Re: [PATCH v5 6/7] dax: add recovery_write to dax_iomap_iter in failure path

[Jane Chu]

On 2/2/2022 5:44 AM, Christoph Hellwig wrote:
> On Fri, Jan 28, 2022 at 02:31:49PM -0700, Jane Chu wrote:
>> +typedef size_t (*iter_func_t)(struct dax_device *dax_dev, pgoff_t pgoff,
>> +		void *addr, size_t bytes, struct iov_iter *i);
>>   static loff_t dax_iomap_iter(const struct iomap_iter *iomi,
>>   		struct iov_iter *iter)
>>   {
>> @@ -1210,6 +1212,7 @@ static loff_t dax_iomap_iter(const struct iomap_iter *iomi,
>>   	ssize_t ret = 0;
>>   	size_t xfer;
>>   	int id;
>> +	iter_func_t write_func = dax_copy_from_iter;
> 
> This use of a function pointer causes indirect call overhead.  A simple
> "bool in_recovery" or do_recovery does the trick in a way that is
> both more readable and generates faster code.

Good point, thanks!

> 
>> +		if ((map_len == -EIO) && (iov_iter_rw(iter) == WRITE)) {
> 
> No need for the braces.

Did you mean the outer "( )" ?

> 
>>   		if (iov_iter_rw(iter) == WRITE)
>> -			xfer = dax_copy_from_iter(dax_dev, pgoff, kaddr,
>> -					map_len, iter);
>> +			xfer = write_func(dax_dev, pgoff, kaddr, map_len, iter);
>>   		else
>>   			xfer = dax_copy_to_iter(dax_dev, pgoff, kaddr,
>>   					map_len, iter);
> 
> i.e.
> 
> 		if (iov_iter_rw(iter) == READ)
> 			xfer = dax_copy_to_iter(dax_dev, pgoff, kaddr,
> 					map_len, iter);
> 		else if (unlikely(do_recovery))
> 			xfer = dax_recovery_write(dax_dev, pgoff, kaddr,
> 					map_len, iter);
> 		else
> 			xfer = dax_copy_from_iter(dax_dev, pgoff, kaddr,
> 					map_len, iter);
> 

Will do.

Thanks a lot!

-jane