Robby Cai [Mon, 18 Jun 2012 11:46:41 +0000 (19:46 +0800)]
ENGR00213997: Fix Section Mismatch warning
Fix:
WARNING: vmlinux.o(.data+0x8c28): Section mismatch in reference from the
variable mx6_gpmi_nand_platform_data to the function
.init.text:gpmi_nand_platform_init()
The variable mx6_gpmi_nand_platform_data references
the function __init gpmi_nand_platform_init()
If the reference is valid then annotate the
variable with __init* or __refdata (see linux/init.h) or name the variable:
*_template, *_timer, *_sht, *_ops, *_probe, *_probe_one, *_console
Robby Cai [Mon, 18 Jun 2012 05:14:08 +0000 (13:14 +0800)]
ENGR00213751: imx6sl: Add ELAN touchscreen support on EINK-DC3 board
Add ELAN capacitive TS support on EINK-DC3 stacked on MX6SL_ARM2 board
- configure the iomux setting (need 4.7K Ohm pull up on 'touch_int_b')
- configure the i2c slave addr
- configure the GPIO setting for ELAN ce/int/rst
- update the defconfig
Robby Cai [Tue, 12 Jun 2012 10:24:54 +0000 (18:24 +0800)]
ENGR00213749: imx6sl: Add keypad support on EINK-DC3 board
Add the support for keypad on EINK-DC3 board which is stacked on ARM2 board.
- configure the iomux setting
- add dummy kpp clock to fool imx_keypad driver
- add platform device for keypad
- add key mapping (4x4 array) used on EINK-DC3
- update the defconfig for keypad driver
Chen Liangjun [Mon, 11 Jun 2012 07:08:18 +0000 (15:08 +0800)]
ENGR00212318 ASRC:update to in/out width config
The origin ASRC driver did not support input and output wordwidth
config but an total wordwidth config instead. And the input wordwith
and output wordwidth are all fixed to 24 bit.
In this path, we do things below:
1 Update to use input wordwidth and output wordwidth config seperately
instead of an total wordwidth config.
2 Set corresponding DMA(input/output) buswidth according ASRC's input
and output wordwidth config.
3 Support 16/24 bit input wordwidth and 24 bit output wordwidth.
Anson Huang [Thu, 14 Jun 2012 10:07:30 +0000 (18:07 +0800)]
ENGR00212720 [MX6]Adjust CPU 672M setpoint voltage
Previous voltage for 672M is 1.05V, normal test is OK,
but if CPU is busy in background and do the CPUFreq change
as well, always fail the stress test at 672M setpoint, after
increase it to 1.1V, stress test is OK.
Wayne Zou [Tue, 12 Jun 2012 06:52:21 +0000 (14:52 +0800)]
ENGR00213158-3 FB: Clean up fb interrupt handler
Clean up the fb driver for maintainability:
1. Use completion instead of semaphore API interface.
2. Use IPU oneshot interrupt mode and remove ipu_disable_irq()
function call in interrupt handler.
Larry Li [Tue, 12 Jun 2012 09:11:45 +0000 (17:11 +0800)]
ENGR00213170-2 [MX6SL] Enable GPU driver
Use allocated GPU resource to enable GPU.
Memroy address on imx6sl board starts from 0x80000000
and GC320 can access [baseAddress, baseAddress + 2G) only without MMU.
So to make GC320 work, baseAddres must be set to 0x80000000, and all
address sent to GC320 must be a offset to baseAddress. GC355 doesn't
need this baseAddress, that means it needs a real physcial adress,
rather than the offset to baseAddress.
Original code always change phsysical address to 'offset' before use it,
no matter it is used by GC355 or GC320, so only one of them can work.
Solution is to move address adjustion to arch specific part. So each
core can get what it wants.
Signed-off-by: Larry Li <b20787@freescale.com> Acked-by: Lily Zhang
Anson Huang [Wed, 13 Jun 2012 12:20:01 +0000 (20:20 +0800)]
ENGR00180919 [MX6]Update clock tree if BUS freq is changed
As DDR freq change is by modifying CCM register directly,
we need to update the clock tree as well, or the clock
tree will be broken. Also, we need to make sure the clock
rate counting is right.
Robin Gong [Wed, 13 Jun 2012 06:44:00 +0000 (14:44 +0800)]
ENGR00213336 sabresd pfuze: support 1.2G by param which pass by u-boot
There is no fuse data for distinguish 1.2G or 1G, kernel need support passed
param from u-boot that can know 1.2G or 1G. If 1.2G, will configure VDDSOC_IN
&VDDARM_IN to 1.425V by pfuze and VDDSOC&VDDPU to 1.25V by internal ldo
Sandor Yu [Wed, 13 Jun 2012 11:18:43 +0000 (19:18 +0800)]
ENGR00180937 IPU: Change IPU error message to IPU warning
IPU driver will print unexpect interrupt state in ipu_irq_handler function,
It is for IPU debug and state check, not a IPU error.
So change print function from dev_error to dev_warn.
Anson Huang [Tue, 12 Jun 2012 10:50:42 +0000 (18:50 +0800)]
ENGR00180930 [MX6]Fix low bus mode bug when there is no CPUFreq change
If the CPUFreq change is done before enabling low bus freq driver,
the bus freq will be staying at high freq until there is new request
of entering low bus freq. So we need to put the bus freq into low
point if all the conditions are met when we enable bus freq.
ENGR000212647 MX6 - Fix IPU and AXI default clock frequency
The max freq for IPU and AXI clocks is 264MHz. Hence source
IPU from mmdc_ch0 clock on MX6 to get maximum frequency.
And source AXI from periph_clk on for max freq.
Peter Chen [Mon, 11 Jun 2012 06:31:20 +0000 (14:31 +0800)]
ENGR00212992 usb: fix potential bug with fast plug in/out with usb charger
- Move imx_usb_vbus_disconnect to work queue, and wait or cancel
charger detect worker when vbus is disconnected if charger detect
worker still hasn't finished.
- Move pull down dp from imx_usb_vbus_disconnect to udc vbus
disconnect interrupt handler, the reason is udc->gadget.speed should
to be updated at that ISR.
- Select POWER_SUPPLY if USB_GADGET_ARC is configured to fix the build
error for manufacture tools firmware's build.
Signed-off-by: Peter Chen <peter.chen@freescale.com>
make shi [Mon, 4 Jun 2012 08:28:33 +0000 (16:28 +0800)]
ENGR00212232-02 USB:fix otg can not work without phy power if VBUS removed
- There is no VBUS supply or not supply for USBPHY power 3p0,the USB mouse
enumeration fail during the board booting with a USB mouse connected to
otg port.During system booting ,some error message appear like below :
"usb 1-1: device not accepting address 5, error -71
hub 1-0:1.0: unable to enumerate USB device on port 1"
The otg phy power is needed during normal USB transmitting ,the otg phy
power should be on before calling the usb_add_hcd. So should make sure
usb_add_hcd is called after otg_set_host is called in usb_hcd_fsl_probe().
- Before switch the otg mode form Host mode to Device mode,we must call
dr_discharge_line() to make sure no abnormal usb wakeup interrupt happen.
- Some user case ,the BM_USBPHY_CTRL_ENHOSTDISCONDETECT bit is cleared by
mistake. For example,the otg port connect a extern USB hub and a USB device
connected to the USB hub.Unplug the USB device,the Hub will enter auto suspend
mode,then plug in the device, hub will auto resume by the device plug in ,the
BM_USBPHY_CTRL_ENHOSTDISCONDETECT bit is cleared by mistake.So the function
platform_set_disconnect_det() should be called after usb_disconnect.
make shi [Mon, 4 Jun 2012 08:24:01 +0000 (16:24 +0800)]
ENGR00212232-01 USB:fix otg can not work without phy power if VBUS removed
According to ticket TKT094250 :There is no VBUS supply or not supply for USB
PHY power 3p0,USB dp/dm is floating when there is no connection in device mode:
S/W can run the following flow to ensure the internal DP/DM linestate does not
change. These should be done before set SUSP/PHCD bit after detecting the VBUS
change from high to low.
clear HW_USBPHY_DEBUG_CLKGATE
loopback_utmi_dig_tst1 = 1
loopback_utmi_dig_tst0 = 0
loopback_tsti_tx_en = 1
loopback_tsti_tx_hiz = 0
loopback_tsti_tx_hs_mode = 0;
After detecting the VBUS or ID change (connected to a host or device), SW do:
set HW_USBPHY_DEBUG_CLKGATE back to 1
clear the loopback_XXXX signals above back to 0
Peter Chen [Tue, 5 Jun 2012 03:08:29 +0000 (11:08 +0800)]
ENGR00212322-1: usb: add usb charger support for i.mx6x
- USB charger function is embedded in usb device driver, and only for i.mx6x
- SDP and DCP charger are tested
- Need to enable usb device function (insmod one gadget driver)
to use usb charger detect function
- The power supply interface for usb charger is:
/sys/class/power_supply/imx_usb_charger/
Some useful entries for power supply interface:
- present: whether usb charger is present or not
- type: usb charger type
- current_max: the max charger current for this charger
- online: whether vbus is on or not
Signed-off-by: Peter Chen <peter.chen@freescale.com>
Sandor Yu [Fri, 8 Jun 2012 05:37:17 +0000 (13:37 +0800)]
ENGR00178617 MX6x HDMI hotplug,video discontinuously after do a hotplug
The issue only in play video on DP FG, no such issue on DP BG.
HDMI drvier will call fb_set_var to config framebuffer after HDMI cable plugin.
In Frame buffer driver, the DP FG and BG register as different fb.
The function fb_set_var only update one fb.
As IPU DP module, if DP BG is re-enabled, the FG should reconfig again.
So after HDMI plugin, only BG fb is update, FG fb will work incorrect,
no end of frame interrupt trigger, and print
"mxc_sdc_fb mxc_sdc_fb.0: timeout when waiting for flip irq"
to display.
Change code in function fb_set_var, FG fb will reconfig with DP BG fb.
Anson Huang [Fri, 8 Jun 2012 06:20:36 +0000 (14:20 +0800)]
ENGR00212805 [MX6SL]Add thermal driver device.
Add thermal driver device to support thermal driver interface,
but as this driver is dependent on OCOTP driver and need a
calibrated part, so the temp read from the thermal sysfs
interface maybe inaccurate on those uncalibrated parts.
If we want to use ESAI call ASRC, we need to know the input sample
rate as well as the output sample rate. however, cause of ALSA-LIB,
the input sample rate pass down to ALSA-DRIVER is not accurate. For
example, if the ALSA-DRIVER only support 48KHz playback and we want
to play an 16KHz audio file, the sample rate params the ALSA DRIVER
see is 48KHz but not 16KHz. So, the ALSA-DRIVER do not know the
original sample rate, and thus unable to config ASRC properly.
In this patch, we add an amixer control interface for alsa lib plugin to
config the input sample rate before playback. If no plugin, user
can call the amixer control interface to manually use ASRC for rate
convert.
For example, if we need to playback an 24bit_24K audio wav file without
call the plug. The sound card is 0.
1 Get the ASRC capability: amixer sget "ASRC"
1 Set the input samplerate: amixer sset "ASRC" "24KHz"
2 Play; aplay -Dhw:0 audio_24k24S-S24_LE.wav
Gary Zhang [Wed, 6 Jun 2012 08:22:26 +0000 (16:22 +0800)]
ENGR00212542 SPDIF: move clock operation
Now clock operation occurs in mxc_spdif_trigger(), and
because clock operation is forbidden to exist in interrupt
context, move disabling clock operation to mxc_spdif_shutdown()
the original low memory killer only take care of system memory accounting,
but for so large shared memory occupy by GPU, and each process memory
killer account become unfair, very large 3D game will not killed firstly
if it going to background.
Add this account to let real large memory user get killed if going to background
eg, the "angry bird Space" will acquire 68,215,360 GPU memory for 1-6 toll-gate.
The test show it can quicker recovery from memory shortage situation,
since it's very like to be killed after add so much GPU memory for such 3D game.
ENGR00211670- CPUFREQ-Set CPU to maximum frequency before entering STOP mode
Ensure that the CPUFREQ driver sets the CPU to its maximum
frequency when it is suspended.
Also change the WAIT macro in clock.c to use GPT counter
for the delay instead of getnsdayoftime(). As the kernel
timekeeping driver is suspended before CPUFREQ and this
causes a dump during suspend.
Anson Huang [Mon, 4 Jun 2012 12:51:57 +0000 (20:51 +0800)]
ENGR00211169-2 [SMP]Need to update loops_per_jiffy when cpufreq is changed
Need to update loops_per_jiffy when cpufreq is changed, our platform
only change the loops_per_jiffy of per_cpu cpu_data, we need to update
it to global loops_per_jiffy which is used for udelay, this is for pure
cpu freq change.
Anson Huang [Mon, 4 Jun 2012 12:50:53 +0000 (20:50 +0800)]
ENGR00211169-1 [SMP]Need to update loops_per_jiffy when cpufreq is changed
Need to update loops_per_jiffy when cpufreq is changed, our platform
only change the loops_per_jiffy of per_cpu cpu_data, we need to update
it to global loops_per_jiffy which is used for udelay.
For each CPU, do the calibration delay only once. For subsequent calls,
use the cached per-CPU value of loops_per_jiffy.
This saves about 200ms of resume time on dual core Intel Atom N5xx based
systems. This helps bring down the kernel resume time on such systems
from about 500ms to about 300ms.
[akpm@linux-foundation.org: make cpu_loops_per_jiffy static]
[akpm@linux-foundation.org: clean up message text]
[akpm@linux-foundation.org: fix things up after upstream rmk changes] Signed-off-by: Sameer Nanda <snanda@chromium.org> Cc: Phil Carmody <ext-phil.2.carmody@nokia.com> Cc: Andrew Worsley <amworsley@gmail.com> Cc: David Daney <ddaney@caviumnetworks.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Tony LIU [Fri, 1 Jun 2012 09:56:48 +0000 (17:56 +0800)]
ENGR00211329-2 add mutex for otg wake up enable
- For the system dump issue of usb unhandled irq,
the root cause is the shared resource being lack of protection
add mutex to protect the shared access.
Signed-off-by: Tony LIU <junjie.liu@freescale.com>
Liu Ying [Wed, 30 May 2012 08:14:06 +0000 (16:14 +0800)]
ENGR00211764 IPUv3 fb: support pan display with fb_set_var
Users may call FBIOPUT_VSCREENINFO ioctrl to do pan display.
This ioctrl relies on fb_set_var() to do the job. fb_set_var()
calls custom fb_set_par() method and then calls custom
fb_pan_display() method. The current implementation of ipuv3fb
reinitializes IPU display controller every time the custom
fb_set_par() method is called, which makes the display flash
if fb_set_var() is called to do panning frequently. The custom
fb_pan_display() method checks if the current xoffset and
yoffset are different from previous ones before doing actual
panning, which prevents the panning from happening within the
fb_set_var() context. This patch checks new var info to decide
whether we really need to reinitialize IPU display controller.
We ignore xoffset and yoffset update because it doesn't require
to reinitialize the controller. Users may specify activate field
of var info with FB_ACTIVATE_NOW and FB_ACTIVATE_FORCE to
reinialize the controller by force. Meanwhile, this patch removes
the check in custom fb_pan_display() method mentioned before to
have the panning work within fb_set_var() context. It doesn't
hurt to do panning again if there is no update for xoffset and
yoffset.
make shi [Thu, 31 May 2012 02:24:10 +0000 (10:24 +0800)]
ENGR00210686 USB:fix unplug&plug device during suspend without wakeup enabled
For High speed device ,we need clear BM_USBPHY_CTRL_ENHOSTDISCONDETECT
after suspend.For Low and full speed device, we should power on and power
off the USB port to make sure USB internal state machine work well.
Add a config to enable/disable this code,the code is enabled by default.
Tony LIU [Wed, 30 May 2012 05:56:38 +0000 (13:56 +0800)]
ENGR00211686 mx6 usb: system crash after suspend/resume
- the pre-condition of this issue is:
1. usb gadget must be probed before usb host
2. usb otg must be in host mode
- the root cause of this issue is
because of the issue of week 2p5, a vbus change interrupt will
be issued when system enter into DSM, which will cause system
exit DSM, so we have a walk aroud to disable usb vbus change
interrupt when system enter into DSM.
But this walk around just provent the interrupt generating, the
vbus change status is still on. When usb gadget is probed, the
vbus change interrupt will be enabled by its resume interface by
mistake, and then continuous interrupt will be generated because
usb otg is in host mode, it can't clear the vbus change status.
The system have a protect mechanism that when one IRQ's handler
return IRQ_NONE more than 99000 times, it will through a exception
to inform such situation. That's the reason why system crash.
Signed-off-by: Tony LIU <junjie.liu@freescale.com>
Zhang Jiejing [Mon, 28 May 2012 11:35:46 +0000 (19:35 +0800)]
ENGR00210918-3 elcdif_fb: change update hardware par method.
update the screen var info compare method, it should:
- should update the hardware parameter when FB_ACTIVATE_FORCE
- use bool as return type.
- return true if the new and old var are same.
Sandor Yu [Tue, 29 May 2012 08:54:13 +0000 (16:54 +0800)]
ENGR00178461-02 HDMI1080p: hotplug cause kernel panic. 10%
It is a warning cause by HDMI driver irq enable count mismatch.
The purpose of maintain HDMI irq count is to disable hdmi_iahb_clk
when HDMI cable plugout.
But hdmi_iahb_clk parent ahb clock is always enabled when system run,
so hdmi_iabh_clk power consumption is very low.
The function clk_get_usecount introduce by irq count maintain
is not safety in SMP.
Remove HDMI irq count in HDMI driver, keep hdmi_iahb_clk always run.
and disable hdmi_iahb_clk and hdmi_isfr_clk when FB Blank.
Liu Ying [Fri, 25 May 2012 09:50:19 +0000 (17:50 +0800)]
ENGR00211133 IPUv3 fb:Check boot opt to decide if disp fb'll be present
This patch checks video boot option to decide whether we'll register
certain display framebuffer devices. The user may add video=mxcfbx:off
to kernel bootup command line to disable a display framebuffer
register. The 'x' means that the display number presented in kernel.
Defaultly, all display framebuffers will be registered if the user
doesn't specify related 'off' option in kernel bootup command line.
Zhang Jiejing [Fri, 25 May 2012 09:03:02 +0000 (17:03 +0800)]
ENGR00210915 fb: mxc_elcdfb: fix screen flash when use fb_set_var
fb_set_var() is used by android default display,
we found the screen will flash using android default
display support, found the root case was we init the
hardware (blank and unblank) the screen.
because fb_set_var() will call fb_set_par() every time,
we needs check only set the hardware register when needed,
then the android default image without any modify can works.
Also a build warnning fix.
Signed-off-by: Zhang Jiejing <jiejing.zhang@freescale.com> Signed-off-by: Liu Ying <Ying.Liu@freescale.com>
- Because mode_4 only use for handwriting mode, and
mode_3 use for power off mode, modify the waveform
mode for update requests.
- Use mode_2 for DU and A2, mode_1 for GU:
mode_A2 = mode_2, mode_du = mode_2;
mode_gu = mode_1;
Anson Huang [Fri, 25 May 2012 08:36:56 +0000 (16:36 +0800)]
ENGR00210547 [MX6]Enable I cache and branch prediction early
1. When resume, we can enable I cache and branch prediction
early to speed up the resume process;
2. L2 cache still need clean before suspend to make suspend/resume
modifiedre stable, add it back until we find the root cause.
Xinyu Chen [Fri, 25 May 2012 03:39:15 +0000 (11:39 +0800)]
ENGR00210850 mx6: boot failure with local timer and wait mode enabled
Previous patch only check the condition that GPT broadcast
event is ready or not before doing clock event switch.
It's not enough, as the clock switch from local timer to GPT
broadcast must be happen after GPT broadcast clock event setup
and current cpu's clock device switch to local timer clock event.
Otherwise, we will have chance that cpu exit the wait mode and
switch back clock event without local timer event setup correctly.
Rong Dian [Thu, 24 May 2012 08:05:31 +0000 (16:05 +0800)]
ENGR00210694 MX6 SABRESD:Enable IRQ for max11801
If miss to configure IRQ for max11801, the max11801 driver fails to
probe and returns error,sabresd battery driver also fails to sample
voltage by max11801 ADC.
Anson Huang [Wed, 23 May 2012 07:13:03 +0000 (15:13 +0800)]
ENGR00180495 [mx6]Fix suspend/resume issue caused by hotplug
When we kill a secondary cpu, we need to wait for it
die, then kill it from hardware setting. And to avoid
the cache unalign issue, we use hardware register to
send flag to inform main cpu to kill secondary cpu.
ARM: 7359/2: smp_twd: Only wait for reprogramming on active cpus
During booting of cpu1, there is a short window where cpu1
is online, but not active where cpu1 is occupied by waiting
to become active. If cpu0 then decides to schedule something
on cpu1 and wait for it to complete, before cpu0 has set
cpu1 active, we have a deadlock.
Typically it's this CPU frequency transition that happens at
this time, so let's just not wait for it to happen, it will
happen whenever the CPU eventually comes online instead.
Cc: Peter Zijlstra <peterz@infradead.org> Cc: stable@kernel.org Signed-off-by: Jonas Aaberg <jonas.aberg@stericsson.com> Reviewed-by: Rickard Andersson <rickard.andersson@stericsson.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Will Deacon [Fri, 27 Apr 2012 11:56:24 +0000 (12:56 +0100)]
ARM: 7406/1: hotplug: copy the affinity mask when forcefully migrating IRQs
When a CPU is hotplugged off, we migrate any IRQs currently affine to it
away and onto another online CPU by calling the irq_set_affinity
function of the relevant interrupt controller chip. This function
returns either IRQ_SET_MASK_OK or IRQ_SET_MASK_OK_NOCOPY, to indicate
whether irq_data.affinity was updated.
If we are forcefully migrating an interrupt (because the affinity mask
no longer identifies any online CPUs) then we should update the IRQ
affinity mask to reflect the new CPU set. Failure to do so can
potentially leave /proc/irq/n/smp_affinity identifying only offline
CPUs, which may confuse userspace IRQ balancing daemons.
This patch updates migrate_one_irq to copy the affinity mask when
the interrupt chip returns IRQ_SET_MASK_OK after forcefully changing the
affinity of an interrupt.
Cc: stable@vger.kernel.org Reported-by: Leif Lindholm <leif.lindholm@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Russell King [Thu, 21 Jul 2011 14:14:21 +0000 (15:14 +0100)]
ARM: CPU hotplug: ensure we migrate all IRQs off a downed CPU
Our selection of interrupts to consider for IRQ migration is sub-
standard. We were potentially including per-CPU interrupts in our
migration strategy, but omitting chained interrupts. This caused
some interrupts to remain on a downed CPU.
We were also trying to migrate interrupts which were not migratable,
resulting in an OOPS.
Instead, iterate over all interrupts, skipping per-CPU interrupts
or interrupts whose affinity does not include the downed CPU, and
attempt to set the affinity for every one else if their chip
implements irq_set_affinity().
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Russell King [Thu, 21 Jul 2011 14:07:56 +0000 (15:07 +0100)]
ARM: CPU hotplug: pass in proper affinity mask on IRQ migration
Now that the GIC takes care of selecting a target interrupt from the
affinity mask, we don't need all this complexity in the core code
anymore. Just detect when we need to break affinity.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Russell King [Thu, 21 Jul 2011 13:51:13 +0000 (14:51 +0100)]
ARM: CPU hotplug: fix abuse of irqdesc->node
irqdesc's node member is supposed to mark the numa node number for the
interrupt. Our use of it is non-standard. Remove this, replacing the
functionality with a test of the affinity mask.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
ARM: introduce handle_IRQ() not to dump exception stack
On Mon, Jul 11, 2011 at 3:52 PM, Russell King - ARM Linux
<linux@arm.linux.org.uk> wrote:
...
> The __exception annotation on a function causes this to happen:
>
> [<c002406c>] (asm_do_IRQ+0x6c/0x8c) from [<c0024b84>]
> (__irq_svc+0x44/0xcc)
> Exception stack(0xc3897c78 to 0xc3897cc0)
> 7c60: 4022d3204022e000
> 7c80: 0800007500001000c32273c0c03ce1c0c2b49b784022d000c2b420b400000001
> 7ca0: 00000000c3897cfc00000000c3897cc0c00afc54c002edd800000013ffffffff
>
> Where that stack dump represents the pt_regs for the exception which
> happened. Any function found in while unwinding will cause this to
> be printed.
>
> If you insert a C function between the IRQ assembly and asm_do_IRQ,
> the
> dump you get from asm_do_IRQ will be the stack for your function,
> not
> the pt_regs. That makes the feature useless.
>
When __irq_svc - or any of the other exception handling assembly code -
calls the C code, the stack pointer will be pointing at the pt_regs
structure.
All the entry points into C code from the exception handling code are
marked with __exception or __exception_irq_enter to indicate that they
are one of the functions which has pt_regs above them.
Normally, when you've entered asm_do_IRQ() you will have this stack
layout (higher address towards top):
pt_regs
asm_do_IRQ frame
If you insert a C function between the exception assembly code and
asm_do_IRQ, you end up with this stack layout instead:
pt_regs
your function frame
asm_do_IRQ frame
This means when we unwind, we'll get to asm_do_IRQ, and rather than
dumping out the pt_regs, we'll dump out your functions stack frame
instead, because that's what is above the asm_do_IRQ stack frame
rather than the expected pt_regs structure.
The fix is to introduce handle_IRQ() for no exception stack dump, so
it can be called with MULTI_IRQ_HANDLER is selected and a C function
is between the assembly code and the actual IRQ handling code.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Eric Miao <eric.y.miao@gmail.com>
Russell King [Wed, 18 Jan 2012 15:59:45 +0000 (15:59 +0000)]
ARM: SMP: use a timing out completion for cpu hotplug
Rather than open-coding the jiffy-based wait, and polling for the
secondary CPU to come online, use a completion instead. This
removes the need to poll, instead we will be notified when the
secondary CPU has initialized.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Heiko Carstens [Tue, 23 Aug 2011 11:20:46 +0000 (13:20 +0200)]
nohz: Remove "Switched to NOHz mode" debugging messages
When performing cpu hotplug tests the kernel printk log buffer gets flooded
with pointless "Switched to NOHz mode..." messages. Especially when afterwards
analyzing a dump this might have removed more interesting stuff out of the
buffer.
Assuming that switching to NOHz mode simply works just remove the printk.
Sourcing AXI_CLK from PLL3_PFD_540M causes the system to
hang on resuming from STOP mode.
The main issue is that PFDs may sometimes hang/freeze
when their parent PLLs are powered on and then relocked
when exiting from STOP mode. To avoid this, PFDs must
be disabled before entering STOP and enabled after resume.
The fix is to move axi_clk to periph_clk before system
enters STOP and then restore it back to PLL3_PFD_540M
after resume.
Anson Huang [Wed, 23 May 2012 01:38:29 +0000 (09:38 +0800)]
ENGR00174974 [MX6]Fix CPU hotplug platform related issue
We need to turn of cache coherency of secondary core before
it is disable by core0, otherwise, the secondary core may be
waked by cache sync, and if it exit from wfi and access BUS,
meanwhile, core0 disable it from hardware, the whole SOC would
hang.
projects / karo-tx-linux.git / log