On 1/9/19 6:53 AM, Stefano Garzarella wrote:
> Hi Liam,
>
> On Tue, Jan 8, 2019 at 3:47 PM Liam Merwick <liam.merw...@oracle.com> wrote:
>> QEMU sets the hvm_modlist_entry in load_linux() after the call to
>> load_elfboot() and then qboot loads it in boot_pvh_from_fw_cfg()
>>
>> But the current PVH patches don't handle initrd (they have
>> start_info.nr_modules == 1).
> Looking in the linux kernel (arch/x86/platform/pvh/enlighten.c) I saw:
> /* The first module is always ramdisk. */
> if (pvh_start_info.nr_modules) {
> struct hvm_modlist_entry *modaddr =
> __va(pvh_start_info.modlist_paddr);
> pvh_bootparams.hdr.ramdisk_image = modaddr->paddr;
> pvh_bootparams.hdr.ramdisk_size = modaddr->size;
> }
>
> So, putting start_info.nr_modules = 1, means that the first
> hvm_modlist_entry should have the ramdisk paddr and size. Is it
> correct?
>
>
>> During (or after) the call to load_elfboot() it looks like we'd need to
>> do something like what load_multiboot() does below (along with the
>> associated initialisation)
>>
>> 400 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_ADDR, ADDR_MBI);
>> 401 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_SIZE, sizeof(bootinfo));
>> 402 fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, mb_bootinfo_data,
>> 403 sizeof(bootinfo));
>>
> In this case I think they used the FW_CFG_INITRD_* to pass bootinfo
> varibales to the guest, maybe we could add something like what
> linux_load() does:
>
> /* load initrd */
> if (initrd_filename) {
> ...
> initrd_addr = (initrd_max-initrd_size) & ~4095;
>
> fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_ADDR, initrd_addr);
> fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_SIZE, initrd_size);
> fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, initrd_data,
> initrd_size);
> ...
> }
>
> Then we can load the initrd in qboot or in the optionrom that I'm writing.
>
> What do you think?
Why not specify this in pvh_start_info? This will be much faster for
everyone, no need to go through fw_cfg.
-boris
