24 Jun, 2015
2 commits
-
Pull fbdev updates from Tomi Valkeinen:
- ssd1307fb: various fixes and improvements, SSD1305 support
- use architecture agnostic functions instead of MTRR functions in
various fbdev drivers- TI DRA7xx SoC display support (arch/arm/ side)
- OMAPDSS componentization to fix probing order issues
- OMAPDSS scaling fixes
- msm_fb: remove obsoleted driver
* tag 'fbdev-4.2' of git://git.kernel.org/pub/scm/linux/kernel/git/tomba/linux: (77 commits)
msm: msm_fb: Remove dead code
OMAPDSS: HDMI: wait for framedone when stopping video
OMAPDSS: HDMI4: fix error handling
OMAPDSS: DISPC: scaler debug print
OMAPDSS: DISPC: do only y decimation on OMAP3
OMAPDSS: DISPC: check if scaling setup failed
OMAPDSS: DISPC: fix 64 bit issue in 5-tap
OMAPDSS: DISPC: fix row_inc for OMAP3
OMAPDSS: DISPC: add check for scaling limits
OMAPDSS: DISPC: fix check_horiz_timing_omap3 args
OMAPDSS: DISPC: fix predecimation for YUV modes
OMAPDSS: DISPC: work-around for errata i631
OMAPDSS: simplify submodule reg/unreg code
OMAPDSS: componentize omapdss
OMAPDSS: reorder uninit calls
OMAPDSS: remove uses of __init/__exit
OMAPDSS: fix dss_init_ports error handling
OMAPDSS: refactor dss probe function
OMAPDSS: move 'dss_initialized' to dss driver
fbdev: propagate result of fb_videomode_from_videomode()
... -
Pull clkdev updates from Russell King:
"This series addresses some breakage in clkdev caused by a previous
patch set from the clk tree which introduced per-user clk structures.
This basically renamed the existing 'struct clk' to 'struct clk_hw',
and introduced a new 'struct clk'.This change will break anyone using clk_add_alias() with the common
clk code enabled. Thankfully, the intersection of users of
clk_add_alias() and those using the common clk code is practically
zero, but this is something which should be fixed to keep the code
sane.The problem is that clk_add_alias() does this:
r = clk_get(...);
l = clkdev_alloc(r, ...);
clk_put(...);which causes the alias to store a pointer to 'r', which has been
freed.The original patch set tried to work around this problem incorrectly -
at clk_get() time, it tried to convert the struct clk to a struct
clk_hw, and then creating a new struct clk from that. Clearly, if the
original struct clk has been freed, then we have a use-after-free bug.We have other places in the tree which do something similar, so this
series also addresses those locations too.This series addresses this problem by converting clkdev to store and
use the clk_hw pointer. This allows clk_get() to only have to create
it's per-user struct clk from the clk_hw. We can also get to the
desired clk_hw at clk_add_alias() or clk lookup creation time, when
the struct clk is "alive".We also perform some cleanups of the code:
- replacing looped calls to clkdev_add() with clkdev_add_table()
- replacing open-coded lookup allocation (which should have been
using clkdev_alloc()) and subsequent clkdev_add() with
clkdev_create()- replacing open-coded clk_add_alias() with clk_add_alias()"
* 'for-linus-clk' of git://ftp.arm.linux.org.uk/~rmk/linux-arm:
clk: s2mps11: use clkdev_create()
ASoC: migor: use clkdev_create()
ARM: omap2: use clkdev_add_alias()
ARM: omap2: use clkdev_create()
ARM: orion: use clkdev_create()
ARM: lpc32xx: convert to use clkdev_add_table()
SH: use clkdev_add_table()
clkdev: add clkdev_create() helper
clkdev: const-ify connection id to clk_add_alias()
clkdev: get rid of redundant clk_add_alias() prototype in linux/clk.h
clkdev: drop __init from clkdev_add_table()
clk: update clk API documentation to clarify clk_round_rate()
clkdev: use clk_hw internally
23 Jun, 2015
2 commits
-
Pull timer updates from Thomas Gleixner:
"A rather largish update for everything time and timer related:- Cache footprint optimizations for both hrtimers and timer wheel
- Lower the NOHZ impact on systems which have NOHZ or timer migration
disabled at runtime.- Optimize run time overhead of hrtimer interrupt by making the clock
offset updates smarter- hrtimer cleanups and removal of restrictions to tackle some
problems in sched/perf- Some more leap second tweaks
- Another round of changes addressing the 2038 problem
- First step to change the internals of clock event devices by
introducing the necessary infrastructure- Allow constant folding for usecs/msecs_to_jiffies()
- The usual pile of clockevent/clocksource driver updates
The hrtimer changes contain updates to sched, perf and x86 as they
depend on them plus changes all over the tree to cleanup API changes
and redundant code, which got copied all over the place. The y2038
changes touch s390 to remove the last non 2038 safe code related to
boot/persistant clock"* 'timers-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (114 commits)
clocksource: Increase dependencies of timer-stm32 to limit build wreckage
timer: Minimize nohz off overhead
timer: Reduce timer migration overhead if disabled
timer: Stats: Simplify the flags handling
timer: Replace timer base by a cpu index
timer: Use hlist for the timer wheel hash buckets
timer: Remove FIFO "guarantee"
timers: Sanitize catchup_timer_jiffies() usage
hrtimer: Allow hrtimer::function() to free the timer
seqcount: Introduce raw_write_seqcount_barrier()
seqcount: Rename write_seqcount_barrier()
hrtimer: Fix hrtimer_is_queued() hole
hrtimer: Remove HRTIMER_STATE_MIGRATE
selftest: Timers: Avoid signal deadlock in leap-a-day
timekeeping: Copy the shadow-timekeeper over the real timekeeper last
clockevents: Check state instead of mode in suspend/resume path
selftests: timers: Add leap-second timer edge testing to leap-a-day.c
ntp: Do leapsecond adjustment in adjtimex read path
time: Prevent early expiry of hrtimers[CLOCK_REALTIME] at the leap second edge
ntp: Introduce and use SECS_PER_DAY macro instead of 86400
... -
Pull x86 core updates from Ingo Molnar:
"There were so many changes in the x86/asm, x86/apic and x86/mm topics
in this cycle that the topical separation of -tip broke down somewhat -
so the result is a more traditional architecture pull request,
collected into the 'x86/core' topic.The topics were still maintained separately as far as possible, so
bisectability and conceptual separation should still be pretty good -
but there were a handful of merge points to avoid excessive
dependencies (and conflicts) that would have been poorly tested in the
end.The next cycle will hopefully be much more quiet (or at least will
have fewer dependencies).The main changes in this cycle were:
* x86/apic changes, with related IRQ core changes: (Jiang Liu, Thomas
Gleixner)- This is the second and most intrusive part of changes to the x86
interrupt handling - full conversion to hierarchical interrupt
domains:[IOAPIC domain] -----
|
[MSI domain] --------[Remapping domain] ----- [ Vector domain ]
| (optional) |
[HPET MSI domain] ----- |
|
[DMAR domain] -----------------------------
|
[Legacy domain] -----------------------------This now reflects the actual hardware and allowed us to distangle
the domain specific code from the underlying parent domain, which
can be optional in the case of interrupt remapping. It's a clear
separation of functionality and removes quite some duct tape
constructs which plugged the remap code between ioapic/msi/hpet
and the vector management.- Intel IOMMU IRQ remapping enhancements, to allow direct interrupt
injection into guests (Feng Wu)* x86/asm changes:
- Tons of cleanups and small speedups, micro-optimizations. This
is in preparation to move a good chunk of the low level entry
code from assembly to C code (Denys Vlasenko, Andy Lutomirski,
Brian Gerst)- Moved all system entry related code to a new home under
arch/x86/entry/ (Ingo Molnar)- Removal of the fragile and ugly CFI dwarf debuginfo annotations.
Conversion to C will reintroduce many of them - but meanwhile
they are only getting in the way, and the upstream kernel does
not rely on them (Ingo Molnar)- NOP handling refinements. (Borislav Petkov)
* x86/mm changes:
- Big PAT and MTRR rework: making the code more robust and
preparing to phase out exposing direct MTRR interfaces to drivers -
in favor of using PAT driven interfaces (Toshi Kani, Luis R
Rodriguez, Borislav Petkov)- New ioremap_wt()/set_memory_wt() interfaces to support
Write-Through cached memory mappings. This is especially
important for good performance on NVDIMM hardware (Toshi Kani)* x86/ras changes:
- Add support for deferred errors on AMD (Aravind Gopalakrishnan)
This is an important RAS feature which adds hardware support for
poisoned data. That means roughly that the hardware marks data
which it has detected as corrupted but wasn't able to correct, as
poisoned data and raises an APIC interrupt to signal that in the
form of a deferred error. It is the OS's responsibility then to
take proper recovery action and thus prolonge system lifetime as
far as possible.- Add support for Intel "Local MCE"s: upcoming CPUs will support
CPU-local MCE interrupts, as opposed to the traditional system-
wide broadcasted MCE interrupts (Ashok Raj)- Misc cleanups (Borislav Petkov)
* x86/platform changes:
- Intel Atom SoC updates
... and lots of other cleanups, fixlets and other changes - see the
shortlog and the Git log for details"* 'x86-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (222 commits)
x86/hpet: Use proper hpet device number for MSI allocation
x86/hpet: Check for irq==0 when allocating hpet MSI interrupts
x86/mm/pat, drivers/infiniband/ipath: Use arch_phys_wc_add() and require PAT disabled
x86/mm/pat, drivers/media/ivtv: Use arch_phys_wc_add() and require PAT disabled
x86/platform/intel/baytrail: Add comments about why we disabled HPET on Baytrail
genirq: Prevent crash in irq_move_irq()
genirq: Enhance irq_data_to_desc() to support hierarchy irqdomain
iommu, x86: Properly handle posted interrupts for IOMMU hotplug
iommu, x86: Provide irq_remapping_cap() interface
iommu, x86: Setup Posted-Interrupts capability for Intel iommu
iommu, x86: Add cap_pi_support() to detect VT-d PI capability
iommu, x86: Avoid migrating VT-d posted interrupts
iommu, x86: Save the mode (posted or remapped) of an IRTE
iommu, x86: Implement irq_set_vcpu_affinity for intel_ir_chip
iommu: dmar: Provide helper to copy shared irte fields
iommu: dmar: Extend struct irte for VT-d Posted-Interrupts
iommu: Add new member capability to struct irq_remap_ops
x86/asm/entry/64: Disentangle error_entry/exit gsbase/ebx/usermode code
x86/asm/entry/32: Shorten __audit_syscall_entry() args preparation
x86/asm/entry/32: Explain reloading of registers after __audit_syscall_entry()
...
22 Jun, 2015
1 commit
-
… to merge last updates
Signed-off-by: Ingo Molnar <mingo@kernel.org>
16 Jun, 2015
2 commits
-
When CONFIG_SND_HDA_I915=n, we get a compile warning:
sound/pci/hda/hda_intel.c: In function ‘azx_probe_continue’:
sound/pci/hda/hda_intel.c:1882:2: warning: label ‘skip_i915’ defined but not used [-Wunused-label]Fix it by putting again ifdef to it. Sigh.
Fixes: bf06848bdbe5 ('ALSA: hda - Continue probing even if i915 binding fails')
Reported-by: Borislav Petkov
Signed-off-by: Takashi Iwai -
The new Dell XPS13 also requires the similar quirk for fixing the
noisy outputs. (But, as the codec was changed, now the fixup for
Latitude is used instead.)Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=99851
Signed-off-by: Takashi Iwai
15 Jun, 2015
3 commits
-
Fix the missing dependency on PCM stuff.
[Add the same fix for HAL2, too -- tiwai]
Signed-off-by: Nicholas Mc Guire
Signed-off-by: Takashi Iwai -
We still got a report that the audio crackles and noises occur with
the recent 4.1 kernels on Dell machines. These machines seem to need
similar workarounds that have been applied to the recent Dell XPS 13
models. Since the codec of these machines (Dell Latitute E7240 and
E7440) is different from XPS 13's one, we need a new fixup entry.Also, it was confirmed that the previous workaround to disable the
widget power-save (commit [219f47e4f964: ALSA: hda - Disable widget
power-saving for ALC292 & co]) is no longer needed after this fix.
So, this patch includes the partial revert of the commit, too.Reported-and-tested-by: Mihai Donțu
Tested-by: Jonathan McDowell
Signed-off-by: Takashi Iwai -
On a HP Envy TouchSmart laptop, there are 2 speakers (main speaker
and subwoofer speaker), 1 headphone and 2 DACs, without this fixup,
the headphone will be assigned to a DAC and the 2 speakers will be
assigned to another DAC, this assignment makes the surround-2.1
channels invalid.To fix it, here using a DAC/pin preference map to bind the main
speaker to 1 DAC and the subwoofer speaker will be assigned to another
DAC.Cc:
Signed-off-by: Hui Wang
Signed-off-by: Takashi Iwai
12 Jun, 2015
1 commit
-
The previous patch tried to continue the probe if i915 binding fails.
For for simplicity reason, we haven't implemented abort even for
controller chips that are dedicated for HDMI/DP on HSW and BDW.
However, Mengdong suggested that this can be dangerous; BIOS may
disable gfx power well although the PCI entry for HD-audio is left,
and this may result in the unexpected behavior, kernel errors, etc.For avoiding this situation, abort the probe at i915 binding failure
only for HSW/BDW chips selectively. For other chips, it still
continues.Fixes: bf06848bdbe5 ('ALSA: hda - Continue probing even if i915 binding fails')
Reported-by: Mengdong Lin
Signed-off-by: Takashi Iwai
11 Jun, 2015
2 commits
-
Yet another regression by the transition to regmap cache; for better
usability, we had the fake mute control using the zero amp value for
Conexant codecs, and this was forgotten in the new hda core code.Since the bits 4-7 are unused for the amp registers (as we follow the
syntax of AMP_GET verb), the bit 4 is now used to indicate the fake
mute. For setting this flag, snd_hda_codec_amp_update() becomes a
function from a simple macro. The bonus is that it gained a proper
function description.Signed-off-by: Takashi Iwai
-
Currently snd-hda-intel driver aborts the probing of Intel HD-audio
controller with i915 power well management when binding with i915
driver via hda_i915_init() fails. This is no big problem for Haswell
and Broadwell where the HD-audio controllers are dedicated to
HDMI/DP, thus i915 link is mandatory. However, Skylake, Baytrail and
Braswell have only one controller and both HDMI/DP and analog codecs
share the same bus. Thus, even if HDMI/DP isn't usable, we should
keep the controller working for other codecs.For fixing this, this patch simply allows continuing the probing even
if hda_i915_init() call fails. This may leave stale sound components
for HDMI/DP devices that are unbound with graphics. We could abort
the probing selectively, but from the code simplicity POV, it's better
to continue in all cases.Reported-by: Libin Yang
Signed-off-by: Takashi Iwai
10 Jun, 2015
1 commit
-
Along with the transition to regmap for managing the cached parameter
reads, the caps overwrite was also moved to regmap cache. The cache
change itself works, but it still tries to write the non-existing verb
(the HDA parameter is read-only) wrongly. It's harmless in most
cases, but some chips are picky and may result in the codec
communication stall.This patch avoids it just by adding the missing flag check in
reg_write ops.Signed-off-by: Takashi Iwai
09 Jun, 2015
1 commit
-
The new regmap code seems to cache this, which isn't helpful
for the hotplug dock situation where this gets updated.Use the uncached query for this.
Signed-off-by: Dave Airlie
Signed-off-by: Takashi Iwai
08 Jun, 2015
1 commit
-
This patch adds native DSD support for the XMOS based JLsounds I2SoverUSB board
Signed-off-by: Jurgen Kramer
Cc:
Signed-off-by: Takashi Iwai
05 Jun, 2015
1 commit
-
Merge arch/ changes for TI's DRA7 SoC Display Subsystem.
03 Jun, 2015
6 commits
-
DM_INH = 1 (stereo downmix prohibited) and CA = 0x00 (Channel
Allocation: FR, FL) is an invalid combination according to the
HDMI Compliance Test 7.31 "Audio InfoFrame".Signed-off-by: Misael Lopez Cruz
Signed-off-by: Jyri Sarha
Signed-off-by: Tomi Valkeinen
Acked-by: Mark Brown -
There is a constraint in the OMAP4 HDMI IP that requires to use
the 8-channel code when transmitting more than two channels.The constraint doesn't apply for OMAP5 so don't force the channel
allocation in the sound driver as it can be done specifically for
OMAP4 later in the hdmi4 core.Signed-off-by: Misael Lopez Cruz
Signed-off-by: Jyri Sarha
Signed-off-by: Tomi Valkeinen
Acked-by: Mark Brown -
Nothing in uses anything from , so
remove it from there and fix up the resulting build problems
triggered on x86 {64|32}-bit {def|allmod|allno}configs.The breakages were triggering in places where x86 builds relied
on vmalloc() facilities but did not include
explicitly and relied on the implicit inclusion via .Also add:
- to
- to... which were two other implicit header file dependencies.
Suggested-by: David Miller
Signed-off-by: Stephen Rothwell
[ Tidied up the changelog. ]
Acked-by: David Miller
Acked-by: Takashi Iwai
Acked-by: Viresh Kumar
Acked-by: Vinod Koul
Cc: Andrew Morton
Cc: Anton Vorontsov
Cc: Boris Ostrovsky
Cc: Colin Cross
Cc: David Vrabel
Cc: H. Peter Anvin
Cc: Haiyang Zhang
Cc: James E.J. Bottomley
Cc: Jaroslav Kysela
Cc: K. Y. Srinivasan
Cc: Kees Cook
Cc: Konrad Rzeszutek Wilk
Cc: Kristen Carlson Accardi
Cc: Len Brown
Cc: Linus Torvalds
Cc: Peter Zijlstra
Cc: Rafael J. Wysocki
Cc: Suma Ramars
Cc: Thomas Gleixner
Cc: Tony Luck
Signed-off-by: Ingo Molnar -
The driver worked around an error in the MAYA44 USB(+)'s mixer unit
descriptor by aborting before parsing the missing field. However,
aborting parsing too early prevented parsing of the other units
connected to this unit, so the capture mixer controls would be missing.Fix this by moving the check for this descriptor error after the parsing
of the unit's input pins.Reported-by: nightmixes
Tested-by: nightmixes
Signed-off-by: Clemens Ladisch
Cc:
Signed-off-by: Takashi Iwai -
Add mixer control names for the ESI Maya44 USB+ (which appears to be
identical width the AudioTrak Maya44 USB).Reported-by: nightmixes
Signed-off-by: Clemens Ladisch
Cc:
Signed-off-by: Takashi Iwai -
Acer Aspire 9420 with ALC883 (1025:0107) needs the fixup for EAPD to
make the sound working like other Aspire models.Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=94111
Cc:
Signed-off-by: Takashi Iwai
31 May, 2015
1 commit
-
VT202x codecs seem requiring some delay after the resume D0 power
transition for making the jack detection working again. Without the
delay soon after D0, the jack is always detected as unplugged.Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=98921
Signed-off-by: Takashi Iwai
30 May, 2015
1 commit
-
This quirk allows us to avoid the noisy:
current rate 0 is different from the runtime rate
message every time playback starts. While USB DAC in the RR2150
supports reading the sample rate, it never returns a sample rate
other than zero in my observation with common sample rates.Signed-off-by: Eric Wong
Cc: Joe Turner
Cc:
Signed-off-by: Takashi Iwai
29 May, 2015
3 commits
-
Fix this from the logs:
usb 7-1: New USB device found, idVendor=046d, idProduct=08ca
...
usb 7-1: Warning! Unlikely big volume range (=3072), cval->res is probably wrong.
usb 7-1: [5] FU [Mic Capture Volume] ch = 1, val = 4608/7680/1Signed-off-by: Wolfram Sang
Cc:
Signed-off-by: Takashi Iwai -
Dell SSID 0x0706 support headset mode.
Signed-off-by: Kailang Yang
Signed-off-by: Takashi Iwai -
The commit [49fb18972581: ALSA: hda - Set stream_pm ops automatically
by generic parser] resulted in regressions on some Realtek and VIA
codecs because these drivers set patch_ops after calling the generic
parser, thus stream_pm got cleared to NULL again. I haven't noticed
since I tested with IDT codec.Restore (partial revert) the stream_pm ops for them to fix the
regression.Fixes: 49fb18972581 ('ALSA: hda - Set stream_pm ops automatically by generic parser')
Reported-by: Jeremiah Mahler
Signed-off-by: Takashi Iwai
28 May, 2015
2 commits
-
When headphone mic boost is above zero, some 10 - 20 second delay
might occur before the headphone mic is operational.
Therefore disable the headphone mic boost control (recording gain is
sufficient even without it).(Note: this patch is not about the headset mic, it's about the less
common mic-in only mode.)BugLink: https://bugs.launchpad.net/bugs/1454235
Suggested-by: Kailang Yang
Signed-off-by: David Henningsson
Signed-off-by: Takashi Iwai -
We've got a regression report that 4.1-rc causes noises on a Dell
laptop. Similar like Realtek codec, this seems also triggered by the
recent power_save_node feature. As this kind of issue is quite hard
to debug without actual hardware, disable the power_save_node flag for
this codec as a workaround.Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=98971
Signed-off-by: Takashi Iwai
27 May, 2015
2 commits
-
A new AMD controller [1002:aac8] seems to need the quirk for other AMD
NS HDMI stuff, otherwise it gives noisy sounds.Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=99021
Cc:
Signed-off-by: Takashi Iwai -
This allows user to test power_save_node feature via sysfs or patch
firmware even on the codecs that don't specify it. It'll also save a
few lines.Signed-off-by: Takashi Iwai
26 May, 2015
1 commit
-
Add ALC3246 for Dell platform.
Signed-off-by: Kailang Yang
Signed-off-by: Takashi Iwai
24 May, 2015
2 commits
-
This reverts commit 7290006d8c0900c56d8c58428134f02c35109d17.
Through the regression report, it was revealed that the
tpacpi_led_set() call to thinkpad_acpi helper doesn't only toggle the
mute LED but actually mutes the sound. This is contradiction to the
expectation, and rather confuses user.According to Henrique, it's not trivial to judge which TP model
behaves "LED-only" and which model does whatever more intrusive, as
Lenovo's implementations vary model by model. So, from the safety
reason, we should revert the patch for now.Reported-by: Martin Steigerwald
Cc: Pali Rohár
Cc:
Signed-off-by: Takashi Iwai -
Microsoft LifeCam HD-3000 (045e:0779) needs a similar quirk for
suppressing the unsupported sample rate inquiry.Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=98481
Cc:
Signed-off-by: Vittorio Gambaletta
Signed-off-by: Takashi Iwai
20 May, 2015
3 commits
-
We've got reports that ALC3226 (a Dell variant of ALC292) gives click
noises at transition from D3 to D0 when the widget power-saving is
enabled. Further debugging session showed that avoiding it isn't
trivial, unfortunately, since paths are basically activated
dynamically while the pins have been already enabled.This patch disables the widget power-saving for such codecs.
Reported-by: Jonathan McDowell
Signed-off-by: Takashi Iwai -
The widget (node) power-saves restore the widget states at each
transition from D3 to D0 on each node. This was added in the commit
[d545a57c5f84:ALSA: hda - Sync node attributes at resume from widget
power saving]. However, the test was rater false-positive; this
wasn't needed for any codecs.Since the resync may take significant number of additional verbs to be
executed, it's better to reduce it. Let's disable it for now again.Signed-off-by: Takashi Iwai
-
The recently added DT support for the ac97 driver is causing
a gcc warning:sound/atmel/ac97c.c: In function 'atmel_ac97c_probe_dt':
sound/atmel/ac97c.c:919:29: warning: unused variable 'match' [-Wunused-variable]
const struct of_device_id *match;The variable is clearly unused, so we can remove it.
Signed-off-by: Arnd Bergmann
Acked-by: Alexander Stein
Signed-off-by: Takashi Iwai
19 May, 2015
2 commits
-
Make sure the upstream fixes are applied before adding further
modifications. -
Microsoft LifeCam Studio (045e:0772) needs a similar quirk for
suppressing the wrong sample rate inquiry.Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=98481
Cc:
Signed-off-by: Takashi Iwai
