On 3.03.2018 02:24, Andrew Morton wrote:
> On Mon, 26 Feb 2018 11:54:30 +0200 Nikolay Borisov <nborisov@xxxxxxxx> wrote:
>
>> We already get the block counts and the calculate the end block at the
>> beginning of the function. Let's use the local variables for consistency and
>> readability. No functional changes
>>
>
> Fair enough.
>
>> --- a/fs/direct-io.c
>> +++ b/fs/direct-io.c
>> @@ -1200,7 +1200,7 @@ do_blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
>> }
>>
>> /* watch out for a 0 len io from a tricksy fs */
>> - if (iov_iter_rw(iter) == READ && !iov_iter_count(iter))
>> + if (iov_iter_rw(iter) == READ && !count)
>> return 0;
>>
>> dio = kmem_cache_alloc(dio_cache, GFP_KERNEL);
>> @@ -1316,8 +1316,7 @@ do_blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
>>
>> dio->should_dirty = (iter->type == ITER_IOVEC);
>> sdio.iter = iter;
>> - sdio.final_block_in_request =
>> - (offset + iov_iter_count(iter)) >> blkbits;
>> + sdio.final_block_in_request = end >> blkbits;
>>
>> /*
>> * In case of non-aligned buffers, we may need 2 more
>
> It's always scary to see a local being used in this fashion so far away
> from where it was initialized. However, we can get our warm fuzzy
> feelings by doing this:
>
Good point !
> --- a/fs/direct-io.c~direct-io-minor-cleanups-in-do_blockdev_direct_io-fix
> +++ a/fs/direct-io.c
> @@ -1178,9 +1178,9 @@ do_blockdev_direct_IO(struct kiocb *iocb
> unsigned blkbits = i_blkbits;
> unsigned blocksize_mask = (1 << blkbits) - 1;
> ssize_t retval = -EINVAL;
> - size_t count = iov_iter_count(iter);
> + const size_t count = iov_iter_count(iter);
> loff_t offset = iocb->ki_pos;
> - loff_t end = offset + count;
> + const loff_t end = offset + count;
> struct dio *dio;
> struct dio_submit sdio = { 0, };
> struct buffer_head map_bh = { 0, };
> _
I assume you don't want me to resend but will just fold the changes
yourself?
>
>
