[RFC DO NOT MERGE] treewide: use __xchg in most obvious places

Andy Shevchenko andriy.shevchenko at linux.intel.com
Tue Jan 10 03:07:02 PST 2023


On Tue, Jan 10, 2023 at 11:53:06AM +0100, Andrzej Hajda wrote:
> This patch tries to show usability of __xchg helper.
> It is not intended to be merged, but I can convert
> it to proper patchset if necessary.
> 
> There are many more places where __xchg can be used.
> This demo shows the most spectacular cases IMHO:
> - previous value is returned from function,
> - temporary variables are in use.
> 
> As a result readability is much better and diffstat is quite
> nice, less local vars to look at.
> In many cases whole body of functions is replaced
> with __xchg(ptr, val), so as further refactoring the whole
> function can be removed and __xchg can be called directly.

...

>  arch_uretprobe_hijack_return_addr(unsigned long trampoline_vaddr,
>  				  struct pt_regs *regs)
>  {
> -	unsigned long orig_ret_vaddr;
> -
> -	orig_ret_vaddr = regs->ARM_lr;
> -	/* Replace the return addr with trampoline addr */
> -	regs->ARM_lr = trampoline_vaddr;
> -	return orig_ret_vaddr;
> +	return __xchg(&regs->ARM_lr, trampoline_vaddr);
>  }

If it's not a callback, the entire function can be killed.
And this is a good example of the function usage.
OTOH, these places might have a side effect (if it's in deep CPU
handlers), means we need to do this carefully.

...

>  static inline void *qed_chain_produce(struct qed_chain *p_chain)
>  {
> -	void *p_ret = NULL, *p_prod_idx, *p_prod_page_idx;
> +	void *p_prod_idx, *p_prod_page_idx;
>  
>  	if (is_chain_u16(p_chain)) {
>  		if ((p_chain->u.chain16.prod_idx &
> @@ -390,11 +391,8 @@ static inline void *qed_chain_produce(struct qed_chain *p_chain)
>  		p_chain->u.chain32.prod_idx++;
>  	}
>  
> -	p_ret = p_chain->p_prod_elem;
> -	p_chain->p_prod_elem = (void *)(((u8 *)p_chain->p_prod_elem) +
> -					p_chain->elem_size);
> -
> -	return p_ret;
> +	return __xchg(&p_chain->p_prod_elem,
> +		      (void *)(((u8 *)p_chain->p_prod_elem) + p_chain->elem_size));

Wondering if you still need a (void *) casting after the change. Ditto for the
rest of similar cases.

>  }

...

Btw, is it done by coccinelle? If no, why not providing the script?

-- 
With Best Regards,
Andy Shevchenko





More information about the linux-arm-kernel mailing list