ac100:marvin24s-kernel.git
5 years agoARM: tegra: paz00: update defconfig linux-tegra-nv-ac100-3.1-exp
Marc Dietrich [Fri, 13 Jul 2012 17:34:22 +0000 (19:34 +0200)]
ARM: tegra: paz00: update defconfig

- remove some useless sound / camera drivers
- enable zsmalloc

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
5 years agoARM: tegra: paz00: skip framebuffer copy if no boot framebuffer is available
Marc Dietrich [Fri, 13 Jul 2012 17:17:47 +0000 (19:17 +0200)]
ARM: tegra: paz00: skip framebuffer copy if no boot framebuffer is available

This fixes boots using u-boot which does not yet supply a boot framebuffer.

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
5 years agostaging: nvec: fix clock enable on resume
Marc Dietrich [Wed, 11 Jul 2012 19:59:34 +0000 (21:59 +0200)]
staging: nvec: fix clock enable on resume

I don't know why the clock was disabled and enabled shortly after,
but we failed to enable it on resume. So remove the clock disable.

This fixes 2nd and later suspend/resume fails.

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
5 years agoARM: tegra: paz00: add more otg support
Marc Dietrich [Tue, 10 Jul 2012 19:22:10 +0000 (21:22 +0200)]
ARM: tegra: paz00: add more otg support

Just copy-n-paste for now. Still no volunteers?

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
5 years agoRevert "staging: nvec: fix mouse suspend/resume"
Marc Dietrich [Mon, 9 Jul 2012 20:01:48 +0000 (22:01 +0200)]
Revert "staging: nvec: fix mouse suspend/resume"

This reverts commit 3eb9a36ec985b0a5aef0c35ca6c4f7860384c507.

Well, while this patch did the "right thing(R)", it somehow broke
resume. So revert it for now and check later on ;-)

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
5 years agoRevert "ARM: tegra: paz00: disable usb otg for now"
Marc Dietrich [Mon, 9 Jul 2012 19:59:35 +0000 (21:59 +0200)]
Revert "ARM: tegra: paz00: disable usb otg for now"

This reverts commit 5b5deeb68d20a99077df96591e95f1e973cbd9e6.

As zombah mentioned on IRC, this was actually working and resume was broken
by something else (see next revert). Also this seems to need the old freescale
driver instead of the tegra one to get it working ...

(HINT TO MYSELF: defconfig update required)

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
5 years agovideo: tegra: make carveout heap relocations less noisy
Marc Dietrich [Sun, 8 Jul 2012 18:27:39 +0000 (20:27 +0200)]
video: tegra: make carveout heap relocations less noisy

Reduce kern log spam.

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
5 years agoASoC: tegra: only stereo is supported
Marc Dietrich [Sun, 8 Jul 2012 17:30:25 +0000 (19:30 +0200)]
ASoC: tegra: only stereo is supported

Report min/max_channels = 2 to limit output to stereo only. Otherwise
we get a lot of errors in the kernel log.

*** This is an upstream candidate! ***

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
5 years agort2x00: Add support for BUFFALO WLI-UC-GNM2 to rt2800usb.
Jeongdo Son [Thu, 14 Jun 2012 17:28:01 +0000 (02:28 +0900)]
rt2x00: Add support for BUFFALO WLI-UC-GNM2 to rt2800usb.

This is a RT3070 based device.

Signed-off-by: Jeongdo Son <sohn9086@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agort2x00 : RT3290 chip support v4
Woody Hung [Wed, 13 Jun 2012 07:01:16 +0000 (15:01 +0800)]
rt2x00 : RT3290 chip support v4

This patch support the new chipset rt3290 wifi implementation in rt2x00.
It initailize the related mac, bbp and rf register in startup phase.
And this patch modify the efuse read/write method for the different efuse data offset of rt3290.

Signed-off-by: Woody Hung <Woody.Hung@mediatek.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agort2x00:Add RT5392 chipset support
Zero.Lin [Tue, 29 May 2012 08:11:09 +0000 (16:11 +0800)]
rt2x00:Add RT5392 chipset support

Signed-off-by: Zero.Lin <Zero.Lin@mediatek.com>
Reviewed-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agort2x00: claim RSN IBSS support
Stanislaw Gruszka [Mon, 28 May 2012 09:53:11 +0000 (11:53 +0200)]
rt2x00: claim RSN IBSS support

Similar like other drivers, do not configure group keys to the hardware
(on Ad-Hoc mode) to make IBSS RSN work.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Acked-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agort2x00: Add support for RT5360 based PCI devices.
villacis@palosanto.com [Wed, 16 May 2012 19:07:12 +0000 (21:07 +0200)]
rt2x00: Add support for RT5360 based PCI devices.

This chip is used at least by the D-Link DWA-525 adapter.

Signed-off-by: Alex Villac<ED>s Lasso <a_villacis@palosanto.com>
Signed-off-by: Gertjan van Wingerde <gwingerde@gmail.com>
Acked-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agort2800usb: 2001:3c17 is an RT3370 device
Albert Pool [Mon, 14 May 2012 16:08:32 +0000 (18:08 +0200)]
rt2800usb: 2001:3c17 is an RT3370 device

D-Link DWA-123 rev A1

Signed-off-by: Albert Pool<albertpool@solcon.nl>
Cc: <stable@vger.kernel.org>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agort2x00: use atomic variable for seqno
Stanislaw Gruszka [Fri, 1 Jun 2012 09:29:40 +0000 (11:29 +0200)]
rt2x00: use atomic variable for seqno

Remove spinlock as atomic_t can be used instead. Note we use only 16
lower bits, upper bits are changed but we impilcilty cast to u16.

This fix possible deadlock on IBSS mode reproted by lockdep:

=================================
[ INFO: inconsistent lock state ]
3.4.0-wl+ #4 Not tainted
---------------------------------
inconsistent {IN-SOFTIRQ-W} -> {SOFTIRQ-ON-W} usage.
kworker/u:2/30374 [HC0[0]:SC0[0]:HE1:SE1] takes:
 (&(&intf->seqlock)->rlock){+.?...}, at: [<f9979a20>] rt2x00queue_create_tx_descriptor+0x380/0x490 [rt2x00lib]
{IN-SOFTIRQ-W} state was registered at:
  [<c04978ab>] __lock_acquire+0x47b/0x1050
  [<c0498504>] lock_acquire+0x84/0xf0
  [<c0835733>] _raw_spin_lock+0x33/0x40
  [<f9979a20>] rt2x00queue_create_tx_descriptor+0x380/0x490 [rt2x00lib]
  [<f9979f2a>] rt2x00queue_write_tx_frame+0x1a/0x300 [rt2x00lib]
  [<f997834f>] rt2x00mac_tx+0x7f/0x380 [rt2x00lib]
  [<f98fe363>] __ieee80211_tx+0x1b3/0x300 [mac80211]
  [<f98ffdf5>] ieee80211_tx+0x105/0x130 [mac80211]
  [<f99000dd>] ieee80211_xmit+0xad/0x100 [mac80211]
  [<f9900519>] ieee80211_subif_start_xmit+0x2d9/0x930 [mac80211]
  [<c0782e87>] dev_hard_start_xmit+0x307/0x660
  [<c079bb71>] sch_direct_xmit+0xa1/0x1e0
  [<c0784bb3>] dev_queue_xmit+0x183/0x730
  [<c078c27a>] neigh_resolve_output+0xfa/0x1e0
  [<c07b436a>] ip_finish_output+0x24a/0x460
  [<c07b4897>] ip_output+0xb7/0x100
  [<c07b2d60>] ip_local_out+0x20/0x60
  [<c07e01ff>] igmpv3_sendpack+0x4f/0x60
  [<c07e108f>] igmp_ifc_timer_expire+0x29f/0x330
  [<c04520fc>] run_timer_softirq+0x15c/0x2f0
  [<c0449e3e>] __do_softirq+0xae/0x1e0
irq event stamp: 18380437
hardirqs last  enabled at (18380437): [<c0526027>] __slab_alloc.clone.3+0x67/0x5f0
hardirqs last disabled at (18380436): [<c0525ff3>] __slab_alloc.clone.3+0x33/0x5f0
softirqs last  enabled at (18377616): [<c0449eb3>] __do_softirq+0x123/0x1e0
softirqs last disabled at (18377611): [<c041278d>] do_softirq+0x9d/0xe0

other info that might help us debug this:
 Possible unsafe locking scenario:

       CPU0
       ----
  lock(&(&intf->seqlock)->rlock);
  <Interrupt>
    lock(&(&intf->seqlock)->rlock);

 *** DEADLOCK ***

4 locks held by kworker/u:2/30374:
 #0:  (wiphy_name(local->hw.wiphy)){++++.+}, at: [<c045cf99>] process_one_work+0x109/0x3f0
 #1:  ((&sdata->work)){+.+.+.}, at: [<c045cf99>] process_one_work+0x109/0x3f0
 #2:  (&ifibss->mtx){+.+.+.}, at: [<f98f005b>] ieee80211_ibss_work+0x1b/0x470 [mac80211]
 #3:  (&intf->beacon_skb_mutex){+.+...}, at: [<f997a644>] rt2x00queue_update_beacon+0x24/0x50 [rt2x00lib]

stack backtrace:
Pid: 30374, comm: kworker/u:2 Not tainted 3.4.0-wl+ #4
Call Trace:
 [<c04962a6>] print_usage_bug+0x1f6/0x220
 [<c0496a12>] mark_lock+0x2c2/0x300
 [<c0495ff0>] ? check_usage_forwards+0xc0/0xc0
 [<c04978ec>] __lock_acquire+0x4bc/0x1050
 [<c0527890>] ? __kmalloc_track_caller+0x1c0/0x1d0
 [<c0777fb6>] ? copy_skb_header+0x26/0x90
 [<c0498504>] lock_acquire+0x84/0xf0
 [<f9979a20>] ? rt2x00queue_create_tx_descriptor+0x380/0x490 [rt2x00lib]
 [<c0835733>] _raw_spin_lock+0x33/0x40
 [<f9979a20>] ? rt2x00queue_create_tx_descriptor+0x380/0x490 [rt2x00lib]
 [<f9979a20>] rt2x00queue_create_tx_descriptor+0x380/0x490 [rt2x00lib]
 [<f997a5cf>] rt2x00queue_update_beacon_locked+0x5f/0xb0 [rt2x00lib]
 [<f997a64d>] rt2x00queue_update_beacon+0x2d/0x50 [rt2x00lib]
 [<f9977e3a>] rt2x00mac_bss_info_changed+0x1ca/0x200 [rt2x00lib]
 [<f9977c70>] ? rt2x00mac_remove_interface+0x70/0x70 [rt2x00lib]
 [<f98e4dd0>] ieee80211_bss_info_change_notify+0xe0/0x1d0 [mac80211]
 [<f98ef7b8>] __ieee80211_sta_join_ibss+0x3b8/0x610 [mac80211]
 [<c0496ab4>] ? mark_held_locks+0x64/0xc0
 [<c0440012>] ? virt_efi_query_capsule_caps+0x12/0x50
 [<f98efb09>] ieee80211_sta_join_ibss+0xf9/0x140 [mac80211]
 [<f98f0456>] ieee80211_ibss_work+0x416/0x470 [mac80211]
 [<c0496d8b>] ? trace_hardirqs_on+0xb/0x10
 [<c077683b>] ? skb_dequeue+0x4b/0x70
 [<f98f207f>] ieee80211_iface_work+0x13f/0x230 [mac80211]
 [<c045cf99>] ? process_one_work+0x109/0x3f0
 [<c045d015>] process_one_work+0x185/0x3f0
 [<c045cf99>] ? process_one_work+0x109/0x3f0
 [<f98f1f40>] ? ieee80211_teardown_sdata+0xa0/0xa0 [mac80211]
 [<c045ed86>] worker_thread+0x116/0x270
 [<c045ec70>] ? manage_workers+0x1e0/0x1e0
 [<c0462f64>] kthread+0x84/0x90
 [<c0462ee0>] ? __init_kthread_worker+0x60/0x60
 [<c083d382>] kernel_thread_helper+0x6/0x10

Cc: stable@vger.kernel.org
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Acked-by: Helmut Schaa <helmut.schaa@googlemail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agort2x00:Add RT539b chipset support
Zero.Lin [Thu, 10 May 2012 02:06:31 +0000 (10:06 +0800)]
rt2x00:Add RT539b chipset support

Signed-off-by: Zero.Lin <Zero.Lin@mediatek.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agort2x00: Add debugfs access for rfcsr register
Anisse Astier [Thu, 19 Apr 2012 13:53:10 +0000 (15:53 +0200)]
rt2x00: Add debugfs access for rfcsr register

RFCSR is only used in rt2800. For other chipsets, the debug struct
for rfcsr should be zeroed, which isn't be an issue, since the code
can now cope with that.

Signed-off-by: Anisse Astier <anisse@astier.eu>
Acked-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agort2x00: debugfs support - allow a register to be empty
Anisse Astier [Thu, 19 Apr 2012 13:04:52 +0000 (15:04 +0200)]
rt2x00: debugfs support - allow a register to be empty

Allow a register to be unspecified, therefore not creating its debugfs
file entry.

Signed-off-by: Anisse Astier <anisse@astier.eu>
Acked-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agort2800: add chipset revision RT5390R support
Anisse Astier [Mon, 23 Apr 2012 10:33:11 +0000 (12:33 +0200)]
rt2800: add chipset revision RT5390R support

About 70% of the chips with revision RT5390R initialize incorrectly, using
the auxiliary antenna instead of the main one. The net result is that
signal reception is very poor (no AP further than 1M).

This chipset differs from RT5390 and RT5390F by its support of hardware
antenna diversity. Therefore antenna selection should be done
differently, by disabling software features and previously selected
antenna.

This changeset does just that, and makes all RT5390R work properly.

This is based on Ralink's 2012_03_22_RT5572_Linux_STA_v2.6.0.0_DPO
driver.

Signed-off-by: Anisse Astier <anisse@astier.eu>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agort2x00: Use GFP_KERNEL for rx buffer allocation on USB devices
Helmut Schaa [Thu, 19 Apr 2012 11:24:10 +0000 (13:24 +0200)]
rt2x00: Use GFP_KERNEL for rx buffer allocation on USB devices

Since the RX path on USB devices is handled in process context we can
use GFP_KERNEL for RX buffer allocation. This should reduce the
likelihood of allocation failures.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Tested-By: Marc Dietrich <marvin24@gmx.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agort2800: debugfs register access: BBP is 256 bytes big
Anisse Astier [Thu, 19 Apr 2012 09:20:32 +0000 (11:20 +0200)]
rt2800: debugfs register access: BBP is 256 bytes big

We're already using BBP for values > 128. Make that explicit and allow
debugfs access.

Signed-off-by: Anisse Astier <anisse@astier.eu>
Acked-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
5 years agoASoC: alc5632: sync with mainline
Marc Dietrich [Fri, 6 Jul 2012 19:15:49 +0000 (21:15 +0200)]
ASoC: alc5632: sync with mainline

I don't know where this I2S Pwr stuff came from.

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
5 years agoASoC: tegra+alc5632: Swap MICBIAS1 and MIC1 dapm widgets
Andrey Danin [Fri, 25 May 2012 20:27:02 +0000 (00:27 +0400)]
ASoC: tegra+alc5632: Swap MICBIAS1 and MIC1 dapm widgets

MICBIAS1 is internal widget. If MICBIAS1 is connected to board driver
instead of MIC1 asoc core enables widgets even if capture stream isn't
active

Signed-off-by: Andrey Danin <danindrey@mail.ru>
Conflicts:

sound/soc/codecs/alc5632.c

5 years agoASoC: alc5632: Convert to devm_regmap_init_i2c()
Axel Lin [Tue, 8 May 2012 15:44:55 +0000 (23:44 +0800)]
ASoC: alc5632: Convert to devm_regmap_init_i2c()

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
5 years agoASoC: alc5632: Convert to module_i2c_driver()
Axel Lin [Tue, 8 May 2012 15:43:33 +0000 (23:43 +0800)]
ASoC: alc5632: Convert to module_i2c_driver()

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
5 years agoASoC: alc5632: Fix I2S digital interface power for recording
Andrey Danin [Sat, 11 Feb 2012 21:26:34 +0000 (23:26 +0200)]
ASoC: alc5632: Fix I2S digital interface power for recording

Fix I2S digital interface power for recording.

Signed-off-by: Andrey Danin <danindrey@mail.ru>
Signed-off-by: Leon Romanovsky <leon@leon.nu>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
5 years agoASoC: alc5632: Refactored DAPM routes to add voice support
Andrey Danin [Sat, 11 Feb 2012 21:26:32 +0000 (23:26 +0200)]
ASoC: alc5632: Refactored DAPM routes to add voice support

Refactored DAPM routes to add voice support.

- Added undocumented register
- Used AIF in/out
- Added missed voice items and routes
- Added DMIC input
- Romoved unrelevant items

Signed-off-by: Andrey Danin <danindrey@mail.ru>
Signed-off-by: Leon Romanovsky <leon@leon.nu>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Conflicts:

sound/soc/codecs/alc5632.c

5 years agoASoC: Add __devinit/__devexit annotations at necessary places
Axel Lin [Mon, 23 Jan 2012 03:18:17 +0000 (11:18 +0800)]
ASoC: Add __devinit/__devexit annotations at necessary places

Fix below build warning when CONFIG_HOTPLUG is not set.

  CC      sound/soc/codecs/alc5623.o
sound/soc/codecs/alc5623.c:1062: warning: â\80\98alc5623_i2c_removeâ\80\99 defined but not used
  CC      sound/soc/codecs/alc5632.o
sound/soc/codecs/alc5632.c:1112: warning: â\80\98alc5632_i2c_removeâ\80\99 defined but not used

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Acked-by: Leon Romanovsky <leon@leon.nu>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
5 years agoASoC: Constify snd_soc_dai_ops structs
Marc Dietrich [Fri, 6 Jul 2012 16:17:14 +0000 (18:17 +0200)]
ASoC: Constify snd_soc_dai_ops structs

From: Author: Lars-Peter Clausen <lars@metafoo.de>

upstream: 85e7652d89293a6dab42bfd31f276f8bc072d4c5

Commit 1ee46ebd("ASoC: Make the DAI ops constant in the DAI structure")
introduced the possibility to have constant DAI ops structures, yet this is
arley used in both existing drivers and also new drivers being submitted,
although none of them modifies its DAI ops structure. The later is not
urprising since existing drivers are often used as templates for new drivers.
So this patch just constifies all existing snd_soc_dai_ops structs to eliminate
the issue altogether.

The patch was generated with the following coccinelle semantic patch:
// <smpl>
@@
identifier ops;
@@
-struct snd_soc_dai_ops ops =
+const struct snd_soc_dai_ops ops =
{ ... };
// </smpl>

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
5 years agoASoC: alc5632: Remove volatile registers from regmap defaults
Leon Romanovsky [Thu, 17 Nov 2011 16:48:42 +0000 (18:48 +0200)]
ASoC: alc5632: Remove volatile registers from regmap defaults

There is no need to provide defaults for the volatile
registers and doing so might cause confusion.

Signed-off-by: Leon Romanovsky <leon@leon.nu>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
5 years agoASoC: alc5632: Update of i2c_probe function to use regmap API only
Leon Romanovsky [Thu, 17 Nov 2011 10:01:30 +0000 (12:01 +0200)]
ASoC: alc5632: Update of i2c_probe function to use regmap API only

Signed-off-by: Leon Romanovsky <leon@leon.nu>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Conflicts:

sound/soc/codecs/alc5632.c

5 years agoASoC: alc5632: Remove unrelevant registers and name the relevant
Leon Romanovsky [Thu, 17 Nov 2011 10:01:28 +0000 (12:01 +0200)]
ASoC: alc5632: Remove unrelevant registers and name the relevant

Signed-off-by: Leon Romanovsky <leon@leon.nu>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
5 years agoASoC: Convert ALC5632 codec to use regmap API
Leon Romanovsky [Wed, 16 Nov 2011 10:06:58 +0000 (12:06 +0200)]
ASoC: Convert ALC5632 codec to use regmap API

Signed-off-by: Leon Romanovsky <leon@leon.nu>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Conflicts:

sound/soc/codecs/alc5632.c

5 years agoARM: tegra: paz00: disable usb otg for now
Marc Dietrich [Fri, 6 Jul 2012 13:56:41 +0000 (15:56 +0200)]
ARM: tegra: paz00: disable usb otg for now

This prevents resume and needs fixing. I theory, we could have our
mini usb network back - volunteers ?

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
5 years agostaging: add zsmalloc to Kconfig/Makefile
Seth Jennings [Mon, 9 Jan 2012 22:51:57 +0000 (16:51 -0600)]
staging: add zsmalloc to Kconfig/Makefile

Adds the new zsmalloc library to the staging Kconfig and Makefile

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agox86: add local_tlb_flush_kernel_range()
Seth Jennings [Mon, 25 Jun 2012 16:14:38 +0000 (11:14 -0500)]
x86: add local_tlb_flush_kernel_range()

This patch adds support for a local_tlb_flush_kernel_range()
function for the x86 arch.  This function allows for CPU-local
TLB flushing, potentially using invlpg for single entry flushing,
using an arch independent function name.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
5 years agozsmalloc: add generic path and remove x86 dependency
Seth Jennings [Mon, 25 Jun 2012 16:14:37 +0000 (11:14 -0500)]
zsmalloc: add generic path and remove x86 dependency

This patch adds generic pages mapping methods that
work on all archs in the absence of support for
local_tlb_flush_kernel_range() advertised by the
arch through __HAVE_LOCAL_TLB_FLUSH_KERNEL_RANGE

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
5 years agostaging: zram/zcache: swtich Kconfig dependency from X86 to ZSMALLOC
Seth Jennings [Mon, 25 Jun 2012 16:14:36 +0000 (11:14 -0500)]
staging: zram/zcache: swtich Kconfig dependency from X86 to ZSMALLOC

This patch switches zcache and zram dependency to ZSMALLOC
rather than X86.  There is no net change since ZSMALLOC
depends on X86, however, this prevent further changes to
these files as zsmalloc dependencies change.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zsmalloc: Finish conversion to a separate module
Ben Hutchings [Wed, 20 Jun 2012 01:31:11 +0000 (02:31 +0100)]
staging: zsmalloc: Finish conversion to a separate module

ZSMALLOC is tristate, but the code has no MODULE_LICENSE and since it
depends on GPL-only symbols it cannot be loaded as a module.  This in
turn breaks zram which now depends on it.  I assume it's meant to be
Dual BSD/GPL like the other z-stuff.

There is also no module_exit, which will make it impossible to unload.
Add the appropriate module_init and module_exit declarations suggested
by comments.

Reported-by: Christian Ohm <chr.ohm@gmx.net>
References: http://bugs.debian.org/677273
Cc: stable@vger.kernel.org # v3.4
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agomm/frontswap: cleanup doc and comment error
Wanpeng Li [Sat, 16 Jun 2012 12:37:48 +0000 (20:37 +0800)]
mm/frontswap: cleanup doc and comment error

Signed-off-by: Wanpeng Li <liwp.linux@gmail.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
5 years agostaging: zcache: don't limit number of pools per client
Sasha Levin [Wed, 6 Jun 2012 19:05:48 +0000 (21:05 +0200)]
staging: zcache: don't limit number of pools per client

Currently the amount of pools each client can use is limited to 16, this is
and arbitrary limit which isn't really required by current implementation.

This places and arbitrary limit on the number of mounted filesystems that
can use cleancache.

This patch removes that limit and uses IDR to do sparse mapping of pools
in each client.

Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zsmalloc: fix uninit'ed variable warning
Seth Jennings [Wed, 13 Jun 2012 21:03:42 +0000 (16:03 -0500)]
staging: zsmalloc: fix uninit'ed variable warning

This patch fixes an uninitialized variable warning in
alloc_zspage().  It also fixes the secondary issue of
prev_page leaving scope on each loop iteration.  The only
reason this ever worked was because prev_page was occupying
the same space on the stack on each iteration.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Reported-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agomm: frontswap: remove unneeded headers
Marc Dietrich [Sat, 30 Jun 2012 21:22:34 +0000 (23:22 +0200)]
mm: frontswap: remove unneeded headers

Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Conflicts:

mm/frontswap.c

5 years agomm: frontswap: split out function to clear a page out
Sasha Levin [Sun, 10 Jun 2012 10:51:07 +0000 (12:51 +0200)]
mm: frontswap: split out function to clear a page out

Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
5 years agomm: frontswap: remove unnecessary check during initialization
Sasha Levin [Sun, 10 Jun 2012 10:51:05 +0000 (12:51 +0200)]
mm: frontswap: remove unnecessary check during initialization

The check whether frontswap is enabled or not is done in the API functions in
the frontswap header, before they are passed to the internal
double-underscored frontswap functions.

Remove the check from __frontswap_init for consistency.

Reviewed-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
5 years agomm: frontswap: make all branches of if statement in put page consistent
Sasha Levin [Sun, 10 Jun 2012 10:51:04 +0000 (12:51 +0200)]
mm: frontswap: make all branches of if statement in put page consistent

Currently it has a complex structure where different things are compared
at each branch. Simplify that and make both branches look similar.

Reviewed-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
5 years agomm: frontswap: split frontswap_shrink further to simplify locking
Sasha Levin [Sun, 10 Jun 2012 10:51:03 +0000 (12:51 +0200)]
mm: frontswap: split frontswap_shrink further to simplify locking

Split frontswap_shrink to simplify the locking in the original code.

Also, assert that the function that was split still runs under the
swap spinlock.

Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
5 years agomm: frontswap: split out __frontswap_unuse_pages
Sasha Levin [Sun, 10 Jun 2012 10:51:02 +0000 (12:51 +0200)]
mm: frontswap: split out __frontswap_unuse_pages

An attempt at making frontswap_shrink shorter and more readable. This patch
splits out walking through the swap list to find an entry with enough
pages to unuse.

Also, assert that the internal __frontswap_unuse_pages is called under swap
lock, since that part of code was previously directly happen inside the lock.

Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
5 years agomm: frontswap: split out __frontswap_curr_pages
Sasha Levin [Sun, 10 Jun 2012 10:51:01 +0000 (12:51 +0200)]
mm: frontswap: split out __frontswap_curr_pages

Code was duplicated in two functions, clean it up.

Also, assert that the deduplicated code runs under the swap spinlock.

Reviewed-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
5 years agomm: frontswap: trivial coding convention issues
Sasha Levin [Sun, 10 Jun 2012 10:51:00 +0000 (12:51 +0200)]
mm: frontswap: trivial coding convention issues

Reviewed-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
5 years agomm: frontswap: remove casting from function calls through ops structure
Marc Dietrich [Sat, 30 Jun 2012 21:19:13 +0000 (23:19 +0200)]
mm: frontswap: remove casting from function calls through ops structure

Removes unneeded casts.

Reviewed-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Conflicts:

mm/frontswap.c

Conflicts:

mm/frontswap.c

5 years agostaging: zram: conventions, __aligned() attribute
Sam Hansen [Thu, 7 Jun 2012 23:03:48 +0000 (16:03 -0700)]
staging: zram: conventions, __aligned() attribute

Using the __aligned() attribute in favor of __attribute__((aligned(size)))

Signed-off-by: Sam Hansen <solid.se7en@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zram: conventions pr_warning -> pr_warn()
Sam Hansen [Thu, 7 Jun 2012 23:03:47 +0000 (16:03 -0700)]
staging: zram: conventions pr_warning -> pr_warn()

Porting zram to use the pr_warn() function instead of the deprecated
pr_warning().

Signed-off-by: Sam Hansen <solid.se7en@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zsmalloc documentation
Nitin Gupta [Sun, 10 Jun 2012 00:41:14 +0000 (17:41 -0700)]
staging: zsmalloc documentation

Documentation of various struct page fields
used by zsmalloc.

Changes for v2:
- Regroup descriptions as suggested by Konrad

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reviewed-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zram: fix random data read
Minchan Kim [Fri, 8 Jun 2012 06:39:26 +0000 (15:39 +0900)]
staging: zram: fix random data read

fd1a30de makes a bug that it uses (struct page *) as zsmalloc's handle
although it's a uncompressed page so that it can access random page,
return random data or even crashed by get_first_page in zs_map_object.

Cc: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Jerome Marchand <jmarchan@redhat.com>
Signed-off-by: Minchan Kim <minchan@kernel.org>
Acked-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Conflicts:

drivers/staging/zram/zram_drv.c

5 years agostaging: zsmalloc: zsmalloc: use unsigned long instead of void *
Minchan Kim [Fri, 8 Jun 2012 06:39:25 +0000 (15:39 +0900)]
staging: zsmalloc: zsmalloc: use unsigned long instead of void *

We should use unsigned long as handle instead of void * to avoid any
confusion. Without this, users may just treat zs_malloc return value as
a pointer and try to deference it.

This patch passed compile test(zram, zcache and ramster) and zram is
tested on qemu.

changelog
  * from v2
- remove hval pointed out by Nitin
- based on next-20120607
  * from v1
- change zcache's zv_create return value
- baesd on next-20120604

Cc: Dan Magenheimer <dan.magenheimer@oracle.com>
Acked-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Minchan Kim <minchan@kernel.org>
Acked-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Conflicts:

drivers/staging/zram/zram_drv.c

5 years agofrontswap: s/put_page/store/g s/get_page/load
Marc Dietrich [Sat, 30 Jun 2012 20:44:08 +0000 (22:44 +0200)]
frontswap: s/put_page/store/g s/get_page/load

Sounds so much more natural.

Suggested-by: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Modified-By: Marc Dietrich <marvin24@gmx.de>
5 years agomm: frontswap: core swap subsystem hooks and headers
Dan Magenheimer [Mon, 9 Apr 2012 23:08:06 +0000 (17:08 -0600)]
mm: frontswap: core swap subsystem hooks and headers

This patch, 2of4, contains the changes to the core swap subsystem.
This includes:

(1) makes available core swap data structures (swap_lock, swap_list and
swap_info) that are needed by frontswap.c but we don't need to expose them
to the dozens of files that include swap.h so we create a new swapfile.h
just to extern-ify these and modify their declarations to non-static

(2) adds frontswap-related elements to swap_info_struct.  Frontswap_map
points to vzalloc'ed one-bit-per-swap-page metadata that indicates
whether the swap page is in frontswap or in the device and frontswap_pages
counts how many pages are in frontswap.

(3) adds hooks in the swap subsystem and extends try_to_unuse so that
frontswap_shrink can do a "partial swapoff".

Note that a failed frontswap_map allocation is safe... failure is noted
by lack of "FS" in the subsequent printk.

---

[v14: rebase to 3.4-rc2]
[v10: no change]
[v9: akpm@linux-foundation.org: mark some statics __read_mostly]
[v9: akpm@linux-foundation.org: add clarifying comments]
[v9: akpm@linux-foundation.org: no need to loop repeating try_to_unuse]
[v9: error27@gmail.com: remove superfluous check for NULL]
[v8: rebase to 3.0-rc4]
[v8: kamezawa.hiroyu@jp.fujitsu.com: change counter to atomic_t to avoid races]
[v8: kamezawa.hiroyu@jp.fujitsu.com: comment to clarify informational counters]
[v7: rebase to 3.0-rc3]
[v7: JBeulich@novell.com: add new swap struct elements only if config'd]
[v6: rebase to 3.0-rc1]
[v6: lliubbo@gmail.com: fix null pointer deref if vzalloc fails]
[v6: konrad.wilk@oracl.com: various checks and code clarifications/comments]
[v5: no change from v4]
[v4: rebase to 2.6.39]
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Reviewed-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Jan Beulich <JBeulich@novell.com>
Acked-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nitin Gupta <ngupta@vflare.org>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Rik Riel <riel@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
[v11: Rebased, fixed mm/swapfile.c context change]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Conflicts:

mm/swapfile.c

5 years agomm: frontswap: config and doc files
Dan Magenheimer [Mon, 9 Apr 2012 23:10:34 +0000 (17:10 -0600)]
mm: frontswap: config and doc files

This patch 4of4 adds configuration and documentation files including a FAQ.

[v14: updated docs/FAQ to use zcache and RAMster as examples]
[v10: no change]
[v9: akpm@linux-foundation.org: sysfs->debugfs; no longer need Doc/ABI file]
[v8: rebase to 3.0-rc4]
[v7: rebase to 3.0-rc3]
[v6: rebase to 3.0-rc1]
[v5: change config default to n]
[v4: rebase to 2.6.39]
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Acked-by: Jan Beulich <JBeulich@novell.com>
Acked-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nitin Gupta <ngupta@vflare.org>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Rik Riel <riel@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
5 years agomm: frontswap: core frontswap functionality
Dan Magenheimer [Mon, 9 Apr 2012 23:09:27 +0000 (17:09 -0600)]
mm: frontswap: core frontswap functionality

This patch, 3of4, provides the core frontswap code that interfaces between
the hooks in the swap subsystem and a frontswap backend via frontswap_ops.

---
New file added: mm/frontswap.c

[v14: add support for writethrough, per suggestion by aarcange@redhat.com]
[v11: sjenning@linux.vnet.ibm.com: s/puts/failed_puts/]
[v10: sjenning@linux.vnet.ibm.com: fix debugfs calls on 32-bit]
[v9: akpm@linux-foundation.org: change "flush" to "invalidate", part 1]
[v9: akpm@linux-foundation.org: mark some statics __read_mostly]
[v9: akpm@linux-foundation.org: add clarifying comments]
[v9: akpm@linux-foundation.org: no need to loop repeating try_to_unuse]
[v9: error27@gmail.com: remove superfluous check for NULL]
[v8: rebase to 3.0-rc4]
[v8: kamezawa.hiroyu@jp.fujitsu.com: add comment to clarify find_next_to_unuse]
[v7: rebase to 3.0-rc3]
[v7: JBeulich@novell.com: use new static inlines, no-ops if not config'd]
[v6: rebase to 3.1-rc1]
[v6: lliubbo@gmail.com: use vzalloc]
[v6: lliubbo@gmail.com: fix null pointer deref if vzalloc fails]
[v6: konrad.wilk@oracl.com: various checks and code clarifications/comments]
[v4: rebase to 2.6.39]
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Acked-by: Jan Beulich <JBeulich@novell.com>
Acked-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nitin Gupta <ngupta@vflare.org>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Rik Riel <riel@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
[v12: Squashed s/flush/invalidate/ in]
[v15: A bit of cleanup and seperate DEBUGFS]
Signed-off-by: Konrad Wilk <konrad.wilk@oracle.com>
5 years agomm: frontswap: add frontswap header file
Dan Magenheimer [Mon, 9 Apr 2012 23:06:54 +0000 (17:06 -0600)]
mm: frontswap: add frontswap header file

Frontswap is the alter ego of cleancache, the "yang" to cleancache's
"yin"... and more precisely frontswap is the provider of anonymous
pages to transcendent memory to nicely complement cleancache's providing
of clean pagecache pages to transcendent memory.  For optimal use
of transcendent memory, both are necessary... because a kernel
under memory pressure first reclaims clean pagecache pages and,
when under more memory pressure, starts swapping anonymous pages.

Frontswap and cleancache (which was merged at 3.0) are the "frontends"
and the only necessary changes to the core kernel for transcendent memory;
all other supporting code -- the "backends" -- is implemented as drivers.
See the LWN.net article "Transcendent memory in a nutshell" for a detailed
overview of frontswap and related kernel parts:
https://lwn.net/Articles/454795/

Frontswap code was first posted publicly in January 2009 and on LKML in
May 2009, and has remained functionally stable for nearly three years now.
It is barely invasive, touching only the swap subsystem and adds less
than 100 lines of code to existing swap subsystem code files.
It has improved syntactically substantially between V1 and this posting
of V14, thanks to the review of a few kernel developers, and has adapted
easily to at least one major swap subsystem change.  As of 3.4, there are
three in-tree users of frontswap patiently waiting for this patchset and
for CONFIG_FRONTSWAP to be enabled: zcache (staging driver merged at
2.6.39), Xen tmem (merged at 3.0 and 3.1) and RAMster (staging driver
merged at 3.4).  In addition, a RFC has been posted for a KVM backend.
The frontswap patchset has been in linux-next since next-110603.  Earlier
versions of frontswap already ship in the Oracle Unbreakable Enterprise Kernel
and SuSE SLES.

This patch, 1of4, provides the header file for the core code for frontswap
that interfaces between the hooks in the swap subsystem and a frontswap
backend via frontswap_ops.
---
New file added: include/linux/frontswap.h

[v14: add support for writethrough, per suggestion by aarcange@redhat.com]
[v14: rebase to 3.4-rc2]
[v11: konrad.wilk@oracle.com: squashed s/flush/invalidate/ in]
[v10: no change]
[v9: akpm@linux-foundation.org: change "flush" to "invalidate", part 1]
[v8: rebase to 3.0-rc4]
[v7: rebase to 3.0-rc3]
[v7: JBeulich@novell.com: new static inlines resolve to no-ops if not config'd]
[v7: JBeulich@novell.com: avoid redundant shifts/divides for *_bit lib calls]
[v6: rebase to 3.1-rc1]
[v5: no change from v4]
[v4: rebase to 2.6.39]
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Jan Beulich <JBeulich@novell.com>
Acked-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nitin Gupta <ngupta@vflare.org>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Rik Riel <riel@redhat.com>
[v15: int/bool on some functions]
Signed-off-by: Konrad Wilk <konrad.wilk@oracle.com>
5 years agostaging: zsmalloc: add/fix function comment
Minchan Kim [Thu, 3 May 2012 06:40:40 +0000 (15:40 +0900)]
staging: zsmalloc: add/fix function comment

Add/fix the comment.

Signed-off-by: Minchan Kim <minchan@kernel.org>
Acked-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zsmalloc: rename zspage_order with zspage_pages
Minchan Kim [Thu, 3 May 2012 06:40:39 +0000 (15:40 +0900)]
staging: zsmalloc: rename zspage_order with zspage_pages

zspage_order defines how many pages are needed to make a zspage.
So _order_ is rather awkward naming. It already deceive Jonathan
- http://lwn.net/Articles/477067/
" For each size, the code calculates an optimum number of pages (up to 16)"

Let's change from _order_ to _pages_ and some function names.

Signed-off-by: Minchan Kim <minchan@kernel.org>
Acked-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agozsmalloc: use PageFlag macro instead of [set|test]_bit
Minchan Kim [Wed, 25 Apr 2012 06:23:09 +0000 (15:23 +0900)]
zsmalloc: use PageFlag macro instead of [set|test]_bit

MM code always uses PageXXX to handle page flags.
Let's keep the consistency.

Signed-off-by: Minchan Kim <minchan@kernel.org>
Acked-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zcache: fix Kconfig crypto dependency
Seth Jennings [Tue, 24 Apr 2012 01:33:50 +0000 (20:33 -0500)]
staging: zcache: fix Kconfig crypto dependency

ZCACHE is a boolean in the Kconfig.  When selected, it
should require that CRYPTO be builtin (=y).

Currently, ZCACHE=y and CRYPTO=m is a valid configuration
when it should not be.

This patch changes the zcache Kconfig to enforce this
dependency.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zsmalloc: fix memory leak
Nitin Gupta [Mon, 2 Apr 2012 14:13:56 +0000 (09:13 -0500)]
staging: zsmalloc: fix memory leak

This patch fixes a memory leak in zsmalloc where the first
subpage of each zspage is leaked when the zspage is freed.

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Acked-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging/zmem: Use lockdep_assert_held instead of spin_is_locked
Andi Kleen [Fri, 16 Mar 2012 19:00:57 +0000 (12:00 -0700)]
staging/zmem: Use lockdep_assert_held instead of spin_is_locked

WARN_ON(!spin_is_locked()) will always trigger on UP.
Use lockdep_assert_held instead.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zsmalloc: remove SPARSEMEM dep from Kconfig
Seth Jennings [Mon, 5 Mar 2012 17:33:24 +0000 (11:33 -0600)]
staging: zsmalloc: remove SPARSEMEM dep from Kconfig

This patch removes the SPARSEMEM from the zsmalloc
Kconfig

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zsmalloc: change ZS_MIN_ALLOC_SIZE
Seth Jennings [Mon, 5 Mar 2012 17:33:23 +0000 (11:33 -0600)]
staging: zsmalloc: change ZS_MIN_ALLOC_SIZE

This patch ensures that the value of ZS_MIN_ALLOC_SIZE, for the
PAGE_SIZE and MAX_PHYSMEM_BITS on the system, allows for all
possible object ids in the lowest storage class to be encoded
in the object handle.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zsmalloc: calculate MAX_PHYSMEM_BITS if not defined
Seth Jennings [Mon, 5 Mar 2012 17:33:22 +0000 (11:33 -0600)]
staging: zsmalloc: calculate MAX_PHYSMEM_BITS if not defined

This patch provides a way to determine or "set a
reasonable value for" MAX_PHYSMEM_BITS in the case that
it is not defined (i.e. !SPARSEMEM)

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zsmalloc: add ZS_MAX_PAGES_PER_ZSPAGE
Seth Jennings [Mon, 5 Mar 2012 17:33:21 +0000 (11:33 -0600)]
staging: zsmalloc: add ZS_MAX_PAGES_PER_ZSPAGE

This patch moves where max_zspage_order is declared and
changes its meaning.  "Order" typically implies 2^order
of something; however, it is currently being used as the
"maximum number of single pages in a zspage".  To add clarity,
ZS_MAX_ZSPAGE_ORDER is now used to calculate ZS_MAX_PAGES_PER_ZSPAGE,
which is 2^ZS_MAX_ZSPAGE_ORDER and is the upper bound on the number
of pages in a zspage.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zsmalloc: move object/handle masking defines
Seth Jennings [Mon, 5 Mar 2012 17:33:20 +0000 (11:33 -0600)]
staging: zsmalloc: move object/handle masking defines

This patch moves the definitions of _PFN_BITS, OBJ_INDEX_BITS
and OBJ_INDEX_MASK from zsmalloc-main.c to zsmalloc_int.h

They will be needed to determine ZS_MIN_ALLOC_SIZE in the next
patch

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zcache: make zcache builtin only
Seth Jennings [Mon, 5 Mar 2012 16:46:38 +0000 (10:46 -0600)]
staging: zcache: make zcache builtin only

zcache cannot currently be loaded as a module.  However
the Kconfig allows it to be built as a module; something that
the user probably does not intend since the module is not
loadable.

This patch switches zcache from a tristate to a bool in the Kconfig

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zcache: fix memory corruption bug
Seth Jennings [Tue, 28 Feb 2012 22:02:23 +0000 (16:02 -0600)]
staging: zcache: fix memory corruption bug

This patch fixes a bug where the zv code writes before the allocated
buffer, resulting in system memory corruption. This was introduced
during the switch from xvmalloc to zsmalloc.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zcache: fix length type mismatch
Seth Jennings [Tue, 28 Feb 2012 22:01:37 +0000 (16:01 -0600)]
staging: zcache: fix length type mismatch

This fixes a type mismatch in the compression code where
a size_t pointer was cast to a unsigned int pointer.  On
little endian archs, there is no issue.  However on big
endian archs, the value is incorrect, taking the high
order bits and truncating the lower order bits.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zcache: avoid AB-BA deadlock condition
Andrea Righi [Mon, 20 Feb 2012 12:11:49 +0000 (13:11 +0100)]
staging: zcache: avoid AB-BA deadlock condition

Commit 9256a47 fixed a deadlock condition, being sure that the buddy
list spinlock is always taken before the page spinlock.

However in zbud_free_and_delist() locking order is the opposite
(page lock -> list lock).

Possible unsafe locking scenario (reported by lockdep):

        CPU0                    CPU1
        ----                    ----
   lock(&(&zbpg->lock)->rlock);
                                lock(zbud_budlists_spinlock);
                                lock(&(&zbpg->lock)->rlock);
   lock(zbud_budlists_spinlock);

Fix by grabbing the locks in opposite order in zbud_free_and_delist().

Signed-off-by: Andrea Righi <andrea@betterlinux.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zram: Rename module parameter
Nitin Gupta [Mon, 13 Feb 2012 04:04:45 +0000 (23:04 -0500)]
staging: zram: Rename module parameter

zram accepts number of devices to be created
as a module parameter. This was renamed from
num_devices to zram_num_devices (without updating
the documentation!) since num_devices was declared
as a non-static global variable, polluting the global
namespace. Now, we declare it as a static variable
and revert back the name change.

The documentation (zram.txt) already mentions
num_devices as the module parameter name.

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: fix powerpc linux-next break on zsmalloc
Seth Jennings [Mon, 13 Feb 2012 14:47:49 +0000 (08:47 -0600)]
staging: fix powerpc linux-next break on zsmalloc

linux/vmalloc.h added to zsmalloc-main.c to resolve implicit
declaration errors.

X86 dependency added to zsmalloc and dependent drivers zcache and zram.

This X86 only requirement is not ideal.  Working to find portable
functions for __flush_tlb_one and set_pte.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zcache: replace xvmalloc with zsmalloc
Seth Jennings [Thu, 9 Feb 2012 18:25:03 +0000 (12:25 -0600)]
staging: zcache: replace xvmalloc with zsmalloc

Replaces xvmalloc with zsmalloc as the persistent memory allocator
for zcache

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zram: remove xvmalloc
Nitin Gupta [Mon, 9 Jan 2012 22:52:00 +0000 (16:52 -0600)]
staging: zram: remove xvmalloc

Removes the xvmalloc allocator code from the zram driver

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Acked-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zram: replace xvmalloc with zsmalloc
Nitin Gupta [Mon, 9 Jan 2012 22:51:59 +0000 (16:51 -0600)]
staging: zram: replace xvmalloc with zsmalloc

Replaces xvmalloc with zsmalloc as the compressed page allocator
for zram

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Acked-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zsmalloc: zsmalloc memory allocation library
Nitin Gupta [Mon, 9 Jan 2012 22:51:56 +0000 (16:51 -0600)]
staging: zsmalloc: zsmalloc memory allocation library

This patch creates a new memory allocation library named
zsmalloc.

NOTE: zsmalloc currently depends on SPARSEMEM for the MAX_PHYSMEM_BITS
value needed to determine the format of the object handle. There may
be a better way to do this.  Feedback is welcome.

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zcache: crypto API support
Seth Jennings [Tue, 3 Jan 2012 22:31:34 +0000 (16:31 -0600)]
staging: zcache: crypto API support

This patch allow zcache to use the crypto API for page compression.
It replaces the direct LZO compress/decompress calls with calls
into the crypto compression API. The compressor to be used is
specified in the kernel boot line with the zcache parameter like:
zcache=lzo or zcache=deflate.  If the specified compressor can't
be loaded, zcache uses lzo as the default compressor.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: zcache: fix serialization bug in zv stats
Seth Jennings [Fri, 30 Dec 2011 16:42:15 +0000 (10:42 -0600)]
staging: zcache: fix serialization bug in zv stats

In a multithreaded workload, the zv_curr_dist_counts
and zv_cumul_dist_counts statistics are being corrupted
because the increments and decrements in zv_create
and zv_free are not atomic.

This patch converts these statistics and their corresponding
increments/decrements/reads to atomic operations.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agozcache: Set SWIZ_BITS to 8 to reduce tmem bucket lock contention.
Dan Magenheimer [Mon, 23 Jan 2012 21:52:20 +0000 (16:52 -0500)]
zcache: Set SWIZ_BITS to 8 to reduce tmem bucket lock contention.

SWIZ_BITS > 8 results in a much larger number of "tmem_obj"
allocations, likely one per page-placed-in-frontswap.  The
tmem_obj is not huge (roughly 100 bytes), but it is large
enough to add a not-insignificant memory overhead to zcache.

The SWIZ_BITS=8  will get roughly the same lock contention
without the space wastage.

The effect of SWIZ_BITS can be thought of as "2^SWIZ_BITS is
the number of unique oids that be generated" (This concept is
limited to frontswap's use of tmem).

Acked-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
5 years agozcache: fix deadlock condition
Dan Magenheimer [Wed, 25 Jan 2012 22:32:51 +0000 (14:32 -0800)]
zcache: fix deadlock condition

I discovered this deadlock condition awhile ago working on RAMster
but it affects zcache as well.  The list spinlock must be
locked prior to the page spinlock and released after.  As
a result, the page copy must also be done while the locks are held.

Applies to 3.2.  Konrad, please push (via GregKH?)...
this is definitely a bug fix so need not be pushed during
a -rc0 window.

Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
5 years agomm: cleancache: Use __read_mostly as appropiate.
Dan Magenheimer [Wed, 21 Sep 2011 16:21:20 +0000 (12:21 -0400)]
mm: cleancache: Use __read_mostly as appropiate.

The values are rarely changed so might as well put them in the
appropiate section.

Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Signed-off-by: Konrad Wilk <konrad.wilk@oracle.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Jan Beulich <JBeulich@novell.com>
Acked-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nitin Gupta <ngupta@vflare.org>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Rik Riel <riel@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
5 years agomm: cleancache: report statistics via debugfs instead of sysfs.
Dan Magenheimer [Wed, 21 Sep 2011 16:28:04 +0000 (12:28 -0400)]
mm: cleancache: report statistics via debugfs instead of sysfs.

[v9: akpm@linux-foundation.org: sysfs->debugfs; no longer need Doc/ABI file]

Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Signed-off-by: Konrad Wilk <konrad.wilk@oracle.com>
Cc: Jan Beulich <JBeulich@novell.com>
Acked-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nitin Gupta <ngupta@vflare.org>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Rik Riel <riel@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
5 years agomm: zcache/tmem/cleancache: s/flush/invalidate/
Dan Magenheimer [Thu, 12 Jan 2012 19:03:25 +0000 (14:03 -0500)]
mm: zcache/tmem/cleancache: s/flush/invalidate/

Complete the renaming from "flush" to "invalidate" across
both tmem frontends (cleancache and frontswap) and both tmem backends
(Xen and zcache), as required by akpm.

This change is completely cosmetic.

[v10: no change]
[v9: akpm@linux-foundation.org: change "flush" to "invalidate", part 3]
Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Jan Beulich <JBeulich@novell.com>
Acked-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nitin Gupta <ngupta@vflare.org>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Rik Riel <riel@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
[v11: Remove the frontswap part]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
5 years agomm: cleancache: s/flush/invalidate/
Dan Magenheimer [Wed, 21 Sep 2011 15:56:28 +0000 (11:56 -0400)]
mm: cleancache: s/flush/invalidate/

Per akpm suggestions alter the use of the term flush to be
invalidate. The next patch will do this across all MM.

This change is completely cosmetic.

[v9: akpm@linux-foundation.org: change "flush" to "invalidate", part 3]

Signed-off-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Jan Beulich <JBeulich@novell.com>
Reviewed-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nitin Gupta <ngupta@vflare.org>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Rik Riel <riel@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
[v10: Fixed  fs: move code out of buffer.c conflict change]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Conflicts:

fs/block_dev.c

5 years agoStaging: zram: Add a missing GFP_KERNEL specifier in zram_init_device()
Jerome Marchand [Wed, 30 Nov 2011 13:16:16 +0000 (14:16 +0100)]
Staging: zram: Add a missing GFP_KERNEL specifier in zram_init_device()

The allocation of zram->compress_buffer is misssing a GFP_* specifier.
This is equivalent to GFP_NOWAIT but it is more likely a omission.
Since the allocation just above it uses GFP_KERNEL, there is no reason
to use GFP_NOWAIT here. Therefore, add GFP_KERNEL.

Signed-off-by: Jerome Marchand <jmarchan@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
5 years agoStaging: zram/zram_sysfs.c: Fixed call of obsolete function strict_strtoX
Sergey Datsevich [Wed, 7 Dec 2011 18:06:37 +0000 (19:06 +0100)]
Staging: zram/zram_sysfs.c: Fixed call of obsolete function strict_strtoX

As reported by checkpatch.pl strict_strtoX is obsolet and should be
replaced by kstrtoX.

Signed-off-by: Sergey Datsevich <srgdts@gmail.com>
Signed-off-by: Bjoern Meier <bjoernmeier@hotmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
5 years agoStaging: zcache: Fix calls to obsolete function
Bernhard Heinloth [Wed, 23 Nov 2011 15:39:53 +0000 (16:39 +0100)]
Staging: zcache: Fix calls to obsolete function

Function "strict_strtol" replaced by "kstrtol" as suggested by the checkpatch script

Signed-off-by: Bernhard Heinloth <bernhard@heinloth.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
5 years agostaging: zcache: remove zcache_direct_reclaim_lock
Seth Jennings [Wed, 12 Oct 2011 19:41:00 +0000 (14:41 -0500)]
staging: zcache: remove zcache_direct_reclaim_lock

zcache_do_preload() currently does a spin_trylock() on the
zcache_direct_reclaim_lock. Holding this lock intends to prevent
shrink_zcache_memory() from evicting zbud pages as a result
of a preload.

However, it also prevents two threads from
executing zcache_do_preload() at the same time.  The first
thread will obtain the lock and the second thread's spin_trylock()
will fail (an aborted preload) causing the page to be either lost
(cleancache) or pushed out to the swap device (frontswap). It
also doesn't ensure that the call to shrink_zcache_memory() is
on the same thread as the call to zcache_do_preload().

Additional, there is no need for this mechanism because all
zcache_do_preload() calls that come down from cleancache already
have PF_MEMALLOC set in the process flags which prevents
direct reclaim in the memory manager. If the zcache_do_preload()
call is done from the frontswap path, we _want_ reclaim to be
done (which it isn't right now).

This patch removes the zcache_direct_reclaim_lock and related
statistics in zcache.

Based on v3.1-rc8

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Reviewed-by: Dave Hansen <dave@linux.vnet.ibm.com>
Acked-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
5 years agostaging: zcache: reduce tmem bucket lock contention
Seth Jennings [Tue, 4 Oct 2011 13:21:32 +0000 (08:21 -0500)]
staging: zcache: reduce tmem bucket lock contention

tmem uses hash buckets each with their own rbtree and lock to
quickly lookup tmem objects.  tmem has TMEM_HASH_BUCKETS (256)
buckets per pool.  However, because of the way the tmem_oid is
generated for frontswap pages, only 16 unique tmem_oids are being
generated, resulting in only 16 of the 256 buckets being used.
This cause high lock contention for the per bucket locks.

This patch changes SWIZ_BITS to include more bits of the offset.
The result is that all 256 hash buckets are potentially used resulting in a
95% drop in hash bucket lock contention.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
5 years agostaging: zcache: fix crash on cpu remove
Seth Jennings [Thu, 6 Oct 2011 19:28:26 +0000 (14:28 -0500)]
staging: zcache: fix crash on cpu remove

In the case that a cpu is taken offline before zcache_do_preload() is
ever called on the cpu, the per-cpu zcache_preloads structure will
be uninitialized.  In the CPU_DEAD case for zcache_cpu_notifier(),
kp->obj is not checked before calling kmem_cache_free() on it.
If it is NULL, a crash results.

This patch ensures that both kp->obj and kp->page are not NULL before
calling the respective free functions. In practice, just checking
one or the other should be sufficient since they are assigned together
in zcache_do_preload(), but I check both for safety.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Dave Hansen <dave@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
5 years agozram: Fix sparse warnings
Nitin Gupta [Fri, 9 Sep 2011 23:01:00 +0000 (19:01 -0400)]
zram: Fix sparse warnings

Fixes sparse warning:
zram_drv.c:666:6: warning: symbol 'zram_slot_free_notify' was not
declared. Should it be static?

Also, max_zpage_size is now size_t just to be consistent with data-type
of other variables maintaining sizes of various kinds.

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
5 years agostaging: zram: prevent accessing an unallocated table when init fails early
Jerome Marchand [Tue, 6 Sep 2011 13:02:12 +0000 (15:02 +0200)]
staging: zram: prevent accessing an unallocated table when init fails early

When the allocation of zram->table fails, we set zram->disksize to zero
to prevent accessing the unallocated table entries during cleanup.
However, we currently don't take this precaution when the initialization
fails earlier.

Signed-off-by: Jerome Marchand <jmarchan@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
5 years agostaging: zram: fix zram locking
Jerome Marchand [Tue, 6 Sep 2011 13:02:11 +0000 (15:02 +0200)]
staging: zram: fix zram locking

Currently init_lock only prevents concurrent execution of zram_init_device()
and zram_reset_device() but not zram_make_request() nor sysfs store functions.

This patch changes init_lock into a rw_semaphore. A write lock is taken by
init, reset and store functions, a read lock is taken by zram_make_request().
Also, avoids to release the lock before calling __zram_reset_device() for
cleaning after a failed init, thus preventing any concurrent task to see an
inconsistent state of zram.

Signed-off-by: Jerome Marchand <jmarchan@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
5 years agostaging: zcache: fix typos
Seth Jennings [Mon, 22 Aug 2011 18:50:08 +0000 (13:50 -0500)]
staging: zcache: fix typos

The patch fixes two typos in zcache-main.c

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Acked-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>