scst_cmd_atomic() -- the 2.6.29 checkpatch script issues a warning about it.
At the same time, the scst_cmd_atomic() comment has been converted to
kernel-doc style. See also Documentation/kernel-doc-nano-HOWTO.txt for more
information.
Signed-off-by: Bart Van Assche <bart.vanassche@gmail.com>
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@723 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Improve logging on write access on read-only devices
- Make the same initiator coming through different sessions use shared IO context
- Cleanups
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@721 d57e44dd-8a1f-0410-8b47-8ef2f437770f
While writing DKMS support for the SCST and iSCST-SCST kernel modules, I ran
into a (common) problem with regard to out-of-tree build detection. A common
method to detect out-of-tree building is by using the KERNELRELEASE variable.
The usage of this variable poses a problem when using DKMS. DKMS itself uses
the KERNELRELEASE variable to specify the kernel for which the driver to
build. A suggested alternative would to use the PATCHLEVEL variable as part
of the detection mechanism instead. I attached a patch reflecting this
suggested change.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@720 d57e44dd-8a1f-0410-8b47-8ef2f437770f
I replaced all occurences of DISTDIR with DESTDIR and also introduced SBINDIR (and friends) to some other Makefiles.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@719 d57e44dd-8a1f-0410-8b47-8ef2f437770f
checkpatch issues reported by the checkpatch script included with the 2.6.29
kernel and that were not yet reported by the 2.6.28 checkpatch script
(trailing statements should be on the next line / struct should normally be
const).
The patch below has been tested as follows:
- Reran scripts/run-regression-tests -k 2.6.24.7 -k 2.6.25.20 -k 2.6.26.8 -k 2.6.27.21 -k 2.6.28.9 -k 2.6.29 and verified the output.
- Rebuilt, installed and loaded scst, iscsi-scst and srpt as follows:
make -s clean && make -s -C scst install && make -s -C iscsi-scst install && make -s -C srpt install && cd scstadmin && make -s && make -s install && modprobe scst_vdisk && modprobe iscsi-scst && dmesg
Signed-off-by: Bart Van Assche <bart.vanassche@gmail.com>
with minor cleanups and corrections in put_page_callback-2.6.29.patch
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@717 d57e44dd-8a1f-0410-8b47-8ef2f437770f
temporary regression test files.
- Added workaround for the problem that the branch profiling code in the
2.6.29 kernel triggers a huge number of sparse warnings.
- Made sure that all tests print some output, even if successful.
- Improved summarizing of sparse warnings.
- Command-line option -j now works.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@716 d57e44dd-8a1f-0410-8b47-8ef2f437770f
simultaneously).
- scripts/generate-kernel-patch is now only run once instead of three
times per kernel version.
- Context imbalance warnings are no longer filtered from the output of
sparse.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@714 d57e44dd-8a1f-0410-8b47-8ef2f437770f
* use tabs not spaces
and the following recently introduced sparse warnings:
* drivers/scst/scst_lib.c:437:6: warning: symbol 'scst_queue_report_luns_changed_UA' was not declared. Should it be static?
* drivers/scst/scst_lib.c:3909:21: warning: symbol 'tm_dbg_tgt_dev' was not declared. Should it be static?
* drivers/scst/iscsi-scst/session.c:170:9: warning: cast to non-scalar
* drivers/scst/iscsi-scst/session.c:180:24: warning: cast to non-scalar
The patch below has been verified by reviewing it carefully and by verifying the output of the following command: scripts/run-regression-tests -k 2.6.28.8
Signed-off-by: Bart Van Assche <bart.vanassche@gmail.com>
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@712 d57e44dd-8a1f-0410-8b47-8ef2f437770f
I prepared it for kernel 2.6.27.7-9 which is used by openSuSE11.1-0 but it can
also be applied to 2.6.27.19-3.2.1 (with some offset/fuzz) which is going to
be deployed in SLE11.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@711 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Added command-line option -f, which enables a full check. Changed
behavior when -f has not been specified to running checkpatch, sparse
and headers_check only. This makes the script complete a lot faster.
- A summary of the errors and warnings reported by checkpatch, sparse
and make headers_check is now printed.
- Made the functions kernel_version and patchlevel more robust.
- The function duplicate_source_tree now only copies the files
administered by Subversion. Other files, e.g. .o files, are skipped.
- Moved the make flag -j3 from individual make commands to a single
assignment to the variable MAKEFLAGS, such that the make flags are
easier to modify.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@709 d57e44dd-8a1f-0410-8b47-8ef2f437770f
* use tabs not spaces
* Macros with complex values should be enclosed in parenthesis.
* printk() should include KERN_ facility level.
The patch below has been verified by reviewing it carefully and by verifying the output of the following command: scripts/run-regression-tests -k 2.6.28.8
Signed-off-by: Bart Van Assche <bart.vanassche@gmail.com>
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@707 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Now sense data are sent without additional memory allocation and copy
- Sending and receiving padding bytes reimplemented
- Cleanups
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@699 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Implemented "plug-and-play" notifications about new devices in security groups and changed size of a device both through AENs and Unit Attentions
- New command SCST_USER_DEVICE_CAPACITY_CHANGED added to scst_user interface to notify SCST core that the corresponding device has changed its capacity
- New command "resync_size" added to scst_vdisk proc interface to tell scst_vdisk to reread size of the corresponding device.
- Docs update
- Minor fixes and cleanups
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@697 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- scst_set_initial_UA() added. It sets initial Unit Attention for session, replacing default scst_sense_reset_UA
- task_mgmt_affected_cmds_done() callback added in struct scst_tgt_template. It's informs target driver that a received task management fun
ction has been completed.
- Now devices blocked only in reset TM commands. ABORTs, etc. now done without devices blocking
- scst_unregister_session_ex() removed, since there are no users of it
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@684 d57e44dd-8a1f-0410-8b47-8ef2f437770f
There are 16 characters of the port_name + 7 characters of ":" + "\0" of
sprintf = total 24 bytes.
wwn_str has only 16 bytes.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@681 d57e44dd-8a1f-0410-8b47-8ef2f437770f
we don't, somehow, inadvertantly overrun the total
number of queue entries for a command + continuation
segments.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@680 d57e44dd-8a1f-0410-8b47-8ef2f437770f
I realized yesterday that the fix missed one rare case
still. if the hiwmk_check() returned with error and no_fail override was set
then the counter would still cross 0 on free.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@677 d57e44dd-8a1f-0410-8b47-8ef2f437770f
i noticed this issue running simple iscsi connection test to a few targets.
without included patch the active_pages_total counter wraps around 0 thus it
breaks logic for releasing memory on "hi watermark" just like it is depicted
in:
Inactive/active pages 18/-25
Hi/lo watermarks [pages] 119808/0
Hi watermark releases/failures 4/0
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@676 d57e44dd-8a1f-0410-8b47-8ef2f437770f
