Hi Adam，
Thank you for your professional insights. Following your suggestion, I added "DEFAULT_RELOC_arm64 = 0x0ec00000" to the Makeconf.local file in the root directory and then rebuilt the system. Upon checking the "Start address: 0x86000000", I found that after rebuilding with 0x0ec00000, the start address became 0x8ee00000. So, I changed DEFAULT_RELOC_arm64 to 0x05e00000, and after rebuilding, the start address was 0x86000000, which meets the requirement.
The contents of Makeconf.local are as follows:
`/l4re-snapshot-24.08.0/obj/l4/builddir$ cat Makeconf.local CROSS_COMPILE:=aarch64-linux-gnu- DEFAULT_RELOC_arm64 = 0x05e00000`
Build information: `/l4re-snapshot-24.08.0/obj/l4/builddir$ make uimage E=uvmm-2vm PT=s32g make[1]: Entering directory '/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/src/l4' [bootstrap] ... Building Dependencies [bootstrap - nofpu] ... Building Dependencies Building entry "uvmm-2vm". Merging images: mod00: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/fiasco/builddir/fiasco [796kB] mod01: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir/bin/arm64_armv8a/l4f/sigma0 [24kB] mod02: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir/bin/arm64_armv8a/l4f/moe [202kB] mod03: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir/bin/arm64_armv8a/l4f/uvmm [974kB] mod04: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir/bin/arm64_armv8a/l4f/l4re [110kB] mod05: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir/bin/arm64_armv8a/l4f/ned [547kB] mod06: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/conf/s32g2xxa-evb.dtb [53kB] mod07: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/conf/fsl-image-dom0less-s32g274aevb.cpio.gz [25679kB] mod08: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/src/l4/pkg/uvmm/configs/vmm.lua [5kB] mod09: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/conf/uvmm-2vm.ned [1kB] mod10: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/conf/Image [14211kB] mod11: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir/bin/arm64_armv8a/l4f/cons [280kB] [bootstrap - nofpu] ... Generating bootstrap.ld [bootstrap - nofpu] ... Compiling startup.o [bootstrap - nofpu] ==> Linking l4re.elf [bootstrap] ==> Image post-processing l4re.elf [bootstrap - nofpu] ==> l4re.elf built ==> Installing l4re.elf in image directory ==> Installing l4re_uvmm-2vm in image directory ==> Installing l4re_uvmm-2vm.elf in image directory [bootstrap - nofpu] ... Generating l4re.raw ==> Installing l4re.raw in image directory ==> Installing l4re_uvmm-2vm.raw in image directory [bootstrap - nofpu] ... Generating l4re.uimage Image Name: L4 Image #14 Created: Wed Feb 12 10:50:58 2025 Image Type: AArch64 Linux Kernel Image (uncompressed) Data Size: 44056496 Bytes = 43023.92 KiB = 42.02 MiB Load Address: 86000000 Entry Point: 86000000 ==> Installing l4re.uimage in image directory ==> Installing l4re_uvmm-2vm.uimage in image directory Image size(s) in bytes: l4re_uvmm-2vm.elf: 44692328 l4re.raw: 44056496 l4re.uimage: 44056560 Start address: --> Build-Nr: 14 [bootstrap - nofpu] ==> Installing l4re.elf to local build-tree [bootstrap - nofpu] ==> Installing l4re_uvmm-2vm to local build-tree [bootstrap - nofpu] ==> Installing l4re_uvmm-2vm.elf to local build-tree make[1]: Leaving directory '/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/src/l4'
`
I then placed it back onto the SD card to boot. Although this solved the previous overlap issue, it seems to hang and does not boot properly. I am unsure if this is related to the message: `Rev: unknown compiled with gcc 11.4.0 for NXP S32G`
because my Linux kernel image was built on another computer using the Yocto toolchain provided by NXP for S32G, while this machine uses`CROSS_COMPILE:=aarch64-linux-gnu-`to compile Fiasco and the L4 uImage.
Boot log:
`fatload mmc 0 0x8000e000 l4re_uvmm-2vm.uimage 44056560 bytes read in 1871 ms (22.5 MiB/s) => fatload mmc 0 0x80000000 s32g2xxa-evb.dtb 53414 bytes read in 17 ms (3 MiB/s) => boot boot bootd bootefi bootelf booti bootm bootp bootvx => bootm 0x8000e000 - 0x80000000 ## Booting kernel from Legacy Image at 8000e000 ... Image Name: L4 Image #14 Image Type: AArch64 Linux Kernel Image (uncompressed) Data Size: 44056496 Bytes = 42 MiB Load Address: 86000000 Entry Point: 86000000 Verifying Checksum ... OK ## Flattened Device Tree blob at 80000000 Booting using the fdt blob at 0x80000000 Loading Kernel Image Using Device Tree in place at 0000000080000000, end 00000000800100a5 fixup: pfe0 set to 00:01:be:be:ef:11 fixup: pfe1 set to 00:01:be:be:ef:22 fixup: pfe2 set to 00:01:be:be:ef:33
Starting kernel ...
L4 Bootstrapper Build: #14 Wed Feb 12 10:50:58 CST 2025, 11.4.0 ... Cache config: ON // Stuck here, no further execution. `
I noticed that JDB had started, so I entered the t command and got the following output:
`jdb: t thread : 6 <0xffff8d5c0000> CPU: 0:0 prio: 00 state : 001 ready wait for: --- polling: rcv descr: timeout : cpu time: 0 timeslice: 10000 us pager : [C: 0] task : kernel exc-hndl: [C: 0] UTCB : 00000000/00000000 vCPU : --- vCPU : ---
Regs (before debug entry from kernel mode): brk #0x0 x0 0000ffff40286930 0000000000000000 ldp x19, x30, [sp], #0x10 x2 0000000000000014 0000ffff402801e0 x4 0000ffff40286ab8 0000000000000000 0000000000000088 0000000000000014 x8 0000000000000040 0000000000000014 0000ffff001c8000 0000000000000014 x12 0000ffff4022c010 0000ffff4022bfd0 0000ffff4022bfd0 0000000000000000 x16 0000ffff001c8000 0000000000000010 0000000000000010 0000ffff4023ef70 psr=00000000600003c9 tpidr: urw=0000000000000000 uro=0000000000000000 pc=0000ffff4022ba98 ksp=0000ffff8d5c1d40 x30=0000ffff4022ba8c 1c00 0000000000000000 0000ffff40280980 0000000000000000 0000ffff40286930 1c20 0000000000000000 0000000000000014 0000ffff402801e0 0000ffff40286ab8 1c40 0000000000000000 0000000000000088 0000000000000014 0000000000000040
1c60 0000000000000014 0000ffff001c8000 0000000000000014 0000ffff4022c010`
Could you please help analyze this issue? Many thanks!
Regards, Qiang
On Tue, Feb 11, 2025 at 8:43 PM Adam Lackorzynski adam@os.inf.tu-dresden.de wrote:
Hi Qiang,
looks to me that bootstrap image to load is just too big as it interferes with the reserved PFE areas of the platform. It seems unfortunate that u-boot allows this to be loaded and that those areas are there at those locations.
Anyway, I would recommend to change the loading address of bootstrap. Please do this by creating a file Makeconf.local in the root directory of your build-tree (/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir) with the content: DEFAULT_RELOC_arm64 = 0x0ec00000
Then build the image again. Please check the output as it needs to state "Start address: 0x86000000"
Adam
On Mon Feb 10, 2025 at 17:25:22 +0800, qiang xu wrote:
Hi Adam,
- I followed your suggestion and changed *"kernel fiasco.debug
-serial_esc"* to *"kernel fiasco -serial_esc"*. 2. The error *"DT: FDT sanity check failed: FDT_ERR_BADMAGIC (-9)
Total
RAM: 0MB"* disappeared after reloading the device tree (dtb). 3. Below is the boot command I used.
`=> fatload mmc 0 0x80000000 s32g2xxa-evb.dtb 53414 bytes read in 17 ms (3 MiB/s) => fatload mmc 0 0x8000e000 l4re_uvmm-2vm.uimage 44056560 bytes read in 1871 ms (22.5 MiB/s) => bootm 0x8000e000 - 0x80000000 ## Booting kernel from Legacy Image at 8000e000 ... Image Name: L4 Image #10 Image Type: AArch64 Linux Kernel Image (uncompressed) Data Size: 44056496 Bytes = 42 MiB Load Address: 81200000 Entry Point: 81200000 Verifying Checksum ... OK ## Flattened Device Tree blob at 80000000 Booting using the fdt blob at 0x80000000 Loading Kernel Image Using Device Tree in place at 0000000080000000, end 00000000800100a5 fixup: pfe0 set to 00:01:be:be:ef:11 fixup: pfe1 set to 00:01:be:be:ef:22 fixup: pfe2 set to 00:01:be:be:ef:33
Starting kernel ...
L4 Bootstrapper Build: #10 Mon Feb 10 10:06:07 CST 2025, 11.4.0 RAM: 0000000080000000 - 00000000ffffffff: 2097152kB RAM: 0000000880000000 - 00000008dfffffff: 1572864kB Total RAM: 3584MB New region for list regions: [ 82ddd000, 83bbd807] { de0808}
Boot
Module overlaps with: [ 83200000, 835dffff] { 3e0000}
Arch
pfebufs@83200000 Regions of list 'regions' [ 34000000, 3407ffff] { 80000} Arch pfebufs@34000000 [ 34080000, 3409ffff] { 20000} Arch pfebufs@34080000 [ 80000000, 8000dfff] { e000} Root dtb [ 81200000, 8121d777] { 1d778} Boot bootstrap [ 81200158, 81200187] { 30} Root cpu_boot [ 8121e1d0, 8121eac7] { 8f8} Boot modinfo [ 8121f000, 812e5fe7] { c6fe8} Boot Module [ 812e6000, 812ebcd7] { 5cd8} Boot Module [ 812ec000, 8131e6cf] { 326d0} Boot Module [ 8131f000, 8141258f] { f3590} Boot Module [ 81413000, 8142e537] { 1b538} Boot Module [ 8142f000, 814b7aa7] { 88aa8} Boot Module [ 814b8000, 814c50a5] { d0a6} Boot Module [ 814c6000, 82dd9a52] { 1913a53} Boot Module [ 82dda000, 82ddb218] { 1219} Boot Module [ 82ddc000, 82ddc224] { 225} Boot Module [ 83200000, 835dffff] { 3e0000} Arch pfebufs@83200000 [ 835e0000, 835fffff] { 20000} Arch pfebufs@835e0000 [ 84000000, 843fffff] { 400000} Arch shm@84000000 [ 85000000, 85000fff] { 1000} Arch ddr@85000000 [ c0000000, c03fffff] { 400000} Arch shm@c0000000 [ c0400000, c07fffff] { 400000} Arch shm@c0400000 [ d0000000, d000007f] { 80} Arch shm@d0000000 [ d0000080, d00000ff] { 80} Arch shm@d0000080 [ ff800000, ff9fffff] { 200000} Arch atf@ff800000 region overlap
Key press reboots... ` Please help me analyze where the issue might be. Thank you very much!
Regards, Qiang
On Mon, Feb 10, 2025 at 3:03 AM Adam Lackorzynski adam@l4re.org wrote:
Hi Qiang,
thanks for the description and all the logs. Could you also please
share
the command given in u-boot (bootm...)?
From the logs, the following lines look suspicious:
DT: FDT sanity check failed: FDT_ERR_BADMAGIC (-9) Total RAM: 0MB
Indicating that something with the device tree is not ok.
kernel fiasco.debug -serial_esc
Please use "fiasco" only. The fiasco.debug is considerably bigger to
load
and could contribute to the region overlap.
Adam
On Thu Feb 06, 2025 at 19:53:49 +0800, qiang xu wrote:
Hi Adam, I encountered the error "region overlap" while running L4Re on
S32G.
First, I built the Linux BSP 38.0 using Yocto to generate the Image, s32g2xxa-evb.dtb, and fsl-image-dom0less-s32g274aevb.cpio.gz. Then, I compiled the S32G platform's L4Re image using the snapshots https://l4re.org/download/snapshots/README.html method. In the modules.list, I added the following:
` entry uvmm-2vm kernel fiasco.debug -serial_esc roottask moe rom/uvmm-2vm.ned module uvmm module l4re module ned module s32g2xxa-evb.dtb module fsl-image-dom0less-s32g274aevb.cpio.gz module[shell] echo $SRC_BASE_ABS/pkg/uvmm/configs/vmm.lua module uvmm-2vm.ned module Image module cons `
Finally, I ran the command make uimage E=uvmm-2vm PT=s32g in the /l4re-snapshot-24.08.0/obj/l4/builddir directory, which showed the following output:
` make[1]: Entering directory '/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/src/l4' [bootstrap - nofpu] ... Building Dependencies Building entry "uvmm-2vm". Merging images: mod00:
/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/fiasco/builddir/fiasco.debug
[27393kB -> 796kB] mod01:
/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir/bin/arm64_armv8a/l4f/sigma0
[24kB] mod02:
/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir/bin/arm64_armv8a/l4f/moe
[202kB] mod03:
/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir/bin/arm64_armv8a/l4f/uvmm
[974kB] mod04:
/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir/bin/arm64_armv8a/l4f/l4re
[110kB] mod05:
/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir/bin/arm64_armv8a/l4f/ned
[547kB] mod06:
/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/conf/s32g2xxa-evb.dtb
[53kB] mod07:
/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/conf/fsl-image-dom0less-s32g274aevb.cpio.gz
[25679kB] mod08:
/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/src/l4/pkg/uvmm/configs/vmm.lua
[5kB] mod09:
/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/conf/uvmm-2vm.ned
[1kB] mod10: /home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/conf/Image [14211kB] mod11:
/home/qiangxu/prebuildl4/l4re-snapshot-24.08.0/obj/l4/builddir/bin/arm64_armv8a/l4f/cons
[280kB] [bootstrap - nofpu] ... Generating bootstrap.ld [bootstrap - nofpu] ... Compiling startup.o [bootstrap - nofpu] ==> Linking l4re.elf [bootstrap] ==> Image post-processing l4re.elf [bootstrap - nofpu] ==> l4re.elf built ==> Installing l4re.elf in image directory ==> Installing l4re_uvmm-2vm in image directory ==> Installing l4re_uvmm-2vm.elf in image directory [bootstrap - nofpu] ... Generating l4re.raw ==> Installing l4re.raw in image directory ==> Installing l4re_uvmm-2vm.raw in image directory [bootstrap - nofpu] ... Generating l4re.uimage Image Name: L4 Image #9 Created: Thu Feb 6 19:29:02 2025 Image Type: AArch64 Linux Kernel Image (uncompressed) Data Size: 44056496 Bytes = 43023.92 KiB = 42.02 MiB Load Address: 81200000 Entry Point: 81200000 ==> Installing l4re.uimage in image directory ==> Installing l4re_uvmm-2vm.uimage in image directory Image size(s) in bytes: l4re_uvmm-2vm.elf: 44692328 l4re.raw: 44056496 l4re.uimage: 44056560 Start address: --> Build-Nr: 9 [bootstrap - nofpu] ==> Installing l4re.elf to local build-tree [bootstrap - nofpu] ==> Installing l4re_uvmm-2vm to local build-tree [bootstrap - nofpu] ==> Installing l4re_uvmm-2vm.elf to local
build-tree
` Finally, I tried loading the l4re_uvmm-2vm.uimage and
s32g2xxa-evb.dtb
files in U-Boot for booting, but it resulted in an error. Can you
help me
identify where the issue might be?
` L4 Bootstrapper Build: #8 Thu Feb 6 14:58:40 CST 2025, 11.4.0 DT: FDT sanity check failed: FDT_ERR_BADMAGIC (-9)
L4 Bootstrapper Build: #8 Thu Feb 6 14:58:40 CST 2025, 11.4.0 Total RAM: 0MB New region for list regions: [ 82ddd000, 855979ff] {
27baa00}
Boot
Module overlaps with: [ 83200000, 835dffff] {
3e0000}
Arch
pfebufs@83200000 Regions of list 'regions' [ 34000000, 3407ffff] { 80000} Arch pfebufs@34000000 [ 34080000, 3409ffff] { 20000} Arch pfebufs@34080000 [ 80000000, 8000d0a5] { d0a6} Root dtb [ 81200000, 8121d777] { 1d778} Boot bootstrap [ 81200158, 81200187] { 30} Root cpu_boot [ 8121e1d0, 8121eadf] { 910} Boot modinfo [ 8121f000, 812e5f7f] { c6f80} Boot Module [ 812e6000, 812ebcd7] { 5cd8} Boot Module [ 812ec000, 8131e6cf] { 326d0} Boot Module [ 8131f000, 8141258f] { f3590} Boot Module [ 81413000, 8142e537] { 1b538} Boot Module [ 8142f000, 814b7aa7] { 88aa8} Boot Module [ 814b8000, 814c50a5] { d0a6} Boot Module [ 814c6000, 82dd9a52] { 1913a53} Boot Module [ 82dda000, 82ddb218] { 1219} Boot Module [ 82ddc000, 82ddc22a] { 22b} Boot Module [ 83200000, 835dffff] { 3e0000} Arch pfebufs@83200000 [ 835e0000, 835fffff] { 20000} Arch pfebufs@835e0000 [ 84000000, 843fffff] { 400000} Arch shm@84000000 [ 85000000, 85000fff] { 1000} Arch ddr@85000000 [ c0000000, c03fffff] { 400000} Arch shm@c0000000 [ c0400000, c07fffff] { 400000} Arch shm@c0400000 [ d0000000, d000007f] { 80} Arch shm@d0000000 [ d0000080, d00000ff] { 80} Arch shm@d0000080 region overlap
Key press reboots...
` Regards, Qiang
On Wed, Jan 8, 2025 at 10:13 AM qiang xu qiangxu431@gmail.com
wrote:
Hi Adam,
Awesome, thanks for the info!
Regards, Qiang
On Tue, Jan 7, 2025 at 10:03 PM Adam Lackorzynski adam@l4re.org
wrote:
Hi Qiang,
thanks for the great feedback.
It's correct, the L4Re microerkernel Fiasco runs in EL2 only. It
runs
both VMs and L4Re applications side by side, so yes, it is a
hypervisor
too.
Best regards, Adam
On Tue Jan 07, 2025 at 19:34:20 +0800, qiang xu wrote: > Hi Adam, > > Thank you very much for your support. The explanation was very
clear.
From > the printed information, it seems to be running fine. I have one
more
> question: Can Fiasco be used as a hypervisor? From the example
of
> l4re_VM-multi, it seems to have the ability to run multiple
operating
> systems on a single platform, and in L4Re, only Fiasco runs in
EL2
> privilege level. I'm not sure if I understand this correctly,
so I
wanted > to confirm with you. If it cannot be used as a hypervisor, what > functionality does it lack that a hypervisor would typically
have?
> > Regards, > Qiang > > > On Tue, Jan 7, 2025 at 7:53 AM Adam Lackorzynski <adam@l4re.org
wrote:
> > > Hi Qiang, > > > > If I'm not mistaken, the provided BSP is using u-boot. With
u-boot,
the > > uimage file is a good fit. When booting up, you can stop at
the
u-boot
> > prompt and do something like this: > > # fatload mmc 0 0x80f00000 fsl-s32g274a-evb.dtb > > # fatload mmc 0 0x80ffffc0 l4re_vm-multi-p2p_s32g.uimage > > # bootm 0x80ffffc0 - 0x80f00000 > > > > You need to copy the l4re_vm-multi-p2p_s32g.uimage file to
the mmc
first > > of course, for example, through Linux. > > Alternative you could also load the elf-image or raw-image
into
memory
> > via your hardware debugger, whatever method the debugger
supports. Do
it > > by intercepting at the u-boot prompt. > > > > A third option is to use the tftp network capabilities of
u-boot
to
load > > the uimage via network. > > > > > > > > Adam > > > > On Mon Jan 06, 2025 at 11:04:35 +0800, qiang xu wrote: > > > Hi Adam, > > > > > > Thank you for your response. > > > > > > I found the following images in the L4Re pre-built images
folder:
> > > “l4re_vm-multi-p2p_s32g.efi,” “l4re_vm-multi-p2p_s32g.elf,”
and
> > > “l4re_vm-multi-p2p_s32g.uimage.” I would like to understand
how
these > > > images can be combined with the BSP built using Yocto and
placed on
the > > SD > > > card for booting. > > > > > > I have been following the examples from the L4Re wiki > > > https://github.com/kernkonzept/manifest/wiki/MultipleVMs
to
get
> > started > > > with L4Re. These examples run very well on QEMU, but there
is an
issue—I > > am > > > unable to use GDB to debug the Fiasco kernel.I noticed that
L4Re
can also > > > run on the S32G platform, so I would like to run L4Re on
S32G
and
use a > > > hardware debugging tool to debug the Fiasco kernel during
its
execution. > > > However, the wiki does not provide instructions on how to
run
L4Re
on > > real > > > hardware.Therefore, I would like your guidance on how to
run the
examples > > > from the wiki < https://github.com/kernkonzept/manifest/wiki/MultipleVMs%3E > > on > > > the S32G platform and boot from an SD card. > > > > > > Regards, > > > Qiang > > > > > > On Mon, Jan 6, 2025 at 9:56 AM Adam Lackorzynski <
adam@l4re.org
wrote: > > > > > > > Hi Qiang, > > > > > > > > On Sat Jan 04, 2025 at 17:03:21 +0800, qiang xu wrote: > > > > > I would like to know how to run L4Re on the S32G. I
couldn’t
find any > > > > > related documentation. I’m a beginner in this technical
field,
so > > could > > > > you > > > > > provide friendly step-by-step instructions? > > > > > > > > One option is to use the snapshot from l4re.org as it
has a
> > menu-driven > > > > target selection which includes the S32G. This hides
quite a
few
of the > > > > details but gives examples for the S32G right away. > > > > How do you boot your S32G?
Adam
Adam adam@os.inf.tu-dresden.de Lackorzynski http://os.inf.tu-dresden.de/~adam/
_______________________________________________ l4-hackers mailing list -- l4-hackers@os.inf.tu-dresden.de To unsubscribe send an email to l4-hackers-leave@os.inf.tu-dresden.de