linux-toradex.git/drivers/target, branch v3.2-rc5

iscsi-target: Fix hex2bin warn_unused compile message

2011-12-06T06:00:59+00:00

Fix the following compile warning with hex2bin() usage:

drivers/target/iscsi/iscsi_target_auth.c: In function ‘chap_string_to_hex’:
drivers/target/iscsi/iscsi_target_auth.c:35: warning: ignoring return value of ‘hex2bin’, declared with attribute warn_unused_result

Signed-off-by: Nicholas Bellinger

target: Don't return an error if disabling unsupported features

2011-12-06T06:00:58+00:00

If an attribute is present (but not yet supported) it should be OK
to write 0 (a no-op) to the attribute.

This is an issue because userspace should be able to save and restore all
set attribute values without error.

Signed-off-by: Andy Grover 
Signed-off-by: Nicholas Bellinger

target/rd: fix or rewrite the copy routine

2011-12-06T06:00:58+00:00

So the code assumes that the sg list is only a array while in reality
loopback SGL memory via scsi_cmnd into target-core may be already
chained.  This patch converts ramdisk code to use sg_miter logic from
scatterlist.h in order to properly support passthrough SGL usage with
transport_generic_map_mem_to_cmd() via loopback.

With this patch the bug goes away. However after umount/mount of the
device my files are gone. So something is still not right. After looking
at it for a while I decided to rewrite the that part of the code and now
things do work for me.

For reference:
- http://article.gmane.org/gmane.linux.scsi.target.devel/595
  the sg_next() conversion
- http://article.gmane.org/gmane.linux.scsi.target.devel/602
  the rewrite of the copy code

(nab: Fix compile warning in rd_MEMCPY)

Signed-off-by: Sebastian Andrzej Siewior 
Cc: stable@kernel.org
Signed-off-by: Nicholas Bellinger

target/rd: simplify the page/offset computation

2011-12-06T06:00:57+00:00

Breakout rd_MEMCPY_do_task() usage of do_div() to tmp value during
rd_request->rd_page assignment.

Signed-off-by: Sebastian Andrzej Siewior 
Signed-off-by: Nicholas Bellinger

target: remove the unused se_dev_list

2011-12-06T06:00:57+00:00

Signed-off-by: Christoph Hellwig 
Signed-off-by: Nicholas Bellinger

target/file: walk properly over sg list

2011-12-06T06:00:57+00:00

This patch changes fileio to use for_each_sg() when walking se_task->task_sg
memory passed into from loopback LLD struct scsi_cmnd scatterlist memory.

This addresses an issue where FILEIO backends with loopback where hitting the
following OOPs with mkfs.ext2:

|kernel BUG at include/linux/scatterlist.h:97!
|invalid opcode: 0000 [#1] PREEMPT SMP
|Modules linked in: sd_mod tcm_loop target_core_stgt scsi_tgt target_core_pscsi target_core_file target_core_iblock target_core_mod configfs scsi_mod
|
|Pid: 671, comm: LIO_fileio Not tainted 3.1.0-rc10+ #139 Bochs Bochs
|EIP: 0060:[] EFLAGS: 00010202 CPU: 0
|EIP is at fd_do_task+0x396/0x420 [target_core_file]
| [] __transport_execute_tasks+0xd4/0x190 [target_core_mod]
| [] transport_execute_tasks+0x3c/0xf0 [target_core_mod]
|EIP: [] fd_do_task+0x396/0x420 [target_core_file] SS:ESP 0068:dea47e90

Signed-off-by: Sebastian Andrzej Siewior 
Cc: Christoph Hellwig 
Cc: stable@kernel.org
Signed-off-by: Nicholas Bellinger

target: remove unused struct fields

2011-12-06T06:00:56+00:00

Some are never used, some are set but never read, dev_hoq_count is
incremented and decremented, but never read.

Signed-off-by: Joern Engel 
Signed-off-by: Nicholas Bellinger

target: Fix page length in emulated INQUIRY VPD page 86h

2011-12-06T06:00:56+00:00

The LSB of the page length is at offset 3, not 2.

Signed-off-by: Roland Dreier 
Cc: stable@kernel.org
Signed-off-by: Nicholas Bellinger

target: Handle 0 correctly in transport_get_sectors_6()

2011-12-06T06:00:55+00:00

SBC-3 says:

    A TRANSFER LENGTH field set to zero specifies that 256 logical
    blocks shall be written.  Any other value specifies the number
    of logical blocks that shall be written.

The old code was always just returning the value in the TRANSFER LENGTH
byte.  Fix this to return 256 if the byte is 0.

Signed-off-by: Roland Dreier 
Cc: stable@kernel.org
Signed-off-by: Nicholas Bellinger

target: Don't return an error status for 0-length READ and WRITE

2011-12-06T06:00:55+00:00

IO commands with a TRANSFER LENGTH of 0 are not an error; for example,
for READ (10) and WRITE (10), SBC-3 says:

    A TRANSFER LENGTH field set to zero specifies that no logical blocks
    shall be read. This condition shall not be considered an error.

In case we have nothing to do, just complete the command with good status.

Signed-off-by: Roland Dreier 
Cc: stable@kernel.org
Signed-off-by: Nicholas Bellinger