Re: Deprecate the ppc405 boards in QEMU?

[Cédric Le Goater]

On 10/19/21 18:12, Christophe Leroy wrote:
> Le 19/10/2021 à 16:56, BALATON Zoltan a écrit :
> > On Tue, 19 Oct 2021, Christophe Leroy wrote:
> > > Le 19/10/2021 à 15:44, Christophe Leroy a écrit :
> > > > There is something:
> > > >
> > > > => bootm 0
> > > > Wrong Image Format for bootm command
> > > > ERROR: can't get kernel image!
> > > >
> > > > => md 0
> > > > 00000000: 00000000 b497aae1 616e9207 003227a4    '..V....an...2'.
> > > > 00000010: 00000000 00000000 ee36255f 05070201    .........6%_....
> > > > 00000020: 4c696e75 782d352e 31352e30 2d726335    Linux-5.15.0-rc5
> > > > 00000030: 2d303232 32342d67 61336330 30376164    -02224-ga3c007ad
> > > > 00000040: 1f8b0800 00000000 0203ec5c 0f7013e7    ...........\.p..
> > > >
> > > > => mw.l 0 0x27051956
> > > >
> > > > => bootm 0
> > > > ## Booting kernel from Legacy Image at 00000000 ...
> > > >     Image Name:   Linux-5.15.0-rc5-02224-ga3c007ad
> > > >     Image Type:   PowerPC Linux Kernel Image (gzip compressed)
> > > >     Data Size:    3286948 Bytes = 3.1 MiB
> > > >     Load Address: 00000000
> > > >     Entry Point:  00000000
> > > >     Verifying Checksum ... Bad Data CRC
> > > > ERROR: can't get kernel image!
> > > >
> > > >
> > > > So we have something but it seems it gets overwritten by stuff.
> > > >
> > > > Anyway loading a uImage at 0 is just bad because it is a gzipped image that 
> > > > should get gunzipped at address 0.
> > > >
> > > > Or should we just copy the raw kernel at 0 and jump to the entry point ? But 
> > > > then we also need a device tree somewhere.
> > >
> > > If I change KERNEL_LOAD_ADDR to 0x1000000, I can bootm at that address, and it 
> > > seems it properly decompress the kernel:
> > >
> > > => bootm 0x1000000
> > > ## Booting kernel from Legacy Image at 01000000 ...
> > >   Image Name:   Linux-5.15.0-rc5-02224-ga3c007ad
> > >   Image Type:   PowerPC Linux Kernel Image (gzip compressed)
> > >   Data Size:    3296789 Bytes = 3.1 MiB
> > >   Load Address: 00000000
> > >   Entry Point:  00000000
> > >   Verifying Checksum ... OK
> > >   Uncompressing Kernel Image ... OK
> > >
> > >
> > > And it initialises the MMU properly.
> > >
> > > Then it gets stuck because there is no devicetree.
> > >
> > > (gdb) bt
> > > #0  0xc00094cc in udelay ()
> > > #1  0xc0025d34 in panic ()
> > > #2  0xc06415a0 in early_init_devtree ()
> > > #3  0xc0641da8 in machine_init ()
> > > #4  0xc0002200 in start_here ()
> >
> > Maybe you need to embed a dtb in your kernel if that's possible somehow? Or QEMU 
> > has a -dtb option that sets machine->dtb but you need board code to do 
> > something with it. See how it's done in other boards like virtex_ml507 and 
> > sam460ex. But maybe you'd be better off not using -kernel option as it seems to 
> > not really working for these 405 boards but load and start the kernel from u-boot. 
> > Not sure what device does u-boot support but QEMU can emulate usb-storage, 
> > network, different disks so something might work with u-boot if this board has any 
> > peripherals. If it doesn't emulate any peripherals what do you expect to do with 
> > the kernel once it boots?
>
> I should be able to build a multi-FIT image that embeds the kernel and the 
> device tree.
>
> I don't know about the peripherals, what I need it a kernel that is able to 
> boot and run some user exe. I'm working on low level functionnalities like 
> VMAP_STACK, STRICT_KERNEL_RWX, Userspace protection, etc ... I want to be able 
> to check that after some modifications the kernel can still boot on every CPU 
> sub-family, and I need to run LKDTM tests.
>
> For this a kernel + initrd is enough.
>
> Thanks
> Christophe

If we don't need a loader, we are better off simplifying the 405 boards with
a simple init sequence such as :

    if (machine->kernel_filename) {
        hwaddr kernel_base = 0;
        int kernel_size = 0;
        hwaddr initrd_base = 0;
        int initrd_size = 0;

        kernel_size = load_elf(machine->kernel_filename, NULL, NULL, NULL,
                               &boot_entry, &kernel_base, NULL, NULL,
                               0 /* LE */, PPC_ELF_MACHINE, 0, 0);
        if (kernel_size < 0) {
            error_report("Could not load kernel '%s' : %s",
                         machine->kernel_filename, 
load_elf_strerror(kernel_size));
            exit(1);
        }

        if (machine->initrd_filename) {
            initrd_base = QEMU_ALIGN_UP(kernel_base + kernel_size, 0x10000);
            initrd_size = load_image_targphys(machine->initrd_filename,
                                              initrd_base, 16 * MiB /* Some 
value */);
            if (initrd_size < 0) {
                error_report("Could not load initial ram disk '%s'",
                             machine->initrd_filename);
                exit(1);
            }
        }

        if (machine->dtb) {
            dt_base = mw_dtb_load(machine, kernel_base, kernel_size, 
initrd_base,
                                  initrd_size);
        }
    }

We need to set the nip to 'boot_entry' and gpr[3] to 'dt_base'.

unless some pre-initialization of hw is required before running Linux ?

Thanks,

C.