Re: [PATCH v7 4/7] xen/console: switch conring runtime allocation to xvmalloc

On 13.07.2026 20:16, [email protected] wrote:
> @@ -474,20 +475,26 @@ void __init console_init_ring(void)
>  {
>      char *ring;
>      XENCONS_RING_IDX done, size, n;
> -    unsigned int order, memflags;
>      unsigned long flags;
>  
>      if ( !opt_conring_size )
>          return;
> -

Please can we keep this blank line and ...

> -    order = get_order_from_bytes(max(opt_conring_size, conring_size));
> -    memflags = MEMF_bits(crashinfo_maxaddr_bits);
> -    while ( (ring = alloc_xenheap_pages(order, memflags)) == NULL )
> +    else if ( opt_conring_size > GB(2) )

... have this be just "if ()"? (I'd also suggest to make the "positive" case
the "if()" part and have the out-of-range one in the "else" body.)

>      {
> -        BUG_ON(order == 0);
> -        order--;
> +        printk(XENLOG_WARNING "Limiting user-configured console ring size to 2 GiB\n");

Nit: Too long line. Yes, format strings shouldn't be split, but XENLOG_* can
every well live on a separate line.

> +        opt_conring_size = GB(2);
>      }
> -    opt_conring_size = PAGE_SIZE << order;
> +    else
> +    {
> +        unsigned int order = get_order_from_bytes(max(opt_conring_size,
> +                                                      conring_size));
> +        opt_conring_size = PAGE_SIZE << order;

Nit: Blank line please between declaration(s) and statement(s).

> +    }
> +
> +    ring = xvmalloc_array(char, opt_conring_size);
> +    if ( !ring )
> +        panic("Unable to allocate console ring of %u KiB\n",
> +              opt_conring_size >> 10);

Why do you lose the prior behavior of halving the size until allocation
succeeds (or we can't even allocate a single page)?

Jan