[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH v1 2/3] xen/dom0less: refactor architecture-specific DomU construction
- To: Oleksii Kurochko <oleksii.kurochko@xxxxxxxxx>, xen-devel@xxxxxxxxxxxxxxxxxxxx
- From: Julien Grall <julien@xxxxxxx>
- Date: Wed, 14 May 2025 08:25:52 +0100
- Cc: Stefano Stabellini <sstabellini@xxxxxxxxxx>, Bertrand Marquis <bertrand.marquis@xxxxxxx>, Michal Orzel <michal.orzel@xxxxxxx>, Volodymyr Babchuk <Volodymyr_Babchuk@xxxxxxxx>, Andrew Cooper <andrew.cooper3@xxxxxxxxxx>, Anthony PERARD <anthony.perard@xxxxxxxxxx>, Jan Beulich <jbeulich@xxxxxxxx>, Roger Pau Monné <roger.pau@xxxxxxxxxx>
- Delivery-date: Wed, 14 May 2025 07:25:57 +0000
- List-id: Xen developer discussion <xen-devel.lists.xenproject.org>
Hi Oleksii,
On 13/05/2025 15:29, Oleksii Kurochko wrote:
Refactor construct_domU() to improve architecture separation and reduce
reliance on ARM-specific logic in common code:
- Drop set_domain_type() from generic code. This function is specific
to ARM and serves no purpose on other architectures like RISC-V,
which lack the arch.type field in kernel_info.
So you will only ever boot one type of domain on RISC-V? IOW, no 32-bit
or else?
- Introduce arch_construct_domU() to encapsulate architecture-specific
DomU construction steps.
- Implement arch_construct_domU() for ARM. This includes:
- Setting the domain type for CONFIG_ARM64.
- Handling static memory allocation if xen,static-mem is present in
the device tree.
- Processing static shared memory.
- Move call of make_resv_memory_node() to Arm's make_arch_nodes() as
this call is specific to CONFIG_STATIC_SHM which is ARM specific,
at least, now.
This looks shortsighted. If there is a plan to use CONFIG_STATIC_SHM on
RISC-V (I don't see why not today), then
I think the code should stick in common/ even if it is not fully usable
yet (that's the whole point of have CONFIG_* options).
Cheers,
--
Julien Grall
|