mirrors/scst
1
0
Fork 0
mirror of https://github.com/SCST-project/scst.git synced 2026-08-17 12:46:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

Web updates

Browse Source 
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@1011 d57e44dd-8a1f-0410-8b47-8ef2f437770f
This commit is contained in:
Vladislav Bolkhovitin
2009-08-04 17:50:43 +00:00
parent cef4a0ec28
commit b06a932d2c
1 changed files with 25 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
www/contributing.html
+25
View File
@@ -445,6 +445,31 @@
InfiniBand) bypassing the regular heavy weighted and CPU consuming TCP/IP data transfers path.</p>
<p>It would be good to add support for iSER in iSCSI-SCST.</p>
<A NAME="GET_CONFIGURATION"></A><h3>GET CONFIGURATION command</h3>
<p>SCSI command GET CONFIGURATION is mandatory for SCSI multimedia devices, like CD/DVD-ROMs or
recorders, see MMC standard. Currently SCST lacks support for it, which leads to problems
with some programs depending on the result of GET CONFIGURATION command execution.</p>
<p>It would be good to add support for it in the SCST core.</p>
<A NAME="Per_Device_Suspend"></A><h3>Per-device suspending</h3>
<p>Currently before doing any management operations SCST core performs so called "activities suspending", i.e.
it suspends new coming SCSI commands and wait until currently being executed ones finished. It allows to
simplify internal locking and reference counting a lot, but has a drawback that it is global, i.e. affects
all devices and SCSI commands, even ones which don't participate in the management operation. In the majority
of regular cases it works pretty well, but sometimes it can be a problem.
For instance, if a SCSI command needs a big amount of execution time (hours for some tapes operations),
the management command and all other SCSI commands will wait until it's finished. Even worse, if a user space
dev handler hangs and stops processing commands, any SCST management command will not be able to complete and fail
with timeout until the user space dev handler gets killed.</p>
<p>The global suspending should be changed to more fine-grained per-device suspending
and only for cases where it's really needed, like device unregistration. This is a very tricky task, because
all the internal SCST locking should be reimplemented.</p>
</div>
</div>
</div>