linux-scsi.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH 0/3] tcm: Add TPE=1 / UNMAP emulation + passthrough support
@ 2010-09-27 22:51 Nicholas A. Bellinger
  0 siblings, 0 replies; only message in thread
From: Nicholas A. Bellinger @ 2010-09-27 22:51 UTC (permalink / raw)
  To: linux-scsi, linux-kernel
  Cc: Christoph Hellwig, Martin K. Petersen, Douglas Gilbert,
	Jens Axboe, FUJITA Tomonori, Mike Christie, Hannes Reinecke,
	James Bottomley, Konrad Rzeszutek Wilk, Boaz Harrosh,
	Richard Sharpe, Nicholas Bellinger

From: Nicholas Bellinger <nab@linux-iscsi.org>

Greetings all,

This series adds generic TPE=1 / UNMAP support into TCM Core and TCM/IBLOCK +
TCM/FILEIO subsystem plugin code code to issue blkdev_issue_discard() for
struct block_device that support QUEUE_FLAG_DISCARD.  Using TPE=1 / UNMAP
is also supported with direct struct request passthrough using TCM/pSCSI
and everything required for this to function is included in patch #1.

Patch #1 also includes the addition of the SBC-3 Block Limits VPD (0xb0) page
and adds the following UNMAP related knobs returned into configfs attributes in
/sys/kernel/config/target/core/$HBA/$DEV/attrib, the new ones include:

	*) optimal_sectors (currently set to our ->max_sectors)
	*) max_unmap_block_desc_count (set to 0 by default)
	*) max_unmap_lba_count (set to 0 by default)
	*) unmap_granularity (set to 0 by default)
	*) unmap_granularity_alignment (set to 0 by default)

Patch #2 includes the necessary changes to IBLOCK to detect QUEUE_FLAG_DISCARD
and call transport_generic_unmap() for each UNMAP op.

Patch #3 includes the necessary changes to FILEIO w/ a valid S_ISBLK() to
detect QUEUE_FLAG_DISCARD and call transport_generic_unmap() for each UNMAP
op.  This is the one item I am unsure about, eg: using iget() + iput() to locate
struct block_device for blk_issue_discard()..?  hch..?

So far these patches have been tested with scsi_debug LUNs w/ TPE=1 with both
IBLOCK and FILEIO TPE=1 emulation modes, along with TCM/pSCSI passthrough mode on
v2.6.36-rc4 HVM.  Here is what Block Limits VPD output looks like via TCM_Loop:

	target# sg_inq --page=0xb0 /dev/sdh
	VPD INQUIRY: Block limits page (SBC)
	  Maximum compare and write length: 0 blocks
	  Optimal transfer length granularity: 1 blocks
	  Maximum transfer length: 1024 blocks
	  Optimal transfer length: 1024 blocks
	  Maximum prefetch, xdread, xdwrite transfer length: 0 blocks
	  Maximum unmap LBA count: 0
	  Maximum unmap block descriptor count: 0
	  Optimal unmap granularity: 0
	  Unmap granularity alignment valid: 0
	  Unmap granularity alignment: 0

and with sg_unmap:

	target# sg_unmap -v --lba=78 --num=4 --verbose /dev/sdh
	open /dev/sdh with flags=0x802
	    unmap cdb: 42 00 00 00 00 00 00 00 18 00 
	    unmap parameter list:
	        00 16 00 10 00 00 00 00  00 00 00 00 00 00 00 4e    
	        00 00 00 04 00 00 00 00                         

Comments are welcome!

Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>

Nicholas Bellinger (3):
  tcm: Add Thin Provisioning / UNMAP emulation and Block Limits VPD
    page
  tcm/iblock: Add UNMAP / Block DISCARD support
  tcm/fileio: Add UNMAP / Block DISCARD support

 drivers/target/target_core_configfs.c  |   24 +++++
 drivers/target/target_core_device.c    |   90 +++++++++++++++++++
 drivers/target/target_core_file.c      |   53 +++++++++++-
 drivers/target/target_core_iblock.c    |   19 ++++-
 drivers/target/target_core_transport.c |  152 ++++++++++++++++++++++++++++++++
 include/target/target_core_base.h      |    9 ++-
 include/target/target_core_device.h    |    6 ++
 include/target/target_core_transport.h |   11 +++
 8 files changed, 361 insertions(+), 3 deletions(-)

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2010-09-27 22:51 UTC | newest]

Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-09-27 22:51 [PATCH 0/3] tcm: Add TPE=1 / UNMAP emulation + passthrough support Nicholas A. Bellinger

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).