On 03/17/2017 01:04 AM, Xiubo Li wrote:
[...]
These days what I have gotten is that the unmap_mapping_range() could
be used.
At the same time I have deep into the mm code and fixed the double
usage of
the data blocks and possible page fault call trace bugs mentioned above.
Following is the V3 p
On 03/08/2017 12:45 AM, lixi...@cmss.chinamobile.com wrote:
From: Xiubo Li
For each target there will be one ring, when the target number
grows larger and larger, it could eventually runs out of the
system memories.
In this patch for each target ring, the cmd area size will be
limited to 8M an
On 03/05/2017 09:48 PM, Xiubo Li wrote:
Maybe fix this by exiting aasda() with iov pointing at the next unused
iov in the array?
May it shouldn't be the aasda()'s duty to increment the iov ptr.
Sure, your call.
[snip text where we agree current solution allows data block sharing,
and new appr
On 02/27/2017 09:47 PM, lixi...@cmss.chinamobile.com wrote:
From: Xiubo Li
If there has BIDI data, its first iov[] will overwrite the last
iov[] for se_cmd->t_data_sg.
(+CCing orig BIDI and data block code authors)
Yeah. It looks like because alloc_and_scatter_data_area() (hereafter
"aasda"
On 02/26/2017 09:59 PM, Xiubo Li wrote:
>> But, We likely don't want to release memory from the data area anyways
>> while active, in any case. How about if we set a timer when active
>> commands go to zero, and then reduce data area to some minimum if no new
>> cmds come in before timer expires?
>
On 02/23/2017 06:07 PM, Xiubo Li wrote:
>> Cool. This is a good approach for an initial patch but this raises
>> concerns about efficiently managing kernel memory usage -- the data area
>> grows but never shrinks, and total possible usage increases per
>> backstore. (What if there are 1000?) Any id
On 02/17/2017 01:24 AM, lixi...@cmss.chinamobile.com wrote:
> From: Xiubo Li
>
> Currently for the TCMU, the ring buffer size is fixed to 64K cmd
> area + 1M data area, and this will be bottlenecks for high iops.
Hi Xiubo, thanks for your work.
daynmic -> dynamic
Have you benchmarked this patc
On 02/13/2017 09:50 PM, Xiubo Li wrote:
> The struct tcmu_cmd_entry {} size is fixed 44 bytes without iovec[], and
> the size of struct iovec[N] is about 16 bytes * N.
>
> The cmd entry size will be [44B, N *16 + 44B], and the data size will be
> [0, N * 4096], so the ratio of sizeof(cmd entry): s
On 02/13/2017 01:50 AM, Xiubo Li wrote:
There is one new scheme in my mind:
Yes I think it's now clear we need more buffer space to avoid
bottlenecks for high iops. The initial design kept it simple with the
1MB vmalloc'd space but anticipated greater would be needed. It should
not be necessar
On 02/12/2017 05:38 PM, Xiubo Li wrote:
On 2017年02月11日 02:02, Andy Grover wrote:
1. increase the region mmap()ed by userspace, TCMU_RING_SIZE, from 1MB
to 1GB or larger
2. Don't vmalloc() the whole thing, instead vmalloc for the cmd ring
portion, and dynamically alloc pages for the data ar
On 02/09/2017 10:48 PM, lixi...@cmss.chinamobile.com wrote:
For now the tcmu is based on UIO framework and only using the
map0 with fixed ring buffer size. This will work fine mostly,
but when we are using the 10GBASE NIC, the ring buffer is too
small to deal with the incoming iscsi cmds.
We can
Hi Martin, (you're the last person to touch this? :)
If I create a local SCSI device using LIO's loopback fabric, and backed
by a file, /sys/block/sda/queue/max_hw_sectors_kb reports 32767. This
reflects the loopback scsi_host_template reporting 65536 max_sectors,
but that's the "controller" -
Hi Varun,
On 04/19/2016 11:30 AM, Varun Prakash wrote:
Add void (*iscsit_get_rx_pdu)() to
struct iscsit_transport, iscsi-target
uses this callback to receive and
process Rx iSCSI PDUs.
cxgbit.ko needs this callback to
reuse iscsi-target Rx thread.
Signed-off-by: Varun Prakash
Acked-by: Sagi G
On 06/06/2016 11:23 PM, Nicholas A. Bellinger wrote:
Hi HCH & Co,
On Mon, 2016-06-06 at 23:22 +0200, Christoph Hellwig wrote:
This patch set adds a generic NVMe over Fabrics target. The
implementation conforms to the NVMe 1.2b specification (which
includes Fabrics) and provides the NVMe over Fa
On 04/01/2016 11:01 AM, Lee Duncan wrote:
On 04/01/2016 12:58 AM, Johannes Thumshirn wrote:
On 2016-03-31 20:05, Lee Duncan wrote:
This commit updates the target core ALUA and PR
modules to use the new "dbroot" attribute instead
of assuming the target database is in "/var/target".
Same goes f
On 06/09/2015 11:41 PM, Hannes Reinecke wrote:
Hi Nic,
here's now the patchset for making LIO-target support 64-bit LUNs.
Pretty straightforward, plus an additional patch to remove the
now obsolete limitation on 256 LUNs per TPG. There had been a
comment in the header that REPORT LUN emulation w
On 05/13/2015 03:12 PM, Chris Leech wrote:
This is only about the structures and functionality involved in maintaining the
iSCSI session, the SCSI host along with it's discovered targets and devices has
no association with network namespaces.
These patches are functional, but not complete. Ther
On 05/12/2015 02:25 AM, Nicholas A. Bellinger wrote:
From: Nicholas Bellinger
This patch converts se_node_acl->device_list[] table for mappedluns
to modern RCU hlist_head usage in order to support an arbitrary number
of node_acl lun mappings.
This includes changes to core_[enable,disable]_devi
On 04/12/2015 07:17 AM, Sebastian Herbszt wrote:
Add target hooks.
Signed-off-by: Sebastian Herbszt
+#ifdef LPFC_TARGET_MODE
+#include "lpfc_target_api.h"
+#include "lpfc_target_api_base.h"
+#endif
Hi, great to see new fabrics posted.
There are a great many ifdefs like this. Are they stri
On 03/30/2015 12:21 PM, Christoph Hellwig wrote:
On Mon, Mar 30, 2015 at 12:16:04PM -0700, Andy Grover wrote:
I dug my patches up and rebased on top of target-pending/for-next. Pushed
here:
git://git.kernel.org/pub/scm/linux/kernel/git/grover/linux.git
mar30-dynalloc-deve
https
On 03/30/2015 05:08 AM, Christoph Hellwig wrote:
Fortunately there is the old patch from Andy to make the se_dev_entry
dynamically allocated, which comes in useful here. With that we might
only change the rdonly flag on a live dev entry, or assign an ACL when
it previously was NULL, something th
On 03/06/2015 10:25 PM, Nicholas A. Bellinger wrote:
That's what I was getting at above: configfs works great for
exactly what we're using it for, but there are other things sysfs does
better so we may want to use both.
Just posted a small series for allowing fabric drivers to expose
this u
On 03/05/2015 11:25 PM, Bart Van Assche wrote:
On 03/05/15 19:39, Andy Grover wrote:
On 03/05/2015 08:06 AM, Bart Van Assche wrote:
If we would do what Nic proposed - modify SCST such that it uses
configfs instead of sysfs - then that would result in the removal of at
least one SCST feature
On 03/05/2015 08:06 AM, Bart Van Assche wrote:
If we would do what Nic proposed - modify SCST such that it uses
configfs instead of sysfs - then that would result in the removal of at
least one SCST feature that is important to its users, namely automatic
population of the configuration filesyst
On 01/30/2015 09:28 AM, Douglas Gilbert wrote:
At the moment many user space programs in Linux access the SG_IO
ioctl related declarations with:
#include
I suspect that file is maintained by the glib folks (because it
does not match the kernel file but is logically equivalent). Now
if this
Signed-off-by: Andy Grover
---
include/scsi/scsi.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/include/scsi/scsi.h b/include/scsi/scsi.h
index d0a66aa..cf778fe 100644
--- a/include/scsi/scsi.h
+++ b/include/scsi/scsi.h
@@ -150,6 +150,7 @@ enum scsi_timeouts {
#define WRITE_16
Make all scsi ioctl-related definitions available via a uapi header.
Change files referencing scsi/scsi_ioctl.h to refer to
uapi/linux/scsi_ioctl.h.
Signed-off-by: Andy Grover
---
block/bsg.c| 2 +-
block/scsi_ioctl.c | 2 +-
drivers/block/cciss.c
tch adds a missing opcode.
Builds with allmodconfig. Thanks for taking a look, and any
suggestions for further improvement.
Regards -- Andy
Andy Grover (5):
scsi: add WRITE_VERIFY_16 to scsi.h
scsi: Move ioctl defs from scsi/scsi.h to scsi/scsi_ioctl.h
scsi: Move user-shareable stuff in scsi
also includes uapi/scsi/scsi.h so kernel code need not update
includes.
Signed-off-by: Andy Grover
---
include/scsi/scsi.h | 497 +
include/uapi/linux/Kbuild | 1 +
include/uapi/linux/scsi.h | 506 ++
3
This will enable user programs to have access to the most current
definitions.
Modify include sites for new path.
Signed-off-by: Andy Grover
---
block/bio.c| 2 +-
block/blk-map.c| 2 +-
block/bsg.c| 2 +-
block/scsi_ioctl.c
These can be consolidated into scsi_ioctl.h with the other ones.
Signed-off-by: Andy Grover
---
include/scsi/scsi.h | 21 -
include/scsi/scsi_ioctl.h | 20
2 files changed, 20 insertions(+), 21 deletions(-)
diff --git a/include/scsi/scsi.h b
On 01/10/2015 02:10 AM, Christoph Hellwig wrote:
On Fri, Jan 09, 2015 at 09:38:50AM -0800, Andy Grover wrote:
Hello glibc people,
This concerns sysdeps/unix/sysv/linux/scsi/{scsi, scsi_ioctl, sg}.h
They define common SCSI values, as well as Linux's SCSI-related ioctls.
Apparently they
On 01/20/2015 04:35 PM, Nicholas A. Bellinger wrote:
As a proof-of-concept, I've implemented a Gluster backend handler. I'm
looking for code review and collaborators, as well as suggestions on
what other userspace handlers we might want to work on. Tape or optical
jukebox emulation? VMDK files??
Hi all,
tcmu-runner is a userspace daemon that simplifies the configuration and
processing of SCSI commands from LIO to userspace handlers, via the new
TCMU userspace passthrough backstore.
https://github.com/agrover/tcmu-runner
As a proof-of-concept, I've implemented a Gluster backend handl
Hello glibc people,
This concerns sysdeps/unix/sysv/linux/scsi/{scsi, scsi_ioctl, sg}.h
They define common SCSI values, as well as Linux's SCSI-related ioctls.
Apparently they were copied from the Linux kernel tree back in 1999, so
they're pretty stale.
I'm wondering if I should submit a pat
On 01/08/2015 03:11 PM, Douglas Gilbert wrote:
Andy,
XCOPY is huge and keeps getting bigger (apart from the recent
trimming of the "LID1" generation mentioned above). Most SCSI
commands are sent to the LU involved, but for EXTENDED COPY
command itself, should it be sent to the source or destinati
On 01/08/2015 12:35 PM, James Bottomley wrote:
On Thu, 2015-01-08 at 11:47 -0800, Andy Grover wrote:
Now that we have uapi/scsi/scsi.h, that is the logical place for SCSI ioctl
definitions to go. The kernel-specific stuff that remains is so little that
it can be folded into scsi/scsi.h, and
ing opcode.
Builds with allmodconfig.
Thanks -- Andy
Andy Grover (3):
scsi: add WRITE_VERIFY_16 to scsi.h
scsi: Move user-shareable stuff in scsi/scsi.h to uapi/scsi/scsi.h
scsi: Remove scsi_ioctl.h
block/bsg.c| 1 -
block/scsi_ioctl.c | 1 -
dri
for ide-floppy_ioctl.c where we just replace it with including
scsi/scsi.h.
Signed-off-by: Andy Grover
---
block/bsg.c| 1 -
block/scsi_ioctl.c | 1 -
drivers/block/cciss.c | 1 -
drivers/block/pktcdvd.c| 1 -
drivers/ide/ide
Signed-off-by: Andy Grover
---
include/scsi/scsi.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/include/scsi/scsi.h b/include/scsi/scsi.h
index 8a7f8ad..faf2118 100644
--- a/include/scsi/scsi.h
+++ b/include/scsi/scsi.h
@@ -150,6 +150,7 @@ enum scsi_timeouts {
#define WRITE_16
A great many SCSI codes and ioctl values can be made available to userspace
in a uapi header, while the kernel-only definitions stay in scsi/scsi.h.
scsi/scsi.h also includes uapi/scsi/scsi.h so kernel code need not update
includes.
Signed-off-by: Andy Grover
---
include/scsi/scsi.h | 522
On 01/07/2015 02:11 PM, Douglas Gilbert wrote:
T10 have now dropped the LID1 and LID4 stuff (its the length
of the LIST IDENTIFIER field: 1 byte or 4 bytes) by obsoleting
all LID1 variants in SPC-5 revision 01. So the LID1 variants
are now gone and the LID4 appendage is dropped from the remaining
On 11/13/2014 11:06 PM, Christoph Hellwig wrote:
On Thu, Nov 13, 2014 at 12:50:58PM -0800, Andy Grover wrote:
The tpg_check_* functions in the fabric_ops struct return 1 or 0,
so we can make their return type bool. Change fabrics to match.
All these true/false stub sounds like some of them
Only used as a param to core_disable_device_list_for_node(), but that
function doesn't use it.
Change function signature to remove that param, and change all callsites
to not pass LUNFLAGS_NO_ACCESS.
Signed-off-by: Andy Grover
---
drivers/target/target_core_device.c | 8 +++-
dr
The tpg_check_* functions in the fabric_ops struct return 1 or 0,
so we can make their return type bool. Change fabrics to match.
Signed-off-by: Andy Grover
---
drivers/infiniband/ulp/srpt/ib_srpt.c| 8
drivers/scsi/qla2xxx/tcm_qla2xxx.c | 10 +-
drivers
LUNFLAGS_READ_WRITE is always the inverse of LUNFLAGS_READ_ONLY.
Removing this enum value resulted in some spots where a parameter's value
can be just true or false, which we can represent with a bool instead of
u32. Change to a bool named "lun_access_ro".
Signed-off-
On 10/13/2014 12:50 AM, Michael S. Tsirkin wrote:
virtio spec 0.9.X requires DRIVER_OK to be set before
VQs are used, but some drivers use VQs before probe
function returns.
Since DRIVER_OK is set after probe, this violates the spec.
Even though under virtio 1.0 transitional devices support this
* The use of idrs may be overkill -- we maybe can replace them with a
simple counter to generate cmd_ids, and a hash table to get a cmd_id's
associated pointer.
* Use of a free-running counter for cmd ring instead of explicit modulo
math. This would require power-of-2 cmd ring siz
The check of SCF_SCSI_DATA_CDB seems to be a remnant from before hch's
refactoring of this function. There are no places where that flag is set
that cmd->execute_cmd isn't also set.
Signed-off-by: Andy Grover
---
drivers/target/target_core_sbc.c | 2 +-
1 file changed, 1 in
Describes the driver and its interface to make it possible for user
programs to back a LIO-exported LUN.
Thanks to Richard W. M. Jones for review, and supplementing this doc
with the first two paragraphs.
Signed-off-by: Andy Grover
---
Documentation/target/tcmu-design.txt | 378
In order to prevent a O(n) search of the filesystem to link up its uio
node with its target configuration, TCMU needs to know the minor number
that UIO assigned. Expose the definition of this struct so TCMU can
access this field.
Signed-off-by: Andy Grover
---
drivers/uio/uio.c | 12
le if device is unconfigured in tcmu_free_device
Thanks -- Andy
Andy Grover (4):
target: Remove unneeded check in sbc_parse_cdb
uio: Export definition of struct uio_device
target: Add documentation on the target userspace pass-through driver
target: Add a user-passthrough backstore
Docum
On 09/30/2014 02:45 AM, Jerome Martin wrote:
On 09/30/2014 11:41 AM, Luigi Tarenga wrote:
On 09/30/2014 11:28 AM, Jerome Martin wrote:
Luigi,
Out of curiosity, does it work as expected with the stock fedora
packaged kernel ?
Best,
--
Jerome
didn't test that configuration and since I can
On 09/18/2014 04:17 PM, Nicholas A. Bellinger wrote:
I'm currently reviewing #2, #4, #5 and #7 and will consider merging
these. These cleanups account for most of the LOC reduction, and avoid
most of the larger concerns.
Also for patches like this, they really need testing on your end before
I
On 09/22/2014 01:58 PM, Nicholas A. Bellinger wrote:
So I'd still like to start for an initial merge with the two different
modes mentioned earlier. The pure-passthrough mode where everything is
handled by user-space, and an I/O passthrough mode where only
SCF_SCSI_DATA_CDB is passed along to us
On 09/19/2014 04:51 PM, Alex Elsayed wrote:
Not sure I follow.. How does the proposed passthrough mode prevent
someone from emulating OSDs, media changers, optical disks or anything
else in userspace with TCMU..?
The main thing that the above comments highlight is why attempting to
combine the
On 09/18/2014 12:38 AM, Nicholas A. Bellinger wrote:
On Sat, 2014-09-13 at 21:55 +0200, Christoph Hellwig wrote:
ping again. We're getting closer to the end of the 3.18 merge window
and there still hasn't been a response. Should Andy just send the patches
directly to Linus once 3.18 opens give
lly is hindered until the kernel code is accepted :)
Changes since version 2:
* Incorporate doc improvements from Richard W. M. Jones
* Correct off-by-1 error in tcmu_get_blocks
* execute_rw must be set in the sbc_ops passed to sbc_parse_cdb
Thanks -- Andy
Andy Grover (4):
target: Remove un
Describes the driver and its interface to make it possible for user
programs to back a LIO-exported LUN.
Thanks to Richard W. M. Jones for review, and supplementing this doc
with the first two paragraphs.
Signed-off-by: Andy Grover
---
Documentation/target/tcmu-design.txt | 239
In order to prevent a O(n) search of the filesystem to link up its uio
node with its target configuration, TCMU needs to know the minor number
that UIO assigned. Expose the definition of this struct so TCMU can
access this field.
Signed-off-by: Andy Grover
---
drivers/uio/uio.c | 12
The check of SCF_SCSI_DATA_CDB seems to be a remnant from before hch's
refactoring of this function. There are no places where that flag is set
that cmd->execute_cmd isn't also set.
Signed-off-by: Andy Grover
---
drivers/target/target_core_sbc.c | 2 +-
1 file changed, 1 in
* The use of idrs may be overkill -- we maybe can replace them with a
simple counter to generate cmd_ids, and a hash table to get a cmd_id's
associated pointer.
* Use of a free-running counter for cmd ring instead of explicit modulo
math. This would require power-of-2 cmd ring siz
On 08/31/2014 02:22 PM, Richard W.M. Jones wrote:
Reading this several times, I now think I get what it's trying to say,
but I think it needs to introduces the terms (as the Economist style
does). Something like this:
"TCM is the new name for LIO, an in-kernel iSCSI target (server).
Exist
On 08/30/2014 10:35 AM, Richard W.M. Jones wrote:
On Tue, Jul 01, 2014 at 12:11:14PM -0700, Andy Grover wrote:
Describes the driver and its interface to make it possible for user
programs to back a LIO-exported LUN.
Signed-off-by: Andy Grover
---
Documentation/target/tcmu-design.txt | 210
On 07/29/2014 06:15 AM, Christoph Hellwig wrote:
Nic,
any progress on looking over these? Seems like there's actually
nothing at all queued up for 3.17 in the target tree, or am І missing
something?
ping?
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of
rintks in code, still.
Signed-off-by: Andy Grover
---
drivers/target/Kconfig |5 +
drivers/target/Makefile|1 +
drivers/target/target_core_transport.c |4 +
drivers/target/target_core_user.c | 1158
include/uapi/li
Describes the driver and its interface to make it possible for user
programs to back a LIO-exported LUN.
Signed-off-by: Andy Grover
---
Documentation/target/tcmu-design.txt | 229 +++
1 file changed, 229 insertions(+)
create mode 100644 Documentation/target/tcmu
s/userback, create a user
storage object and give it a size, and 'file/foo.img' as config
parameter
* Start tcmu-runner (manually for now)
* Create a loopback fabric and link the above storage object to it
Andy Grover (4):
target: Remove unneeded check in sbc_parse_cdb
uio: Export d
In order to prevent a O(n) search of the filesystem to link up its uio
node with its target configuration, TCMU needs to know the minor number
that UIO assigned. Expose the definition of this struct so TCMU can
access this field.
Signed-off-by: Andy Grover
---
drivers/uio/uio.c | 12
The check of SCF_SCSI_DATA_CDB seems to be a remnant from before hch's
refactoring of this function. There are no places where that flag is set
that cmd->execute_cmd isn't also set.
Signed-off-by: Andy Grover
---
drivers/target/target_core_sbc.c | 2 +-
1 file changed, 1 in
[re-adding individual CCs that were dropped]
On 07/05/2014 04:29 AM, Alex Elsayed wrote:
+Device Discovery:
+
+Other devices may be using UIO besides TCMU. Unrelated user processes
+may also be handling different sets of TCMU devices. TCMU userspace
+processes must find their devices by scanning
Hi, here's some additional beginning bits for the configuration and use
of this code:
Changes to targetcli-fb and rtslib-fb:
https://github.com/agrover/targetcli-fb/tree/userback
https://github.com/agrover/rtslib-fb/tree/userback
An incredibly-crappy stand-in for the user process that will han
rintks in code, still.
Signed-off-by: Andy Grover
---
drivers/target/Kconfig |5 +
drivers/target/Makefile|1 +
drivers/target/target_core_transport.c |4 +
drivers/target/target_core_user.c | 1078
drivers/target/
Describes the driver and its interface to make it possible for user
programs to back a LIO-exported LUN.
Signed-off-by: Andy Grover
---
Documentation/target/tcmu-design.txt | 210 +++
1 file changed, 210 insertions(+)
create mode 100644 Documentation/target/tcmu
is the actual code, along with
implementation rationale.
Thanks -- Andy
[1] http://thread.gmane.org/gmane.linux.scsi.target.devel/5044
Andy Grover (2):
target: Add documentation on the target userspace pass-through driver
target: Add a user-passthrough backstore
Documentation/target/tcmu
Reported-by: Maurizio Lombardi
Signed-off-by: Andy Grover
---
Hi Maurizio, yup, thanks!
Here's a patch that fixes the issue - it can be applied as 9/8 or
alternatively squashed into 1/8.
Thanks again -- Regards -- Andy
drivers/target/target_core_ua.c | 4 +++-
1 file changed, 3 inser
ons
referencing deve inside the spinlock, since it's not safe to access deve
outside it. Skip reinitializing, since the deve is actually being freed.
Do not allocate device_list array separately, but include it in
se_node_acl.
Signed-off-by: Andy Grover
---
drivers/target/target_core_
hange the other place enable_device_list_for_node is called,
add_initiator_node_lun_acl. Hold tpg_lun_lock across all uses of lun.
Change error handling to release lock from error paths. Remove
core_dev_get_lun.
Signed-off-by: Andy Grover
---
drivers/target/target_core_devi
Nothing in it can raise an error.
Reviewed-by: Christoph Hellwig
Signed-off-by: Andy Grover
---
drivers/target/target_core_internal.h | 2 +-
drivers/target/target_core_tpg.c | 5 +
2 files changed, 2 insertions(+), 5 deletions(-)
diff --git a/drivers/target/target_core_internal.h
b
hanks -- Regards -- Andy
Andy Grover (8):
target: Add locking to some accesses to nacl.device_list
target: Don't unlock/relock tpg_lun_lock in loop in add_node_to_devs
target: Allocate se_dev_entrys in device list only when used
target: core_tpg_post_dellun can return void
targ
igned-off-by: Andy Grover
---
drivers/target/target_core_device.c | 18 --
drivers/target/target_core_fabric_configfs.c | 2 +-
drivers/target/target_core_internal.h| 5 ++--
drivers/target/target_core_tpg.c | 36 +++-
4 files ch
Make sure all accesses of deve elements in device_list are protected. This
will become critically important once device_list entries are potentially
NULL.
Reported-by: Christoph Hellwig
Signed-off-by: Andy Grover
---
drivers/target/target_core_device.c | 7 ++---
drivers/target
Simple and just called from one place.
Reviewed-by: Christoph Hellwig
Signed-off-by: Andy Grover
---
drivers/target/target_core_tpg.c | 9 +
1 file changed, 1 insertion(+), 8 deletions(-)
diff --git a/drivers/target/target_core_tpg.c b/drivers/target/target_core_tpg.c
index a63a3d7
_list, put it directly in se_portal_group.
Reviewed-by: Christoph Hellwig
Signed-off-by: Andy Grover
---
drivers/target/sbp/sbp_target.c | 6 +-
drivers/target/target_core_device.c | 43 ++--
drivers/target/target_core_fabric_configfs.c | 21 +++---
drivers/t
Create helper functions to alloc a deve and to transition it from
demo mode to explicit.
Signed-off-by: Andy Grover
---
drivers/target/target_core_device.c | 121 ++--
1 file changed, 74 insertions(+), 47 deletions(-)
diff --git a/drivers/target
Fix comment to make it clear that check_dev_wce can return true if
emulation is in use, but also if it's *not* emulated but the underlying
device supports it.
Reviewed-by: Chris Leech
Signed-off-by: Andy Grover
---
drivers/target/target_core_spc.c | 2 +-
1 file changed, 1 insertion(
Simple and just called from one place.
Signed-off-by: Andy Grover
---
drivers/target/target_core_tpg.c | 9 +
1 file changed, 1 insertion(+), 8 deletions(-)
diff --git a/drivers/target/target_core_tpg.c b/drivers/target/target_core_tpg.c
index a63a3d7..bc0299a 100644
--- a/drivers
Hi nab, hch, and all,
This patchset reduces the amount of memory for se_dev_entry and se_lun
arrays by waiting to allocate array members until they are
created. This patch saves up to 261KB per TPG, and up to 65KB per
ACL. It also fixes a number of locking bugs around these data
structures.
Andy
hange the other place enable_device_list_for_node is called,
add_initiator_node_lun_acl. Hold tpg_lun_lock across all uses of lun.
Change error handling to release lock from error paths. Remove
core_dev_get_lun.
Signed-off-by: Andy Grover
---
drivers/target/target_core_devi
_list, put it directly in se_portal_group.
Signed-off-by: Andy Grover
---
drivers/target/sbp/sbp_target.c | 6 +-
drivers/target/target_core_device.c | 43 ++--
drivers/target/target_core_fabric_configfs.c | 21 +++---
drivers/target/target_core_internal.h
Create helper functions to alloc a deve and to transition it from
demo mode to explicit.
Signed-off-by: Andy Grover
---
drivers/target/target_core_device.c | 121 ++--
1 file changed, 74 insertions(+), 47 deletions(-)
diff --git a/drivers/target
Make sure all accesses of deve elements in device_list are protected. This
will become critically important once device_list entries are potentially
NULL.
Reported-by: Christoph Hellwig
Signed-off-by: Andy Grover
---
drivers/target/target_core_device.c | 7 ++---
drivers/target
Nothing in it can raise an error.
Signed-off-by: Andy Grover
---
drivers/target/target_core_internal.h | 2 +-
drivers/target/target_core_tpg.c | 5 +
2 files changed, 2 insertions(+), 5 deletions(-)
diff --git a/drivers/target/target_core_internal.h
b/drivers/target
Remove core_tpg_pre_dellun entirely, since we don't need to get/check
a pointer we already have.
Nothing else can return an error, so core_dev_del_lun can return void.
Signed-off-by: Andy Grover
---
drivers/target/target_core_device.c | 18 +---
drivers/t
ons
referencing deve inside the spinlock, since it's not safe to access deve
outside it. Skip reinitializing, since the deve is actually being freed.
Do not allocate device_list array separately, but include it in
se_node_acl.
Signed-off-by: Andy Grover
---
drivers/target/target_core_
A clearer name, now that pre_dellun is gone.
Signed-off-by: Andy Grover
---
drivers/target/target_core_device.c | 2 +-
drivers/target/target_core_internal.h | 2 +-
drivers/target/target_core_tpg.c | 4 ++--
3 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/target
On 05/14/2014 05:57 PM, Nicholas A. Bellinger wrote:
On Wed, 2014-05-14 at 17:22 -0700, Andy Grover wrote:
On 05/14/2014 05:07 PM, Nicholas A. Bellinger wrote:
On Wed, 2014-05-14 at 15:48 -0700, Andy Grover wrote:
Just like for pSCSI, if the transport sets get_write_cache, then it is
not
On 05/14/2014 05:11 PM, Nicholas A. Bellinger wrote:
On Wed, 2014-05-14 at 15:51 -0700, Andy Grover wrote:
Whether this is set or not is important for whether good performance is
possible, when using the block device as a backstore for the kernel
target. Exposing this will let kernel target
On 05/14/2014 05:07 PM, Nicholas A. Bellinger wrote:
On Wed, 2014-05-14 at 15:48 -0700, Andy Grover wrote:
Just like for pSCSI, if the transport sets get_write_cache, then it is
not valid to enable write cache emulation for it. Return an error.
see https://bugzilla.redhat.com/show_bug.cgi?id
Signed-off-by: Andy Grover
---
block/genhd.c | 10 ++
1 file changed, 10 insertions(+)
diff --git a/block/genhd.c b/block/genhd.c
index 791f419..258fadc 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -958,6 +958,14 @@ static ssize_t disk_capability_show(struct device *dev
1 - 100 of 234 matches
Mail list logo
