Kernel 5.10.36 and the Jornada 720

Hello!
I finally had time to start working on an attempt to get the 5.10 Linux Kernel booting on the Jornada 720. I first like to get this Kernel to boot before I try to fix USB and maybe fiddle around with the audio driver ...

I started with a fresh copy from kernel.org. Everything looked promising, i did not apply any of my patches - i wanted to start with everything that is included in mainline. It compiled well, but sadly I am out of luck.

The kernel boots, but it doesnt load init ... I tried it in various combinations. With Debian Lenny (which supports armv4) and Debian Stretch and the BX emulation applied - with sysvinit or systemd. But it always gets stuck at loading init. So I think there is something wrong with the kernel itself (or my kernel config).

Maybe you guys can help me ...



Here is the console output:

--------linexec-------- 
LinExec: UART Initialized. 

LinExec: UART Initialized. 

LinExec: Beginning boot_linux. 

LinExec: Passing the point of no return.. Now. 

LinExec: Entering DRAMloader... 

Booting Linux on physical CPU 0x0 

Linux version 5.10.36 (stefan@spnub1804work) (arm-linux-gnueabi-gcc (Ubuntu/Linaro 7.5.0-3ubuntu1~18.04) 7.5.0, GNU ld (GNU Binutils for Ubuntu) 2.30) #1 Fri May 14 06:56:00 UTC 2021 

CPU: StrongARM-1110 [6901b118] revision 8 (ARMv4), cr=c020717f 

CPU: VIVT data cache, VIVT instruction cache 

Machine: HP Jornada 720 

Memory policy: Data cache writeback 

Zone ranges: 

  DMA      [mem 0x00000000c0000000-0x00000000c00fefff] 

  Normal   [mem 0x00000000c00ff000-0x00000000c7ffffff] 

Movable zone start for each node 

Early memory node ranges 

  node   0: [mem 0x00000000c0000000-0x00000000c7ffffff] 

Initmem setup node 0 [mem 0x00000000c0000000-0x00000000c7ffffff] 

  DMA zone: 32513 pages in unavailable ranges 

Built 1 zonelists, mobility grouping on.  Total pages: 32511 

Kernel command line: root=/dev/sda2 mem=128m rootdelay=5 console=ttySA0,115200n8 

Dentry cache hash table entries: 16384 (order: 4, 65536 bytes, linear) 

Inode-cache hash table entries: 8192 (order: 3, 32768 bytes, linear) 

mem auto-init: stack:off, heap alloc:off, heap free:off 

Memory: 125160K/131072K available (3582K kernel code, 187K rwdata, 716K rodata, 136K init, 86K bss, 5912K reserved, 0K cma-reserved) 

SLUB: HWalign=32, Order=0-3, MinObjects=0, CPUs=1, Nodes=1 

NR_IRQS: 16, nr_irqs: 61, preallocated irqs: 61 

irq: Cannot allocate irq_descs @ IRQ1, assuming pre-allocated 

irq: Cannot allocate irq_descs @ IRQ33, assuming pre-allocated 

random: get_random_bytes called from start_kernel+0x244/0x3c4 with crng_init=0 

sched_clock: 32 bits at 3686kHz, resolution 271ns, wraps every 582542222200ns 

clocksource: oscr0: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 518462577799 ns 

Console: colour dummy device 30x40 

printk: console [ttySA0] enabled 

Calibrating delay loop... 136.60 BogoMIPS (lpj=683008) 

pid_max: default: 32768 minimum: 301 

Mount-cache hash table entries: 1024 (order: 0, 4096 bytes, linear) 

Mountpoint-cache hash table entries: 1024 (order: 0, 4096 bytes, linear) 

CPU: Testing write buffer coherency: ok 

Setting up static identity map for 0xc0208240 - 0xc0208298 

clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns 

futex hash table entries: 256 (order: -1, 3072 bytes, linear) 

NET: Registered protocol family 16 

DMA: preallocated 256 KiB pool for atomic coherent allocations 

SA1111 Microprocessor Companion Chip: silicon revision 1, metal revision 1 

sa1111 sa1111.0: Providing IRQ61-115 

sa1111-rab 0400: dmabounce: registered device 

sa1111-rab 0600: dmabounce: registered device 

SCSI subsystem initialized 

clocksource: Switched to clocksource oscr0 

genirq: Setting trigger mode 3 for irq 112 failed (sa1111_type_irq+0x0/0x84) 

genirq: Setting trigger mode 3 for irq 114 failed (sa1111_type_irq+0x0/0x84) 

jornada720_pcmcia_configure_socket(): config socket 0 vcc 0 vpp 0 

genirq: Setting trigger mode 3 for irq 113 failed (sa1111_type_irq+0x0/0x84) 

genirq: Setting trigger mode 3 for irq 115 failed (sa1111_type_irq+0x0/0x84) 

jornada720_pcmcia_configure_socket(): config socket 1 vcc 0 vpp 0 

NET: Registered protocol family 2 

tcp_listen_portaddr_hash hash table entries: 512 (order: 0, 4096 bytes, linear) 

TCP established hash table entries: 1024 (order: 0, 4096 bytes, linear) 

TCP bind hash table entries: 1024 (order: 0, 4096 bytes, linear) 

TCP: Hash tables configured (established 1024 bind 1024) 

UDP hash table entries: 256 (order: 0, 4096 bytes, linear) 

UDP-Lite hash table entries: 256 (order: 0, 4096 bytes, linear) 

NET: Registered protocol family 1 

NetWinder Floating Point Emulator V0.97 (double precision) 

SSP: device initialized with irq 

workingset: timestamp_bits=30 max_order=15 bucket_order=0 

io scheduler mq-deadline registered 

io scheduler kyber registered 

Epson S1D13XXX FB Driver 

s1d13xxxfb: chip production id 4 = S1D13506 

s1d13xxxfb: chip revision 1 

s1d13xxxfb: regs mapped at 0x(ptrval), fb 512 KiB mapped at 0x(ptrval) 

Console: switching to colour frame buffer device 80x30 

fb0: S1D13xxx frame buffer device 

Serial: SA11x0 driver 

sa11x0-uart.1: ttySA1 at MMIO 0x80010000 (irq = 16, base_baud = 230400) is a SA1100 

sa11x0-uart.3: ttySA0 at MMIO 0x80050000 (irq = 18, base_baud = 230400) is a SA1100 

loop: module loaded 

jornada720_pcmcia_configure_socket(): config socket 1 vcc 33 vpp 33 

input: HP Jornada 720 keyboard as /devices/platform/jornada720_kbd/input/input0 

input: HP Jornada 7xx Touchscreen as /devices/platform/jornada_ts/input/input1 

sa1100-rtc sa1100-rtc: registered as rtc0 

sa1100-rtc sa1100-rtc: setting system clock to 1970-01-01T00:01:00 UTC (60) 

NET: Registered protocol family 17 

Waiting 5 sec before mounting root device... 

jornada720_pcmcia_configure_socket(): config socket 1 vcc 33 vpp 33 

jornada720_pcmcia_configure_socket(): config socket 1 vcc 33 vpp 33 

pcmcia_socket pcmcia_socket1: pccard: PCMCIA card inserted into slot 1 

pcmcia 1.0: pcmcia: registering new device pcmcia1.0 (IRQ: 111) 

jornada720_pcmcia_configure_socket(): config socket 1 vcc 33 vpp 0 

jornada720_pcmcia_configure_socket(): config socket 1 vcc 33 vpp 0 

scsi host0: pata_pcmcia 

ata1: PATA max PIO0 cmd 0xc8870000 ctl 0xc887000e irq 111 

ata1.00: CFA: CF Card, Ver7.02K, max PIO6 

ata1.00: 7806960 sectors, multi 1: LBA48 

scsi 0:0:0:0: Direct-Access     ATA      CF Card          .02K PQ: 0 ANSI: 5 

sd 0:0:0:0: [sda] 7806960 512-byte logical blocks: (4.00 GB/3.72 GiB) 

sd 0:0:0:0: [sda] Write Protect is off 

sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA 

 sda: sda1 sda2 sda3 

sd 0:0:0:0: [sda] Attached SCSI removable disk 

random: fast init done 

EXT2-fs (sda2): warning: mounting unchecked fs, running e2fsck is recommended 

VFS: Mounted root (ext2 filesystem) on device 8:2. 

Freeing unused kernel memory: 136K 

Kernel memory protection not selected by kernel config. 

Run /sbin/init as init process 

random: crng init done

Regards
Stefan

Postings on the net by some of the Raspberry Pi guys who get this message "random: crng init done" usually revolve around a couple of issues involving the system waiting on key presses or mouse movements or discrepancies in the blkid and uuid assigned to the sd card.

_dev-null

Hope this helps... makes sense to me: https://stackoverflow.com/questions/59249941/what-is-random-crng-ini...

here it appears to be dhcp related: https://github.com/OP-TEE/build/issues/306

something about a patch at the end of this, give you any ideas? https://patchwork.kernel.org/project/linux-crypto/patch/201804130130...

joval

Edited by joval 2021-06-02 4:21 AM

Hello!
Thanks for this, but I think it has nothing to do with crng, because this bug was fixed a long time ago. The message appears also in the other working kernels.
Sadly nobody seems to be interested in the arm kernel mailing lists ...

Regards
Stefan

Hi Stefan!

Do you have a copy of the zimage file for kernal 4.9.8 available for download?? I would like to tinker around with it and see how it runs on SD Cards if I can get a copy... it's not available on your Jornadaj720 site as far as I can see. Just point me in the right direction. Thanks in advance.

Regarding kernel 5.10 it looks like the bootloader mounts the drive as sda2. This is different from kernel 3.1.6.where the drive is listed/named hda2. Any clue as to why would this be different between kernels, seems it mounts sda2 and then crashes?

Thanks again for all your great work and sharing it! I am really enjoying being online with the J720!!

best regards,

joval

Edited by joval 2021-07-04 5:36 PM

_dev-null
Actually I think it was the 4.8.7.kernel (not 4.9) that you had working but suffered from swap file instability... so that's the one I would like to download if still available.

joval (posted from his Amazing J720/Devuan/RasPi hybrid... again!!)

Edited by joval 2021-07-04 6:54 PM

Hello!
Here you go: http://www.sp-net.at/nextcloud/s/SQbqbX64QdZarop (inside the folder "kernel4.8.7_should_not_be_used_(swap_bug)"

The reason why it is "sda" is because they abandoned the pata disk name convention ... sadly this has nothing to do with the crash - the kernel can access the disk.

Regards
Stefan

Edited by _dev-null 2021-07-05 6:59 AM

Hi Stefan!

I have the Kernel 4.8.7 working using Fat32/ext2/swap....many hours running several youtube videos etc with no crashes (via RasPi 4) using ultra speed microSD cards of 2, 16, 32, 64 Gb. Things are faster and video frame rate increased IMO, although mouse pointer can be a bit squirrelly at times with higher resolution.

Only crash occurred with my first try of an old slow 2 GB SD card with swap size of 158MB... when I decreased it to 100 Mb no crash occurred. But the faster cards have swap sizes up to 1Gb and seem rock solid stable AFAIK. I just replaced the 3.x.x kernel zImage with the 4.8.7. I thought you might like to know.you really have done it right!

best regards, herr Maestro!

joval

Edited by joval 2021-07-07 5:46 AM

Hello!
This is great news. Can you try to do an "apt update" with the 4.8.7 Kernel? And also try to install some packages with apt. By running apt the system uses a lot of RAM and SWAP.

Thanks!

Stefan

Hi Stefan!

FYI, no success with any "apt install or apt update"... downloads packages 25%( with swap file) to 100% (swap file deleted) then stalls/crashes. Interesting that X-forwarding works so well, though, which is great for the RasPi hybrid use. Since changing zImage is so easy, can one do "apt update" in 3.x.x then boot it later with 4.x.x zImage... can that work? Am I mixing apples with oranges???

I suppose TimBo's qemu emulator is one workaround, I may give it a try... But I'll keep hammering away at swap file variations, and hope the problem isn't in the "apt" program (anyway to check that out?). Maybe I will try an install with dpkg and a .deb file... start narrowing things down somewhat. Never a dull moment, eh. Oh, now I know more about armel(j720) vs armhf(RasPi) differences regarding floating point, etc.

Much appreciate your great and never ending efforts, Stefan!

best regards,

joval

Edited by joval 2021-07-08 2:20 AM

Stefan,

One other thing. J720 with stock 32mb can't really run apt commands either....even with swap file. Every time I tried it, it would be really slow and always fail. That's when I started using qemu emulator TimoBO suggested.

And yet, X-forwarding works as well on J720 32Mb as it does with J720 64Mb...watching youtube videos via the RasPi was the same speed and access for most part. Surfing the web, etc the same...which surprised me. So kernel 4.8.7 is still very useful if advantages outweigh all the work-arounds sort of like J720 32Mb with the 3.x.x kernel


I have upgraded my j720 rom/ram card to 64 Mb and apt commands run really quite well... with the 3.x.x kernel. Works but not so well (happy?) with the 4.8.7==> no apt functionality.

I will try the 4.8.7 kernel on a j720 32Mb for comparison. I don't have a J720 128Mb yet (I have all the parts but not yet the time or courage). Maybe you can try it with one of yours and see if apt works with a high speed SD card and ext2. Thanks in advance.

So, I see its still good news... as you said!

joval

Puzzling, my J820 easily ran 'apt' with 32mb and no swap file. Of course I don't have X running... yet.

Yes, ntware showed if you use the CLI command "free" with fluxbox running there is not enough ram left to do much, say run the apt unpacking program, unlike with 64Mb

Joval

Great news Stefan, i hope you continued success

Hello Stefan,
Thank you for working on a new kernel. Is there any update regarding this project? it would be interesting to see this new kernel in J720.

How do people like us get started with porting Linux?
I am interested in programming but when I want to port Linux, I found it overwhelming. Can you please provide some guidelines, share books, online resources, online courses on Coursera/udacity etc. Seeing your amazing work inspires me to become a better engineer (both hardware and software engineer).
Thank you!