On Thu, Feb 21, 2019 at 05:09:01PM +0200, Nikolay Borisov wrote:
> On 21.02.19 г. 17:07 ч., Johannes Thumshirn wrote:
> > On 21/02/2019 14:25, Nikolay Borisov wrote:
> >> On 21.02.19 г. 15:15 ч., Johannes Thumshirn wrote:
> >>> On 21/02/2019 12:57, Nikolay Borisov wrote:
> >>>>
> >>>> static int cow_file_range_async(struct inode *inode, struct page
> >>>> *locked_page,
> >>>> @@ -1190,45 +1201,68 @@ static int cow_file_range_async(struct inode
> >>>> *inode, struct page *locked_page,
> >>>> unsigned int write_flags)
> >>>> {
> >>>> struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
> >>>> - struct async_cow *async_cow;
> >>>> + struct async_cow *ctx;
> >>>> + struct async_chunk *async_cow;
> >>>
> >>> In case you have to re-send the patch you could maybe rename the
> >>> async_cow variable to async_chunk or sth like that. Would make the
> >>> resulting code a little bit clearer but no strong opinions here.
> >>
> >> The reason I left it like that is to minimize the resulting diff.
> >
> > Understood, but now we have a 'ctx' of type 'struct async_cow' and a
> > '*async_cow' of type 'struct asyn_chunk'. This is the type of code that
> > will make people go WTF when they read it in 3-4 cycles.
>
> I also thought of that but then sometimes David has strange tastes re.
> code cleanups that add noise. If he is fine with doing the rename in
> this patch then I'm happy to do it and resend.
You must be joking, the naming you chose is quite confusing exactly as
Johannes points out.
I'm not sure what noise patches or changes you mean. There are commits
that just rename something for better readability or free the
type/variable name for another purpose, similar what you do.
Your concern about diff size is ok, though it should result in a
separate patch doing just the rename and another patch using the new
async_cow.
