, some MMC host controllers don't support
multiple-block reads[1].  To resolve, add a new MMC capability flag,
MMC_CAP2_NO_MULTI_READ, which can be set by affected host controller
drivers.  When this capability is set, all reads will be issued one
sector at a time.

1. See for example Advisory 2.1.1.128 "MMC: Multiple Block Read
Operation Issue" in _OMAP3530/3525/3515/3503 Silicon Errata_
Revision F (October 2010) (SPRZ278F), available from
http://focus.ti.com/lit/er/sprz278f/sprz278f.pdf

Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Dave Hylands <dhylands@gmail.com>
Tested-by: Steve Sakoman <sakoman@gmail.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
13 years agoMerge branch 'topic/asoc' into for-linus
Takashi Iwai [Thu, 27 Oct 2011 08:43:40 +0000 (10:43 +0200)]
Merge branch 'topic/asoc' into for-linus

13 years agoASoC: wm8940: Properly set codec->dapm.bias_level
Axel Lin [Wed, 26 Oct 2011 01:53:41 +0000 (09:53 +0800)]
ASoC: wm8940: Properly set codec->dapm.bias_level

Reported-by: Chris Paulson-Ellis <chris@edesix.com>
Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Cc: stable@vger.kernel.org