Re: [PATCH v2] linux irq: remove suer_intr from linux_action

[Samuel Thibault]

Hello,

In this patch,

Junling Ma, le mar. 04 août 2020 14:07:45 -0700, a ecrit:
> diff --git a/linux/dev/arch/i386/kernel/irq.c 
> b/linux/dev/arch/i386/kernel/irq.c
> index aee10462..de7d6c6a 100644
> --- a/linux/dev/arch/i386/kernel/irq.c
> +++ b/linux/dev/arch/i386/kernel/irq.c
> @@ -116,14 +115,6 @@ linux_intr (int irq)
>        action = action->next;
>      }
>  
> -  if (!irq_action[irq])
> -    {
> -      /* No handler any more, disable interrupt */
> -      mask_irq (irq);
> -      ivect[irq] = intnull;
> -      iunit[irq] = irq;
> -    }
> -
>    restore_flags (flags);
>  
>    intr_count--;

This should really go along the patch that removes the deletion within
the loop, since *that* is why we don't need this any more.

And then, I would say this could as well move along:

> @@ -78,7 +78,6 @@ struct linux_action
>    void *dev_id;
>    struct linux_action *next;
>    unsigned long flags;
> -  user_intr_t *user_intr;
>  };
>  
>  static struct linux_action *irq_action[16] =
> @@ -235,7 +226,6 @@ request_irq (unsigned int irq, void (*handler) (int, void 
> *, struct pt_regs *),
>    action->next = NULL;
>    action->dev_id = dev_id;
>    action->flags = flags;
> -  action->user_intr = NULL;
>    
>    retval = setup_x86_irq (irq, action);
>    if (retval)

The two resulting patches make full sense by themselves and nice to
read.