On Wed, 2013-02-20 at 11:17 +0800, Asias He wrote:
> On 02/20/2013 09:43 AM, Nicholas A. Bellinger wrote:
> > From: Nicholas Bellinger <n...@linux-iscsi.org>
> >
> > Hi mkp, hch, & Co,
> >
> > Please review this patch to add support for WRITE_SAME w/ UNMAP=0
> > emulation into FILEIO, as I'd really like to include this in the
> > upcoming for-3.9 target merge.
> >
> > Thank you,
> >
> > --nab
> >
> > ---------------------------------------------------------------------
> >
> > This patch adds support for emulation of WRITE_SAME w/ UNMAP=0 within
> > fd_execute_write_same() backend code.
> >
> > The emulation uses vfs_writev() to submit a locally populated buffer
> > from the received WRITE_SAME scatterlist block for duplication, and by
> > default enforces a limit of max_write_same_len=0x1000 (8192) sectors up
> > to the limit of 1024 iovec entries for the single call to vfs_writev().
> >
> > It also sets max_write_same_len to the operational default at setup ->
> > fd_configure_device() time.
> >
> > Tested with 512, 1k, 2k, and 4k block_sizes.
> >
> > Cc: Martin K. Petersen <martin.peter...@oracle.com>
> > Cc: Christoph Hellwig <h...@lst.de>
> > Cc: Asias He <as...@redhat.com>
> > Signed-off-by: Nicholas Bellinger <n...@linux-iscsi.org>
> > ---
> > drivers/target/target_core_file.c | 115
> > +++++++++++++++++++++++++++++++++++++
> > 1 files changed, 115 insertions(+), 0 deletions(-)
> >
<SNIP>
> > +
> > +static sense_reason_t
> > +fd_execute_write_same(struct se_cmd *cmd)
> > +{
> > + struct se_device *se_dev = cmd->se_dev;
> > + struct fd_dev *fd_dev = FD_DEV(se_dev);
> > + struct file *f = fd_dev->fd_file;
> > + struct scatterlist *sg;
> > + struct iovec *iov;
> > + mm_segment_t old_fs;
> > + sector_t nolb = spc_get_write_same_sectors(cmd);
> > + loff_t pos = cmd->t_task_lba * se_dev->dev_attrib.block_size;
> > + unsigned int len, len_tmp, iov_num;
> > + int i, rc;
> > + unsigned char *buf;
> > +
> > + if (!nolb) {
> > + target_complete_cmd(cmd, SAM_STAT_GOOD);
> > + return 0;
> > + }
> > + sg = &cmd->t_data_sg[0];
> > +
> > + if (cmd->t_data_nents > 1 ||
> > + sg->length != cmd->se_dev->dev_attrib.block_size) {
> > + pr_err("WRITE_SAME: Illegal SGL t_data_nents: %u length: %u"
> > + " block_size: %u\n", cmd->t_data_nents, sg->length,
> > + cmd->se_dev->dev_attrib.block_size);
> > + return TCM_INVALID_CDB_FIELD;
> > + }
> > +
> > + len = len_tmp = nolb * se_dev->dev_attrib.block_size;
> > + iov_num = DIV_ROUND_UP(len, PAGE_SIZE);
> > +
> > + buf = fd_setup_write_same_buf(cmd, sg, len);
> > + if (!buf)
> > + return TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE;
> > +
> > + iov = vmalloc(sizeof(struct iovec) * iov_num);
> > + if (!iov) {
> > + pr_err("Unable to allocate fd_execute_write_same iovecs\n");
> > + kfree(buf);
> > + return TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE;
> > + }
> > + memset(iov, 0, sizeof(struct iovec) * iov_num);
>
> Use vzalloc()?
Of course.. Changed to vzalloc + dropped memset.
Thanks for reviewing!
--nab
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
