What I was worry about was: CPU_STATE is too much. While we only need very few registers (Defined as hypercall parameter now) exported for DM. Another thing is: we do hypercall from the SOS cpu and

From: Jiangbo Wu <jiangbo.wu@...> Add sample script launch option to launch AliOS as a guest. Signed-off-by: Jiangbo Wu <jiangbo.wu@...> --- devicemodel/samples/apl-mrb/launch_uos.sh |

SET_BOOT_CONTEXT is hard to understand from semantics point of view. I think we should do with CPU_SET_STATE. It is fine if we set state more than we need for boot_ctx.

Arindam Roy <arindam.roy@...> writes: s/0/0U Do we need ''end'' any more? There are some other violations here: * ''entry'' is struct acpi_subtable_header *, but ''first'' is void *.

Hi Wensheng, "Wang, WenshengX" <wenshengx.wang@...> writes: The reason behind the introductino of CONFIG_RELOC is that it requires a linker option supported by binutils >= 2.27, leading to

Yes. We should do like this. Otherwise, we can't copy the structure from DM to arch_vm.init_ctx directly. That's my patch's fault. I will adjust the boot_ctx to this type in v3. Regards Yin,

Yes. It's used to setup the UOS BSP state. About the AP, the initial state is fixed. Good point. The thing here is boot context is only part of UOS BSP state. Not sure whether the set_init_state fit

Hi Geoffroy, About the condition of options existing, there may be some reasons. I know one piece of reason is about compatibility. For example -fstatck-protector-strong is only supported gcc 4.9 or

This patch slightly modifies two functions to avoid violating MISRA C R.18.3. The modifications removes comparison between pointers of structures, instead using the length to reach to the end of the

Agree. Let's do it with ACRN_ prefix first. And do the cleanup in future.

Can you please point out detail kernel virtio headers? Likes linux/virtio_config.h The VRING_AVAIL_F_NO_INTERRUPT, VRING_USED_F_NO_NOTIFY, VIRTIO_RING_F_EVENT_IDX are still appeared in the comments

We didn't enable relocation by default because the noreloc-overflow option is not available for binutil < 2.27.

Zide/Gen, Can you help review this patch? Thanks, Anthony

You can have struct boot_ctx for uos, boot_ctx only includes CPU registers which are needed to support different GuestOS/GuestBIOS. And struct vm0_boot_ctx for SOS, Vm0_boot_ctx is superset of

Acked-by: Anthony Xu <anthony.xu@...>

The boot context seems to be the initial CPU state to me, is it what we want? If YES, then we should use more general hcall. It is kind of set_cpu_state (we already have a RESET hcall -- this can be

Hi Wensheng, Thanks for sharing these additional details. I'm not disputing the introduction of these options. What I am wondering is if it needs to be conditional. If we are not ever going to

Acked-by: Yu Wang <yu1.wang@...>

This patch is used to fix a potential issue resulted from typo. Signed-off-by: Xiaoguang Wu <xiaoguang.wu@...> Reviewed-by: Liang Yang <liang3.yang@...> --- devicemodel/hw/pci/xhci.c | 2