Re: [Qemu-devel] [PATCH 6/8] exec: mark unassigned_io_ops as unlocked

[Jan Kiszka <jan.kiszka@siemens.com>]

On 2015-03-18 14:21, Paolo Bonzini wrote:
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

A commit log should briefly state why we can do this safely (no state
dependency, no need for locking). Yes, it's trivial in this case but
already a little bit less trivial in case of the PM timer.

> ---
>  exec.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/exec.c b/exec.c
> index d60abfc..c1a2e9e 100644
> --- a/exec.c
> +++ b/exec.c
> @@ -2197,6 +2197,7 @@ static void memory_map_init(void)
>      system_io = g_malloc(sizeof(*system_io));
>      memory_region_init_io(system_io, NULL, &unassigned_io_ops, NULL, "io",
>                            65536);
> +    memory_region_clear_global_locking(system_io);
>      address_space_init(&address_space_io, system_io, "I/O");
>  
>      memory_listener_register(&core_memory_listener, &address_space_memory);
> 

Jan

-- 
Siemens AG, Corporate Technology, CT RTC ITP SES-DE
Corporate Competence Center Embedded Linux