[PATCH 01/18] dbtr: Add consistent range checks to trigger ecalls

[Ilya Mamay]

Hi, Nicholas
This is a significant improvement over the previous version!

I have a few comments below, though they may be off the mark since they're
based on my own reading of the SBI and Sdtrig specifications.

On Fri, Mar 13, 2026 at 03:19:30PM +1000, Nicholas Piggin wrote:
> @@ -594,6 +597,9 @@ int sbi_dbtr_install_trig(unsigned long smode,
>  	if (!hs)
>  		return SBI_ERR_FAILED;
>  
> +	if (trig_count >= hs->total_trigs)
> +		return SBI_ERR_BAD_RANGE;

According to the SBI specification (section 19.4):
- "The sbiret.value is set to zero upon success or if shared memory is
  disabled". I interpret this to mean that sbi_dbtr_install_trig must return a
  trigger index along with an SBI_ERR_BAD_RANGE error.

- "sbiret.value is set to the array index i of the failing
  trigger configuration". In this case, sbiret.value should be hs->total_trigs,
  because that is the index of the first failing trigger configuration.

Same comment applies to sbi_dbtr_update_trig.

> @@ -710,14 +726,14 @@ int sbi_dbtr_update_trig(unsigned long smode,
>  	if (!hs)
>  		return SBI_ERR_FAILED;
>  
> +	if (trig_count >= hs->total_trigs)
> +		return SBI_ERR_BAD_RANGE;
> +

Regards
Ilya

> -- 
> opensbi mailing list
> opensbi at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/opensbi