[PATCH v4 03/11] fs: Initial atomic write support

[John Garry]

...

>>
>> Helper function atomic_write_valid() can be used by FSes to verify
>> compliant writes.
>>
>> Signed-off-by: Prasad Singamsetty <prasad.singamsetty at oracle.com>
>> #jpg: merge into single patch and much rewrite
> 
> ^^^ this might be a miss I guess.

I'm not sure what you mean. Here I am just briefly commenting on much 
changes which I made.

> 
>> Signed-off-by: John Garry <john.g.garry at oracle.com>
>> ---
>>   fs/aio.c                |  8 ++++----
>>   fs/btrfs/ioctl.c        |  2 +-
>>   fs/read_write.c         |  2 +-
>>   include/linux/fs.h      | 36 +++++++++++++++++++++++++++++++++++-
>>   include/uapi/linux/fs.h |  5 ++++-
>>   io_uring/rw.c           |  4 ++--
>>   6 files changed, 47 insertions(+), 10 deletions(-)
>>
>> diff --git a/fs/aio.c b/fs/aio.c
>> index bb2ff48991f3..21bcbc076fd0 100644
>> --- a/fs/aio.c
>> +++ b/fs/aio.c
>> @@ -1502,7 +1502,7 @@ static void aio_complete_rw(struct kiocb *kiocb, long res)
>>   	iocb_put(iocb);
>>   }
>>   
>> -static int aio_prep_rw(struct kiocb *req, const struct iocb *iocb)
>> +static int aio_prep_rw(struct kiocb *req, const struct iocb *iocb, int type)
> 
> maybe rw_type?

ok

> 
>>   {
>>   	int ret;
>>   
>> @@ -1528,7 +1528,7 @@ static int aio_prep_rw(struct kiocb *req, const struct iocb *iocb)
>>   	} else

...

>> +
>>   /* 32bit hashes as llseek() offset (for directories) */
>>   #define FMODE_32BITHASH         ((__force fmode_t)0x200)
>>   /* 64bit hashes as llseek() offset (for directories) */
>> @@ -328,6 +333,7 @@ enum rw_hint {
>>   #define IOCB_SYNC		(__force int) RWF_SYNC
>>   #define IOCB_NOWAIT		(__force int) RWF_NOWAIT
>>   #define IOCB_APPEND		(__force int) RWF_APPEND
>> +#define IOCB_ATOMIC		(__force int) RWF_ATOMIC
>>   
> 
> You might also want to add this definition in here too
> 
> #define TRACE_IOCB_STRINGS \
> <...>
> <...>
> { IOCB_ATOMIC, "ATOMIC" }

ok

I suppose that new flag RWF_NOAPPEND in linux-next also should have this

>>   
>> +static inline bool atomic_write_valid(loff_t pos, struct iov_iter *iter,
>> +			   unsigned int unit_min, unsigned int unit_max)
>> +{
>> +	size_t len = iov_iter_count(iter);
>> +
>> +	if (!iter_is_ubuf(iter))
>> +		return false;
> 
> There is no mention about this limitation in the commit message of this
> patch. Maybe it will be good to capture why this limitation to only
> support ubuf and/or any plans to lift this restriction in future
> in the commit message?

ok, I can mention this in the commit message.

> 
> 
>> +
>> +	if (len == unit_min || len == unit_max) {
>> +		/* ok if exactly min or max */
>> +	} else if (len < unit_min || len > unit_max) {
>> +		return false;
>> +	} else if (!is_power_of_2(len)) {
>> +		return false;
>> +	}
> 
> Checking for len == unit_min || len == unit_max is redundant when
> unit_min and unit_max are already power of 2.

Sure, but it was an optimization, considering that typically we will be 
issuing unit_max in anticipated FS scenario.

Anyway, I will be changing this according to an earlier comment.

Thanks,
John