Devin,

Could you please provide full log for your suspend/resume flow.

Are you using the target # /etc/init.d/pvr-init stop, target # /etc/init.d/matrix-gui-e stop and target # /etc/init.d/load-hd-firmware.sh stop before suspend to RAM?

Are you using the target # /etc/init.d/pvr-init start, target # /etc/init.d/matrix-gui-e start and target # /etc/init.d/load-hd-firmware.sh start after resume and before using the LCD?

If not, please try.

BR
Pavel

Hello Pavel,

Thanks for taking the time to reply.

Yes, if I kill my Qt application, unload ti81xxfb and vpss, and unload the firmware (via load-hd-firmware.sh), and then bring everything back up after the resume, then yes it does appear to work after the resume.  Of course this isn't actually vpss being power management aware - it's an awful workaround that adds seconds to both the suspend and resume processes (the need to start our application back up on resume adds nearly ten seconds to the process).

I need to be able to suspend/resume without tearing down the entire driver and application stack in the system.  The suggestion offered is not acceptable due to the time required.

Regards,

Devin

Devin,

Are you using EZSDK 5.05.02.00? By EVK do you mean EVM? If not, what is EVK? Could you please provide full log (u-boot + linux kernel + user space) and the steps to reproduce the issue on my side (I am using DM8148 EVM).

Best regards,
Pavel

Hello Pavel,

Yes, I was referring to the EVM (specifically the TI DM814x Evaluation Module by Mistral).

I'm still running 5.05.01.04.  The release notes for 5.05.02 didn't suggest there was anything we were interested in that warranted going through the work to do an upgrade.

Regarding getting you the logs, I can certainly boot back to the original EVM filesystem and collect the logs as requested.  However it should be trivial to reproduce by simply booting the EVM, running "echo mem > /sys/power/state" (without unloading the firmware or applications), and then watch the system hang/crash.  It seems like you're already quite aware of the nature of the underlying problem since you suggested killing the application and unloading the firmware prior to putting the hardware to sleep.

Devin

Devin,

Devin Heitmueller said:

However it should be trivial to reproduce by simply booting the EVM, running "echo mem > /sys/power/state" (without unloading the firmware or applications), and then watch the system hang/crash.

Yes, I was suspecting this is the issue, but I need to double check with you. Now it is clear.

Hi Pavel,

Ok, so you have indeed confirmed my worst fear - that TI shipped a product claiming it supports suspend/resume, but for some reason they don't think that should include the LCD.  Wow.  That's absolutely amazing.

I'm familiar with the example you pointed to for disabling the clocks on the DSP.  If I were building some custom DSP application, that would be quite useful.  However since we're talking about TI's binary firmware running on the M3, it isn't really any help.  You suggested using syslink/rpmsg to put the core into suspend/resume.  Does the FVID2 API for talking to TI's M3 firmware provide a call I can make to suspend the firmware?   As far as I can tell, the answer is "no".

Do you have any idea what's going on inside the M3 when the suspend occurs?  As far as I can tell, the power domains are not being touched and the clocks are in an identical state before the suspend compared to after the resume.  Any information you can provide on *why* the core stops responding to syslink requests?  Does it not recover properly when the clocks are restored?  Is the firmware no longer running at all on the MPU?

If I have to reload the firmware, I'm going to lose all the internal state of the M3, and hence won't be able to have the graphics pipelines pick up where they left off.  It would be good to know whether this is a case where I simply need to be configuring the clocks on the VPSS driver (e.g. adding suspend/resume PM events to the VPSS driver to manage the clocks), or whether the driver would have to tear down all the internal state inside of the VPSS driver, the M3 core and the FB driver, and then be able to recreate it on resume.

Again, I am absolutely amazed that I'm expected to do this work and not some employee of TI who wrote and maintains the VPSS driver.

Devin

Devin,

I also think FVID2 API does not provide a call to M3 for suspend. At least I can not find anything in the {EZSDK}/board-support/linux-2.6.37-pspxx.xx.xx.xx/drivers/video/ti81xx/vpss/fvid2.c file.

The first place of stuck of the suspend process, when M3 firmware is not unloaded/stopped is right after the "PM: Syncing filesystems ... done." line. I workaround this "stuck" with removing the following function:

ti-ezsdk_dm814x-evm_5_05_02_00/board-support/linux-2.6.37-psp04.04.00.01/drivers/tty/vt/vt_ioctl.c

static void vt_event_wait(struct vt_event_wait *vw)
{
    unsigned long flags;
     
    /* Prepare the event */
    INIT_LIST_HEAD(&vw->list);
    vw->done = 0;
    /* Queue our event */
     
    spin_lock_irqsave(&vt_event_lock, flags);
    list_add(&vw->list, &vt_events);
    spin_unlock_irqrestore(&vt_event_lock, flags);
    /* Wait for it to pass */
     
    //wait_event_interruptible_tty(vt_event_waitqueue, vw->done);  -----> I comment this function
    /* Dequeue it */
     
    spin_lock_irqsave(&vt_event_lock, flags);
    list_del(&vw->list);
    spin_unlock_irqrestore(&vt_event_lock, flags);
}

I tested the suspend/resume with and without the M3 firmware. In both cases the suspend/resume works fine without executing the wait_event_interruptible_tty() function.

Regards,
Pavel

Hello Pavel,

That sounds very promising.  I'll give that a try and report back the results.

I understand about the LCD clocking, pin-muxing, calibration, etc. and have no problem dealing with those issues.

Thanks,

Devin

Hi Pavel,

So I did a bit of testing on this, and while the system doesn't hang when you run "cat /img.bin >/dev/fb0", I'm not confident it's actually working either.

Did you actually see any output on the LCD after doing the resume?  Or are you assuming that it should be outputting to the LCD but you didn't actually seen anything because you don't think the LCD is enabled.

I'm testing on the EVM and not seeing any output, but I'm also testing on our proprietary platform and I'm not seeing output there either (even though I *know* the LCD is being reinitialized after the resume).

Further, if I run the following, I get a VPSS timeout.

root@dm814x-evm:/bin# cat /sys/devices/platform/vpss/display1/enabled
[  178.367187] VPSS_FVID2: contrl event 0x1004001a timeout
[  178.375000] VPSS_DCTRL: failed to get venc info.
[  178.375000] VPSS_DCTRL:  Failed to get venc infor
cat: read error: Invalid argument

In addition, the change you're proposing seems pretty dangerous.  That code path is used in a variety of places unrelated to the VPSS subsystem and it seems more likely that you should be waking up that work queue rather than commenting out the code that tells it to wait.

I'm just trying to understand whether you've actually confirmed your change results in the VPSS subsystem working properly after the resume, or whether you simply concluded that the change was correct because the system no longer hangs after commenting out that line of code.  I'm worried that the change you've proposed is just obscuring the actual problem (which is that the VPSS firmware is no longer responding to requests).

Devin

Devin,

Devin Heitmueller said:

Did you actually see any output on the LCD after doing the resume?  Or are you assuming that it should be outputting to the LCD but you didn't actually seen anything because you don't think the LCD is enabled.

No, I am providing you some hints that might help you. I received these hints from our PSP team.

Devin Heitmueller said:

In addition, the change you're proposing seems pretty dangerous.  That code path is used in a variety of places unrelated to the VPSS subsystem and it seems more likely that you should be waking up that work queue rather than commenting out the code that tells it to wait.

If you do not comment the wait_event_interruptible_tty(vt_event_waitqueue, vw->done); than how do you workaround the stuck after the "PM: Syncing filesystems ... done." message?

When I keep the wait_event_interruptible_tty(vt_event_waitqueue, vw->done); function, I can workaround the stuck with pressing Ctrl+C, but even after that I have the VPSS timeout:

root@dm814x-evm:/bin# cat /sys/devices/platform/vpss/display1/enabled
 VPSS_FVID2: contrl event 0x1004001a timeout
 VPSS_DCTRL: failed to get venc info.
 VPSS_DCTRL:  Failed to get venc infor
cat: read error: Invalid argument

So I doubt that commenting wait_event_interruptible_tty(vt_event_waitqueue, vw->done); is causing this VPSS timeout.

Which HDVPSS clock signals you are gating at suspend and then enable again at resume?

I can provide you a presentation for the DSP <-> Cortex-A8 communication at suspend/resume: 2480.DSP_Susp_design.pptx

I think that the M3 <-> Cortex A8 communication should be similar.

Regards,
Pavel

Devin,

I have also found that when using the 'core' test mode, the LCD suspend/resume fine. See the log below:

root@dm814x-evm:~# echo core > /sys/power/pm_test
root@dm814x-evm:~# echo mem > /sys/power/state
PM: Syncing filesystems ... done.
PM: Preparing system for mem sleep
PM: Adding info for No Bus:vcs63
PM: Adding info for No Bus:vcsa63
Freezing user space processes ... (elapsed 0.01 seconds) done.
Freezing remaining freezable tasks ... (elapsed 0.01 seconds) done.
PM: Entering mem sleep
Suspending console(s) (use no_console_suspend to debug)
platform reg-dummy: preparing suspend
platform mpu.0: preparing suspend
platform l3_slow.0: preparing suspend
omap_gpio omap_gpio.0: preparing suspend
omap_gpio omap_gpio.1: preparing suspend
omap_gpio omap_gpio.2: preparing suspend
omap_gpio omap_gpio.3: preparing suspend
omap_uart omap_uart.0: preparing suspend
omap_uart omap_uart.1: preparing suspend
omap_uart omap_uart.2: preparing suspend
omap_uart omap_uart.3: preparing suspend
omap_uart omap_uart.4: preparing suspend
omap_uart omap_uart.5: preparing suspend
omap_i2c omap_i2c.1: preparing suspend
omap_i2c omap_i2c.3: preparing suspend
davinci-mcasp davinci-mcasp.2: preparing suspend
mmci-omap-hs mmci-omap-hs.0: preparing suspend
omap2-nand omap2-nand.0: preparing suspend
ti81xx-usbss ti81xx-usbss: preparing suspend
omap-mailbox omap-mailbox: preparing suspend
omap2_mcspi omap2_mcspi.1: preparing suspend
omap2_mcspi omap2_mcspi.2: preparing suspend
omap2_mcspi omap2_mcspi.3: preparing suspend
omap2_mcspi omap2_mcspi.4: preparing suspend
omap2_elm omap2_elm.1: preparing suspend
arm-pmu arm-pmu.0: preparing suspend
davinci_mdio davinci_mdio.0: preparing suspend
cpsw cpsw.0: preparing suspend
ti81xx_pcie ti81xx_pcie.0: preparing suspend
edma edma: preparing suspend
davinci-pcm-audio davinci-pcm-audio: preparing suspend
d_can d_can: preparing suspend
ahci ahci.0: preparing suspend
omap_rtc omap_rtc: preparing suspend
vpss vpss: preparing suspend
platform t81xx_vidout: preparing suspend
TI81XX_HDMI TI81XX_HDMI: preparing suspend
ti81xxfb ti81xxfb: preparing suspend
platform ti81xxvin: preparing suspend
omap_wdt omap_wdt: preparing suspend
musb-hdrc musb-hdrc.0: preparing suspend
musb-hdrc musb-hdrc.1: preparing suspend
tps65910-pmic tps65910-pmic: preparing suspend
platform tps65910-rtc: preparing suspend
platform tps65910-power: preparing suspend
nop_usb_xceiv nop_usb_xceiv.0: preparing suspend
usb usb1: preparing type suspend, may wakeup
nop_usb_xceiv nop_usb_xceiv.1: preparing suspend
usb usb2: preparing type suspend, may wakeup
omap-iommu omap-iommu.0: preparing suspend
omap-iommu omap-iommu.1: preparing suspend
serial8250 serial8250: preparing suspend
soc-audio soc-audio.0: preparing suspend
pvrsrvkm pvrsrvkm: preparing suspend
pvrsrvkm pvrsrvkm: suspend
backlight tlc59108-bl: legacy class suspend
mmcblk mmc0:ac19: suspend
soc-audio soc-audio.0: suspend
rtc rtc0: legacy class suspend
input input0: type suspend
Generic PHY 0:01: suspend
Generic PHY 0:00: suspend
mtd mtd5ro: legacy class suspend
mtd mtd5: legacy class suspend
mtd mtd4ro: legacy class suspend
mtd mtd4: legacy class suspend
mtd mtd3ro: legacy class suspend
mtd mtd3: legacy class suspend
mtd mtd2ro: legacy class suspend
mtd mtd2: legacy class suspend
mtd mtd1ro: legacy class suspend
mtd mtd1: legacy class suspend
mtd mtd0ro: legacy class suspend
mtd mtd0: legacy class suspend
serial8250 serial8250: suspend
omap-iommu omap-iommu.1: suspend
omap-iommu omap-iommu.0: suspend
nop_usb_xceiv nop_usb_xceiv.1: suspend
nop_usb_xceiv nop_usb_xceiv.0: suspend
i2c 3-0021: suspend
i2c i2c-3: suspend
platform tps65910-power: suspend
platform tps65910-rtc: suspend
tps65910-pmic tps65910-pmic: suspend
tps65910 1-002d: suspend
qt602240_ts 1-004a: suspend
scsi host0: suspend
hub 2-0:1.0: suspend
usb usb2: type suspend, may wakeup
usb usb2: suspend, may wakeup
hub 1-0:1.0: suspend
usb usb1: type suspend, may wakeup
usb usb1: suspend, may wakeup
pcf857x 1-0021: suspend
tlc59108 1-0040: suspend
i2c 1-0020: suspend
tlv320aic3x-codec 1-0018: suspend
i2c 1-0023: suspend
i2c 1-0050: suspend
i2c i2c-1: suspend
musb-hdrc musb-hdrc.1: suspend
musb-hdrc musb-hdrc.0: suspend
spi spi1.0: legacy suspend
omap_wdt omap_wdt: suspend
platform ti81xxvin: suspend
ti81xxfb ti81xxfb: suspend
TI81XX_HDMI TI81XX_HDMI: suspend
platform t81xx_vidout: suspend
vpss vpss: suspend
omap_rtc omap_rtc: suspend
ahci ahci.0: suspend
d_can d_can: suspend
davinci-pcm-audio davinci-pcm-audio: suspend
edma edma: suspend
ti81xx_pcie ti81xx_pcie.0: suspend
cpsw cpsw.0: suspend
davinci_mdio davinci_mdio.0: suspend
arm-pmu arm-pmu.0: suspend
omap2_elm omap2_elm.1: suspend
omap2_mcspi omap2_mcspi.4: suspend
omap2_mcspi omap2_mcspi.3: suspend
omap2_mcspi omap2_mcspi.2: suspend
omap2_mcspi omap2_mcspi.1: suspend
omap-mailbox omap-mailbox: suspend
ti81xx-usbss ti81xx-usbss: suspend
omap2-nand omap2-nand.0: suspend
mmci-omap-hs mmci-omap-hs.0: suspend
davinci-mcasp davinci-mcasp.2: suspend
omap_i2c omap_i2c.3: suspend
omap_i2c omap_i2c.1: suspend
omap_uart omap_uart.5: suspend
omap_uart omap_uart.4: suspend
omap_uart omap_uart.3: suspend
omap_uart omap_uart.2: suspend
omap_uart omap_uart.1: suspend
omap_uart omap_uart.0: suspend
omap_gpio omap_gpio.3: suspend
omap_gpio omap_gpio.2: suspend
omap_gpio omap_gpio.1: suspend
omap_gpio omap_gpio.0: suspend
platform l3_slow.0: suspend
platform mpu.0: suspend
platform reg-dummy: suspend
PM: suspend of devices complete after 5.521 msecs
pvrsrvkm pvrsrvkm: LATE suspend
mmcblk mmc0:ac19: LATE suspend
soc-audio soc-audio.0: LATE suspend
input input0: LATE type suspend
mdio_bus 0:01: LATE suspend
mdio_bus 0:00: LATE suspend
scsi host0: LATE suspend
serial8250 serial8250: LATE suspend
omap-iommu omap-iommu.1: LATE suspend
omap-iommu omap-iommu.0: LATE suspend
hub 2-0:1.0: LATE suspend
usb usb2: LATE type suspend, may wakeup
usb usb2: LATE suspend, may wakeup
nop_usb_xceiv nop_usb_xceiv.1: LATE suspend
hub 1-0:1.0: LATE suspend
usb usb1: LATE type suspend, may wakeup
usb usb1: LATE suspend, may wakeup
nop_usb_xceiv nop_usb_xceiv.0: LATE suspend
i2c 3-0021: LATE suspend
i2c i2c-3: LATE suspend
platform tps65910-power: LATE suspend
platform tps65910-rtc: LATE suspend
tps65910-pmic tps65910-pmic: LATE suspend
tps65910 1-002d: LATE suspend
qt602240_ts 1-004a: LATE suspend
pcf857x 1-0021: LATE suspend
tlc59108 1-0040: LATE suspend
i2c 1-0020: LATE suspend
tlv320aic3x-codec 1-0018: LATE suspend
i2c 1-0023: LATE suspend
i2c 1-0050: LATE suspend
i2c i2c-1: LATE suspend
musb-hdrc musb-hdrc.1: LATE suspend
musb-hdrc musb-hdrc.0: LATE suspend
omap_wdt omap_wdt: LATE suspend
platform ti81xxvin: LATE suspend
ti81xxfb ti81xxfb: LATE suspend
TI81XX_HDMI TI81XX_HDMI: LATE suspend
platform t81xx_vidout: LATE suspend
vpss vpss: LATE suspend
omap_rtc omap_rtc: LATE suspend
ahci ahci.0: LATE suspend
d_can d_can: LATE suspend
davinci-pcm-audio davinci-pcm-audio: LATE suspend
edma edma: LATE suspend
ti81xx_pcie ti81xx_pcie.0: LATE suspend
cpsw cpsw.0: LATE suspend
davinci_mdio davinci_mdio.0: LATE suspend
arm-pmu arm-pmu.0: LATE suspend
omap2_elm omap2_elm.1: LATE suspend
omap2_mcspi omap2_mcspi.4: LATE suspend
omap2_mcspi omap2_mcspi.3: LATE suspend
omap2_mcspi omap2_mcspi.2: LATE suspend
omap2_mcspi omap2_mcspi.1: LATE suspend
omap-mailbox omap-mailbox: LATE suspend
ti81xx-usbss ti81xx-usbss: LATE suspend
omap2-nand omap2-nand.0: LATE suspend
mmci-omap-hs mmci-omap-hs.0: LATE suspend
davinci-mcasp davinci-mcasp.2: LATE suspend
omap_i2c omap_i2c.3: LATE suspend
omap_i2c omap_i2c.1: LATE suspend
omap_uart omap_uart.5: LATE suspend
omap_uart omap_uart.4: LATE suspend
omap_uart omap_uart.3: LATE suspend
omap_uart omap_uart.2: LATE suspend
omap_uart omap_uart.1: LATE suspend
omap_uart omap_uart.0: LATE suspend
omap_gpio omap_gpio.3: LATE suspend
omap_gpio omap_gpio.2: LATE suspend
omap_gpio omap_gpio.1: LATE suspend
omap_gpio omap_gpio.0: LATE suspend
platform l3_slow.0: LATE suspend
platform mpu.0: LATE suspend
platform reg-dummy: LATE suspend
PM: late suspend of devices complete after 0.977 msecs
suspend debug: Waiting for 5 seconds.
platform reg-dummy: EARLY resume
platform mpu.0: EARLY resume
platform l3_slow.0: EARLY resume
omap_gpio omap_gpio.0: EARLY resume
omap_gpio omap_gpio.1: EARLY resume
omap_gpio omap_gpio.2: EARLY resume
omap_gpio omap_gpio.3: EARLY resume
omap_uart omap_uart.0: EARLY resume
omap_uart omap_uart.1: EARLY resume
omap_uart omap_uart.2: EARLY resume
omap_uart omap_uart.3: EARLY resume
omap_uart omap_uart.4: EARLY resume
omap_uart omap_uart.5: EARLY resume
omap_i2c omap_i2c.1: EARLY resume
omap_i2c omap_i2c.3: EARLY resume
davinci-mcasp davinci-mcasp.2: EARLY resume
mmci-omap-hs mmci-omap-hs.0: EARLY resume
omap2-nand omap2-nand.0: EARLY resume
ti81xx-usbss ti81xx-usbss: EARLY resume
omap-mailbox omap-mailbox: EARLY resume
omap2_mcspi omap2_mcspi.1: EARLY resume
omap2_mcspi omap2_mcspi.2: EARLY resume
omap2_mcspi omap2_mcspi.3: EARLY resume
omap2_mcspi omap2_mcspi.4: EARLY resume
omap2_elm omap2_elm.1: EARLY resume
arm-pmu arm-pmu.0: EARLY resume
davinci_mdio davinci_mdio.0: EARLY resume
cpsw cpsw.0: EARLY resume
ti81xx_pcie ti81xx_pcie.0: EARLY resume
edma edma: EARLY resume
davinci-pcm-audio davinci-pcm-audio: EARLY resume
d_can d_can: EARLY resume
ahci ahci.0: EARLY resume
omap_rtc omap_rtc: EARLY resume
vpss vpss: EARLY resume
platform t81xx_vidout: EARLY resume
TI81XX_HDMI TI81XX_HDMI: EARLY resume
ti81xxfb ti81xxfb: EARLY resume
platform ti81xxvin: EARLY resume
omap_wdt omap_wdt: EARLY resume
musb-hdrc musb-hdrc.0: EARLY resume
musb-hdrc musb-hdrc.1: EARLY resume
i2c i2c-1: EARLY resume
i2c 1-0050: EARLY resume
i2c 1-0023: EARLY resume
tlv320aic3x-codec 1-0018: EARLY resume
i2c 1-0020: EARLY resume
tlc59108 1-0040: EARLY resume
pcf857x 1-0021: EARLY resume
qt602240_ts 1-004a: EARLY resume
tps65910 1-002d: EARLY resume
tps65910-pmic tps65910-pmic: EARLY resume
platform tps65910-rtc: EARLY resume
platform tps65910-power: EARLY resume
i2c i2c-3: EARLY resume
i2c 3-0021: EARLY resume
nop_usb_xceiv nop_usb_xceiv.0: EARLY resume
usb usb1: EARLY resume
usb usb1: EARLY type resume
hub 1-0:1.0: EARLY resume
nop_usb_xceiv nop_usb_xceiv.1: EARLY resume
usb usb2: EARLY resume
usb usb2: EARLY type resume
hub 2-0:1.0: EARLY resume
omap-iommu omap-iommu.0: EARLY resume
omap-iommu omap-iommu.1: EARLY resume
serial8250 serial8250: EARLY resume
scsi host0: EARLY resume
mdio_bus 0:00: EARLY resume
mdio_bus 0:01: EARLY resume
input input0: EARLY type resume
soc-audio soc-audio.0: EARLY resume
mmcblk mmc0:ac19: EARLY resume
pvrsrvkm pvrsrvkm: EARLY resume
PM: early resume of devices complete after 0.701 msecs
platform reg-dummy: resume
platform mpu.0: resume
platform l3_slow.0: resume
omap_gpio omap_gpio.0: resume
omap_gpio omap_gpio.1: resume
omap_gpio omap_gpio.2: resume
omap_gpio omap_gpio.3: resume
omap_uart omap_uart.0: resume
omap_uart omap_uart.1: resume
omap_uart omap_uart.2: resume
omap_uart omap_uart.3: resume
omap_uart omap_uart.4: resume
omap_uart omap_uart.5: resume
omap_i2c omap_i2c.1: resume
omap_i2c omap_i2c.3: resume
davinci-mcasp davinci-mcasp.2: resume
mmci-omap-hs mmci-omap-hs.0: resume
omap2-nand omap2-nand.0: resume
ti81xx-usbss ti81xx-usbss: resume
omap-mailbox omap-mailbox: resume
omap2_mcspi omap2_mcspi.1: resume
omap2_mcspi omap2_mcspi.2: resume
omap2_mcspi omap2_mcspi.3: resume
omap2_mcspi omap2_mcspi.4: resume
omap2_elm omap2_elm.1: resume
arm-pmu arm-pmu.0: resume
davinci_mdio davinci_mdio.0: resume
cpsw cpsw.0: resume

CPSW phy found : id is : 0x4dd074

CPSW phy found : id is : 0x4dd074
ti81xx_pcie ti81xx_pcie.0: resume
edma edma: resume
davinci-pcm-audio davinci-pcm-audio: resume
d_can d_can: resume
ahci ahci.0: resume
omap_rtc omap_rtc: resume
vpss vpss: resume
platform t81xx_vidout: resume
TI81XX_HDMI TI81XX_HDMI: resume
ti81xxfb ti81xxfb: resume
platform ti81xxvin: resume
omap_wdt omap_wdt: resume
spi spi1.0: legacy resume
musb-hdrc musb-hdrc.0: resume
musb-hdrc musb-hdrc.1: resume
i2c i2c-1: resume
i2c 1-0050: resume
i2c 1-0023: resume
tlv320aic3x-codec 1-0018: resume
i2c 1-0020: resume
tlc59108 1-0040: resume
usb usb1: resume
usb usb1: type resume
usb usb2: resume
usb usb2: type resume
scsi host0: resume
pcf857x 1-0021: resume
qt602240_ts 1-004a: resume
hub 1-0:1.0: resume
hub 2-0:1.0: resume
tps65910 1-002d: resume
tps65910-pmic tps65910-pmic: resume
platform tps65910-rtc: resume
platform tps65910-power: resume
i2c i2c-3: resume
i2c 3-0021: resume
nop_usb_xceiv nop_usb_xceiv.0: resume
nop_usb_xceiv nop_usb_xceiv.1: resume
omap-iommu omap-iommu.0: resume
omap-iommu omap-iommu.1: resume
serial8250 serial8250: resume
mtd mtd0: legacy class resume
mtd mtd0ro: legacy class resume
mtd mtd1: legacy class resume
mtd mtd1ro: legacy class resume
mtd mtd2: legacy class resume
mtd mtd2ro: legacy class resume
mtd mtd3: legacy class resume
mtd mtd3ro: legacy class resume
mtd mtd4: legacy class resume
mtd mtd4ro: legacy class resume
mtd mtd5: legacy class resume
mtd mtd5ro: legacy class resume
Generic PHY 0:00: resume
Generic PHY 0:01: resume
input input0: type resume
rtc rtc0: legacy class resume
soc-audio soc-audio.0: resume
mmcblk mmc0:ac19: resume
backlight tlc59108-bl: legacy class resume
pvrsrvkm pvrsrvkm: resume
PM: resume of devices complete after 182.936 msecs
pvrsrvkm pvrsrvkm: completing resume
soc-audio soc-audio.0: completing resume
serial8250 serial8250: completing resume
omap-iommu omap-iommu.1: completing resume
omap-iommu omap-iommu.0: completing resume
usb usb2: completing type resume
nop_usb_xceiv nop_usb_xceiv.1: completing resume
usb usb1: completing type resume
nop_usb_xceiv nop_usb_xceiv.0: completing resume
platform tps65910-power: completing resume
platform tps65910-rtc: completing resume
tps65910-pmic tps65910-pmic: completing resume
musb-hdrc musb-hdrc.1: completing resume
musb-hdrc musb-hdrc.0: completing resume
omap_wdt omap_wdt: completing resume
platform ti81xxvin: completing resume
ti81xxfb ti81xxfb: completing resume
TI81XX_HDMI TI81XX_HDMI: completing resume
platform t81xx_vidout: completing resume
vpss vpss: completing resume
omap_rtc omap_rtc: completing resume
ahci ahci.0: completing resume
d_can d_can: completing resume
davinci-pcm-audio davinci-pcm-audio: completing resume
edma edma: completing resume
ti81xx_pcie ti81xx_pcie.0: completing resume
cpsw cpsw.0: completing resume
davinci_mdio davinci_mdio.0: completing resume
arm-pmu arm-pmu.0: completing resume
omap2_elm omap2_elm.1: completing resume
omap2_mcspi omap2_mcspi.4: completing resume
omap2_mcspi omap2_mcspi.3: completing resume
omap2_mcspi omap2_mcspi.2: completing resume
omap2_mcspi omap2_mcspi.1: completing resume
omap-mailbox omap-mailbox: completing resume
ti81xx-usbss ti81xx-usbss: completing resume
omap2-nand omap2-nand.0: completing resume
mmci-omap-hs mmci-omap-hs.0: completing resume
davinci-mcasp davinci-mcasp.2: completing resume
omap_i2c omap_i2c.3: completing resume
omap_i2c omap_i2c.1: completing resume
omap_uart omap_uart.5: completing resume
omap_uart omap_uart.4: completing resume
omap_uart omap_uart.3: completing resume
omap_uart omap_uart.2: completing resume
omap_uart omap_uart.1: completing resume
omap_uart omap_uart.0: completing resume
omap_gpio omap_gpio.3: completing resume
omap_gpio omap_gpio.2: completing resume
omap_gpio omap_gpio.1: completing resume
omap_gpio omap_gpio.0: completing resume
platform l3_slow.0: completing resume
platform mpu.0: completing resume
platform reg-dummy: completing resume
PM: Finishing wakeup.
Restarting tasks ... done.
root@dm814x-evm:~#

With the 'core' test mode, the LCD turn off then on with success.

The 'core' test mode tests the freezing of processes, suspending of devices, platform global control methods, the disabling of nonboot CPUs and suspending of platform/system devices. The "core" level tests the hardware and drivers as deeply as possible without creating a hibernation image.

For more information about the "core" test mode, see linux-src/Documentation/power/basic-pm-debugging.txt

I will further check about the difference between the normal mode and the "core" test mode.

Regards,
Pavel

Hi Pavel,

Thanks for taking the time to continue investigating this.

When you say "the LCD turns off then on with success", do you mean you actually see output/graphics on the LCD after the resume other than just the backlight turning on?  In my testing with your patch, the backlight turns on, but there is no actual graphics output to the LCD (which in this case is the only thing I would categorize as "success").

If you do see graphics, then certainly understanding why the core test works but the actual suspend doesn't would be a very useful thing to know.

Devin

Devin,

Devin Heitmueller said:

 When you say "the LCD turns off then on with success", do you mean you actually see output/graphics on the LCD after the resume other than just the backlight turning on?

I mean I see output/graphics on the LCD after the resume. The touch screen works fine. I run the 3D Graphics -> Chameleon demo (OpenGL ES Demos) and 2D Graphics -> Moving Blocks demo. Both work fine.

Also the VPSS-M3 firmware seems to behave correct after the resume:

omap_gpio omap_gpio.0: completing resume
platform l3_slow.0: completing resume
platform mpu.0: completing resume
platform reg-dummy: completing resume
PM: Finishing wakeup.
Restarting tasks ... done.
root@dm814x-evm:~# cat /sys/devices/platform/vpss/display1/enabled
1
root@dm814x-evm:~#

The difference between the "core" test and the actual suspend should be something related to the VPSS-M3 firmware (/usr/share/ti/ti-media-controller-utils/dm814x_hdvpss.xem3).In the basic-pm-debugging.txt we have:

In particular, the "core" test allows you to test everything except for the actual invocation of the platform firmware in order to put the system into the sleep state.

The error that we have after resume is the same as just unload/stop the VPSS-M3 firmware, without running any suspend/resume test:

Arago 2011.09 dm814x-evm ttyO0

dm814x-evm login: root
root@dm814x-evm:~# cat /sys/devices/platform/vpss/display1/enabled
1
root@dm814x-evm:~# firmware_loader 2 /usr/share/ti/ti-media-controller-utils/dm814x_hdvpss.xem3 stop
FIRMWARE: I2cInit will be done by M3
FIRMWARE: Memory map bin file not passed
Usage : firmware_loader <Processor Id> <Location of Firmware> <start|stop> [-mmap <memory_map_file>] [-i2c <0|1>]
===Mandatory arguments===
<Processor Id>         0: DSP, 1: Video-M3, 2: Vpss-M3
<Location of Firmware> firmware binary file
<start|stop>           to start/stop the firmware
===Optional arguments===
-mmap                  input memory map bin file name
-i2c                   0: i2c init not done by M3, 1(default): i2c init done by M3
FIRMWARE: isI2cInitRequiredOnM3: 1
FIRMWARE: Default memory configuration is used
Firmware Loader debugging not configured
Default FL_DEBUG: warning
Allowed FL_DEBUG levels: error, warning, info, debug, log
MemCfg: DCMM (Dynamically Configurable Memory Map) Version :  2.1.2.1
FIRMWARE: 2 stop Successful
root@dm814x-evm:~# cat /sys/devices/platform/vpss/display1/enabled
VPSS_FVID2: contrl event 0x1004001a timeout
VPSS_DCTRL: failed to get venc info.
VPSS_DCTRL:  Failed to get venc infor
cat: read error: Invalid argument
root@dm814x-evm:~#

I will further investigate on this.

Best regards,
Pavel

Hi Pavel,

Ah, excellent.  It sounds like you've successfully reproduced the behavior I'm seeing and narrowed it down to a specific case you can test with.  I look forward to you identifying the issue.

For my information, are you currently testing with the code change you mentioned in your response on Aug 14th where you commented out the call to wait_event_interruptible_tty()?

Devin

Devin,

Devin Heitmueller said:

For my information, are you currently testing with the code change you mentioned in your response on Aug 14th where you commented out the call to wait_event_interruptible_tty()?

Yes, I am testing with that change. Everything seems to work fine, I can not observe any side effects with that change.

When I test without this change, the "core" test hang at the "PM : Syncing filesystems... done" line. When I press Ctrl+C, the flow continues and finish with success:

dm814x-evm login: root
root@dm814x-evm:~# echo core > /sys/power/pm_test
root@dm814x-evm:~# echo mem > /sys/power/state
PM: Syncing filesystems ... done.  ------> stuck here, press Ctrl+C to continue
Freezing user space processes ... (elapsed 0.01 seconds) done.
Freezing remaining freezable tasks ... (elapsed 0.01 seconds) done.
PM: suspend of devices complete after 5.483 msecs
PM: late suspend of devices complete after 1.071 msecs
suspend debug: Waiting for 5 seconds.
PM: early resume of devices complete after 0.784 msecs

CPSW phy found : id is : 0x4dd074

CPSW phy found : id is : 0x4dd074
PM: resume of devices complete after 175.374 msecs
Restarting tasks ... done.

mem
root@dm814x-evm:~# cat /sys/devices/platform/vpss/display1/enabled
1
root@dm814x-evm:~#

When I test with the normal suspend/resume, the flow also stuck at the "PM : Syncing filesystems... done" line. When I press Ctrl+C, the flow continues and finish with black screen on the LCD, as the VPSS-M3 firmware is not re-loaded properly:

root@dm814x-evm:~# echo none > /sys/power/pm_test
root@dm814x-evm:~# echo mem > /sys/power/state
PM: Syncing filesystems ... done.    ------> stuck here, press Ctrl+C to continue
Freezing user space processes ... (elapsed 0.01 seconds) done.
Freezing remaining freezable tasks ... (elapsed 0.01 seconds) done.
PM: suspend of devices complete after 5.430 msecs
PM: late suspend of devices complete after 1.069 msecs
PM: early resume of devices complete after 1.029 msecs
 
CPSW phy found : id is : 0x4dd074

CPSW phy found : id is : 0x4dd074
PM: resume of devices complete after 175.144 msecs
Restarting tasks ... done.

mem
root@dm814x-evm:~# cat /sys/devices/platform/vpss/display1/enabled
VPSS_FVID2: contrl event 0x1004001a timeout
VPSS_DCTRL: failed to get venc info.
VPSS_DCTRL:  Failed to get venc infor
cat: read error: Invalid argument
root@dm814x-evm:~#

To sum up, the only difference when I leave the wait_event_interruptible_tty() function not commented in the code, is that I need to press Ctrl+C to workaround the stuck of the flow. All rest is the same.

Regards,
Pavel

Devin,

I found that other possible solution for workaround the stuck at the "PM: Syncing filesystems ... done." line is just stopping the Matrix application:

root@dm814x-evm:~# /etc/init.d/matrix-gui-e stop

When stopping the Matrix application, you do not need nor to comment the wait_event_interruptible_tty() function, neither to press Ctrl+C.

Regards,
Pavel

Hi Pavel,

That is an interesting observation, and hopefully it will lead to identifying the underlying problem in the TI platform.

I suspect what's likely happening is the matrix application is actively making calls into ti81xx_fb when the suspend happens, and whatever queued request the application has to the driver is what's getting stuck.  I would suspect though that it would happen with *any* application that's actively using the framebuffer.

As we discussed before, our expectation is that the TI platform will behave like every other Linux system out there - in that we should be able to suspend and resume the hardware without having to kill all the applications in userland which might be using the graphics stack.

I hope this provides some insight as to the real problem.

Devin

Hello Devin:

              Have you resolved your problem?I have the same question as you,After suspending the system,then resume it ,The LCD can't wok normally,and

          Thanks.

                                                                                                                                     zhichao.

Hello :

  I have the same problem,now the M3 VPSS cant work well after resuming the system,So  wheather I can prevent the LCD(FB or vpss model)not geting into suspend model when execute echo -n "mem" > /sys/power/state?

Thansk.

 zhichao.

Hello:

  And I see the  AM335x  is support the  Cortex-M3 get int to standby model by upgrating M3 source  binary ,According to "

AM335x Power Management Standby User's Guide

"