- not acknowledge when fail to register
- unregistration when no one can call functions from HBA
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@324 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Make pre-1240 cards be able to do 64 bit dma as well, making sure that
they don't cross a 16MB boundary.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@316 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Make FC_SCRATCH_ACQUIRE a macro that returns true/false
- Fixes from Stanislaw for 2.6.23 & 2.6.24. Many thanks.
- A quick hack to isp_kthread to use waitqueues instead of semaphores
to have less blockage. More rewrites later
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@314 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Do some whitespace changes. Solaris related compilation fix. Remove
XS_SET_STATE_STAT macro.
- Remove XS_SET_STATE_STATE macro.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@313 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- whitespace and other minor fixes
- Remove F/W crash dump support. We only worked it out for 2200 and 2300
cards. It's also true that QLogic would not support us even if we *did*
give them a crash dump, so the usage here is not really worth it.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@311 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Allow different platforms/environments to specificy the size of the
high and low private storage for tmd_cmd_t structures
- Add a scan scsi thread event so that we can attach devices as we
discover them
- Fix some tabs
- Make things work correctly when CONFIG_FW_LOADER is *not* defined
- Get rid of sema and use wait_interruptible
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@307 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- pass commands to SCST after assigning initiator to them
- enable lun's on proper channel
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@302 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Connection close improved to be less agressive and honor TCP TIME_WATE state
- Docs updates
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@301 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Fixed dev_cdrom and dev_modisk load failures if there are no media in the drives
- Other minor fixes and cleanups
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@298 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Optimizations and send timeout introduced in r295 temporary disabled
- Big connection shutdown times improvements. Great thanks to Alexey Kuznetsov for suggestions!
- RESERVE/RELEASE serialization now made with TST field 0 (i.e. single task set)
- Other minor fixes and improvements
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@297 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Response data send timeout implemented: now if initiator doesn't accept data for too long (7 sec), target closes connection
- Connction closing improved: now session cleared without need to wait for all outstanding commands data fully transmitted
- On session unregistration now all outstanding commands are implicitely aborted
- TM processing in SCST core made independant from other TM commands (before they were serialized)
- Few bug fixes
- Other minor improvements and cleanups
- Docs update
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@295 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Important reference counting and barriers usage cleanups
- Sense buffer made dynamic
- Other minor improvements and cleanups
- Docs updates
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@287 d57e44dd-8a1f-0410-8b47-8ef2f437770f
Clean up MSI-X implementation and make sure we snag at least *3* vectors
so that the ATIO-Q vector is snagged.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@284 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Add a missing MSI disable and provide for pre-2.6.8 kernels.
- Fix bugs- forgot to actually compile (shame on me).
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@283 d57e44dd-8a1f-0410-8b47-8ef2f437770f
I don't know why I thought this would work- forcing a re-login to the fabric
controller just barfs.
Make down count reflect jiffies elapsed rather than depend on calls to
ISP_FC_RESCAN.
Shorten deadloop time default to 10 seconds.
At least enable MSI (if not MSI-X) for 24XX cards.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@282 d57e44dd-8a1f-0410-8b47-8ef2f437770f
We can (somehow) get commands on an atio7 for which we've never had a PLOGI/PRLI.
How this happens, I don't know. Anyway, we stick a tentative entry in the PDB
and call the task thread to find the rest of the info. This includes, badly and
stupidly, searching for the N-port handle by asking the chip. Sometimes it's not
there. In that case, we terminate the command.
In restarting commands that arrive while we start up, have the watchdog
time call isp_task_thread- this gives us a bit of hysteresis.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@281 d57e44dd-8a1f-0410-8b47-8ef2f437770f
- Need to include linux/vmalloc.h. Showed up on PPC port.
- Fix complaints about using an uninitialized pointer.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@280 d57e44dd-8a1f-0410-8b47-8ef2f437770f
Make 2100/2200 and 2300 cards also maintain a target mode port database.
For the 24XX, we have to find the IID, N-port handle and channel when we
get an ATIO7. For the other cards, we already have the channel and N-port
handle and Port IID of the initiator, but have to find the S_ID.
Do the isp_find_pdb_by_{sid,loopid} functions search backwards to minimize
search time.
Don't load fw images more than once per device type!
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@279 d57e44dd-8a1f-0410-8b47-8ef2f437770f
