9 years agoirq_work: fix compile failure on tile from missing include
Chris Metcalf [Wed, 11 Apr 2012 16:21:39 +0000]
irq_work: fix compile failure on tile from missing include

Building with IRQ_WORK configured results in

kernel/irq_work.c: In function ‘irq_work_run’:
kernel/irq_work.c:110: error: implicit declaration of function ‘irqs_disabled’

The appropriate header just needs to be included.

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

9 years agoia64: populate the cmpxchg header with appropriate code
Paul Gortmaker [Tue, 3 Apr 2012 17:51:35 +0000]
ia64: populate the cmpxchg header with appropriate code

commit 93f378883cecb9dcb2cf5b51d9d24175906659da

    "Fix ia64 build errors (fallout from system.h disintegration)"

introduced arch/ia64/include/asm/cmpxchg.h as a temporary
build fix and stated:

    "... leave the migration of xchg() and cmpxchg() to this new
     header file for a future patch."

Migrate the appropriate chunks from asm/intrinsics.h and fix
the whitespace issues in the migrated chunk.

Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: David Howells <dhowells@redhat.com>
Acked-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

9 years agoalpha: fix build failures from system.h dismemberment
Paul Gortmaker [Mon, 2 Apr 2012 20:04:13 +0000]
alpha: fix build failures from system.h dismemberment

commit ec2212088c42ff7d1362629ec26dda4f3e8bdad3

    "Disintegrate asm/system.h for Alpha"

combined with commit b4816afa3986704d1404fc48e931da5135820472

    "Move the asm-generic/system.h xchg() implementation to asm-generic/cmpxchg.h"

introduced the concept of asm/cmpxchg.h but the alpha arch
never got one.  Fork the cmpxchg content out of the asm/atomic.h
file to create one.

Some minor whitespace fixups were done on the block of code that
created the new file.

Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Matt Turner <mattst88@gmail.com>
Cc: David Howells <dhowells@redhat.com>
Acked-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

9 years agovideo:uvesafb: Fix oops that uvesafb try to execute NX-protected page
Wang YanQing [Sun, 1 Apr 2012 00:54:02 +0000]
video:uvesafb: Fix oops that uvesafb try to execute NX-protected page

This patch fix the oops below that catched in my machine

[   81.560602] uvesafb: NVIDIA Corporation, GT216 Board - 0696a290, Chip Rev   , OEM: NVIDIA, VBE v3.0
[   81.609384] uvesafb: protected mode interface info at c000:d350
[   81.609388] uvesafb: pmi: set display start = c00cd3b3, set palette = c00cd40e
[   81.609390] uvesafb: pmi: ports = 3b4 3b5 3ba 3c0 3c1 3c4 3c5 3c6 3c7 3c8 3c9 3cc 3ce 3cf 3d0 3d1 3d2 3d3 3d4 3d5 3da
[   81.614558] uvesafb: VBIOS/hardware doesn't support DDC transfers
[   81.614562] uvesafb: no monitor limits have been set, default refresh rate will be used
[   81.614994] uvesafb: scrolling: ypan using protected mode interface, yres_virtual=4915
[   81.744147] kernel tried to execute NX-protected page - exploit attempt? (uid: 0)
[   81.744153] BUG: unable to handle kernel paging request at c00cd3b3
[   81.744159] IP: [<c00cd3b3>] 0xc00cd3b2
[   81.744167] *pdpt = 00000000016d6001 *pde = 0000000001c7b067 *pte = 80000000000cd163
[   81.744171] Oops: 0011 [#1] SMP
[   81.744174] Modules linked in: uvesafb(+) cfbcopyarea cfbimgblt cfbfillrect
[   81.744178]
[   81.744181] Pid: 3497, comm: modprobe Not tainted 3.3.0-rc4NX+ #71 Acer            Aspire 4741                    /Aspire 4741
[   81.744185] EIP: 0060:[<c00cd3b3>] EFLAGS: 00010246 CPU: 0
[   81.744187] EIP is at 0xc00cd3b3
[   81.744189] EAX: 00004f07 EBX: 00000000 ECX: 00000000 EDX: 00000000
[   81.744191] ESI: f763f000 EDI: f763f6e8 EBP: f57f3a0c ESP: f57f3a00
[   81.744192]  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
[   81.744195] Process modprobe (pid: 3497, ti=f57f2000 task=f748c600 task.ti=f57f2000)
[   81.744196] Stack:
[   81.744197]  f82512c5 f759341c 00000000 f57f3a30 c124a9bc 00000001 00000001 000001e0
[   81.744202]  f8251280 f763f000 f7593400 00000000 f57f3a40 c12598dd f5c0c000 00000000
[   81.744206]  f57f3b10 c1255efe c125a21a 00000006 f763f09c 00000000 c1c6cb60 f7593400
[   81.744210] Call Trace:
[   81.744215]  [<f82512c5>] ? uvesafb_pan_display+0x45/0x60 [uvesafb]
[   81.744222]  [<c124a9bc>] fb_pan_display+0x10c/0x160
[   81.744226]  [<f8251280>] ? uvesafb_vbe_find_mode+0x180/0x180 [uvesafb]
[   81.744230]  [<c12598dd>] bit_update_start+0x1d/0x50
[   81.744232]  [<c1255efe>] fbcon_switch+0x39e/0x550
[   81.744235]  [<c125a21a>] ? bit_cursor+0x4ea/0x560
[   81.744240]  [<c129b6cb>] redraw_screen+0x12b/0x220
[   81.744245]  [<c128843b>] ? tty_do_resize+0x3b/0xc0
[   81.744247]  [<c129ef42>] vc_do_resize+0x3d2/0x3e0
[   81.744250]  [<c129efb4>] vc_resize+0x14/0x20
[   81.744253]  [<c12586bd>] fbcon_init+0x29d/0x500
[   81.744255]  [<c12984c4>] ? set_inverse_trans_unicode+0xe4/0x110
[   81.744258]  [<c129b378>] visual_init+0xb8/0x150
[   81.744261]  [<c129c16c>] bind_con_driver+0x16c/0x360
[   81.744264]  [<c129b47e>] ? register_con_driver+0x6e/0x190
[   81.744267]  [<c129c3a1>] take_over_console+0x41/0x50
[   81.744269]  [<c1257b7a>] fbcon_takeover+0x6a/0xd0
[   81.744272]  [<c12594b8>] fbcon_event_notify+0x758/0x790
[   81.744277]  [<c10929e2>] notifier_call_chain+0x42/0xb0
[   81.744280]  [<c1092d30>] __blocking_notifier_call_chain+0x60/0x90
[   81.744283]  [<c1092d7a>] blocking_notifier_call_chain+0x1a/0x20
[   81.744285]  [<c124a5a1>] fb_notifier_call_chain+0x11/0x20
[   81.744288]  [<c124b759>] register_framebuffer+0x1d9/0x2b0
[   81.744293]  [<c1061c73>] ? ioremap_wc+0x33/0x40
[   81.744298]  [<f82537c6>] uvesafb_probe+0xaba/0xc40 [uvesafb]
[   81.744302]  [<c12bb81f>] platform_drv_probe+0xf/0x20
[   81.744306]  [<c12ba558>] driver_probe_device+0x68/0x170
[   81.744309]  [<c12ba731>] __device_attach+0x41/0x50
[   81.744313]  [<c12b9088>] bus_for_each_drv+0x48/0x70
[   81.744316]  [<c12ba7f3>] device_attach+0x83/0xa0
[   81.744319]  [<c12ba6f0>] ? __driver_attach+0x90/0x90
[   81.744321]  [<c12b991f>] bus_probe_device+0x6f/0x90
[   81.744324]  [<c12b8a45>] device_add+0x5e5/0x680
[   81.744329]  [<c122a1a3>] ? kvasprintf+0x43/0x60
[   81.744332]  [<c121e6e4>] ? kobject_set_name_vargs+0x64/0x70
[   81.744335]  [<c121e6e4>] ? kobject_set_name_vargs+0x64/0x70
[   81.744339]  [<c12bbe9f>] platform_device_add+0xff/0x1b0
[   81.744343]  [<f8252906>] uvesafb_init+0x50/0x9b [uvesafb]
[   81.744346]  [<c100111f>] do_one_initcall+0x2f/0x170
[   81.744350]  [<f82528b6>] ? uvesafb_is_valid_mode+0x66/0x66 [uvesafb]
[   81.744355]  [<c10c6994>] sys_init_module+0xf4/0x1410
[   81.744359]  [<c1157fc0>] ? vfsmount_lock_local_unlock_cpu+0x30/0x30
[   81.744363]  [<c144cb10>] sysenter_do_call+0x12/0x36
[   81.744365] Code: f5 00 00 00 32 f6 66 8b da 66 d1 e3 66 ba d4 03 8a e3 b0 1c 66 ef b0 1e 66 ef 8a e7 b0 1d 66 ef b0 1f 66 ef e8 fa 00 00 00 61 c3 <60> e8 c8 00 00 00 66 8b f3 66 8b da 66 ba d4 03 b0 0c 8a e5 66
[   81.744388] EIP: [<c00cd3b3>] 0xc00cd3b3 SS:ESP 0068:f57f3a00
[   81.744391] CR2: 00000000c00cd3b3
[   81.744393] ---[ end trace 18b2c87c925b54d6 ]---

Signed-off-by: Wang YanQing <udknight@gmail.com>
Cc: Michal Januszewski <spock@gentoo.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: stable@vger.kernel.org

9 years agofbdev: fix au1*fb builds
Manuel Lauss [Sat, 24 Mar 2012 10:38:02 +0000]
fbdev: fix au1*fb builds

Commit 1c16697bf9d5b206cb0d2b905a54de5e077296be
("drivers/video/au*fb.c: use devm_ functions) introduced 2 build failures
in the au1100fb and au1200fb drivers, fix them.

Signed-off-by: Manuel Lauss <manuel.lauss@googlemail.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>

9 years agokyrofb: fix on x86_64
Ondrej Zary [Wed, 21 Mar 2012 22:36:50 +0000]
kyrofb: fix on x86_64

kyrofb is completely broken on x86_64 because the registers are defined as
unsigned long. Change them to u32 to make the driver work.
Tested with Hercules 3D Prophet 4000XT.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Acked-by: Paul Mundt <lethal@linux-sh.org>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>

9 years agoBtrfs: use commit root when loading free space cache
Josef Bacik [Thu, 12 Apr 2012 20:03:57 +0000]
Btrfs: use commit root when loading free space cache

A user reported that booting his box up with btrfs root on 3.4 was way
slower than on 3.3 because I removed the ideal caching code.  It turns out
that we don't load the free space cache if we're in a commit for deadlock
reasons, but since we're reading the cache and it hasn't changed yet we are
safe reading the inode and free space item from the commit root, so do that
and remove all of the deadlock checks so we don't unnecessarily skip loading
the free space cache.  The user reported this fixed the slowness.  Thanks,

Tested-by: Calvin Walton <calvin.walton@kepstin.ca>
Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

9 years agoBtrfs: fix use-after-free in __btrfs_end_transaction
Dave Jones [Thu, 12 Apr 2012 20:03:56 +0000]
Btrfs: fix use-after-free in __btrfs_end_transaction

49b25e0540904be0bf558b84475c69d72e4de66e introduced a use-after-free bug
that caused spurious -EIO's to be returned.

Do the check before we free the transaction.

Cc: David Sterba <dsterba@suse.cz>
Cc: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

9 years agoBtrfs: check return value of bio_alloc() properly
Tsutomu Itoh [Thu, 12 Apr 2012 20:03:56 +0000]
Btrfs: check return value of bio_alloc() properly

bio_alloc() has the possibility of returning NULL.
So, it is necessary to check the return value.

Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

9 years agoBtrfs: remove lock assert from get_restripe_target()
Ilya Dryomov [Thu, 12 Apr 2012 20:03:56 +0000]
Btrfs: remove lock assert from get_restripe_target()

This fixes a regression introduced by fc67c450.  spin_is_locked() always
returns 0 on UP kernels, which caused assert in get_restripe_target() to
be fired on every call from btrfs_reduce_alloc_profile() on UP systems.
Remove it completely for now, it's not clear if it's going to be needed
in future.

Reported-by: Bobby Powers <bobbypowers@gmail.com>
Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org>
Tested-by: Mitch Harder <mitch.harder@sabayonlinux.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

9 years agoBtrfs: fix eof while discarding extents
Liu Bo [Thu, 12 Apr 2012 20:03:56 +0000]
Btrfs: fix eof while discarding extents

We miscalculate the length of extents we're discarding, and it leads to
an eof of device.

Reported-by: Daniel Blueman <daniel@quora.org>
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

9 years agoBtrfs: fix uninit variable in repair_eb_io_failure
Chris Mason [Thu, 12 Apr 2012 19:55:15 +0000]
Btrfs: fix uninit variable in repair_eb_io_failure

We'd have to be passing bogus extent buffers for this uninit variable to
actually be used, but set it to zero just in case.

Signed-off-by: Chris Mason <chris.mason@oracle.com>

9 years agoblock: mtip32xx: remove HOTPLUG_PCI_PCIE dependancy
Greg Kroah-Hartman [Thu, 12 Apr 2012 06:47:05 +0000]
block: mtip32xx: remove HOTPLUG_PCI_PCIE dependancy

This removes the HOTPLUG_PCI_PCIE dependency on the driver and makes it
depend on PCI.

Cc: Sam Bradshaw <sbradshaw@micron.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agomtip32xx: dump tagmap on failure
Asai Thambi S P [Mon, 9 Apr 2012 06:35:39 +0000]
mtip32xx: dump tagmap on failure

Dump tagmap on failure, instead of individual tags.

Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agomtip32xx: fix handling of commands in various scenarios
Asai Thambi S P [Mon, 9 Apr 2012 06:35:39 +0000]
mtip32xx: fix handling of commands in various scenarios

* If a ncq  command time out and a non-ncq command is active, skip restart port
* Queue(pause) ncq commands during operations spanning more than one non-ncq commands - secure erase, download microcode
* When a non-ncq command is active, allow incoming non-ncq commands to wait instead of failing back
* Changed timeout for download microcode and smart commands
* If the device in write protect mode, fail all writes (do not send to device)
* Set maximum retries to 2

Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agomtip32xx: Shorten macro names
Asai Thambi S P [Mon, 9 Apr 2012 06:35:38 +0000]
mtip32xx: Shorten macro names

Shortened macros used to represent mtip_port->flags and dd->dd_flag

Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agomtip32xx: misc changes
Asai Thambi S P [Mon, 9 Apr 2012 06:35:38 +0000]
mtip32xx: misc changes

* Handle the interrupt completion of polled internal commands
* Do not check remove pending flag for standby command
* On rebuild failure,
    - set corresponding bit dd_flag
    - do not send standby command
* Free ida index in remove path

Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agomtip32xx: Add new sysfs entry 'status'
Asai Thambi S P [Mon, 9 Apr 2012 06:35:38 +0000]
mtip32xx: Add new sysfs entry 'status'

* Add support for detecting the following device status
        - write protect
        - over temp (thermal shutdown)
* Add new sysfs entry 'status', possible values - online, write_protect, thermal_shutdown
* Add new file 'sysfs-block-rssd' to document ABI (Reported-by: Greg Kroah-Hartman)

Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agomtip32xx: make setting comp_time as common
Asai Thambi S P [Mon, 9 Apr 2012 06:35:38 +0000]
mtip32xx: make setting comp_time as common

Moved setting completion time into mtip_issue_ncq_command()

Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agomtip32xx: Add new bitwise flag 'dd_flag'
Asai Thambi S P [Mon, 9 Apr 2012 06:35:38 +0000]
mtip32xx: Add new bitwise flag 'dd_flag'

* Merged the following flags into one variable 'dd_flag':
        * drv_cleanup_done
        * resumeflag
* Added the following flags into 'dd_flag'
        * remove pending
        * init done
* Removed 'ftlrebuildflag' (similar flag is already part of mti_port->flags)

Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agomtip32xx: fix error handling in mtip_init()
Ryosuke Saito [Thu, 5 Apr 2012 14:09:34 +0000]
mtip32xx: fix error handling in mtip_init()

Ensure that block device is properly unregistered, if
pci_register_driver() fails.

Signed-off-by: Ryosuke Saito <raitosyo@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agovirtio-blk: Call revalidate_disk() upon online disk resize
Vivek Goyal [Thu, 29 Mar 2012 08:09:44 +0000]
virtio-blk: Call revalidate_disk() upon online disk resize

If a virtio disk is open in guest and a disk resize operation is done,
(virsh blockresize), new size is not visible to tools like "fdisk -l".
This seems to be happening as we update only part->nr_sects and not
bdev->bd_inode size.

Call revalidate_disk() which should take care of it. I tested growing disk
size of already open disk and it works for me.

Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agoxen/blkback: Make optional features be really optional.
Konrad Rzeszutek Wilk [Wed, 14 Mar 2012 17:04:00 +0000]
xen/blkback: Make optional features be really optional.

They were using the xenbus_dev_fatal() function which would
change the state of the connection immediately. Which is not
what we want when we advertise optional features.

So make 'feature-discard','feature-barrier','feature-flush-cache'
optional.

Suggested-by: Jan Beulich <JBeulich@suse.com>
[v1: Made the discard function void and static]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

9 years agoxen/blkback: Squash the discard support for 'file' and 'phy' type.
Konrad Rzeszutek Wilk [Tue, 13 Mar 2012 22:43:23 +0000]
xen/blkback: Squash the discard support for 'file' and 'phy' type.

The only reason for the distinction was for the special case of
'file' (which is assumed to be loopback device), was to reach inside
the loopback device, find the underlaying file, and call fallocate on it.
Fortunately "xen-blkback: convert hole punching to discard request on
loop devices" removes that use-case and we now based the discard
support based on blk_queue_discard(q) and extract all appropriate
parameters from the 'struct request_queue'.

CC: Li Dongyang <lidongyang@novell.com>
Acked-by: Jan Beulich <JBeulich@suse.com>
[v1: Dropping pointless initializer and keeping blank line]
[v2: Remove the kfree as it is not used anymore]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

9 years agomtip32xx: fix incorrect value set for drv_cleanup_done, and re-initialize and start...
Asai Thambi S P [Fri, 23 Mar 2012 11:33:03 +0000]
mtip32xx: fix incorrect value set for drv_cleanup_done, and re-initialize and start port in mtip_restart_port()

This patch includes two changes:
* fix incorrect value set for drv_cleanup_done
* re-initialize and start port in mtip_restart_port()

Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Sam Bradshaw <sbradshaw@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agocciss: Fix scsi tape io with more than 255 scatter gather elements
Stephen M. Cameron [Thu, 22 Mar 2012 20:40:09 +0000]
cciss: Fix scsi tape io with more than 255 scatter gather elements

The total number of scatter gather elements in the CISS command
used by the scsi tape code was being cast to a u8, which can hold
at most 255 scatter gather elements.  It should have been cast to
a u16.  Without this patch the command gets rejected by the controller
since the total scatter gather count did not add up to the right
value resulting in an i/o error.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Cc: stable@vger.kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agocciss: Initialize scsi host max_sectors for tape drive support
Stephen M. Cameron [Thu, 22 Mar 2012 20:40:08 +0000]
cciss: Initialize scsi host max_sectors for tape drive support

The default is too small (1024 blocks), use h->cciss_max_sectors (8192 blocks)
Without this change, if you try to set the block size of a tape drive above
512*1024, via "mt -f /dev/st0 setblk nnn" where nnn is greater than 524288,
it won't work right.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Cc: stable@vger.kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agoxen-blkfront: make blkif_io_lock spinlock per-device
Steven Noonan [Fri, 17 Feb 2012 20:04:44 +0000]
xen-blkfront: make blkif_io_lock spinlock per-device

This patch moves the global blkif_io_lock to the per-device structure. The
spinlock seems to exists for two reasons: to disable IRQs when in the interrupt
handlers for blkfront, and to protect the blkfront VBDs when a detachment is
requested.

Having a global blkif_io_lock doesn't make sense given the use case, and it
drastically hinders performance due to contention. All VBDs with pending IOs
have to take the lock in order to get work done, which serializes everything
pretty badly.

Signed-off-by: Steven Noonan <snoonan@amazon.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

9 years agoxen/blkfront: don't put bdev right after getting it
Andrew Jones [Thu, 16 Feb 2012 12:16:25 +0000]
xen/blkfront: don't put bdev right after getting it

We should hang onto bdev until we're done with it.

Signed-off-by: Andrew Jones <drjones@redhat.com>
[v1: Fixed up git commit description]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

9 years agoxen-blkfront: use bitmap_set() and bitmap_clear()
Akinobu Mita [Fri, 20 Jan 2012 15:15:26 +0000]
xen-blkfront: use bitmap_set() and bitmap_clear()

Use bitmap_set and bitmap_clear rather than modifying individual bits
in a memory region.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: xen-devel@lists.xensource.com
Cc: virtualization@lists.linux-foundation.org
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

9 years agoxen/blkback: Enable blkback on HVM guests
Daniel De Graaf [Mon, 28 Nov 2011 16:49:05 +0000]
xen/blkback: Enable blkback on HVM guests

Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

9 years agoxen/blkback: use grant-table.c hypercall wrappers
Daniel De Graaf [Mon, 28 Nov 2011 16:49:03 +0000]
xen/blkback: use grant-table.c hypercall wrappers

Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

9 years agoblock: make auto block plug flush threshold per-disk based
Shaohua Li [Fri, 6 Apr 2012 17:37:47 +0000]
block: make auto block plug flush threshold per-disk based

We do auto block plug flush to reduce latency, the threshold is 16
requests. This works well if the task is accessing one or two drives.
The problem is if the task is accessing a raid 0 device and the raid
disk number is big, say 8 or 16, 16/8 = 2 or 16/16=1, we will have
heavy lock contention.

This patch makes the threshold per-disk based. The latency should be
still ok accessing one or two drives. The setup with application
accessing a lot of drives in the meantime uaually is big machine,
avoiding lock contention is more important, because any contention
will actually increase latency.

Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agoDocumentation: Add sysfs ABI change for cfq's target latency.
Tao Ma [Sun, 1 Apr 2012 21:33:40 +0000]
Documentation: Add sysfs ABI change for cfq's target latency.

Cc: Jens Axboe <axboe@kernel.dk>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tao Ma <boyu.mt@taobao.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agoblock: Make cfq_target_latency tunable through sysfs.
Tao Ma [Sun, 1 Apr 2012 21:33:39 +0000]
block: Make cfq_target_latency tunable through sysfs.

In cfq, when we calculate a time slice for a process(or a cfqq to
be precise), we have to consider the cfq_target_latency so that all the
sync request have an estimated latency(300ms) and it is controlled by
cfq_target_latency. But in some hadoop test, we have found that if
there are many processes doing sequential read(24 for example), the
throughput is bad because every process can only work for about 25ms
and the cfqq is switched. That leads to a higher disk seek. We can
achive the good throughput by setting low_latency=0, but then some
read's latency is too much for the application.

So this patch makes cfq_target_latency tunable through sysfs so that
we can tune it and find some magic number which is not bad for both
the throughput and the read latency.

Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Tao Ma <boyu.mt@taobao.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agoblock: use lockdep_assert_held for queue locking
Andi Kleen [Fri, 30 Mar 2012 10:33:28 +0000]
block: use lockdep_assert_held for queue locking

Instead of an ugly open coded variant.

Cc: axboe@kernel.dk
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agoblock: blk_alloc_queue_node(): use caller's GFP flags instead of GFP_KERNEL
Dan Carpenter [Fri, 23 Mar 2012 08:58:54 +0000]
block: blk_alloc_queue_node(): use caller's GFP flags instead of GFP_KERNEL

We should use the GFP flags that the caller specified instead of picking
our own.  All the callers specify GFP_KERNEL so this doesn't make a
difference to how the kernel runs, it's just a cleanup.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 years agocpufreq: OMAP: fix build errors: depends on ARCH_OMAP2PLUS
Kevin Hilman [Fri, 13 Apr 2012 20:32:30 +0000]
cpufreq: OMAP: fix build errors: depends on ARCH_OMAP2PLUS

The OMAP driver needs a 'depends on ARCH_OMAP2PLUS' since it only
builds for OMAP2+ platforms.

This 'depends on' was in the original patch from Russell King, but was
erroneously removed by me when making this option user-selectable in
commit b09db45c (cpufreq: OMAP driver depends CPUfreq tables.)  This
patch remedies that.

Apologies to Russell King for breaking his originally working patch.

Also, thanks to Grazvydas Ignotas for reporting the same problem.

Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: Grazvydas Ignotas <notasas@gmail.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agosparc64: Eliminate obsolete __handle_softirq() function
Paul E. McKenney [Fri, 13 Apr 2012 03:35:13 +0000]
sparc64: Eliminate obsolete __handle_softirq() function

The invocation of softirq is now handled by irq_exit(), so there is no
need for sparc64 to invoke it on the trap-return path.  In fact, doing so
is a bug because if the trap occurred in the idle loop, this invocation
can result in lockdep-RCU failures.  The problem is that RCU ignores idle
CPUs, and the sparc64 trap-return path to the softirq handlers fails to
tell RCU that the CPU must be considered non-idle while those handlers
are executing.  This means that RCU is ignoring any RCU read-side critical
sections in those handlers, which in turn means that RCU-protected data
can be yanked out from under those read-side critical sections.

The shiny new lockdep-RCU ability to detect RCU read-side critical sections
that RCU is ignoring located this problem.

The fix is straightforward: Make sparc64 stop manually invoking the
softirq handlers.

Reported-by: Meelis Roos <mroos@linux.ee>
Suggested-by: David Miller <davem@davemloft.net>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Meelis Roos <mroos@linux.ee>
Cc: stable@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>

9 years agosparc64: Fix bootup crash on sun4v.
David S. Miller [Fri, 13 Apr 2012 18:56:22 +0000]
sparc64: Fix bootup crash on sun4v.

The DS driver registers as a subsys_initcall() but this can be too
early, in particular this risks registering before we've had a chance
to allocate and setup module_kset in kernel/params.c which is
performed also as a subsyts_initcall().

Register DS using device_initcall() insteal.

Signed-off-by: David S. Miller <davem@davemloft.net>
Cc: stable@vger.kernel.org

9 years agohwmon: (pmbus_core) Fix compiler warning
Guenter Roeck [Wed, 28 Mar 2012 16:14:03 +0000]
hwmon: (pmbus_core) Fix compiler warning

Some configurations produce the following compiler warning:

drivers/hwmon/pmbus/pmbus_core.c: In function 'pmbus_show_boolean':
drivers/hwmon/pmbus/pmbus_core.c:752: warning: 'val' may be used uninitialized in this function

While this is a false positive, it can easily be fixed by overloading the return
value from pmbus_get_boolean with both val and error return code (val is a
boolean and thus never negative).

Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Robert Coulson <robert.coulson@ericsson.com>

9 years agohwmon: (smsc47m1) Fix compiler warning
Guenter Roeck [Wed, 28 Mar 2012 15:55:12 +0000]
hwmon: (smsc47m1) Fix compiler warning

Some configurations produce the following compiler warning:

drivers/hwmon/smsc47m1.c: In function 'sm_smsc47m1_init':
drivers/hwmon/smsc47m1.c:938: warning: 'address' may be used uninitialized in this function

While this is a false positive, it can easily be fixed by overloading the return
value from smsc47m1_find with both address and error return code (the address
is an unsigned short and thus never negative). This also reduces module size by
a few bytes (46 bytes for x86_64).

Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Robert Coulson <robert.coulson@ericsson.com>

9 years agohwmon: (acpi_power_meter) Fix compiler warning seen in some configurations
Guenter Roeck [Wed, 28 Mar 2012 16:03:26 +0000]
hwmon: (acpi_power_meter) Fix compiler warning seen in some configurations

In some configurations, BUG() does not result in an endless loop but returns
to the caller. This results in the following compiler warning:

drivers/hwmon/acpi_power_meter.c: In function 'show_str':
drivers/hwmon/acpi_power_meter.c:380: warning: 'val' may be used uninitialized in this function

Fix the warning by setting val to an empty string after BUG().

Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Robert Coulson <robert.coulson@ericsson.com>

9 years agohwmon: (smsc47b397) Fix compiler warning
Guenter Roeck [Wed, 28 Mar 2012 15:35:26 +0000]
hwmon: (smsc47b397) Fix compiler warning

Some configurations produce the following compiler warning:

drivers/hwmon/smsc47b397.c: In function 'smsc47b397_init':
drivers/hwmon/smsc47b397.c:385: warning: 'address' may be used uninitialized in this function

While this is a false positive, it can easily be fixed by overloading the return
value from smsc47b397_find with both address and error return code (the address
is an unsigned short and thus never negative). This also reduces module size by
a few bytes (64 bytes for x86_64).

Cc: Mark M. Hoffman <mhoffman@lightlink.com>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Robert Coulson <robert.coulson@ericsson.com>
Acked-by: Jean Delvare <khali@linux-fr.org>

9 years agogpio/exynos: Fix compiler warning in gpio-samsung.c file
Sachin Kamat [Tue, 3 Apr 2012 08:42:55 +0000]
gpio/exynos: Fix compiler warning in gpio-samsung.c file

Fixes the following warning when "SAMSUNG EXYNOS5" is not selected:

warning: ‘exynos5_gpios_1’ defined but not used [-Wunused-variable]
warning: ‘exynos5_gpios_2’ defined but not used [-Wunused-variable]
warning: ‘exynos5_gpios_3’ defined but not used [-Wunused-variable]
warning: ‘exynos5_gpios_4’ defined but not used [-Wunused-variable]

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

9 years agogpio: Fix range check in of_gpio_simple_xlate()
Roland Stigge [Wed, 4 Apr 2012 00:02:58 +0000]
gpio: Fix range check in of_gpio_simple_xlate()

of_gpio_simple_xlate() has an off-by-one bug where it checks to see if
args[0] is > ngpio instead of >=.  args[0] must always be less than
ngpio because it is a zero-based enumeration.

Signed-off-by: Roland Stigge <stigge@antcom.de>
[grant.likely: beef up commit text]
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

9 years agogpio: Fix uninitialized variable bit in adp5588_irq_handler
Axel Lin [Fri, 6 Apr 2012 12:37:43 +0000]
gpio: Fix uninitialized variable bit in adp5588_irq_handler

The variable 'bit' is uninitialized in the first iteration of for
loop. Fix it.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

9 years agogpio/sodaville: Convert sodaville driver to new irqdomain API
Grant Likely [Wed, 28 Mar 2012 20:55:04 +0000]
gpio/sodaville: Convert sodaville driver to new irqdomain API

The irqdomain api changed significantly in v3.4 which caused a build
failure for this driver.

Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Acked-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Hans J. Koch <hjk@linutronix.de>
Cc: Torben Hohn <torbenh@linutronix.de>

9 years agospi/imx: prevent NULL pointer dereference in spi_imx_probe()
Lothar Waßmann [Tue, 3 Apr 2012 13:03:44 +0000]
spi/imx: prevent NULL pointer dereference in spi_imx_probe()

When no platform_data is present and either 'spi-num-chipselects' is
not defined in the DT or 'cs-gpios' has less entries than
'spi-num-chipselects' specifies, the NULL platform_data pointer is
being dereferenced.

Signed-off-by: Lothar Waßmann <LW@KARO-electronics.de>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

9 years agospi/imx: mark base member in spi_imx_data as __iomem
Uwe Kleine-König [Thu, 29 Mar 2012 19:54:18 +0000]
spi/imx: mark base member in spi_imx_data as __iomem

This fixes 36 sparse warnings like:

drivers/spi/spi-imx.c:143:1: warning: incorrect type in argument 1 (different address spaces)
drivers/spi/spi-imx.c:143:1:    expected void const volatile [noderef] <asn:2>*<noident>
drivers/spi/spi-imx.c:143:1:    got void *

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

9 years agospi/mpc83xx: fix NULL pdata dereference bug
Kenth Eriksson [Fri, 30 Mar 2012 15:05:30 +0000]
spi/mpc83xx: fix NULL pdata dereference bug

Commit 178db7d3, "spi: Fix device unregistration when unregistering
the bus master", changed device initialization to be children of the
bus master, not children of the bus masters parent device. The pdata
pointer used in fsl_spi_chipselect must updated to reflect the changed
initialization.

Signed-off-by: Kenth Eriksson <kenth.eriksson@transmode.com>
Acked-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

9 years agospi/davinci: Fix DMA API usage in davinci
Shubhrajyoti D [Tue, 20 Mar 2012 10:40:09 +0000]
spi/davinci: Fix DMA API usage in davinci

The driver uses NULL for dma_unmap_single instead of
the struct device that the API expects.

Signed-off-by: Shubhrajyoti D <shubhrajyoti@ti.com>
Tested-by: Akshay Shankarmurthy <akshay.s@ti.com>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

9 years agospi/pL022: include types.h to remove compilation warnings
Viresh Kumar [Fri, 23 Mar 2012 07:35:48 +0000]
spi/pL022: include types.h to remove compilation warnings

linux/pl022.h uses definitions like, u8, u16, etc, which have dependency of
types.h file, which isn't included in it. So, we get compilation warnings.

This patch includes types.h there to fix these warnings.

Signed-off-by: Viresh Kumar <viresh.kumar@st.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

9 years agoIB/srpt: Set srq_type to IB_SRQT_BASIC
Roland Dreier [Thu, 12 Apr 2012 14:51:08 +0000]
IB/srpt: Set srq_type to IB_SRQT_BASIC

Since commit 96104eda0169 ("RDMA/core: Add SRQ type field"), kernel
users of SRQs need to specify srq_type = IB_SRQT_BASIC in struct
ib_srq_init_attr, or else most low-level drivers will fail in
when srpt_add_one() calls ib_create_srq() and gets -ENOSYS.

(mlx4_ib works OK nearly all of the time, because it just needs
srq_type != IB_SRQT_XRC.  And apparently nearly everyone using
ib_srpt is using mlx4 hardware)

Reported-by: Alexey Shvetsov <alexxy@gentoo.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Roland Dreier <roland@purestorage.com>

9 years agokconfig: delete last traces of __enabled_ from autoconf.h
Paul Gortmaker [Thu, 12 Apr 2012 23:46:34 +0000]
kconfig: delete last traces of __enabled_ from autoconf.h

We've now fixed IS_ENABLED() and friends to not require any special
"__enabled_" prefixed versions of the normal Kconfig options, so delete
the last traces of them being generated.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoRevert "kconfig: fix __enabled_ macros definition for invisible and un-selected symbols"
Paul Gortmaker [Thu, 12 Apr 2012 23:46:33 +0000]
Revert "kconfig: fix __enabled_ macros definition for invisible and un-selected symbols"

This reverts commit 953742c8fe8ac45be453fee959d7be40cd89f920.

Dumping two lines into autoconf.h for all existing Kconfig options
results in a giant file (~16k lines) we have to process each time we
compile something.  We've weaned IS_ENABLED() and similar off of
requiring the __enabled_ definitions so now we can revert the change
which caused all the extra lines.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agokconfig: fix IS_ENABLED to not require all options to be defined
Paul Gortmaker [Thu, 12 Apr 2012 23:46:32 +0000]
kconfig: fix IS_ENABLED to not require all options to be defined

Using IS_ENABLED() within C (vs.  within CPP #if statements) in its
current form requires us to actually define every possible bool/tristate
Kconfig option twice (__enabled_* and __enabled_*_MODULE variants).

This results in a huge autoconf.h file, on the order of 16k lines for a
x86_64 defconfig.

Fixing IS_ENABLED to be able to work on the smaller subset of just
things that we really have defined is step one to fixing this.  Which
means it has to not choke when fed non-enabled options, such as:

  include/linux/netdevice.h:964:1: warning: "__enabled_CONFIG_FCOE_MODULE" is not defined [-Wundef]

The original prototype of how to implement a C and preprocessor
compatible way of doing this came from the Google+ user "comex ." in
response to Linus' crowdsourcing challenge for a possible improvement on
his earlier C specific solution:

#define config_enabled(x)       (__stringify(x)[0] == '1')

In this implementation, I've chosen variable names that hopefully make
how it works more understandable.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoUSB: update usbtmc api documentation
Stephen Lewis [Mon, 2 Apr 2012 20:38:49 +0000]
USB: update usbtmc api documentation

Correct path names in API documentation for usbtmc

Signed-off-by: Stephen Lewis <lewis@sdf.lonestar.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoxHCI: Correct the #define XHCI_LEGACY_DISABLE_SMI
Alex He [Fri, 30 Mar 2012 02:21:38 +0000]
xHCI: Correct the #define XHCI_LEGACY_DISABLE_SMI

Re-define XHCI_LEGACY_DISABLE_SMI and used it in right way. All SMI enable
bits will be cleared to zero and flag bits 29:31 are also cleared to zero.
Other bits should be presvered as Table 146.

This patch should be backported to kernels as old as 2.6.31.

Signed-off-by: Alex He <alex.he@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org

9 years agoxHCI: use gfp flags from caller instead of GFP_ATOMIC
Dan Carpenter [Wed, 28 Mar 2012 07:30:26 +0000]
xHCI: use gfp flags from caller instead of GFP_ATOMIC

The caller is allowed to specify the GFP flags for these functions.
We should prefer their flags unless we have good reason.  For
example, if we take a spin_lock ourselves we'd need to use
GFP_ATOMIC.  But in this case it's safe to use the callers GFP
flags.

The callers all pass GFP_ATOMIC here, so this change doesn't affect
how the kernel behaves but we may add other callers later and this
is a cleanup.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>

9 years agoxHCI: add XHCI_RESET_ON_RESUME quirk for VIA xHCI host
Elric Fu [Thu, 29 Mar 2012 07:47:50 +0000]
xHCI: add XHCI_RESET_ON_RESUME quirk for VIA xHCI host

The suspend operation of VIA xHCI host have some issues and
hibernate operation works fine, so The XHCI_RESET_ON_RESUME
quirk is added for it.

This patch should base on "xHCI: Don't write zeroed pointer
to xHC registers" that is released by Sarah. Otherwise, the
host system error will ocurr in the hibernate operation
process.

This should be backported to stable kernels as old as 2.6.37,
that contain the commit c877b3b2ad5cb9d4fe523c5496185cc328ff3ae9
"xhci: Add reset on resume quirk for asrock p67 host".

Signed-off-by: Elric Fu <elricfu1@gmail.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org

9 years agoUSB: fix bug of device descriptor got from superspeed device
Elric Fu [Mon, 26 Mar 2012 13:16:02 +0000]
USB: fix bug of device descriptor got from superspeed device

When the Seagate Goflex USB3.0 device is attached to VIA xHCI
host, sometimes the device will downgrade mode to high speed.
By the USB analyzer, I found the device finished the link
training process and worked at superspeed mode. But the device
descriptor got from the device shows the device works at 2.1.
It is very strange and seems like the device controller of
Seagate Goflex has a little confusion.

The first 8 bytes of device descriptor should be:
12 01 00 03 00 00 00 09

But the first 8 bytes of wrong device descriptor are:
12 01 10 02 00 00 00 40

The wrong device descriptor caused the initialization of mass
storage failed. After a while, the device would be recognized
as a high speed device and works fine.

This patch will warm reset the device to fix the issue after
finding the bcdUSB field of device descriptor isn't 0x0300
but the speed mode of device is superspeed.

This patch should be backported to kernels as old as 3.2, or ones that
contain the commit 75d7cf72ab9fa01dc70877aa5c68e8ef477229dc "usbcore:
refine warm reset logic".

Signed-off-by: Elric Fu <elricfu1@gmail.com>
Acked-by: Andiry Xu <Andiry.Xu@amd.com>
Acked-by: Sergei Shtylyov <sshtylyov@mvista.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org

9 years agoxhci: Fix register save/restore order.
Sarah Sharp [Fri, 16 Mar 2012 20:19:35 +0000]
xhci: Fix register save/restore order.

The xHCI 1.0 spec errata released on June 13, 2011, changes the ordering
that the xHCI registers are saved and restored in.  It moves the
interrupt pending (IMAN) and interrupt control (IMOD) registers to be
saved and restored last.  I believe that's because the host controller
may attempt to fetch the event ring table when interrupts are
re-enabled.  Therefore we need to restore the event ring registers
before we re-enable interrupts.

This should be backported to kernels as old as 2.6.37, that contain the
commit 5535b1d5f8885695c6ded783c692e3c0d0eda8ca "USB: xHCI: PCI power
management implementation"

Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Elric Fu <elricfu1@gmail.com>
Cc: Andiry Xu <andiry.xu@amd.com>
Cc: stable@vger.kernel.org

9 years agoxhci: Restore event ring dequeue pointer on resume.
Sarah Sharp [Fri, 16 Mar 2012 20:27:39 +0000]
xhci: Restore event ring dequeue pointer on resume.

The xhci_save_registers() function saved the event ring dequeue pointer
in the s3 register structure, but xhci_restore_registers() never
restored it.  No other code in the xHCI successful resume path would
ever restore it either.  Fix that.

This should be backported to kernels as old as 2.6.37, that contain the
commit 5535b1d5f8885695c6ded783c692e3c0d0eda8ca "USB: xHCI: PCI power
management implementation".

Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Elric Fu <elricfu1@gmail.com>
Cc: Andiry Xu <andiry.xu@amd.com>
Cc: stable@vger.kernel.org

9 years agoxhci: Don't write zeroed pointers to xHC registers.
Sarah Sharp [Fri, 16 Mar 2012 20:09:39 +0000]
xhci: Don't write zeroed pointers to xHC registers.

When xhci_mem_cleanup() is called, we can't be sure if the xHC is
actually halted.  We can ask the xHC to halt by writing to the RUN bit
in the command register, but that might timeout due to a HW hang.

If the host controller is still running, we should not write zeroed
values to the event ring dequeue pointers or base tables, the DCBAA
pointers, or the command ring pointers.  Eric Fu reports his VIA VL800
host accesses the event ring pointers after a failed register restore on
resume from suspend.  The hypothesis is that the host never actually
halted before the register write to change the event ring pointer to
zero.

Remove all writes of zeroed values to pointer registers in
xhci_mem_cleanup().  Instead, make all callers of the function reset the
host controller first, which will reset those registers to zero.
xhci_mem_init() is the only caller that doesn't first halt and reset the
host controller before calling xhci_mem_cleanup().

This should be backported to kernels as old as 2.6.32.

Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Elric Fu <elricfu1@gmail.com>
Cc: stable@vger.kernel.org

9 years agoxhci: Warn when hosts don't halt.
Sarah Sharp [Fri, 16 Mar 2012 19:58:20 +0000]
xhci: Warn when hosts don't halt.

Eric Fu reports a problem with his VIA host controller fetching a zeroed
event ring pointer on resume from suspend.  The host should have been
halted, but we can't be sure because that code ignores the return value
from xhci_halt().  Print a warning when the host controller refuses to
halt within XHCI_MAX_HALT_USEC (currently 16 seconds).

(Update: it turns out that the VIA host controller is reporting a halted
state when it fetches the zeroed event ring pointer.  However, we still
need this warning for other host controllers.)

Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>

9 years agoxhci: don't re-enable IE constantly
Felipe Balbi [Thu, 15 Mar 2012 14:37:08 +0000]
xhci: don't re-enable IE constantly

While we're at that, define IMAN bitfield to aid readability.

The interrupt enable bit should be set once on driver init, and we
shouldn't need to continually re-enable it.  Commit c21599a3 introduced
a read of the irq_pending register, and that allows us to preserve the
state of the IE bit.  Before that commit, we were blindly writing 0x3 to
the register.

This patch should be backported to kernels as old as 2.6.36, or ones
that contain the commit c21599a36165dbc78b380846b254017a548b9de5 "USB:
xhci: Reduce reads and writes of interrupter registers".

Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org

9 years agousb: xhci: fix section mismatch in linux-next
Gerard Snitselaar [Fri, 16 Mar 2012 18:34:11 +0000]
usb: xhci: fix section mismatch in linux-next

xhci_unregister_pci() is called in xhci_hcd_init().

Signed-off-by: Gerard Snitselaar <dev@snitselaar.org>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>

9 years agoxHCI: correct to print the true HSEE of USBCMD
Alex He [Thu, 22 Mar 2012 07:06:59 +0000]
xHCI: correct to print the true HSEE of USBCMD

Correct the print of HSEE of USBCMD in xhci-dbg.c.

Signed-off-by: Alex He <alex.he@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>

9 years agoUSB: serial: fix race between probe and open
Johan Hovold [Tue, 20 Mar 2012 15:59:33 +0000]
USB: serial: fix race between probe and open

Fix race between probe and open by making sure that the disconnected
flag is not cleared until all ports have been registered.

A call to tty_open while probe is running may get a reference to the
serial structure in serial_install before its ports have been
registered. This may lead to usb_serial_core calling driver open before
port is fully initialised.

With ftdi_sio this result in the following NULL-pointer dereference as
the private data has not been initialised at open:

[  199.698286] IP: [<f811a089>] ftdi_open+0x59/0xe0 [ftdi_sio]
[  199.698297] *pde = 00000000
[  199.698303] Oops: 0000 [#1] PREEMPT SMP
[  199.698313] Modules linked in: ftdi_sio usbserial
[  199.698323]
[  199.698327] Pid: 1146, comm: ftdi_open Not tainted 3.2.11 #70 Dell Inc. Vostro 1520/0T816J
[  199.698339] EIP: 0060:[<f811a089>] EFLAGS: 00010286 CPU: 0
[  199.698344] EIP is at ftdi_open+0x59/0xe0 [ftdi_sio]
[  199.698348] EAX: 0000003e EBX: f5067000 ECX: 00000000 EDX: 80000600
[  199.698352] ESI: f48d8800 EDI: 00000001 EBP: f515dd54 ESP: f515dcfc
[  199.698356]  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
[  199.698361] Process ftdi_open (pid: 1146, ti=f515c000 task=f481e040 task.ti=f515c000)
[  199.698364] Stack:
[  199.698368]  f811a9fe f811a9e0 f811b3ef 00000000 00000000 00001388 00000000 f4a86800
[  199.698387]  00000002 00000000 f806e68e 00000000 f532765c f481e040 00000246 22222222
[  199.698479]  22222222 22222222 22222222 f5067004 f5327600 f5327638 f515dd74 f806e6ab
[  199.698496] Call Trace:
[  199.698504]  [<f806e68e>] ? serial_activate+0x2e/0x70 [usbserial]
[  199.698511]  [<f806e6ab>] serial_activate+0x4b/0x70 [usbserial]
[  199.698521]  [<c126380c>] tty_port_open+0x7c/0xd0
[  199.698527]  [<f806e660>] ? serial_set_termios+0xa0/0xa0 [usbserial]
[  199.698534]  [<f806e76f>] serial_open+0x2f/0x70 [usbserial]
[  199.698540]  [<c125d07c>] tty_open+0x20c/0x510
[  199.698546]  [<c10e9eb7>] chrdev_open+0xe7/0x230
[  199.698553]  [<c10e48f2>] __dentry_open+0x1f2/0x390
[  199.698559]  [<c144bfec>] ? _raw_spin_unlock+0x2c/0x50
[  199.698565]  [<c10e4b76>] nameidata_to_filp+0x66/0x80
[  199.698570]  [<c10e9dd0>] ? cdev_put+0x20/0x20
[  199.698576]  [<c10f3e08>] do_last+0x198/0x730
[  199.698581]  [<c10f4440>] path_openat+0xa0/0x350
[  199.698587]  [<c10f47d5>] do_filp_open+0x35/0x80
[  199.698593]  [<c144bfec>] ? _raw_spin_unlock+0x2c/0x50
[  199.698599]  [<c10ff110>] ? alloc_fd+0xc0/0x100
[  199.698605]  [<c10f0b72>] ? getname_flags+0x72/0x120
[  199.698611]  [<c10e4450>] do_sys_open+0xf0/0x1c0
[  199.698617]  [<c11fcc08>] ? trace_hardirqs_on_thunk+0xc/0x10
[  199.698623]  [<c10e458e>] sys_open+0x2e/0x40
[  199.698628]  [<c144c990>] sysenter_do_call+0x12/0x36
[  199.698632] Code: 85 89 00 00 00 8b 16 8b 4d c0 c1 e2 08 c7 44 24 14 88 13 00 00 81 ca 00 00 00 80 c7 44 24 10 00 00 00 00 c7 44 24 0c 00 00 00 00 <0f> b7 41 78 31 c9 89 44 24 08 c7 44 24 04 00 00 00 00 c7 04 24
[  199.698884] EIP: [<f811a089>] ftdi_open+0x59/0xe0 [ftdi_sio] SS:ESP 0068:f515dcfc
[  199.698893] CR2: 0000000000000078
[  199.698925] ---[ end trace 77c43ec023940cff ]---

Reported-and-tested-by: Ken Huang <csuhgw@gmail.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoUHCI: hub_status_data should indicate if ports are resuming
Alan Stern [Tue, 3 Apr 2012 19:24:43 +0000]
UHCI: hub_status_data should indicate if ports are resuming

This patch (as1538) causes uhci_hub_status_data() to return a nonzero
value when any port is undergoing a resume transition while the root
hub is suspended.  This will allow usbcore to handle races between
root-hub suspend and port wakeup.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoEHCI: keep track of ports being resumed and indicate in hub_status_data
Alan Stern [Wed, 18 Apr 2012 14:55:45 +0000]
EHCI: keep track of ports being resumed and indicate in hub_status_data

This patch (as1537) adds a bit-array to ehci-hcd for keeping track of
which ports are undergoing a resume transition.  If any of the bits
are set when ehci_hub_status_data() is called, the routine will return
a nonzero value even if no ports have any status changes pending.
This will allow usbcore to handle races between root-hub suspend and
port wakeup.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: Sarah Sharp <sarah.a.sharp@linux.intel.com>
CC: Chen Peter-B29397 <B29397@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Conflicts:

drivers/usb/host/ehci-tegra.c

Change-Id: I30669c6e4ceb78c26209bcadf4d9647141741898

9 years agoUSB: fix race between root-hub suspend and remote wakeup
Alan Stern [Tue, 3 Apr 2012 19:24:18 +0000]
USB: fix race between root-hub suspend and remote wakeup

This patch (as1533) fixes a race between root-hub suspend and remote
wakeup.  If a wakeup event occurs while a root hub is suspending, it
might not cause the suspend to fail.  Although the host controller
drivers check for pending wakeup events at the start of their
bus_suspend routines, they generally do not check for wakeup events
while the routines are running.

In addition, if a wakeup event occurs any time after khubd is frozen
and before the root hub is fully suspended, it might not cause a
system sleep transition to fail.  For example, the host controller
drivers do not fail root-hub suspends when a connect-change event is
pending.

To fix both these issues, this patch causes hcd_bus_suspend() to query
the controller driver's hub_status_data method after a root hub is
suspended, if the root hub is enabled for wakeup.  Any pending status
changes will count as wakeup events, causing the root hub to be
resumed and the overall suspend to fail with -EBUSY.

A significant point is that not all events are reflected immediately
in the status bits.  Both EHCI and UHCI controllers notify the CPU
when remote wakeup begins on a port, but the port's suspend-change
status bit doesn't get set until after the port has completed the
transition out of the suspend state, some 25 milliseconds later.
Consequently, the patch will interpret any nonzero return value from
hub_status_data as indicating a pending event, even if none of the
status bits are set in the data buffer.  Follow-up patches make the
necessary changes to ehci-hcd and uhci-hcd.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: Sarah Sharp <sarah.a.sharp@linux.intel.com>
CC: Chen Peter-B29397 <B29397@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoUSB: sierra: add support for Sierra Wireless MC7710
Anton Samokhvalov [Wed, 4 Apr 2012 18:26:01 +0000]
USB: sierra: add support for Sierra Wireless MC7710

Just add new device id. 3G works fine, LTE not tested.

Signed-off-by: Anton Samokhvalov <pg83@yandex.ru>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoUSB: ftdi_sio: fix race condition in TIOCMIWAIT, and abort of TIOCMIWAIT when the...
Simon Arlott [Mon, 26 Mar 2012 22:27:59 +0000]
USB: ftdi_sio: fix race condition in TIOCMIWAIT, and abort of TIOCMIWAIT when the device is removed

There are two issues here, one is that the device is generating
spurious very fast modem status line changes somewhere:

CTS becomes high then low 18µs later:
[121226.924373] ftdi_process_packet: prev rng=0 dsr=10 dcd=0 cts=6
[121226.924378] ftdi_process_packet: status=10 prev=00 diff=10
[121226.924382] ftdi_process_packet: now rng=0 dsr=10 dcd=0 cts=7
(wake_up_interruptible is called)
[121226.924391] ftdi_process_packet: prev rng=0 dsr=10 dcd=0 cts=7
[121226.924394] ftdi_process_packet: status=00 prev=10 diff=10
[121226.924397] ftdi_process_packet: now rng=0 dsr=10 dcd=0 cts=8
(wake_up_interruptible is called)

This wakes up the task in TIOCMIWAIT:
[121226.924405] ftdi_ioctl: 19451 rng=0->0 dsr=10->10 dcd=0->0 cts=6->8
(wait from 20:51:46 returns and observes both changes)

Which then calls TIOCMIWAIT again:
20:51:46.400239 ioctl(3, TIOCMIWAIT, 0x20) = 0
22:11:09.441818 ioctl(3, TIOCMGET, [TIOCM_DTR|TIOCM_RTS]) = 0
22:11:09.442812 ioctl(3, TIOCMIWAIT, 0x20) = -1 EIO (Input/output error)
(the second wake_up_interruptible takes effect and an I/O error occurs)

The other issue is that TIOCMIWAIT will wait forever (unless the task is
interrupted) if the device is removed.

This change removes the -EIO return that occurs if the counts don't
appear to have changed. Multiple counts may have been processed as
one or the waiting task may have started waiting after recording the
current count.

It adds a bool to indicate that the device has been removed so that
TIOCMIWAIT doesn't wait forever, and wakes up any tasks so that they can
return -EIO.

Signed-off-by: Simon Arlott <simon@fire.lp0.eu>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoUSB: ftdi_sio: fix status line change handling for TIOCMIWAIT and TIOCGICOUNT
Simon Arlott [Mon, 26 Mar 2012 20:19:40 +0000]
USB: ftdi_sio: fix status line change handling for TIOCMIWAIT and TIOCGICOUNT

Handling of TIOCMIWAIT was changed by commit 1d749f9afa657f6ee9336b2bc1fcd750a647d157
 USB: ftdi_sio.c: Use ftdi async_icount structure for TIOCMIWAIT, as in other drivers

FTDI_STATUS_B0_MASK does not indicate the changed modem status lines,
it indicates the value of the current modem status lines. An xor is
still required to determine which lines have changed.

The count was only being incremented if the line was high. The only
reason TIOCMIWAIT still worked was because the status packet is
repeated every 1ms, so the count was always changing. The wakeup
itself still ran based on the status lines changing.

This change fixes handling of updates to the modem status lines and
allows multiple processes to use TIOCMIWAIT concurrently.

Tested with two processes waiting on different status lines being
toggled independently.

Signed-off-by: Simon Arlott <simon@fire.lp0.eu>
Cc: Uwe Bonnes <bon@elektron.ikp.physik.tu-darmstadt.de>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoUSB: don't ignore suspend errors for root hubs
Alan Stern [Wed, 28 Mar 2012 19:56:17 +0000]
USB: don't ignore suspend errors for root hubs

This patch (as1532) fixes a mistake in the USB suspend code.  When the
system is going to sleep, we should ignore errors in powering down USB
devices, because they don't really matter.  The devices will go to low
power anyway when the entire USB bus gets suspended (except for
SuperSpeed devices; maybe they will need special treatment later).

However we should not ignore errors in suspending root hubs,
especially if the error indicates that the suspend raced with a wakeup
request.  Doing so might leave the bus powered on while the system was
supposed to be asleep, or it might cause the suspend of the root hub's
parent controller device to fail, or it might cause a wakeup request
to be ignored.

The patch fixes the problem by ignoring errors only when the device in
question is not a root hub.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Reported-by: Chen Peter <B29397@freescale.com>
CC: <stable@vger.kernel.org>
Tested-by: Chen Peter <peter.chen@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoUSB: fix bug in serial driver unregistration
Alan Stern [Wed, 28 Mar 2012 20:10:49 +0000]
USB: fix bug in serial driver unregistration

This patch (as1536) fixes a bug in the USB serial core.  Unloading and
reloading a serial driver while a serial device is plugged in causes
errors because of the code in usb_serial_disconnect() that tries to
make sure the port_remove method is called.  With the new order of
driver registration introduced in the 3.4 kernel, this is definitely
not the right thing to do (if indeed it ever was).

The patch removes that whole section code, along with the mechanism
for keeping track of each port's registration state, which is no
longer needed.  The driver core can handle all that stuff for us.

Note: This has been tested only with one or two USB serial drivers.
In theory, other drivers might still run into trouble.  But if they
do, it will be the fault of the drivers, not of this patch -- that is,
the drivers will need to be fixed.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Reported-and-tested-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoUSB: serial: metro-usb: Fix idProduct for Uni-Directional mode.
Aleksey Babahin [Mon, 19 Mar 2012 20:46:31 +0000]
USB: serial: metro-usb: Fix idProduct for Uni-Directional mode.

The right idProduct for Metrologic Bar Code Scanner
in Uni-Directional Serial Emulation mode is 0x0700.

Also rename idProduct for Bi-Directional mode to be a bit more informative.

Signed-off-by: Aleksey Babahin <tamerlan311@gmail.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoUSB: option: re-add NOVATELWIRELESS_PRODUCT_HSPA_HIGHSPEED to option_id array
Santiago Garcia Mantinan [Mon, 19 Mar 2012 17:17:00 +0000]
USB: option: re-add NOVATELWIRELESS_PRODUCT_HSPA_HIGHSPEED to option_id array

Re-add NOVATELWIRELESS_PRODUCT_HSPA_HIGHSPEED to option_id array

Signed-off-by: Santiago Garcia Mantinan <manty@debian.org>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoUSB: pl2303: fix DTR/RTS being raised on baud rate change
Johan Hovold [Fri, 23 Mar 2012 14:23:18 +0000]
USB: pl2303: fix DTR/RTS being raised on baud rate change

DTR/RTS should only be raised when changing baudrate from B0 and not on
any baud rate change (> B0).

Reported-by: Søren Holm <sgh@sgh.dk>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoUSB: remove compile warning on gadget/inode.c
Michael BRIGHT [Thu, 22 Mar 2012 17:42:52 +0000]
USB: remove compile warning on gadget/inode.c

Removed unused "restart:" label, which was causing compiler warning.

Signed-off-by: Michael BRIGHT <mjbrightfr+git-kernel@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoUSB documentation: explain lifetime rules for unlinking URBs
Alan Stern [Wed, 28 Mar 2012 20:13:28 +0000]
USB documentation: explain lifetime rules for unlinking URBs

This patch (as1534c) updates the documentation for usb_unlink_urb and
related functions.  It explains that the caller must prevent the URB
being unlinked from getting deallocated while the unlink is taking
place.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: Ming Lei <tom.leiming@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoUSB: don't clear urb->dev in scatter-gather library
Alan Stern [Thu, 22 Mar 2012 15:00:21 +0000]
USB: don't clear urb->dev in scatter-gather library

This patch (as1517b) fixes an error in the USB scatter-gather library.
The library code uses urb->dev to determine whether or nor an URB is
currently active; the completion handler sets urb->dev to NULL.
However the core unlinking routines need to use urb->dev.  Since
unlinking always racing with completion, the completion handler must
not clear urb->dev -- it can lead to invalid memory accesses when a
transfer has to be cancelled.

This patch fixes the problem by getting rid of the lines that clear
urb->dev after urb has been submitted.  As a result we may end up
trying to unlink an URB that failed in submission or that has already
completed, so an extra check is added after each unlink to avoid
printing an error message when this happens.  The checks are updated
in both sg_complete() and sg_cancel(), and the second is updated to
match the first (currently it prints out unnecessary warning messages
if a device is unplugged while a transfer is in progress).

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Reported-and-tested-by: Illia Zaitsev <I.Zaitsev@adbglobal.com>
CC: Ming Lei <tom.leiming@gmail.com>
CC: <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agousb: storage: fix lockdep warning inside usb_stor_pre_reset(v2)
Ming Lei [Mon, 19 Mar 2012 07:20:57 +0000]
usb: storage: fix lockdep warning inside usb_stor_pre_reset(v2)

This patch fixes one lockdep warning[1] inside usb_stor_pre_reset.

If the current configuration includes multiple mass storage interfaces,
the 'AA' lockdep warning will be triggered since the lock class of
'us->dev_mutex' is acquired two times in .reset path. It isn't
a real deadlock, so just take the lockdep_set_class annotation to
remove the warning.

[1], lockdep warning log
:[ INFO: possible recursive locking detected ]
:3.3.0-0.rc5.git3.1.fc17.x86_64 #1 Tainted: G        W
:---------------------------------------------
:usb-storage/14846 is trying to acquire lock:
: (&(us->dev_mutex)){+.+.+.}, at: [<ffffffffa0481c0c>] usb_stor_pre_reset+0x1c/0x20 [usb_storage]
:but task is already holding lock:
: (&(us->dev_mutex)){+.+.+.}, at: [<ffffffffa0481c0c>] usb_stor_pre_reset+0x1c/0x20 [usb_storage]
:other info that might help us debug this:
: Possible unsafe locking scenario:
:       CPU0
:       ----
:  lock(&(us->dev_mutex));
:  lock(&(us->dev_mutex));
: *** DEADLOCK ***
: May be due to missing lock nesting notation
:2 locks held by usb-storage/14846:
: #0:  (&__lockdep_no_validate__){......}, at: [<ffffffff8147e6a5>] usb_lock_device_for_reset+0x95/0x100
: #1:  (&(us->dev_mutex)){+.+.+.}, at: [<ffffffffa0481c0c>] usb_stor_pre_reset+0x1c/0x20 [usb_storage]
:stack backtrace:
:Pid: 14846, comm: usb-storage Tainted: G        W 3.3.0-0.rc5.git3.1.fc17.x86_64 #1
:Call Trace:
: [<ffffffff810cbdaf>] __lock_acquire+0x168f/0x1bb0
: [<ffffffff81021083>] ? native_sched_clock+0x13/0x80
: [<ffffffff810210f9>] ? sched_clock+0x9/0x10
: [<ffffffff810210f9>] ? sched_clock+0x9/0x10
: [<ffffffff810a2975>] ? sched_clock_local+0x25/0xa0
: [<ffffffff810cc9a1>] lock_acquire+0xa1/0x1e0
: [<ffffffffa0481c0c>] ? usb_stor_pre_reset+0x1c/0x20 [usb_storage]
: [<ffffffff81699c86>] mutex_lock_nested+0x76/0x3a0
: [<ffffffffa0481c0c>] ? usb_stor_pre_reset+0x1c/0x20 [usb_storage]
: [<ffffffffa0481c0c>] ? usb_stor_pre_reset+0x1c/0x20 [usb_storage]
: [<ffffffffa0481c0c>] usb_stor_pre_reset+0x1c/0x20 [usb_storage]
: [<ffffffff8148184d>] usb_reset_device+0x7d/0x190
: [<ffffffffa048119c>] usb_stor_port_reset+0x7c/0x80 [usb_storage]
: [<ffffffffa0481234>] usb_stor_invoke_transport+0x94/0x560 [usb_storage]
: [<ffffffff810cd3b2>] ? mark_held_locks+0xb2/0x130
: [<ffffffff8169dbd0>] ? _raw_spin_unlock_irq+0x30/0x50
: [<ffffffffa047fe3e>] usb_stor_transparent_scsi_command+0xe/0x10 [usb_storage]
: [<ffffffffa0481ae3>] usb_stor_control_thread+0x173/0x280 [usb_storage]
: [<ffffffffa0481970>] ? fill_inquiry_response+0x20/0x20 [usb_storage]
: [<ffffffff8108a3f7>] kthread+0xb7/0xc0
: [<ffffffff816a7d34>] kernel_thread_helper+0x4/0x10
: [<ffffffff8169e0f4>] ? retint_restore_args+0x13/0x13
: [<ffffffff8108a340>] ? kthread_worker_fn+0x1a0/0x1a0
: [<ffffffff816a7d30>] ? gs_change+0x13/0x13

Reported-By: Dave Jones <davej@redhat.com>
Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agousb/usbmon: correct the data interpretation of usbmon's output
Sebastian Andrzej Siewior [Sat, 17 Mar 2012 14:23:49 +0000]
usb/usbmon: correct the data interpretation of usbmon's output

The doc says that the data
55534243 5e000000 00000000 00000600 00000000 00000000 00000000 000000
is the SCSI command 0x5e. According to the usbmon source, it dumps one
byte after the other. The first 4 bytes are US_BULK_CB_SIGN which is
correct. After that we see the TAG which is 0x5e. The cdb is 0x00 in
this example.
In order to correct this, I change the example to a READ_10 command
which is 0x28 so it is not just a zero somewhere in the stream.

Signed-off-by: Sebastian Andrzej Siewior <sebastian@breakpoint.cc>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agousb: Put USB Kconfig items back under USB.
David Daney [Thu, 22 Mar 2012 21:01:07 +0000]
usb: Put USB Kconfig items back under USB.

commit 53c6bc24fdc8db87109a5760579cbb060fa644cf (usb: Don't make
USB_ARCH_HAS_{XHCI,OHCI,EHCI} depend on USB_SUPPORT.)  Removed the
dependency of the USB_ARCH_HAS_* symbols on USB_SUPPORT.  However the
resulting Kconfig somehow caused many of the USB configuration items
to appear under the top level devices menu.

To fix this we reunite the 'menuconfig USB_SUPPORT' with the 'if
USB_SUPPORT', and the config items magically go back to their desired
location.

Reported-by: Julian Wollrath <jwollrath@web.de>
Reported-by: Nobuhiro Iwamatsu <nobuhiro.iwamatsu.yj@renesas.com>
Reported-by: Borislav Petkov <bp@alien8.de>
Reported-by: Rupesh Gujare <rgujare@ozmodevices.com>
Reported-by: Feng King <ronyjin@tencent.com>
Reported-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Signed-off-by: David Daney <david.daney@cavium.com>
Tested-by: Peter Chen <peter.chen@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agopch_uart: Add Kontron COMe-mTT10 uart clock quirk
Michael Brunner [Fri, 23 Mar 2012 10:06:37 +0000]
pch_uart: Add Kontron COMe-mTT10 uart clock quirk

Add UART clock quirk for the Kontron COMe-mTT10 module.

The board has previously been called nanoETXexpress-TT, therefore this
is also checked.

As suggested by Darren Hart the comparison in this patch version is
placed after the FRI2 checks to ensure it will also work with possible
upcoming changes to the FRI2 firmware.

This patch follows the patchset submitted by Darren Hart at
commit a46f5533ecfc7bbdd646d84fdab8656031a715c6.

Signed-off-by: Michael Brunner <mibru@gmx.de>
Acked-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agopch_uart: Fix MSI setting issue
Tomoya MORINAGA [Mon, 2 Apr 2012 05:36:22 +0000]
pch_uart: Fix MSI setting issue

The following patch (MSI setting) is not enough.

commit e463595fd9c752fa4bf06b47df93ef9ade3c7cf0
Author: Alexander Stein <alexander.stein@systec-electronic.com>
Date:   Mon Jul 4 08:58:31 2011 +0200

    pch_uart: Add MSI support

Signed-off-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

To enable MSI mode, PCI bus-mastering must be enabled.
This patch enables the setting.

cc: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Tomoya MORINAGA <tomoya.rohm@gmail.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoserial/8250_pci: add a "force background timer" flag and use it for the "kt" serial...
Dan Williams [Fri, 6 Apr 2012 18:49:50 +0000]
serial/8250_pci: add a "force background timer" flag and use it for the "kt" serial port

Workaround dropped notifications in the iir register.  Register reads
coincident with new interrupt notifications sometimes result in this
device clearing the interrupt event without reporting it in the read
data.

The serial core already has a heuristic for determining when a device
has an untrustworthy iir register.  In this case when we apriori know
that the iir is faulty use a flag (UPF_BUG_THRE) to bypass the test and
force usage of the background timer.

[stable: 3.3.x]
Acked-by: Alan Cox <alan@linux.intel.com>
Cc: stable <stable@vger.kernel.org>
Reported-by: Nhan H Mai <nhan.h.mai@intel.com>
Reported-by: Sudhakar Mamillapalli <sudhakar@fb.com>
Tested-by: Nhan H Mai <nhan.h.mai@intel.com>
Tested-by: Sudhakar Mamillapalli <sudhakar@fb.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoRevert "serial/8250_pci: setup-quirk workaround for the kt serial controller"
Dan Williams [Fri, 6 Apr 2012 18:49:44 +0000]
Revert "serial/8250_pci: setup-quirk workaround for the kt serial controller"

This reverts commit 448ac154c957c4580531fa0c8f2045816fe2f0e7.

The semantic of UPF_IIR_ONCE is only guaranteed to workaround the race
condition in the kt serial's iir register if the only source of
interrupts is THRE (fifo-empty) events.  An modem status event at the
wrong time can again cause an iir read to drop the 'empty' status
leading to a hang.  So, revert this in preparation for using the
existing "I don't trust my iir register" workaround in the 8250 core
(UART_BUG_THRE).

[stable: 3.3.x]
Cc: stable <stable@vger.kernel.org>
Acked-by: Alan Cox <alan@linux.intel.com>
Cc: Sudhakar Mamillapalli <sudhakar@fb.com>
Reported-by: Nhan H Mai <nhan.h.mai@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoRevert "serial/8250_pci: init-quirk msi support for kt serial controller"
Dan Williams [Fri, 6 Apr 2012 18:49:37 +0000]
Revert "serial/8250_pci: init-quirk msi support for kt serial controller"

This reverts commit e86ff4a63c9fdd875ba8492577cd1ad2252f525c.

This tried to enforce the semantics of one interrupt per iir read of the
THRE (transmit-hold empty) status, but events from other sources
(particularly modem status) defeat this guarantee.

This change also broke 8250_pci suspend/resume support as
pciserial_resume_ports() re-runs .init() quirks, but does not run
.exit() quirks in pciserial_suspend_ports() leading to reports like:

  sysfs: cannot create duplicate filename '/devices/pci0000:00/0000:00:16.3/msi_irqs'

...and a subsequent crash.  The mismatch of init/exit at suspend/resume
seems like a bug in its own right.

[stable: 3.3.x]
Cc: stable <stable@vger.kernel.org>
Acked-by: Alan Cox <alan@linux.intel.com>
Cc: Sudhakar Mamillapalli <sudhakar@fb.com>
Reported-by: Nhan H Mai <nhan.h.mai@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agotty/serial/omap: console can only be built-in
Arnd Bergmann [Mon, 2 Apr 2012 22:32:17 +0000]
tty/serial/omap: console can only be built-in

When the omap serial driver is built as a module, we must
not allow the console driver to be selected, because consoles
can not be loadable modules.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Acked-by: Govindraj.R <govindraj.raja@ti.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoserial: samsung: fix omission initialize ulcon in reset port fn()
Kukjin Kim [Wed, 4 Apr 2012 01:14:24 +0000]
serial: samsung: fix omission initialize ulcon in reset port fn()

Fix omission initialize ulcon in s3c24xx_serial_resetport(),
reset port function in drivers/tty/serial/samsung.c. It has
been happened from commit 0dfb3b41("serial: samsung: merge
all SoC specific port reset functions")

Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
Cc: stable <stable@vger.kernel.org> [3.3]
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoprintk(): add KERN_CONT where needed in hpet and vt code
Kay Sievers [Tue, 3 Apr 2012 01:18:23 +0000]
printk(): add KERN_CONT where needed in hpet and vt code

A prototype for kmsg records instead of a byte-stream buffer revealed
a couple of missing printk(KERN_CONT ...) uses. Subsequent calls produce
one record per printk() call, while all should have ended up in a single
record.

Instead of:
  ACPI: (supports S0 S5)
  ACPI: PCI Interrupt Link [LNKA] (IRQs 5 *10 11)
  hpet0: at MMIO 0xfed00000, IRQs 2 , 8 , 0

It prints:
  ACPI: (supports S0
   S5
  )
  ACPI: PCI Interrupt Link [LNKA] (IRQs
   5
   *10
   11
  )
  hpet0: at MMIO 0xfed00000, IRQs
   2
  , 8
  , 0

Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Len Brown <lenb@kernel.org>
Signed-off-by: Kay Sievers <kay@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agotty/serial: atmel_serial: fix RS485 half-duplex problem
Siftar, Gabe [Thu, 29 Mar 2012 13:40:05 +0000]
tty/serial: atmel_serial: fix RS485 half-duplex problem

On our custom board, we are using RS485 in half-duplex mode on an AT91SAM9G45.
SER_RS485_RX_DURING_TX is not set as we do not want to receive the data we
transmit (our transceiver will receive transmitted data).
Although the current driver attempts to disable and enable the receiver at the
appropriate points, incoming data is still loaded into the receive register
causing our code to receive the very last byte that was sent once the receiver
is enabled.

I ran this by Atmel support and they wrote: "The issue comes from the fact
that you disable the PDC/DMA Reception and not the USART Reception channel. In
your case, the[n] you will still receive data into the USART_RHR register, and
maybe you [h]ave the overrun flag set. So please disable the USART reception
channel."

The following patch should force the driver to enable/disable the receiver via
RXEN/RXDIS fields of the USART control register. It fixed the issue I was
having.

Signed-off-by: Gabe Siftar <gabe.siftar@getingeusa.com>
[nicolas.ferre@atmel.com: slightly modify commit message]
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agotty: serial: altera_uart: Check for NULL platform_data in probe.
Yuriy Kozlov [Thu, 29 Mar 2012 07:55:27 +0000]
tty: serial: altera_uart: Check for NULL platform_data in probe.

Follow altera_jtag_uart.  This fixes a crash if there is a mistake in the DTS.

Signed-off-by: Yuriy Kozlov <ykozlov@ptcusa.com>
Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoisdn/gigaset: use gig_dbg() for debugging output
Tilman Schmidt [Sun, 25 Mar 2012 17:12:37 +0000]
isdn/gigaset: use gig_dbg() for debugging output

The "TTY buffer in tty_port" patchset introduced an opencoded
debug message in the Gigaset tty device if_close() function.
Change it to use the gig_dbg() macro like everywhere else in
the driver.

Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Cc: Jiri Slaby <jslaby@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoomap-serial: Fix the error handling in the omap_serial probe
Shubhrajyoti D [Wed, 21 Mar 2012 11:52:22 +0000]
omap-serial: Fix the error handling in the omap_serial probe

The patch does the following

- The pm_runtime_disable is called in the remove not in the error
  case of probe.The patch calls the pm_runtime_disable in the error
  case.
- Calls pm_runtime_put in the error case.
- The  up is not freed in the error path. Fix the memory leak by using
  devm_* so that the memory need not be freed in the driver.
- Also the iounmap is not called fix the same by calling using devm_ioremap.
- Make the name of the error tags more meaningful.

Cc: Mark Brown <broonie@opensource.wolfsonmicro.com>
Cc: Govindraj.R <govindraj.raja@ti.com>
Signed-off-by: Shubhrajyoti D <shubhrajyoti@ti.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

9 years agoserial: PL011: move interrupt clearing
Linus Walleij [Wed, 21 Mar 2012 19:15:18 +0000]
serial: PL011: move interrupt clearing

Commit 360f748b204275229f8398cb2f9f53955db1503b
"serial: PL011: clear pending interrupts"
attempts to clear interrupts by writing to a
yet-unassigned memory address. This fixes the issue.

The breaking patch is marked for stable so should be
carried along with the other patch.

Cc: Shreshtha Kumar Sahu <shreshthakumar.sahu@stericsson.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: stable <stable@vger.kernel.org>
Cc: Nicolas Pitre <nico@fluxnic.net>
Reported-by: Viresh Kumar <viresh.kumar@st.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Tested-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>