I'm working with a Nios II design with an MMU, on a Nios II Development Kit, Cyclone II Edition board (EP2C35). I'm trying to figure out how to get to a system that boots with some sort of root filesystem stored in flash. With help from hippo's wiki page (http://www.nioswiki.com/linux) (http://www.nioswiki.com/linux%29), I can now:Build a kernel Download it with nios2-download Connect with nios2-terminal Get a Linux Command PromptI have also ported U-Boot (starting from the U-Boot code included in the nios-linux tarball from the above-mentioned wiki page). I can burn a kernel to flash, and have U-Boot boot that kernel. I can TFTP a vmImage file, and boot from that (though the system hangs due to lack of a root filesystem). The "bootargs" variable isn't passed from U-Boot to the Linux Kernel. To work around this, I compile the kernel command line arguments into the kernel (CONFIG_COMMANDLINE). By compiling in the kernel command-line arguments, I can boot with root=/dev/nfs. After adding a few devices to uClinux-Dist/romfs/dev, I was able to get a command prompt. Now, I want to boot without using NFS. Here are the files in my uClinux-dist/images directory: -rwxrwxr-x 1 pinta pinta 4442546 2009-09-28 15:21 linux.initramfs.gz -rw-rw-r-- 1 pinta pinta 2478080 2009-09-28 15:20 rootfs.initramfs -rw-rw-r-- 1 pinta pinta 15997 2009-09-28 15:20 rootfs.initramfs.contents -rw-rw-r-- 1 pinta pinta 1059759 2009-09-28 15:20 rootfs.initramfs.gz -rw-r--r-- 1 pinta pinta 579226 2009-09-28 15:21 System.map.initramfs.gz -rw-r--r-- 1 pinta pinta 1221552 2009-09-28 15:20 vmImage -rwxrwxr-x 1 pinta pinta 3383298 2009-09-28 15:20 vmlinux -rwxrwxr-x 1 pinta pinta 2296861 2009-09-28 15:21 zImage.initramfs.gz linux.initramfs.gz is the file I download with nios2-download. vmImage is the file I use to boot Linux from U-Boot. According to the nioswiki Linux page, rootfs.initramfs.gz is compressed initramfs to be used as initrd by u-boot. How do I use rootfs.initramfs.gz? Specifically: 1) How do I pass it to U-Boot? 2) What (if any) kernel command-line arguments should I compile into the Linux Kernel? Pinta

Why using u-Boot ? >>> "images/zImge.initramfs.gz is the compressed elf image with initramfs built-in" AFAIK, same just needs to be copied into RAM and started, now it will spawn the file system automatically. -Michael

There are a few reasons I would like to use U-Boot. It allows me to switch between kernel builds easily. It allows me to boot with an nfs filesystem, which is convenient for application development. I have an I2C-based RTC added to my board, and the Linux RTC driver expects the clock chip to be initialized by pre-Linux code. I'm looking to U-Boot to provide that pre-Linux initialization code. (Getting the I2C working in U-Boot is my next project.) -- I downloaded the zImage.initramfs.gz file to the board, with nios2-download. This runs, and gets me to a command prompt, just like linux.initramfs.gz does. zImage.initramfs.gz does not seem to be an image suitable for U-Boot. I tried downloading it directly to the board via U-Boot tftp, then running it using the U-Boot command "go". I also tried packing it up with mkimage, then downloading the result to the board with tftp, and booting from it with the U-Boot command "bootm". I eventually want to burn the filesystem into flash, and boot from U-Boot without a network connection. -Pinta

Please find some example u-boot scripts in this file, ftp://ftp.altera.com/outgoing/devkit/altera_niosii_wrlinux_deliverables.zip You may enable passing command line from u-boot with this kernel config, "NiosII Configuration"-->"Passed kernel command line from u-boot" It will be more flexible to keep compile-in command line empty and pass it from u-boot. So that you are free to boot from nfs or flash. You may enable jffs2 and mtd in kernel config, (or ubifs etc) so that your can root your fs on it. as stated in the wiki, "images/rootfs.jffs2 is jffs2 image, eg, cp rootfs.jffs2 /dev/mtd0. This is available when jffs2 is selected in kernel. Please note the flash erase sector size on 3c120 dev board is 128KB, you will have to specify "MKFS_JFFS2_FLAGS = -l -p -e 128" at the beginning of your product Makefile." I am working with Altera to prepare a user guide, which should be released soon. - Hippo

--- Quote Start --- It allows me to boot with an nfs filesystem, which is convenient for application development. --- Quote End --- You can mount ntfs within initramfs. It might or might no be sensible to chroot top same. -Michael

Hippo wrote: --- Quote Start --- You may enable passing command line from u-boot with this kernel config, "NiosII Configuration"-->"Passed kernel command line from u-boot" --- Quote End --- When I was trying to figure out what was going on, I verified that this was set in the kernel config. Despite this, when the kernel output prints "Kernel command line: ", it shows no bootargs (even when I set bootargs in U-Boot). I had to do weird things with the addresses in U-Boot, to get it to work with the MMU, as I documented in http://www.nioswiki.com/dasuboot/u-boot_with_mmu. It could be that I missed something, and the bootargs are being written to non-existent memory. Specifically, I prefixed the flash address and all the device addresses with "E". For example, my custom_fpga.h shows JTAG_UART_BASE is 0x21208B0. My EP2C35.h defines JTAG_UART_BASE as 0xE21208B0. (Getting the UART right is essential.... without it, you get no output to tell you what is wrong.) I did not move CFG_SDRAM_BASE and CFG_SRAM_BASE. They are still at 0x06000000 and 0x01400000 respectively (the same as in my custom_fpga.h). I did define TEXT_BASE as 0xC7FC0000, 256k from the end of the RAM, in kernel space. If I didn't do this, nios2-download failed. Should CFG_SDRAM_BASE be a 0xC6000000 (in kernel space) instead of 0x06000000? --- Quote Start --- You may enable jffs2 and mtd in kernel config, (or ubifs etc) so that your can root your fs on it. as stated in the wiki, "images/rootfs.jffs2 is jffs2 image, eg, cp rootfs.jffs2 /dev/mtd0. This is available when jffs2 is selected in kernel. Please note the flash erase sector size on 3c120 dev board is 128KB, you will have to specify "MKFS_JFFS2_FLAGS = -l -p -e 128" at the beginning of your product Makefile." --- Quote End --- The Altera EP2C35 board uses either the AM29LV128M or the S29GL128M10TFIR1, both of which have 64 kB sectors. I assume that means I should switch to "MKFS_JFFS2_FLAGS = -l -p -e 64". I turned on jffs2 support in the kernel build. The kernel builds, but the tools fail: make: Entering directory `/home/nina/projects/micropump/nios_linux/nios2-linux/uClinux-dist/user/mtd-utils' unset CC CFLAGS CPPFLAGS CROSS LDFLAGS; make -C 606f38a2221648ca5c5fa292c9f71d2ddd59fa66/ WITHOUT_XATTR=1 SUBDIRS= BUILDDIR=$PWD/build-606f38a2221648ca5c5fa292c9f71d2ddd59fa66-host/ TARGETS=" mkfs.jffs2" make: Entering directory `/home/nina/projects/micropump/nios_linux/nios2-linux/uClinux-dist/user/mtd-utils/606f38a2221648ca5c5fa292c9f71d2ddd59fa66' mkdir -p /home/nina/projects/micropump/nios_linux/nios2-linux/uClinux-dist/user/mtd-utils/build-606f38a2221648ca5c5fa292c9f71d2ddd59fa66-host/ gcc -I./include -DWITHOUT_XATTR -D_FILE_OFFSET_BITS=64 -O2 -g -Wall -Wextra -Wwrite-strings -Wno-sign-compare -c -o /home/nina/projects/micropump/nios_linux/nios2-linux/uClinux-dist/user/mtd-utils/build-606f38a2221648ca5c5fa292c9f71d2ddd59fa66-host/mkfs.jffs2.o mkfs.jffs2.c -g -Wp,-MD,/home/nina/projects/micropump/nios_linux/nios2-linux/uClinux-dist/user/mtd-utils/build-606f38a2221648ca5c5fa292c9f71d2ddd59fa66-host/.mkfs.jffs2.c.dep mkfs.jffs2.c:73:18: error: zlib.h: No such file or directory mkfs.jffs2.c: In function ‘recursive_populate_directory’: mkfs.jffs2.c:1336: warning: format ‘%9lu’ expects type ‘long unsigned int’, but argument 3 has type ‘__off64_t’ mkfs.jffs2.c:1345: warning: format ‘%9lu’ expects type ‘long unsigned int’, but argument 3 has type ‘__off64_t’ mkfs.jffs2.c:1354: warning: format ‘%9lu’ expects type ‘long unsigned int’, but argument 3 has type ‘__off64_t’ mkfs.jffs2.c:1384: warning: format ‘%9lu’ expects type ‘long unsigned int’, but argument 3 has type ‘__off64_t’ mkfs.jffs2.c:1395: warning: format ‘%9lu’ expects type ‘long unsigned int’, but argument 3 has type ‘__off64_t’ mkfs.jffs2.c: At top level: mkfs.jffs2.c:1519: warning: initialization discards qualifiers from pointer target type mkfs.jffs2.c:1521: warning: initialization discards qualifiers from pointer target type mkfs.jffs2.c: In function ‘main’: mkfs.jffs2.c:1881: warning: passing argument 2 of ‘recursive_add_host_directory’ discards qualifiers from pointer target type make: *** Error 1 make: Leaving directory `/home/nina/projects/micropump/nios_linux/nios2-linux/uClinux-dist/user/mtd-utils/606f38a2221648ca5c5fa292c9f71d2ddd59fa66' make: *** Error 2 make: Leaving directory `/home/nina/projects/micropump/nios_linux/nios2-linux/uClinux-dist/user/mtd-utils' make: *** Error 2 make: Leaving directory `/home/nina/projects/micropump/nios_linux/nios2-linux/uClinux-dist/user' make: *** Error 2 make: Leaving directory `/home/nina/projects/micropump/nios_linux/nios2-linux/uClinux-dist/user' make: *** Error 1 This seems to be due to a lack of a zlib.h. Recommendations?

Linux, U-Boot, and Root File Systems | Altera Community

13 Replies

Altera_Forum
Honored Contributor
16 years ago
The hardware is a standard Altera Development Kit, http://www.altera.com/products/devkits/altera/kit-nios-2c35.html

The Quartus design is probably a slightly tweaked version of the standard design that comes with that board. I directed the hardware team to the Linux for Nios II Wiki page, section "Create a Nios II design with MMU", and asked them to modify the design accordingly.

I don't think I will be able to post the FPGA design. I'll check tomorrow.

Is there any more specific information I could provide to help figure this out?

Or, do you have any recommendations on where I should be looking? Or what questions I should be asking the hardware team?

Something that looked weird to me:
As you had recommended, I copied images/rootfs.jffs2 to the flash, at the start of the area I allocated to jffs2, 0x00200000. I'm concerned that Linux is mucking around with lower addresses (The "CLEANMARKER" references in the output), where I have stored U-Boot and the kernel. Is this ok?

Altera_Forum

Honored Contributor

16 years ago

It seems I'm not done with silly mistakes.

The flash device is 16 bit, but it is being used in 8-bit mode.

Switching to ".width = 1" solves the problem of detecting two devices and twice as much flash.

It also eliminated the CLEANMARKER error.

Now, I think I am down to flash driver problems.

The Spansion part is slightly different from the original AMD part. Specifically, the erase commands need to be written to specific offsets in the flash. For the original part, it was sufficient to write the commands to the start address of the flash.

The older sector erase sequence (for flash starting at address 0x0, and sector to erase at SA) was:

Address Data

---------------------

0x0 0xAA

0x0 0x55

0x0 0x80

0x0 0xAA

0x0 0x55

SA 0x30

New Sector Erase Sequence:

Address Data

---------------------

0xAAA 0xAA

0x555 0x55

0xAAA 0x80

0xAAA 0xAA

0x555 0x55

SA 0x30

Here is the new boot log output. It does eventually get to a command prompt, but it keeps generating errors about erase and write errors.

Am I on the right track here?


Linux version 2.6.30-00494-g84a224b-dirty ([email protected]) (gcc version 4.1.2)# 47 Tue Oct 6 14:08:55 EDT 2009
console  enabled
Early printk initialized
Linux/Nios II-MMU
init_bootmem_node(?,0x6391, 0x6000, 0x8000)
free_bootmem(0x6391000, 0x1c6f000)
reserve_bootmem(0x6391000, 0x400)
Built 1 zonelists in Zone order, mobility grouping on.  Total pages: 8128
Kernel command line:
NR_IRQS:32
PID hash table entries: 128 (order: 7, 512 bytes)
Dentry cache hash table entries: 4096 (order: 2, 16384 bytes)
Inode-cache hash table entries: 2048 (order: 1, 8192 bytes)
We have 32768 pages of RAM
Memory available: 28824k/3648k RAM, 0k/0k ROM (2119k kernel code, 1528k data)
Calibrating delay loop... 41.98 BogoMIPS (lpj=209920)
Mount-cache hash table entries: 512
net_namespace: 492 bytes
NET: Registered protocol family 16
init_BSP(): registering device resources
bio: create slab <bio-0> at 0
NET: Registered protocol family 2
IP route cache hash table entries: 1024 (order: 0, 4096 bytes)
TCP established hash table entries: 1024 (order: 1, 8192 bytes)
TCP bind hash table entries: 1024 (order: 0, 4096 bytes)
TCP: Hash tables configured (established 1024 bind 1024)
TCP reno registered
NET: Registered protocol family 1
JFFS2 version 2.2. (NAND) © 2001-2006 Red Hat, Inc.
JFFS2: default compression mode: priority
msgmni has been set to 56
io scheduler noop registered
io scheduler anticipatory registered
io scheduler deadline registered
io scheduler cfq registered (default)
ttyJ0 at MMIO 0x21208b0 (irq = 1) is a Altera JTAG UART
ttyS0 at MMIO 0x2120840 (irq = 4) is a Altera UART
console handover: boot  -> real 
ttyS1 at MMIO 0x2120b00 (irq = 8) is a Altera UART
smc91x.c: v1.1, sep 22 2004 by Nicolas Pitre <[email protected]>
eth0: SMC91C11xFD (rev 2) at e2110300 IRQ 6 
eth0: Ethernet addr: 00:07:ed:0f:0a:53
physmap platform flash device: 01000000 at 00000000
Physmap Width: 1
physmap-flash.0: Found 1 x16 devices at 0x0 in 8-bit bank in cfi_chip_setup
 Amd/Fujitsu Extended Query Table at 0x0040
physmap-flash.0: CFI does not contain boot bank location. Assuming top.
number of CFI chips: 1
cfi_cmdset_0002: Disabling erase-suspend-program due to code brokenness.
cmdlinepart partition parsing not available
RedBoot partition parsing not available
Using physmap partition information
Creating 3 MTD partitions on "physmap-flash.0":
0x000000000000-0x000000060000 : "U-Boot"
0x000000060000-0x000000200000 : "uImage"
0x000000200000-0x000001000000 : "romfs/jffs2"
TCP cubic registered
NET: Registered protocol family 10
NET: Registered protocol family 17
RPC: Registered udp transport module.
RPC: Registered tcp transport module.
/home/pinta/projects/micropump/nios_linux/nios2-linux/linux-2.6/drivers/rtc/hctosys.c: unable to open rtc device (rtc0)
eth0: link down
ADDRCONF(NETDEV_UP): eth0: link is not ready
IP-Config: Unable to set interface netmask (-22).
eth0: link up, 100Mbps, full-duplex, lpa 0x45E1
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00120000: 0x55ff instead
JFFS2: Erase block at 0x00120000 is not formatted. It will be erased
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x002b0000: 0x55ff instead
JFFS2: Erase block at 0x002b0000 is not formatted. It will be erased
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00440000: 0x55ff instead
JFFS2: Erase block at 0x00440000 is not formatted. It will be erased
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x005d0000: 0x55ff instead
JFFS2: Erase block at 0x005d0000 is not formatted. It will be erased
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00760000: 0x55ff instead
JFFS2: Erase block at 0x00760000 is not formatted. It will be erased
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x008f0000: 0x55ff instead
JFFS2: Erase block at 0x008f0000 is not formatted. It will be erased
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00a00010: 0xd76a instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00a00014: 0x5000 instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00a00018: 0x6a00 instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00a0001c: 0x0040 instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00a00020: 0x0068 instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00a00024: 0x40ff instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00a00028: 0x6880 instead
Empty flash at 0x00a0002c ends at 0x00a0004c
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00a0004c: 0x0400 instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00a00050: 0x0404 instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00a00054: 0x0704 instead
Further such events for this erase block will not be printed

Altera_Forum
Honored Contributor
16 years ago
OK.... so those errors were just initialization stuff. They went away once the partition was formatted, I guess.

I accidentally blew away my FPGA image, but I have that back now. And, I think I've adjusted my partitions so that doesn't happen again.

Hippo, you are a genius (or at least very helpful).

Forum Discussion

Linux, U-Boot, and Root File Systems

13 Replies

Recent Discussions

Ashling RISC Free IDE fails to download ELF file

Dual Port Nios V BSP error

set_output_delay explained for dummies

[Test only]How to verify Agilex 3 in radiation tolerant environment

gdb server problem when debugging