Re: [PATCH 2/3] befs: improve documentation in datastream.c

From: Salah Triki
Date: Tue Aug 16 2016 - 03:35:08 EST


On Sat, Aug 13, 2016 at 06:11:20PM +0100, Luis de Bethencourt wrote:
> Convert function descriptions to kernel-doc style.
>
> Signed-off-by: Luis de Bethencourt <luisbg@xxxxxxxxxxxxxxx>
> ---
> fs/befs/datastream.c | 193 ++++++++++++++++++++++++++-------------------------
> 1 file changed, 98 insertions(+), 95 deletions(-)
>
> diff --git a/fs/befs/datastream.c b/fs/befs/datastream.c
> index b2eb5b5..5ce85cf 100644
> --- a/fs/befs/datastream.c
> +++ b/fs/befs/datastream.c
> @@ -75,7 +75,13 @@ befs_read_datastream(struct super_block *sb, const befs_data_stream *ds,
> return bh;
> }
>
> -/*
> +/**
> + * befs_fblock2brun - give back block run for fblock
> + * @sb: the superblock
> + * @data: datastream to read from
> + * @fblock: the blocknumber with the file position to find
> + * @run: The found run is passed back through this pointer
> + *
> * Takes a file position and gives back a brun who's starting block
> * is block number fblock of the file.
> *
> @@ -212,36 +218,35 @@ befs_count_blocks(struct super_block *sb, const befs_data_stream *ds)
> return blocks;
> }
>
> -/*
> - Finds the block run that starts at file block number blockno
> - in the file represented by the datastream data, if that
> - blockno is in the direct region of the datastream.
> -
> - sb: the superblock
> - data: the datastream
> - blockno: the blocknumber to find
> - run: The found run is passed back through this pointer
> -
> - Return value is BEFS_OK if the blockrun is found, BEFS_ERR
> - otherwise.
> -
> - Algorithm:
> - Linear search. Checks each element of array[] to see if it
> - contains the blockno-th filesystem block. This is necessary
> - because the block runs map variable amounts of data. Simply
> - keeps a count of the number of blocks searched so far (sum),
> - incrementing this by the length of each block run as we come
> - across it. Adds sum to *count before returning (this is so
> - you can search multiple arrays that are logicaly one array,
> - as in the indirect region code).
> -
> - When/if blockno is found, if blockno is inside of a block
> - run as stored on disk, we offset the start and length members
> - of the block run, so that blockno is the start and len is
> - still valid (the run ends in the same place).
> -
> - 2001-11-15 Will Dyson
> -*/
> +/**
> + * befs_find_brun_direct - find a direct block run in the datastream
> + * @sb: the superblock
> + * @data: the datastream
> + * @blockno: the blocknumber to find
> + * @run: The found run is passed back through this pointer
> + *
> + * Finds the block run that starts at file block number blockno
> + * in the file represented by the datastream data, if that
> + * blockno is in the direct region of the datastream.
> + *
> + * Return value is BEFS_OK if the blockrun is found, BEFS_ERR
> + * otherwise.
> + *
> + * Algorithm:
> + * Linear search. Checks each element of array[] to see if it
> + * contains the blockno-th filesystem block. This is necessary
> + * because the block runs map variable amounts of data. Simply
> + * keeps a count of the number of blocks searched so far (sum),
> + * incrementing this by the length of each block run as we come
> + * across it. Adds sum to *count before returning (this is so
> + * you can search multiple arrays that are logicaly one array,
> + * as in the indirect region code).
> + *
> + * When/if blockno is found, if blockno is inside of a block
> + * run as stored on disk, we offset the start and length members
> + * of the block run, so that blockno is the start and len is
> + * still valid (the run ends in the same place).
> + */
> static int
> befs_find_brun_direct(struct super_block *sb, const befs_data_stream *data,
> befs_blocknr_t blockno, befs_block_run * run)
> @@ -273,29 +278,28 @@ befs_find_brun_direct(struct super_block *sb, const befs_data_stream *data,
> return BEFS_ERR;
> }
>
> -/*
> - Finds the block run that starts at file block number blockno
> - in the file represented by the datastream data, if that
> - blockno is in the indirect region of the datastream.
> -
> - sb: the superblock
> - data: the datastream
> - blockno: the blocknumber to find
> - run: The found run is passed back through this pointer
> -
> - Return value is BEFS_OK if the blockrun is found, BEFS_ERR
> - otherwise.
> -
> - Algorithm:
> - For each block in the indirect run of the datastream, read
> - it in and search through it for search_blk.
> -
> - XXX:
> - Really should check to make sure blockno is inside indirect
> - region.
> -
> - 2001-11-15 Will Dyson
> -*/
> +/**
> + * befs_find_brun_indirect - find a block run in the datastream
> + * @sb: the superblock
> + * @data: the datastream
> + * @blockno: the blocknumber to find
> + * @run: The found run is passed back through this pointer
> + *
> + * Finds the block run that starts at file block number blockno
> + * in the file represented by the datastream data, if that
> + * blockno is in the indirect region of the datastream.
> + *
> + * Return value is BEFS_OK if the blockrun is found, BEFS_ERR
> + * otherwise.
> + *
> + * Algorithm:
> + * For each block in the indirect run of the datastream, read
> + * it in and search through it for search_blk.
> + *
> + * XXX:
> + * Really should check to make sure blockno is inside indirect
> + * region.
> + */
> static int
> befs_find_brun_indirect(struct super_block *sb,
> const befs_data_stream *data,
> @@ -365,47 +369,46 @@ befs_find_brun_indirect(struct super_block *sb,
> return BEFS_ERR;
> }
>
> -/*
> - Finds the block run that starts at file block number blockno
> - in the file represented by the datastream data, if that
> - blockno is in the double-indirect region of the datastream.
> -
> - sb: the superblock
> - data: the datastream
> - blockno: the blocknumber to find
> - run: The found run is passed back through this pointer
> -
> - Return value is BEFS_OK if the blockrun is found, BEFS_ERR
> - otherwise.
> -
> - Algorithm:
> - The block runs in the double-indirect region are different.
> - They are always allocated 4 fs blocks at a time, so each
> - block run maps a constant amount of file data. This means
> - that we can directly calculate how many block runs into the
> - double-indirect region we need to go to get to the one that
> - maps a particular filesystem block.
> -
> - We do this in two stages. First we calculate which of the
> - inode addresses in the double-indirect block will point us
> - to the indirect block that contains the mapping for the data,
> - then we calculate which of the inode addresses in that
> - indirect block maps the data block we are after.
> -
> - Oh, and once we've done that, we actually read in the blocks
> - that contain the inode addresses we calculated above. Even
> - though the double-indirect run may be several blocks long,
> - we can calculate which of those blocks will contain the index
> - we are after and only read that one. We then follow it to
> - the indirect block and perform a similar process to find
> - the actual block run that maps the data block we are interested
> - in.
> -
> - Then we offset the run as in befs_find_brun_array() and we are
> - done.
> -
> - 2001-11-15 Will Dyson
> -*/
> +/**
> + * befs_find_brun_dblindirect - find a block run in the datastream
> + * @sb: the superblock
> + * @data: the datastream
> + * @blockno: the blocknumber to find
> + * @run: The found run is passed back through this pointer
> + *
> + * Finds the block run that starts at file block number blockno
> + * in the file represented by the datastream data, if that
> + * blockno is in the double-indirect region of the datastream.
> + *
> + * Return value is BEFS_OK if the blockrun is found, BEFS_ERR
> + * otherwise.
> + *
> + * Algorithm:
> + * The block runs in the double-indirect region are different.
> + * They are always allocated 4 fs blocks at a time, so each
> + * block run maps a constant amount of file data. This means
> + * that we can directly calculate how many block runs into the
> + * double-indirect region we need to go to get to the one that
> + * maps a particular filesystem block.
> + *
> + * We do this in two stages. First we calculate which of the
> + * inode addresses in the double-indirect block will point us
> + * to the indirect block that contains the mapping for the data,
> + * then we calculate which of the inode addresses in that
> + * indirect block maps the data block we are after.
> + *
> + * Oh, and once we've done that, we actually read in the blocks
> + * that contain the inode addresses we calculated above. Even
> + * though the double-indirect run may be several blocks long,
> + * we can calculate which of those blocks will contain the index
> + * we are after and only read that one. We then follow it to
> + * the indirect block and perform a similar process to find
> + * the actual block run that maps the data block we are interested
> + * in.
> + *
> + * Then we offset the run as in befs_find_brun_array() and we are
> + * done.
> + */
> static int
> befs_find_brun_dblindirect(struct super_block *sb,
> const befs_data_stream *data,
> --
> 2.5.1
>

Signed-off-by: Salah Triki <salah.triki@xxxxxxxxx>

Thanks,
Salah