Re: [PATCH] tools/ocaml: fill arch_config for ARM in domain_getinfolist

On 27/07/2026 10:11 am, Julian Vetter wrote:
> In the function alloc_domaininfo() the arch_config field in domaininfo
> is only filled inside #if defined(__i386__) || defined(__x86_64__). On
> ARM the field is left unpopulated. caml_alloc_tuple() does not zero new
> blocks, so di.Xenctrl.arch_config on ARM is uninitialised garbage for
> every caller of Xenctrl.domain_getinfolist/domain_getinfo. Any toolstack
> dereferencing that value, segfaults once a VM is enumerated in a domain
> scan.
>
> Add the missing ARM equivalent, populating xen_arm_arch_domainconfig
> from the raw xc_domaininfo_t the same way the x86 branch does for
> xen_x86_arch_domainconfig.
>
> Signed-off-by: Julian Vetter <[email protected]>
> ---
>  tools/ocaml/libs/xc/xenctrl_stubs.c | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
>
> diff --git a/tools/ocaml/libs/xc/xenctrl_stubs.c b/tools/ocaml/libs/xc/xenctrl_stubs.c
> index 7f6381cdd2..b60ea9f12e 100644
> --- a/tools/ocaml/libs/xc/xenctrl_stubs.c
> +++ b/tools/ocaml/libs/xc/xenctrl_stubs.c
> @@ -415,6 +415,7 @@ static value alloc_domaininfo(xc_domaininfo_t * info)
>  {
>  	CAMLparam0();
>  	CAMLlocal5(result, tmp, arch_config, x86_arch_config, emul_list);
> +	CAMLlocal1(arm_arch_config);
>  	int i;
>  
>  	result = caml_alloc_tuple(17);
> @@ -460,6 +461,19 @@ static value alloc_domaininfo(xc_domaininfo_t * info)
>  
>  	Store_field(arch_config, 0, x86_arch_config);
>  
> +	Store_field(result, 16, arch_config);
> +#elif defined(__arm__) || defined(__aarch64__)
> +	/* xen_arm_arch_domainconfig */
> +	arm_arch_config = caml_alloc_tuple(3);
> +	Store_field(arm_arch_config, 0, Val_int(info->arch_config.gic_version));
> +	Store_field(arm_arch_config, 1, Val_int(info->arch_config.nr_spis));
> +	Store_field(arm_arch_config, 2, caml_copy_int32(info->arch_config.clock_frequency));
> +
> +	/* arch_config: arch_domainconfig */
> +	arch_config = caml_alloc_small(1, 0);
> +
> +	Store_field(arch_config, 0, arm_arch_config);
> +
>  	Store_field(result, 16, arch_config);
>  #endif
>  

Oops, this was unintentional, but obvious now you point it out.

While this patch works, it's continuing a bad pattern I accidentally
started.

I'll do a prep cleanup patch, and rebase this one on top.

~Andrew