No messages starting linux kernel

Hi Yordan,

Our design does not use the PMIC.  We use pre-programmed EXAR power supplies to sequence the the power to the processor and have checked that operation closely - all appears correct with that.  Is it possible the kernel is expecting the PMIC to be there?  I have attached the kernel config file below.

I did also change the bootfile from zImage to uImage but with no change in the results.  With low-level debug messages and earlyprintk options enabled in the kernel should I at least expect to see messages about decompressing the kernel image?  I am still concerned that I do not have configured the kernel to use uart1 properly somewhere but not sure if that is the issue or not.

My config file is attached below.  I noticed that CONFIG_TTY_PRINTK is not set (after reading help on this, I don't think it is necessary).  Is that perhaps why I do not see any of the earlyprintk messages?

Thanks,

Tim

am437x-ndct-debuginitramfs_defconfig.txt

Hi Tim,

Tim Dicke said:

 With low-level debug messages and earlyprintk options enabled in the kernel should I at least expect to see messages about decompressing the kernel image?

Yes, enabling low-level debug & earlyprintk options is a good idea. You should be able to see additional information on your serial console, and setting CONFIG_TTY_PRINTK should help. 

I'll check the config file & dts to see if I find something wrong in your setup.

Regarding your doubt on pmic:  

Tim Dicke said:

Our design does not use the PMIC.  We use pre-programmed EXAR power supplies to sequence the the power to the processor and have checked that operation closely - all appears correct with that.  Is it possible the kernel is expecting the PMIC to be there?  

Kernel does not necessarily expect pmic. But usually some modules nodes (i.e. &cpu, &mmc, ) in dts are set to use some voltage defined in the power supply node (in general case this is the tps chip). So, IMO, you should have some description of your pre-programmed EXAR power supplies in the dts. 

Best Regards, 
Yordan

OK.  I tried enabling CONFIG_TTY_PRINTK but still no new output after starting the kernel.  Just for my own clarification, my understanding is that after the u-boot scripts create the bootargs and load all the relevant pieces into memory (compressed kernel image and device tree), it first starts to execute a small piece of "bootstrap" code at the head of the kernel that is required to decompress the kernel image and in this case the initramfs.  This happens even ahead of processing the device tree correct?  Are the earlyprintk messages buffered until after the kernel configures the uart?  Is this perhaps why I don't see the early printks assuming the kernel hangs before the finish of device tree processing?  Just hoping to understand this better so I know where to start looking for the problem.

I am not sure what definitions are needed for our EXAR power supply in the dts.  Are you able to provide some guidance here?

Thanks again, Tim.

Hi Yordan,

Any new information you can supply here?  It has been a while.  I did some more to try and troubleshoot how for the boot process is progressing.  I am still not able to get any linux  kernel messages to output and not sure why.  I added my own very simple print function in misc.c so I could track the progress of the function decompress_kernel, as well as a few cache init functions and all looks ok.  I see plenty of puts and error functions in the linux preloader but I am not able to get any of them to send output to uart1 for some reason.  I'm sure I'm still missing something in the kernel setup to allow this but don't know what it could be.

#define UART1_BASE 0x48022000
#define UART_THD 0x0
static void myputc(const char ch)
{
 *((unsigned long *)(UART1_BASE+UART_THD)) = ch;
}

I tried to use the same technique to track progress in the kernel initialization but my simple function above no longer works there and causes the kernel to hang I think (not really sure).

Thanks again for your help.

Tim

I made a little bit of progress getting debug messages to come out from the linux preloader.  In the menuconfig output I supplied above, I had selected CONFIG_DEBUG_OMAP2UART2=y to try and enable debug messages to be routed to UART2 (ttyO1).  This did not work.  Instead, I have now selected CONFIG_DEBUG_LL_UART_8250=y.  This also forces one to configure both CONFIG_DEBUG_UART_PHYS and CONFIG_DEBUG_UART_VIRT.  For CONFIG_DEBUG_UART_PHYS it is easy to find the physical address of the appropriate UART in the TRM.

CONFIG_DEBUG_UART_PHYS=0x48022000 (for UART2 - ttyO1)

But I'm not sure what address to use for CONFIG_DEBUG_UART_VIRT.  It looks like there are predefined virtual addresses defined for other processors searching the web but couldn't find values for the AM437x.  Any ideas?

I now get the following messages out before the system hangs...

Starting kernel ...

Uncompressing Linux... done, booting the kernel.

Thanks,

Tim

Hi Yordan,

U-Boot# run args_mmc

U-Boot# printenv bootargs

bootargs=console=ttyO1,115200n8 root=PARTUUID=f6c1ce4b-b669-584e-b679-a5619081822f rw rootfstype=ext4 rootwait

I've also attached my updated menuconfig file for reference.

am437x-ndct_defconfig.txt

Hi Tim,

I'm seeing a similar issue, and I can make it happen on the TI's GP EVM as well. Specifically:

1. Use branch ti2013.12.01 from git://git.ti.com/ti-linux-kernel/ti-linux-kernel.git

2. Use the arch/arm/boot/dts/am437x-gp-evm.dts dts.

3. Use the tisdk_am437x-evm_defconfig config, which is their default config for from the ti2013.12.01 tag, excepting that I've enabled low level debugging and early printk.

Now, if you build all that, it works fine, but you don't get any early printk because you haven't set earlyprintk in U-Boot. If you do so, then you get nothing past "Starting Kernel". Now, I hacked it a bit and got some stuff printing out the serial port, but it doesn't quite work because it causes the system to hang after the call to local_flush_tlb_all (basically, when it dereferences the pointer to initial_boot_params in order to get at anything in that struct).

So, basically, using TI's reference implementation on TI hardware, early printk is broken. Any help that TI can provide to assist in this issue would be greatly appreciated. I expect that, if you fix the above issue with the EVM, the same fix will likely fix Tim's issue.

Hi Joe,

I'm pretty sure that's not it, because:

1. The earlyprintk messages are supposed to come out before init runs.
2. If I don't pass earlyprintk on the command line, I get the "normal" console output messages.
3. This is on the EVM, so UART0 is the console.
4. That option is selected in TI's reference config, and I didn't disable it.

Thanks for the thought though.

Hi Matthew,

I think it may be working for me.  I was missing the earlyprintk in my boot options.  now that I have that the messages below are what I get:

U-Boot SPL 2015.07-00114-gf8d68d1-dirty (Dec 07 2015 - 13:36:50)
USB0:   Register 2000440 NbrPorts 2
Starting the controller
USB XHCI 1.00
scanning bus 0 for devices... 2 USB Device(s) found
       scanning usb for storage devices... 1 Storage Device(s) found
SPL: Please implement spl_start_uboot() for your board
SPL: Direct Linux boot not active!
reading u-boot.img
reading u-boot.img


U-Boot 2015.07-00114-gf8d68d1-dirty (Dec 07 2015 - 13:36:50 -0500)

I2C:   ready
DRAM:  1 GiB
Success probing Si5388 at 0x70 - Initializing...
Si5388 disable all outputs.
Si5388 pause LOL.
Si5388 set register map.
Si5388 validating input clock status.
Si5388 configuring PLL for locking.
Si5388 initiate locking of PLL.
Si5388 restart LOL.
Si5388 waiting for PLL lock...
Si5388 LOS status = 0x8.
Si5388 copying FCAl values.
Si5388 set PLL to use FCAl values.
Si5388 enable outputs.
Si5388 initialization complete.
MMC:   OMAP SD/MMC: 0, OMAP SD/MMC: 1
reading uboot.env

** Unable to read "uboot.env" from mmc1:1 **
Using default environment

Net:   mac addr: c4:be:84:cc:7d:6b
<ethaddr> not set. Validating first E-fuse MAC
Registering cpsw_data...
cpsw, usb_ether
Hit any key to stop autoboot:  0
switch to partitions #0, OK
mmc1(part 0) is current device
SD/MMC found on device 1
reading uEnv.txt
732 bytes read in 2 ms (357.4 KiB/s)
Loaded environment from uEnv.txt
Importing environment from mmc ...
Failed to mount ext2 filesystem...
** Unrecognized filesystem type **
starting USB...
USB0:   Register 2000440 NbrPorts 2
Starting the controller
USB XHCI 1.00
scanning bus 0 for devices... 2 USB Device(s) found
       scanning usb for storage devices... 1 Storage Device(s) found

USB device 0:
    Device 0: Vendor: JetFlash Rev: 1100 Prod: Transcend 2GB
            Type: Removable Hard Disk
            Capacity: 2048.0 MB = 2.0 GB (4194304 x 512)
... is now current device
reading uEnv.txt
738 bytes read in 27 ms (26.4 KiB/s)
Loaded environment from uEnv.txt
Importing environment from mmc ...
3526552 bytes read in 2123 ms (1.6 MiB/s)
35996 bytes read in 285 ms (123 KiB/s)
Booting from usb 0...
Kernel image @ 0x82000000 [ 0x000000 - 0x35cf98 ]
## Flattened Device Tree blob at 88000000
   Booting using the fdt blob at 0x88000000
   Loading Device Tree to 8fff4000, end 8ffffc9b ... OK

Starting kernel ...

Uncompressing Linux... done, booting the kernel.
[    0.000000] Booting Linux on physical CPU 0x0
[    0.000000] Initializing cgroup subsys cpuset
[    0.000000] Initializing cgroup subsys cpu
[    0.000000] Initializing cgroup subsys cpuacct
[    0.000000] Linux version 4.1.6-g52c4aa7 (udsp@udsp-VirtualBox) (gcc version 4.9.3 20150413 (prerelease) (Linaro GCC 4.9-2015.05) ) #18 SMP Mon Dec 7 18:15:12 EST 2015
[    0.000000] CPU: ARMv7 Processor [412fc09a] revision 10 (ARMv7), cr=10c5387d
[    0.000000] CPU: PIPT / VIPT nonaliasing data cache, VIPT aliasing instruction cache
[    0.000000] Machine model: Generic AM43 (Flattened Device Tree)
[    0.000000] bootconsole [earlycon0] enabled
[    0.000000] cma: Reserved 16 MiB at 0xbf000000
[    0.000000] Memory policy: Data cache writeback

My uEnv.txt file: 

# This uEnv.txt file can contain additional environment settings that you
# want to set in U-Boot at boot time.  This can be simple variables such
# as the serverip or custom variables.  The format of this file is:
#    variable=value
# NOTE: This file will be evaluated after the bootcmd is run and the
#       bootcmd must be set to load this file if it exists (this is the
#       default on all newer U-Boot images.  This also means that some
#       variables such as bootdelay cannot be changed by this file since
#       it is not evaluated until the bootcmd is run.
#optargs=video=HDMI-A-1:800x600
optargs=earlyprintk

# Uncomment the following line to enable HDMI display and disable LCD display.
#fdtfile=am437x-gp-evm-hdmi.dtb

Hi Tim,

I don't think those aren't the early printk output. By the time you see the timestamps leading the messages, they're the ones that printk has buffered and then spits out once the main console has been set up. That's why their timestamp is all 0.00000.

Early printk messages look like this:

## Booting kernel from Legacy Image at 90000000 ...
   Image Name:   Linux-3.12.10-00023-g508133e
   Image Type:   ARM Linux Kernel Image (uncompressed)
   Data Size:    4135088 Bytes = 3.9 MiB
   Load Address: 80008000
   Entry Point:  80008000
   Verifying Checksum ... OK
## Flattened Device Tree blob at 85000000
   Booting using the fdt blob at 0x85000000
   Loading Kernel Image ... OK
   Loading Device Tree to bef2f000, end bef3b70c ... OK

Starting kernel ...

6Booting Linux on physical CPU 0x0
5Linux version 3.12.10-00023-g508133e (mattc@E2-06L) (gcc version 4.9.2 20140904 (prerelease) (crosstool-NG linaro-1.13.1-4.9-2014.09 - Linaro GCC 4.9-2014.09) ) #32 Fri Dec 4 13:11:37 EST 2015
CPU: ARMv7 Processor [412fc09a] revision 10 (ARMv7), cr=10c53c7d
CPU: PIPT / VIPT nonaliasing data cache, VIPT aliasing instruction cache
6Machine: Generic AM43 (Flattened Device Tree), model: TI AM437x gp EVM
7search "chosen", depth: 0, uname: 
7search "chosen", depth: 1, uname: chosen
7Looking for initrd properties... 7Command line is: console=0,115200 sxid=sxni-am43xx_cpu9 sxserno= panic=5 ubi.mtd=8 ubi.mtd=9 ubi.mtd=10 root=ubi0_0 mtdparts=nand.0:128k(spl0),128k(spl1),128k(spl2),128k(spl3),1m(uboot),256k(env1),256k(env2),64M(buffer),64M(root1),64M(root2),8M(config),7dt_root_size_cells = 1
7dt_root_addr_cells = 1
7memory scan node memory, reg size 8, data: 80 40 2000000 1000000,
7 - 80000000 ,  40000000
6cma: CMA: reserved 24 MiB at ae000000
Memory policy: ECC disabled, Data cache writeback
7On node 0 totalpages: 261376
7free_area_init_node: node 0, pgdat c08426f0, node_mem_map c087d000
7  Normal zone: 1520 pages used for memmap
7  Normal zone: 0 pages reserved
7  Normal zone: 194560 pages, LIFO batch:31
7  HighMem zone: 528 pages used for memmap
7  HighMem zone: 66816 pages, LIFO batch:15

And then, after that, it should flip the console to the new one and will emit what you see.

I do find it interesting that you don't see a hang when you enable earlyprintk.. but you're also not getting the early messages.

I tried removing earlyprintk option but lost all the messages:

Kernel image @ 0x82000000 [ 0x000000 - 0x35cff0 ]
## Flattened Device Tree blob at 88000000
   Booting using the fdt blob at 0x88000000
   Loading Device Tree to 8fff4000, end 8ffffc9b ... OK

Starting kernel ...

Uncompressing Linux... done, booting the kernel.

See attached...  I think my address for CONFIG_DEBUG_UART_VIRT is almost certainly wrong but don't know what it should be.

7875.am437x-ndct_defconfig.txt

Okay, so I grabbed your config and I noticed that the main difference is that you're using the generic 8250 UART option whereas I was using the TI81XX option. This ends up using different drivers, hence the different behavior. When I enable the generic 8250 UART and give it the address of UART0 (for the EVM), I get the same results as you do, namely:

Kernel image @ 0x80200000 [ 0x000000 - 0x3f1770 ]
## Flattened Device Tree blob at 80f80000
   Booting using the fdt blob at 0x80f80000
   Loading Device Tree to 9fff1000, end 9ffff587 ... OK

Starting kernel ...

[    0.000000] Booting Linux on physical CPU 0x0
[    0.000000] Linux version 3.12.10-00003-ge9bef61 (mattc@E2-06L) (gcc version 4.9.2 20140904 (prerelease) (crosstool-NG linaro-1.13.1-4.9-2014.09 - Linaro GCC 4.9-2014.09) ) #3 Wed Dec 9 12:57:09 EST 2015
[    0.000000] CPU: ARMv7 Processor [412fc09a] revision 10 (ARMv7), cr=10c53c7d
[    0.000000] CPU: PIPT / VIPT nonaliasing data cache, VIPT aliasing instruction cache
[    0.000000] Machine: Generic AM43 (Flattened Device Tree), model: TI AM437x gp EVM
[    0.000000] bootconsole [earlycon0] enabled
[    0.000000] cma: CMA: reserved 24 MiB at ae000000
[    0.000000] Memory policy: ECC disabled, Data cache writeback

So, it appears that I was either in error, or the different drivers have slightly different output formats.

Answering other questions:

1. Yes, I am on the earlier kernel. 3.12 series.
2. Yes, the console UART and the low-level debug UART can be 2 different ports, but they should also be allowed to be the same (at least, they have been in the past). The idea is that the debug one is used until the main one is initialized. I find that the debug output is vital for figuring out problems with device tree misconfiguration - hence why I'm working on getting the debug stuff working on the EVM, as it has a known working device tree. It is quite possible that your board is booting enough that you should be able to see console messages - you just need to have your console boot parameter properly set (something like console=ttyS1,115200.. though looking at the EVM, it looks like (at least on 3.12) the device may be named ttyO1.
3. For what it's worth, I've tried setting the consoles to two different ports in case they're fighting each other - no improvement.

I was wrong before.  While I do have a newer version of the TRM (SPRUHL7D from April 2015) it does reference them the same as you described above.  I think I got them confused with the uboot references to COM1-COM6 and the menuconfig references to debug UARTs being 1 based instead of 0-based.  I'm easily confused though... ;)

I don't know if you found this 

Thought maybe it might offer some clues to you that I've not been able to glean out of it.

Tim

Wow Matthew!  Thanks.  I swear I tried configuring for CONFIG_DEBUG_TI81XXUART2 but maybe that was before I realized I needed to have the earlyprintk in my boot options as well as in the menuconfig setup not sure.  Poking around the code I found what looked like support for CONFIG_DEBUG_AM33XXUART2 but there wasn't an option to select it in the menuconfig.  I tried adding it manually but it did not work.

I did not quite understand what you were referring to for the fdt_high environment variable so I did not make that change.  Maybe I don't need this for my board? These are my boot messages now (I think I'm past the point where the earlyprintk would hang before):

Kernel image @ 0x82000000 [ 0x000000 - 0x35d2f8 ]
## Flattened Device Tree blob at 88000000
   Booting using the fdt blob at 0x88000000
   Loading Device Tree to 8fff4000, end 8ffffc9b ... OK

Starting kernel ...

[    0.000000] Booting Linux on physical CPU 0x0
[    0.000000] Initializing cgroup subsys cpuset
[    0.000000] Initializing cgroup subsys cpu
[    0.000000] Initializing cgroup subsys cpuacct
[    0.000000] Linux version 4.1.6-g52c4aa7 (udsp@udsp-VirtualBox) (gcc version 4.9.3 20150413 (prerelease) (Linaro GCC 4.9-2015.05) ) #43 SMP Thu Dec 10 11:52:17 EST 2015
[    0.000000] CPU: ARMv7 Processor [412fc09a] revision 10 (ARMv7), cr=10c5387d
[    0.000000] CPU: PIPT / VIPT nonaliasing data cache, VIPT aliasing instruction cache
[    0.000000] Machine model: Generic AM43 (Flattened Device Tree)
[    0.000000] bootconsole [earlycon0] enabled
[    0.000000] cma: Reserved 16 MiB at 0xbf000000
[    0.000000] setup_arch: 0
[    0.000000] paging_init: 0[    0.000000] Memory policy: Data cache writeback
[    0.000000] paging_init: 1
[    0.000000] paging_init: 2paging_init: 3
[    0.000000] paging_init: 4[    0.000000] devicemaps_init: 7
[    0.000000] paging_init: 5
[    0.000000] paging_init: 6paging_init: 7
[    0.000000] paging_init: 8paging_init: 9
[    0.000000] On node 0 totalpages: 262144
[    0.000000] free_area_init_node: node 0, pgdat c099f800, node_mem_map eeef9000
[    0.000000]   Normal zone: 1710 pages used for memmap
[    0.000000]   Normal zone: 0 pages reserved
[    0.000000]   Normal zone: 194560 pages, LIFO batch:31
[    0.000000]   HighMem zone: 67584 pages, LIFO batch:15
[    0.000000] paging_init: 10
[    0.000000] paging_init: 11paging_init: 12
[    0.000000] setup_arch: 1setup_arch: 2
[    0.000000] setup_arch: 3setup_arch: 4
[    0.000000] setup_arch: 5[    0.000000] CPU: All CPU(s) started in SVC mode.
[    0.000000] AM437x ES1.2 (sgx neon )
[    0.000000] PERCPU: Embedded 13 pages/cpu @eeeb9000 s22976 r8192 d22080 u53248
[    0.000000] pcpu-alloc: s22976 r8192 d22080 u53248 alloc=13*4096
[    0.000000] pcpu-alloc: [0] 0
[    0.000000] Built 1 zonelists in Zone order, mobility grouping on.  Total pages: 260434
[    0.000000] Kernel command line: console=ttyO1,115200n8 debug earlyprintk root=/dev/sda2 rw rootfstype=ext4 rootwait
[    0.000000] PID hash table entries: 4096 (order: 2, 16384 bytes)
[    0.000000] Dentry cache hash table entries: 131072 (order: 7, 524288 bytes)
[    0.000000] Inode-cache hash table entries: 65536 (order: 6, 262144 bytes)
[    0.000000] mm_init:1
[    0.000000] mm_init:2
[    0.000000] Memory: 1003740K/1048576K available (6329K kernel code, 800K rwdata, 2320K rodata, 436K init, 8221K bss, 28452K reserved, 16384K cma-reserved, 253952K highmem)
[    0.000000] Virtual kernel memory layout:
[    0.000000]     vector  : 0xffff0000 - 0xffff1000   (   4 kB)
[    0.000000]     fixmap  : 0xffc00000 - 0xfff00000   (3072 kB)
[    0.000000]     vmalloc : 0xf0000000 - 0xff000000   ( 240 MB)
[    0.000000]     lowmem  : 0xc0000000 - 0xef800000   ( 760 MB)
[    0.000000]     pkmap   : 0xbfe00000 - 0xc0000000   (   2 MB)
[    0.000000]     modules : 0xbf000000 - 0xbfe00000   (  14 MB)
[    0.000000]       .text : 0xc0008000 - 0xc087ab14   (8651 kB)
[    0.000000]       .init : 0xc087b000 - 0xc08e8000   ( 436 kB)
[    0.000000]       .data : 0xc08e8000 - 0xc09b01f0   ( 801 kB)
[    0.000000]        .bss : 0xc09b3000 - 0xc11ba448   (8222 kB)
[    0.000000] mm_init:3
[    0.000000] mm_init:4
[    0.000000] mm_init:5
[    0.000000] mm_init:6
[    0.000000] mm_init:7
[    0.000000] mm_init:8
[    0.000000] Running RCU self tests
[    0.000000] Hierarchical RCU implementation.
[    0.000000]  RCU lockdep checking is enabled.
[    0.000000]  Additional per-CPU info printed with stalls.
[    0.000000]  RCU restricting CPUs from NR_CPUS=2 to nr_cpu_ids=1.
[    0.000000] RCU: Adjusting geometry for rcu_fanout_leaf=16, nr_cpu_ids=1
[    0.000000] NR_IRQS:16 nr_irqs:16 16
[    0.000000] L2C: platform modifies aux control register: 0x0e030000 -> 0x3e430000
[    0.000000] L2C: DT/platform modifies aux control register: 0x0e030000 -> 0x3e430000
[    0.000000] L2C-310 enabling early BRESP for Cortex-A9
[    0.000000] OMAP L2C310: ROM does not support power control setting
[    0.000000] L2C-310 ID prefetch enabled, offset 1 lines
[    0.000000] L2C-310 dynamic clock gating disabled, standby mode disabled
[    0.000000] L2C-310 cache controller enabled, 16 ways, 256 kB
[    0.000000] L2C-310: CACHE_ID 0x410000c9, AUX_CTRL 0x7e430000
[    0.000000] OMAP clockevent source: timer2 at 24000000 Hz
[    0.000013] sched_clock: 32 bits at 24MHz, resolution 41ns, wraps every 89478484971ns
[    0.008092] clocksource timer1: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 79635851949 ns
[    0.017535] OMAP clocksource: timer1 at 24000000 Hz
[    0.023603] Console: colour dummy device 80x30
[    0.028240] Lock dependency validator: Copyright (c) 2006 Red Hat, Inc., Ingo Molnar
[    0.036245] ... MAX_LOCKDEP_SUBCLASSES:  8
[    0.040503] ... MAX_LOCK_DEPTH:          48
[    0.044856] ... MAX_LOCKDEP_KEYS:        8191
[    0.049382] ... CLASSHASH_SIZE:          4096
[    0.053914] ... MAX_LOCKDEP_ENTRIES:     32768
[    0.058527] ... MAX_LOCKDEP_CHAINS:      65536
[    0.063138] ... CHAINHASH_SIZE:          32768
[    0.067759]  memory used by lock dependency info: 5167 kB
[    0.073355]  per task-struct memory footprint: 1152 bytes
[    0.078961] Calibrating delay loop... 1987.37 BogoMIPS (lpj=9936896)
[    0.136295] pid_max: default: 32768 minimum: 301
[    0.141391] Security Framework initialized
[    0.145824] Mount-cache hash table entries: 2048 (order: 1, 8192 bytes)
[    0.152658] Mountpoint-cache hash table entries: 2048 (order: 1, 8192 bytes)
[    0.162702] Initializing cgroup subsys blkio
[    0.167233] Initializing cgroup subsys memory
[    0.171841] Initializing cgroup subsys devices
[    0.176579] Initializing cgroup subsys freezer
[    0.181313] Initializing cgroup subsys perf_event
[    0.186270] CPU: Testing write buffer coherency: ok
[    0.192680] CPU0: thread -1, cpu 0, socket 0, mpidr 80000000
[    0.198704] Setting up static identity map for 0x80008340 - 0x800083b0
[    0.211191] Brought up 1 CPUs
[    0.214398] SMP: Total of 1 processors activated (1987.37 BogoMIPS).
[    0.220958] CPU: All CPU(s) started in SVC mode.
[    0.228975] devtmpfs: initialized
[    0.258618] VFP support v0.3: implementor 41 architecture 3 part 30 variant 9 rev 4
[    0.312000] omap_hwmod: tptc0 using broken dt data from edma
[    0.318314] omap_hwmod: tptc1 using broken dt data from edma
[    0.324609] omap_hwmod: tptc2 using broken dt data from edma

Sorry, I jumped ahead of myself. On my board. I was seeing the following:

Starting kernel ...

[    0.000000] Booting Linux on physical CPU 0x0
[    0.000000] Linux version 3.12.10 (mattc@E2-06L) (gcc version 4.9.2 20140904 (prerelease) (crosstool-NG linaro-1.13.1-4.9-2014.09 - Linaro GCC 4.9-2014.09) ) #7 PREEMPT Thu Dec 10 08:35:36 EST 2015
[    0.000000] CPU: ARMv7 Processor [412fc09a] revision 10 (ARMv7), cr=10c53c7d
[    0.000000] CPU: PIPT / VIPT nonaliasing data cache, VIPT aliasing instruction cache
[    0.000000] Machine: Generic AM43 (Flattened Device Tree), model: RedLion AM437x CPU9
[    0.000000] bootconsole [earlycon0] enabled
[    0.000000] Memory policy: ECC disabled, Data cache writeback
[    0.000000] Unable to handle kernel paging request at virtual address fef2e000
[    0.000000] pgd = c0004000
[    0.000000] [fef2e000] *pgd=00000000
[    0.000000] Internal error: Oops: 5 [#1] PREEMPT ARM
[    0.000000] Modules linked in:
[    0.000000] CPU: 0 PID: 0 Comm: swapper Not tainted 3.12.10 #7
[    0.000000] task: c05752e8 ti: c056a000 task.ti: c056a000
[    0.000000] PC is at __unflatten_device_tree+0x24/0x15c
[    0.000000] LR is at unflatten_device_tree+0x30/0x3c
[    0.000000] pc : [<c0303db8>]    lr : [<c0551dc8>]    psr: a00001d3
[    0.000000] sp : c056bf20  ip : c056bf58  fp : c056bf54
[    0.000000] r10: c0576218  r9 : c0599b90  r8 : c0576218
[    0.000000] r7 : c0532fcc  r6 : c0de7080  r5 : c055df54  r4 : fef2e000
[    0.000000] r3 : c05dda04  r2 : 0000feed  r1 : c05dd9f4  r0 : fef2e000
[    0.000000] Flags: NzCv  IRQs off  FIQs off  Mode SVC_32  ISA ARM  Segment kernel
[    0.000000] Control: 10c53c7d  Table: 80004059  DAC: 00000015
[    0.000000] Process swapper (pid: 0, stack limit = 0xc056a238)
[    0.000000] Stack: (0xc056bf20 to 0xc056c000)
[    0.000000] bf20: c0599b90 c0576218 c056bf54 c05dd9f4 c056bf5c c0532fcc c055df54 c0de7080
[    0.000000] bf40: 80000200 c0576218 c056bf6c c056bf58 c0551dc8 c0303da0 c05d6344 c055df54
[    0.000000] bf60: c056bfb4 c056bf70 c05329f0 c0551da4 c01c2e20 10c53c7d c056bfa4 c055ee04
[    0.000000] bf80: c052ff2c c04ba37c c04b8e7c 0000d440 c0572000 c056a000 00000001 80004059
[    0.000000] bfa0: 412fc09a 00000000 c056bff4 c056bfb8 c053085c c0532384 00000000 00000000
[    0.000000] bfc0: 00000000 00000000 00000000 c055ee04 10c53c7d c0572044 c055ee00 c0576320
[    0.000000] bfe0: 80004059 412fc09a 00000000 c056bff8 80008070 c05307ac 00000000 00000000
[    0.000000] Backtrace: 
[    0.000000] [<c0303d94>] (__unflatten_device_tree+0x0/0x15c) from [<c0551dc8>] (unflatten_device_tree+0x30/0x3c)
[    0.000000]  r8:c0576218 r7:80000200 r6:c0de7080 r5:c055df54 r4:c0532fcc
[    0.000000] [<c0551d98>] (unflatten_device_tree+0x0/0x3c) from [<c05329f0>] (setup_arch+0x678/0x6e0)
[    0.000000]  r5:c055df54 r4:c05d6344
[    0.000000] [<c0532378>] (setup_arch+0x0/0x6e0) from [<c053085c>] (start_kernel+0xbc/0x2e8)
[    0.000000] [<c05307a0>] (start_kernel+0x0/0x2e8) from [<80008070>] (0x80008070)
[    0.000000]  r9:412fc09a r8:80004059 r7:c0576320 r6:c055ee00 r5:c0572044
r4:10c53c7d
[    0.000000] Code: e1a07002 e50b1028 0a00003c e30f2eed (e5943000) 
[    0.000000] ---[ end trace da227214a82491b7 ]---
[    0.000000] Kernel panic - not syncing: Attempted to kill the idle task!

This crash was happening when trying to access the the device tree in memory after a local_flush_tlb_all. This did not happen on the EVM, and I noticed that the fdt was being loaded to a different place by the EVM's U-Boot. That is controlled by the fdt_high environment variable. Using the value from the EVM fixes it. I expect that this is because there is some internal MMU restriction in accessing that memory.

If your board has a small amount of memory (both the EVM and my board have 1G) then that is likely why you don't have this problem.

Something strange just happened for me.  I had a lot of pr_info() calls sprinkled around to try and find out where things were hanging up and with the success I had earlier, I thought I could remove them.  Now I only get this:

Kernel image @ 0x82000000 [ 0x000000 - 0x35ce90 ]
## Flattened Device Tree blob at 88000000
   Booting using the fdt blob at 0x88000000
   Loading Device Tree to 8fff4000, end 8ffffc9b ... OK

Starting kernel ...

Uncompressing Linux... done, booting the kernel.
[    0.000000] Booting Linux on physical CPU 0x0
[    0.000000] Initializing cgroup subsys cpuset
[    0.000000] Initializing cgroup subsys cpu
[    0.000000] Initializing cgroup subsys cpuacct
[    0.000000] Linux version 4.1.6-g52c4aa7 (udsp@udsp-VirtualBox) (gcc version 4.9.3 20150413 (prerelease) (Linaro GCC 4.9-2015.05) ) #44 SMP Thu Dec 10 12:30:00 EST 2015
[    0.000000] CPU: ARMv7 Processor [412fc09a] revision 10 (ARMv7), cr=10c5387d
[    0.000000] CPU: PIPT / VIPT nonaliasing data cache, VIPT aliasing instruction cache
[    0.000000] Machine model: Generic AM43 (Flattened Device Tree)
[    0.000000] bootconsole [earlycon0] enabled
[    0.000000] cma: Reserved 16 MiB at 0xbf000000
[    0.000000] Memory policy: Data cache writeback

I believe this is where local_flush_tlb_all is called...

The environment variable you are referring to "fdt_high", where exactly does it belong?  In the uenv.txt or is it compiled into your u-boot image?  I tried setting that manually "setenv fdt_high 0xa0000000" before doing "run usbboot" but I didn't get any joy...

Completely ignore this post.  I overwrote my .config file with one that reverted back to the 8250 debug UART settings...  Doh!

Perhaps it is a difference. No, I have not tried the latest SDK on my board.

I have learned a couple of things. First, by debugging via JTAG dumps of the printk log buffer, I figured out that the reason I couldn't get the serial port to work after a time was because the pinmux stuff wasn't working because I had turned off the pinconfig-single driver in my new kernel configuration. OOops.

Second, I note that ttyO1 may be correct, EXCEPT if you've added an alias to your DTS, as in:

serial0 = &uart1;

In that case, at least with my kernel, 

console=ttyO0,115200

gets me nice messages without earlyprintk.

I tried playing around with a few of your suggestions but none of them had any positive effects on my setup (only negative effects like no messages at all).

I seem to be stuck someplace in the device tree processing but having trouble finding exactly where.  I think it gets through omap_hwmod_setup_all() but not sure what comes after that in the call chain.  Tables of pointers to functions...

Interestingly enough, when it finally stopped spitting out messages, here's where I ended up:

[  141.712149] omap_hwmod: timer7: disabling clocks
[  141.716954] omap_hwmod: tpcc: enabling
[  141.720860] powerdomain: completed transition in 0 loops
[  141.726379] clockdomain: l3_clkdm: enabled
[  141.730635] omap_hwmod: tpcc: enabling clocks
[  141.735172] omap_hwmod: tpcc: _omap4_enable_module: 2
[  141.740407] omap_hwmod: tpcc: resetting
[  141.744409] omap_hwmod: tpcc: idling
[  141.748135] omap_hwmod: tpcc: _omap4_disable_module
[  141.753191] omap_hwmod: tpcc: disabling clocks
[  141.757819] omap_hwmod: uart2: enabling
[  141.761813] powerdomain: completed transition in 0 loops
[  141.767332] clockdomain: l4ls_clkdm: enabled
[  141.771766] omap_hwmod: uart2: enabling clocks
[  141.776390] clock: locking DPLL dpll_per_ck
[  141.780738] omap_hwmod: uart2: _omap4_enable_module: 2
[  141.786071] omap_hwmod: uart2: resetting
[  141.790150] omap_hwmod: uart2: resetting via OCP SOFTRESET

uart2.  Now I have uart2 disabled at the moment in the dts file so I wasn't sure why this was even being initialized.  When I looked closer at the dts include am4372.dtsi I found the following:

		uart1: serial@48022000 {
			compatible = "ti,am4372-uart","ti,omap2-uart";
			reg = <0x48022000 0x2000>;
			interrupts = <GIC_SPI 73 IRQ_TYPE_LEVEL_HIGH>;
			ti,hwmods = "uart2";
			status = "disabled";
		};

So...   uart1 is actually uart2 in the omap_hwmod setups...

Going to try and disable all the uarts in my dts for now.

Hey Tim,

Can you attach your current .dts? I wonder if you're not setting up the serial port correctly and the omap_hwmod goes off and tries to use the .dts and that's why it goes kaput.

(It's just a thought, and likely a long shot, but I don't have any better ideas).