Re: of AVR target page size

[Peter Maydell <peter.maydell@linaro.org>]

On Thu, 18 Mar 2021 at 10:25, Dr. David Alan Gilbert
<dgilbert@redhat.com> wrote:
> Oh yes, just:
>
> diff --git a/migration/ram.c b/migration/ram.c
> index 52537f14ac..a7269955b5 100644
> --- a/migration/ram.c
> +++ b/migration/ram.c
> @@ -81,6 +81,8 @@
>  /* 0x80 is reserved in migration.h start with 0x100 next */
>  #define RAM_SAVE_FLAG_COMPRESS_PAGE    0x100
>
> +#define RAM_SAVE_FLAG__MAX RAM_SAVE_FLAG_COMPRESS_PAGE
> +
>  static inline bool is_zero_range(uint8_t *p, uint64_t size)
>  {
>      return buffer_is_zero(p, size);
> @@ -4090,5 +4092,6 @@ static SaveVMHandlers savevm_ram_handlers = {
>  void ram_mig_init(void)
>  {
>      qemu_mutex_init(&XBZRLE.lock);
> +    QEMU_BUILD_BUG_ON(RAM_SAVE_FLAG__MAX >= (1 << TARGET_PAGE_BITS_MIN));
>      register_savevm_live("ram", 0, 4, &savevm_ram_handlers, &ram_state);
>  }
>
>
> works; lets keep that in mind somewhere after Michael fixes AVR.

You don't have a great deal of headroom even after getting AVR
to change, by the way -- TARGET_PAGE_BITS_MIN for Arm is 10.
So you might want to think about ways to eg reclaim usage of
that "obsolete, not used" RAM_SAVE_FLAG_FULL bit.

Also, what does the
 /* 0x80 is reserved in migration.h start with 0x100 next */
comment refer to? migration.h has no instances of "RAM_SAVE"
or 0x80 or 1 << 7...

thanks
-- PMM