Re: [PATCH v5 10/26] xen/riscv: implement make_cpus_node()

[Oleksii Kurochko <oleksii.kurochko@gmail.com>]

On 7/9/26 3:20 PM, Jan Beulich wrote:
> On 06.07.2026 17:57, Oleksii Kurochko wrote:
>> Implement make_cpus_node() to create cpus node for a guest domain.
>>
>> This function is going to be use by common dom0less code during
>> construction domain.
>>
>> Signed-off-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
>> ---
>> Changes in v5:
>> - Drop Acked-by: Jan Beulich <jbeulich@suse.com> as extra changes were done
>>    because of the changed in prev. patch.
>> - Move isa_str allocation and construction out of arch_domain_create() and
>>    into make_cpus_node() as a local variable, since the string is only
>>    needed during FDT generation. Use a two-call build_guest_isa_str()
>>    pattern (size probe, then fill) with xvmalloc_array, and convert all
>>    post-allocation error returns to goto out so xvfree() runs on every path.
>> ---
>> Changes in v4:
>>   - Update the comment in make_cpus_node() to match code style.
>>   - Add Acked-by: Jan Beulich <jbeulich@suse.com>.
>> ---
>> Changes in v3:
>>   - Add blank line above make_cpus_node() function definition.
>>   - Move 'unsigned int cpu' from function-level declarations into the for loop.
>>   - Drop 'uint32_t reg = cpu_to_fdt32(cpu)'; use fdt_property_cell(fdt, "reg", cpu)
>>     instead of fdt_property(fdt, "reg", &reg, sizeof(reg)) so byte-order adjustment
>>     is handled internally.
>>   - Add matching /* interrupt-controller */ start comment; fix end comment to
>>     /* end interrupt-controller */.
>>   - Update d->arch.guest_isa_str to ->isa_str in make_cpus_node() function.
>> ---
>> Changes in v2:
>>   - s/u32/uint32_t for timebase_frequency local variable.
>>   - Drop +1 from BUILD_BUG_ON().
>>   - return fdt_end_node(fdt); instead of res at the end of the function.
>> ---
>> ---
>>   build/tools/fixdep            | Bin 0 -> 13632 bytes
> 
> There was a stray file in the v4 series iirc. Now there again is a stray file
> in v5. Can you please make sure to take a look yourself before posting a
> series? 

Sorry about that, I will double-check next time.

> Are you doing out-of-tree hypervisor builds, but with a build directory
> within the source repo? There is an entry in .gitignore after all, just for
> xen/tools/fixdep.

No, it should in-tree:
CONTAINER=bookworm-riscv64 ./automation/scripts/containerize make V=1 
XEN_TARGET_ARCH=riscv64 -C xen

Sometime I am doing without CONTAINER when I want to check on my system 
with newer compile version but still I don't use O=, just make ... -C xen.

> 
> With this properly dropped:
> Acked-by: Jan Beulich <jbeulich@suse.com>

Thanks.

~ Oleksii