On 16.11.2023 13:04, Oleksii wrote:
> On Thu, 2023-11-16 at 08:36 +0100, Jan Beulich wrote:
>> On 10.11.2023 17:30, Oleksii Kurochko wrote:
>>> --- /dev/null
>>> +++ b/xen/include/asm-generic/percpu.h
>>> @@ -0,0 +1,35 @@
>>> +/* SPDX-License-Identifier: GPL-2.0-only */
>>> +#ifndef __ASM_GENERIC_PERCPU_H__
>>> +#define __ASM_GENERIC_PERCPU_H__
>>> +
>>> +#ifndef __ASSEMBLY__
>>> +
>>> +#include <xen/types.h>
>>> +
>>> +extern char __per_cpu_start[], __per_cpu_data_end[];
>>
>> Can we go one tiny step beyond what Arm presently has and make the
>> latter of the two const?
> I am not sure I think we will have compilation issue with the following
> code in Arm and x86 because of [-Werror=discarded-qualifiers]:
>
> static void cf_check _free_percpu_area(struct rcu_head *head)
> {
> struct free_info *info = container_of(head, struct free_info, rcu);
> unsigned int cpu = info->cpu;
> char *p = __per_cpu_start + __per_cpu_offset[cpu];
>
> free_xenheap_pages(p, PERCPU_ORDER);
> __per_cpu_offset[cpu] = INVALID_PERCPU_AREA;
> }
There's no use of __per_cpu_data_end here; I specifically didn't ask for both
of the declarations to have const added.
> I guess cast can help.
There may not be casts casting away constness, except maybe in very delicate
situations.
Jan
