Booting Linux Kernel on Zynq Devices (Zybo and ZedBoard)

Question

We are trying to boot linux in a xilinx system following the instructions in wiki-xilinx-linux. After a lot of different tests, after had a look (among others) here, we have enabled the option "LOW LEVEL KERNEL DEBUGGING AND EARLYPRINTKs". Thanks to this we were able to see the following:

U-Boot 2016.07-03720-g95e11f6 (Oct 13 2016 - 03:48:21 -0700)

Model: Zynq MMC:   sdhci@e0100000: 0
SF: Detected S25FL256S_64K with page size 256 Bytes, eIn:    serial@e0001000
Out:   serial@e0001000
Err:   serial@e0001000
Model: Z00b000
Hit any key to stop autoboot:  0 
Copying Linux from SD to RAM...
Device: sdhci@e0100000
Manufacturer ID: 2
OEM: 544d
Name: SA04G 
Tran Speed:reading uImage
3842280 bytes read in 342 ms (10.7 MiB/s)
reading devicetree.dtb
8955 bytes read in 20 ms (436.5 KiB/s)
## Booting kernel from Legacy Image at 0008000
   Entry Point:  00008000
   Verifying Checksum ... OK
## Flattened Device Tree blob at 02a00000
   Booting using the fdt blob at OK
   Loading Device Tree to 1eb0e000, end 1eb132fa ... OK

Starting kernel .... done, booting the kernel.
Booting Linux on physical CPU 0x0
Linux version 4.6.0-xilinx-22282-g7d819bd (leProcessor [413fc090] revision 0 (ARMv7), cr=18c5387d
CPU: PIPT / VIPT nonaliasiMemory policy: Data cache writealloc
percpu: Embedded 12 pages/cpu @debcd000 s1115200 root=/dev/mmcblk0p2 rw rootwait
PID hash table entries: 2048 (order: 1, (order: 5, 131072 bytes)
Memory: 494528K/524288K available (5304K kernel code, or  : 0xffff0000 - 0xffff1000   (   4 kB)
    fixmap  : 0xffc00000 - 0xfff00000( 512 MB)
    pkmap   : 0xbfe00000 - 0xc0000000   (   2 MB)
    modules : 0xbf   .data : 0xc0900000 - 0xc0939b60   ( 231 kB)
       .bss : 0xc0939b60 - 0xc09R_CPUS=4 to nr_cpu_ids=2.
RCU: Adjusting geometry for rcu_fanout_leaf=32, nr_cp00 -> 0x72760000
L2C: DT/platform modifies aux control register: 0x72360000 -> ortex-A9
L2C-310 ID prefetch enabled, offset 1 lines
L2C-310 dynamic clock gatnq_clock_init: clkc starts at e0802100
Zynq clock init
sched_clock: 64 bits at_idle_ns: 440795209040 ns
Switching to timer-based delay loop, resolution 3ns
my device 80x30
Calibrating delay loop (skipped), value calculated using timer )
Mountpoint-cache hash table entries: 1024 (order: 0, 4096 bytes)
CPU: Testinx100058
CPU1: thread -1, cpu 1, socket 0, mpidr 80000001
Brought up 2 CPUs
SMP
   VFP support v0.3: implementor 41 architecture 3 part 30 variant 9 rev 4
clocksT: Registered protocol family 16
DMA: preallocated 256 KiB pool for atomic cohe maximum watchpoint size is 4 bytes.
zynq-ocm f800c000.ocmc: ZYNQ OCM pool: 256interface driver usbfs
usbcore: registered new interface driver hub
usbcore: r00
pps_core: LinuxPPS API ver. 1 registered
pps_core: Software ver. 5.3.6 - CoLinux Sound Architecture Driver Initialized.
clocksource: Switched to clocksourle entries: 4096 (order: 3, 32768 bytes)
TCP: Hash tables configured (establish (order: 1, 8192 bytes)
NET: Registered protocol family 1
RPC: Registered nameRegistered tcp NFSv4.1 backchannel transport module.
hw perfevents: enabled witx_order=17 bucket_order=0
jffs2: version 2.2. (NAND) (SUMMARY)  © 2001-2006 Rever for PL330 DMAC-241330
dma-pl330 f8003000.dmac:    DBUFF-128x8bytes Num_Chans-led
xdevcfg f8007000.devcfg: ioremap 0xf8007000 to e086c000
[drm] Initialized brd: module loaded
loop: module loaded
CAN device driver interface
libphy: MACB_mii_bus: probed
macb e000b000.ethernet eth0: Cadence GEM rev 0x00020118 at 0xe000b000 irq 145 (0rne:00, irq=-1)
e1000e: Intel(R) PRO/1000 Network Driver - 3.2.6-k
e1000e: Copci-pci: EHCI PCI platform driver
usbcore: registered new interface driver usb-si2c /dev entries driver
EDAC MC: ECC not enabled
Xilinx Zynq CpuIdle Driver ste Ossman
sdhci-pltfm: SDHCI platform and OF driver helper
mmc0: Invalid maximummc0: SDHCI controller on e0100000.sdhci [e0100000.sdhci] using ADMA
ledtrig-cphid: USB HID core driver
NET: Registered protocol family 10
sit: IPv6 over IPv4 tunneling driver
NET: Registered protocol family 17
can: controller ar protocol (rev 20120528 t)
can: netlink gateway (rev 20130117) max_hops=1
Regards found.
Waiting for root device /dev/mmcblk0p2...
mmc0: new high speed SDHCmmcblk0: p1 p2
VFS: Cannot open root device "mmcblk0p2" or unknown-block(179,2): error -30
Ple4 ram1  (driver?)
0102           16384 ram2  (driver?)
0103           16384 ra     16384 ram7  (driver?)
0108           16384 ram8  (driver?)
0109          0c           16384 ram12  (driver?)
010d           16384 ram13  (driver?)
010e ?)
b300         3813376 mmcblk0  driver: mmcblk
  b301         1048576 mmcblk0: Unable to mount root fs on unknown-block(179,2)
CPU1: stopping
CPU: 1 PID: 0nd_backtrace) from [<c010a608>] (show_stack+0x10/0x14)
[<c010a608>] (show_stackb24>] (ipi_cpu_stop+0x3c/0x70)
[<c010cb24>] (ipi_cpu_stop) from [<c010d394>] (h4>] (__irq_svc+0x54/0x90)
Exception stack(0xde46ff70 to 0xde46ffb8)
ff60:     01 debdc5c0 474d4ec6 00000000 45283f08 00000000 00000000 00000000
ffa0: 0000000/0x1c0)
[<c0492ccc>] (cpuidle_enter_state) from [<c014a72c>] (cpu_startup_entryFS: Unable to mount root fs on unknown-block(179,2)

It seems that the Virtual File System tries to read from some unknown-block:

VFS: Cannot open root device "mmcblk0p2" or unknown-block(179,2): error -30

Googling, no solutions were found. Has somebody some idea on how to solve this? Where can we have a look? Which parameter to play with?

From this output is not visible what you supply as a kernel command line. Do you have `rootwait` there? — 0andriy, Oct 13 '16 at 19:56
I do not thing root wait will help because the kernel has already detected both partitions. One thing you can do to debug it is add busy box to the ramdisk image and use that as root so that you can try to manually mount the second partition. Also, check that the second partition is mountable on another Linux machine. And verify that the appropriate filesystem is configured in your kernel. — Jamey Hicks, Nov 02 '16 at 14:32
I have a similar problem with my Zynq Mini-ITX, though. Trying to remember if I have actually solved it yet. It has the additional symptom that the partition showed up as read-only for no apparent reason. — Jamey Hicks, Nov 02 '16 at 14:36
I have just added an answer with the problems I found (and solved) — Leos313, Nov 02 '16 at 15:30

Leos313 · Accepted Answer · 2018-09-27T10:00:33.257

Finally I solved all the problems I had (running linux on Zybo with a custom hw in the Programmable Logic). Yes, because I am a beginner, I found different problems:

1) One of the problem was in the .dts: the part with eFuse was compleatly missing. Xilinx committed another version of the .dts with the problem fixed.

2) Another big problem was with the bitstream: at the beginning I wanted to program the PL on Zybo with a running linux using: cat bitstream.bit > /dev/xdevcfg. Of course this is compleatly possible (I tested it) as it is explaned in [this wiki-page][1]. Problem: if you have a custom HW in in FPGA you have also its description in the device-tree. The kernel tries to initialize all the periferics connected to the PS but, of course, the PL is not yet connected!! The solution is to download the bitstream before, including it in the boot.bif in order to generate the boot.bin in the right way:

    image : {
        [bootloader]fsbl.elf
        bitstream.bit
        u-boot.elf
        devicetree.dtb
        uramdisk.image.gz
        uImage.bin   // currently bootgen requires a file extension. this is just a renamed uImage
}

of course using: bootgen -image boot.bif -o i boot.bin. There were other little problems and I will try to add a list in future editing-revision of this answer.

UPDATE:

to solve the problem described in the question, if there are no additional problems/errors, should be enough to add the bootargs the option rootdelay=3 (rootdelay — Time to delay before attempting to mount the root filesystem). This can be necessary because u-boot tries to access the SD too fast and the SD is not ready yet.

score 1 · Answer 2 · answered Oct 13 '16 at 20:10

1

You have it set to use the root filesystem on mmcblk0p2 - the second partition of your SD-card. If the filesystem is in fact placed there, you have to verify that the kernel has support for what the partition was formatted with (e.g. ext3). Is this a prebuilt image or one you built yourself? If you use the command petalinux-config before building, you can change settings for where to store the filesystem. Btw, this is a very common problem so you should find lots of info if you use the right terms when you google it.

answered Oct 13 '16 at 20:10

Anders

41
3

I didn't use petalinux: I have used a ramdisk.gz downloaded by the wiki-xilinx-Linux Web page in the section "build and modify system file root". After, the ramdisk was insert in the same partition of the boot.bin, devicetree.dtb and uImage. – Leos313 Oct 13 '16 at 23:09
And I have created (thanks to xilinx -tools) all the file that I needed: the FSBL, U-BOOT, .DTS and so on. I hope that with this info the situation can be clear enough. Thank you so much for your support – Leos313 Oct 13 '16 at 23:26
1

For some reason, it looks for the root on the the second partition. Usually you have to change this by stopping U-boot at "Hit any key to stop autoboot" and examine the settings with "printenv". Look for "root=/dev/mmcblk0p2 rw rootwait". You can also start by looking at "default_bootcmd" and figure out what it does. On ZedBoard you can set the bootmode using jumpers so be sure to check them too. I'll get back to you when I have checked what the bootarguments should be when you use a ramdisk. – Anders Oct 15 '16 at 18:27
Thank you again. Maybe we have found one of the problem: I will give you news on monday – Leos313 Oct 16 '16 at 00:17
Finally we have solved the problems (yes, we were facing more than ones): - Kernel command line: console=ttyPS0,115200 root=/dev/mmcblk0p2 rw earlyprintk rootfstype=ext4 rootwait devtmpfs.mount=1 - The file "zynq-7000.dtsi" was the wrong one: YESTERDAY Xilinx committed another version on git !!!!!!!!!! – Leos313 Oct 18 '16 at 09:26
1

Glad to hear it worked out. You could put that info in an answer and mark it as the correct answer. – Anders Oct 19 '16 at 19:02
I have just added an answer with the problems I found (and solved some days ago). I hadnt the opportunity to write it before – Leos313 Nov 02 '16 at 15:31

Booting Linux Kernel on Zynq Devices (Zybo and ZedBoard)

2 Answers2