Re: [PATCH v5 1/2] xen/console: re-calibrate rate-limiter based on user input

On 12.08.2026 01:38, [email protected] wrote:
> From: Denis Mukhin <[email protected]> 

First: What user input is the subject talking about?

> Current __printk_ratelimit() relies on hardcoded values 5000 and 10
> respectively to program leaky-bucket message limiting.

Which doesn't change, as ...

> Use 'ratelimit_ms' and 'ratelimit_burst' variables to re-calibrate
> limiter.

... the values of the two static variables never change.

> Ensure rate limiter is disabled if either 'ratelimit_ms' or
> 'ratelimit_burst' is 0.
> 
> Fixes: 26cf03554a75 ("[XEN] Implement rate-limited logging.")

I don't think anything is being fixed here.

> --- a/xen/drivers/char/console.c
> +++ b/xen/drivers/char/console.c
> @@ -33,6 +33,7 @@
>  #include <asm/setup.h>
>  #include <xen/sections.h>
>  #include <xen/consoled.h>
> +#include <xen/xvmalloc.h>

This shouldn't be needed, as ...

> @@ -1286,21 +1287,34 @@ bool __printk_ratelimit(unsigned int ratelimit_ms,
>                          unsigned int ratelimit_burst)
>  {
>      static DEFINE_SPINLOCK(ratelimit_lock);
> -    static unsigned long toks = 10 * 5 * 1000;
> +    static unsigned long toks = ~0;
>      static unsigned long last_msg;
>      static unsigned int missed;
> +    unsigned long limit;
> +    unsigned long elapsed;
>      unsigned long flags;
> -    unsigned long long now = NOW(); /* ns */
>      unsigned long ms;
> +    s_time_t now;
>  
> -    do_div(now, 1000000);
> -    ms = (unsigned long)now;
> +    if ( !ratelimit_ms || !ratelimit_burst )
> +        return true;
> +
> +    limit = DIM_MUL2(ratelimit_burst, ratelimit_ms);

... this helper shouldn't be (ab)used here. I further wonder whether we
really want differing behavior here for 32- and 64-bit hosts.

Jan