Documentation/scsi/ChangeLog.lpfc

2 	* Please read the associated RELEASE-NOTES file !!!
8 	* Fixed build warning for 2.6.12-rc2 kernels: mempool_alloc now
19 	* Removed FC_TRANSPORT_PATCHESxxx defines.  They're in 2.6.12-rc1.
26 	* Added PCI ID for LP10000-S.
31 	* Zero-out response sense length in lpfc_scsi_prep_cmnd to prevent
32 	  interpretation of stale sense length when the command completes
33 	  - was causing spurious 0710 messages.
55 	  - stop using volatile.  if you need special ordering use memory
57 	  - switch lpfc_sli_pcimem_bcopy to take void * arguments.
58 	  - remove typecast for constants - a U postfix marks them
60 	  - add a MAILBOX_CMD_SIZE macro, as most users of
62 	  - kill struct lpfc_scsi_dma_buf and embedded the two members
64 	  - don't call dma_sync function on allocations from
65 	    pci_pool_alloc - it's only for streaming mappings (pci_map_*)
66 	* Merged patch from Christoph Hellwig (3/19) - nlp_failMask isn't
68 	  a multi-value file which is against the sysfs guidelines).
71 	* Added #include <linux/dma-mapping.h> as required by the DMA
73 	* Merged patch from Christoph Hellwig (03/19) - fix initialization
74 	  order - scsi_add_host must happen last from scsi POV. Also some
76 	* Fixed use of TRANSPORT_PATCHES_V2 by changing to
92 	* Feedback from Christoph Hellwig (on 2/5) - In the LPFC_EVT_SCAN
102 	* Use DMA_64BIT_MASK and DMA_32BIT_MASK defines instead of
103 	  0xffffffffffffffffULL & 0xffffffffULL respectively.  Use pci
108 	  waking up the els_tmo handler un-necessarily.  The thread was
122 	  semantics - ELS completions worked for the wrong reasons.  Also
123 	  provided new log message number - had two 0326 entries.
129 	* Close humongous memory leak in lpfc_sli.c - driver was losing 13
139 	  premise is that the worker thread is signal-immune.
146 	* Closed large memory leak -- we were losing 13 iocbq structures
148 	* Changing EIO and ENOMEM to -EIO and -ENOMEM respectively.
151 	  macros putLunHigh, putLunLow. Use lpfc_put_lun() inline instead.
153 	  cpu_to_be32(), use swab16((uint16_t)lun). This is the same as
165 	* Integrated Christoph Hellwig's patch from 1/30: Use
175 	* Cleaned up references to list_head->next field in the driver.
188 	  lpfc_alloc_transport_attr - remove shim lpfc_alloc_shost_attrs -
189 	  remove shim lpfc_scsi_host_init - allocate phba mem in scsi's
190 	  hostdata readjust code so that they are no use after free's
191 	  (don't use after scsi_host_put) - make lpfc_alloc_sysfs_attr
202 	  lpfc_get_brd_no and use Linux provided IDR.
207 	* VPD changes: 1) Modify driver to use the model name and
208 	  description from the VPD data if it exists 2) Rework use of DUMP
222 	* Make sure function which processes unsolicited IOCBs on ELS ring
227 	* Implemented fast IOCB processing for FCP ring.
236 	* Remove ELS ring handling leftovers from the lpfc_sli_inter().
237 	* ELS ring (any slow ring) moved from the lpfc_sli_inter() into a
253 	* Fixed a use after free issue in lpfc_init.c.
271 	  ill-formatting multiline comments.
279 	* Removed unpleasant casting in the definition and use of
281 	* Makefile cleanup.  Use ?= operator for setting default
282 	  KERNELVERSION and BASEINCLUDE values.  Use $(PWD) consistently.
299 	  ring.
308 	  such use.
322 	* rport fix: use new fc_remote_port_rolechg() function instead of
331 	  lpfc_sli_handle_mb_event, pmb->context1 points to a waitq. Do
339 	  lpfc_disc_done() Moving all events from phba->dpc_disc to local
347 	  Moved rport delete to freenode - so rport tracks node.
350 	* Added use of wait_event to work with kthread interface.
353 	* In remote port changes: no longer nulling target->pnode when
358 	  (per the pre-remote port changes).
359 	* Merge in support for fc transport remote port use. This removes
361 	  on a per-target basis driven by the discovery engine.
369 	  list handling.  Also clean up lpfc_reset_bus_handler - routine
374 	* Part 1 of the memory allocation rework request by linux-scsi.
376 	  the scatter-gather count a module parameter, builds a linked
390 	  waiting on a semaphore and "modprobe -r lpfc" is not always
393 	* Small Makefile cleanup - Remove remnants of 2.4 vs. 2.6
399 	* Janitorial cleanup after removal of sliinit and ringinit[] ring
400 	  statistic is owned by the ring and SLI stats are in sli
408 	* Integrated patch from Christoph Hellwig <hch@lst.de>: use
414 	  and lpfc_reset_bus_handler.  Also use schedule_timeout() instead
429 	* Fix some sparse warnings -- 0 used as NULL pointer.
442 	  Only major changes is changing of some offsets from word-based to
443 	  byte-based so we can simply do void pointer arithmetic (gcc
446 	  LPFC_SLI_ABORT_IMED, aka 0 - remove dead code.
449 	  clean;make ADVANCED=1 on SMP x86, 2.6.10-rc2 on RHEL 4 Beta
459 	  command, make sure the driver does not complete post-abort
491 	* lpfc_sli_update_ring() is split in to two --
495 	* Simplified lpfc_sli_resume_iocb() and its use.
510 	* Use DID not SCSI target id as a port_id and add some missing
514 	* Fix crash: paging request at virtual address 0000000000100108 -
524 	* Fix setting of upper 32 bits for Host Group Ring Pointers if in
526 	* Use scsi_[activate|deactivate]_tcq calls provided in scsi_tcq.h.
539 	* Fixed discovery bug: rnid acc cmpl, can potentially use ndlp
556 	  events to use these routines.
595 	* Introduce and use less lethal event handler for unexpected
599 	* Added new configuration parameter, lpfc_max_luns range 1-32768,
635 	* Exported lpfc_get_hba_sym_node_name for use by FC_TRANS_VER2
658 	  rid of build warnings on later 2.6.9-rc kernels.  Build
666 	* Update Makefile to account for SLES 9 and scsi-target upstream
671 	* Discovery changes regarding setting targetp->pnode and
672 	  ndlp->nlp_Target Ensure fc_target_* routines are called properly
690 	* Save seg_cnt from dma_map_sg.  Save scatter-gather start address
693 	  change is supported by scsi-target-2.6 kernel and forward only.
708 	  to lpfc_sli_resume_iocb and lpfc_sli_issue_iocb - singificant
709 	  cleanup of those two functions - the unused SLI_IOCB_USE_TXQ is
710 	  gone - lpfc_sli_issue_iocb_wait loses its flags argument
712 	* Fix in lpfc_sli.c: we can not store a 5 bit value in a 4-bit
723 	  i.e. use NULL instead of 0.
725 	  here, the most notable is that the phba->slim2p lpfc_dmabuf goes
734 	  fast-pathing the non-error case.  Also removes tons of dead
735 	  members and defines from lpfc_scsi.h - e.g. lpfc_target is down
764 	  progress with linux-scsi list.  These are name changes and
786 	  only valid for ELS ring.
790 	* Set host->unique_id in lpfc_fcp.c.
799 	* Use mod_timer instead of add_timer for fdmi in lpfc_ct.c.
807 	  lpfc_disc_done/lpfc_do_dpc cleanup - lpfc_disc_done can return
808 	  void - move lpfc_do_dpc and lpfc_disc_done to lpfc_hbadisc.c -
816 	* Removed unused prototypes from lpfc_crtn.h -
818 	* Removed some unused prototypes from lpfc_crtn.h -
825 	  lpfc_crtn.h - fixes build warnings.
827 	* Removed pre-2.6.5 MODULE_VERSION macro from lpfc_compat.h.
836 	* Make nodev-tmo default to 20 seconds.
841 	* Move suspend/resume in lpfc_nlp_list under appropriate case -
843 	* Fix some corner cases for PLOGI receive - simplify error case
845 	* Bug fix for ppc64 EEH MMIO panic - always do readl after
856 	* Fixed link down->up transitions when linkdown tmo expires. Fix
864 	  addressing.  Use 16 bits of LUN address so that flat
874 	* Removed range checking. phba->config[] array elements are now
882 	* Removed scsi_cmnd->timeout_per_command cancelation. SCSI midlayer
883 	  now times out all commands - FW is instructed to not timeout.
892 	* Fixed a use after free case (panic on 2.6.8.1 with
902 	  sdev->queue_depth.  The driver informs the midlayer of its
911 	* Removed per HBA driver lock.  Driver now uses the host->host_lock
915 	  "[PATCH] use scsi host private data in ->proc_info.
939 	* Use msecs_to_jiffies() where applicable.
940 	* Only use queue depth attribute only after SLI HBA setup was
946 	* Updated timer logic: Set timer data after init_timer use
953 	* Changed LPFC_CFG_DFT_HBA_Q_DEPTH -> LPFC_CFG_HBA_Q_DEPTH.
969 	* Removed LUN-tracking in driver.  Removed lpfc_lun struct and
984 	  lpfc_disc_retry_rptlun and their use.
1018 	  un-necessary checks on config[LPFC_CFG_FCP_CLASS].
1056 	* Arranging lpfc_scsiport.c to follow style of use after
1081 	* Change lpfc_printf_log to a #define. Also include phba->brd_no
1100 	* Changing phba->config to kmalloc lpfc_icfgparam and not
1101 	  *phba->config. This is manifesting itself as a panic in
1118 	* Modified lpfc_slave_alloc to only set the scsi_device->hostdata
1131 	* On error in lpfc_pci_probe_one() return -1 and not 1.
1140 	* Removed all (struct list_head *) casts from clkData-related list
1148 	* Rewrite use of FC transport attributes.
1165 	* Changed pHba->phba, pCommand->pcmd
1166 	* Changed plogi_ndlp -> plogindlp, pos_tmp->postmp, pRsp->prsp,
1167 	  pCmd->pcmd
1168 	* Changed pText -> ptext
1169 	* Changed p_tmp_buff -> ptmpbuff
1170 	* Changed pBufList -> pbuflist, pRsp -> prsp, pCmd -> pcmd
1171 	* Changed *pos_tmp -> *postmp, *p_mbuf -> *pmbuf
1195 	* Fixed a large number of overly-long lines.
1227 	* Including dma-mapping.h as one of the include headers.  Also
1265 	* Eliminate one-line function lpfc_find_target()
1268 	  scsi_device->hostdata, which is setup in slave_alloc.
1272 	  lpfc_scsi_buf is hung off of linux's scsi_cmnd->host_scribble
1294 	* Remove all one-line wrappers from lpfc_mem.
1298 	* Use device attributes instead of driver attributes where appropriate
1305 	* Fix the number of cmd / rsp ring entries in lpfc_fcp.c to match
1307 	* Remove phba->iflag reference.
1312 	* Make lpfc_info() use lpfc_get_hba_model_desc() instead of
1329 	* Change lpfc-param names to lpfc_param.
1330 	* Add use of MODULE_VERSION macro for 2.6 kernels.
1358 	  mbox response as a byte-count
1362 	* Completing MODULE_PARM -> module_param changes
1367 	* Fix use of lun-q-depth config param
1371 	* Change lpfc_param to lpfc-param
1387 	  modprobe are killed via ctrl-c
1389 	* Wrap use of log message 311 in macro
1391 	* Fix use-after-free of IOCB in lpfc_sli_process_sol_iocb which
1393 	* Cleanup use of several gotos not used for error exit.
1394 	* Replace memcpy_toio() and memcpy_toio() with endian-dependent
1397 	  time instead of as a byte-stream.
1402 	  return a zero-valued host pointer.  A ctrl-C on 2.6 kernels
1405 	  lpfc_sli_hba_down - it was in the wrong place and the driver
1416 	  - mostly replacing groups of 8 spaces with hard tabs and keeping
1425 	* Clean up formatting in lpfc_sli.c and lpfc_sysfs.c - mostly
1432 	* Fix build warnings on 2.6 kernels - remove no longer used calls
1465 	* Replace mlist->dma.list = dmp->dma.list; to mlist = dmp.
1505 	* Do not use DMABUf_t list to store ndlp context
1509 	* Do not use DMABUf_t list to hold rpi context
1510 	* Replace elx_cfg_* names with lpfc_cfg-*
1511 	* Moved FCP activity to ring 0.  Moved ELS/CT activity to ring 2.
1522 	* Use safe list iterator for ndlp list
1535 	* Use Linux list macros for DMABUF_t
1543 	* Added code to use list macro for ELXSCSITARGET_t.
1556 	  fc_nlpbind_start and _end with fc_nlpbind_list and use
1570 	  3-digit HBAs.  Also changed can_queue so midlayer will only
1571 	  send (HBA_Q_DEPTH - 10) cmds.
1573 	  unused sense-related fields in lun structure.
1596 	  doesn't use them anymore
1615 	* Removed some commented-out code
1618 	* Removed two PCI Sync defines.  Removed includes - not
1633 	* Use Linux swap macros to replace ELX swapping macros
1645 	  lpfc_ip_timeout_handler() now use system timer.
1646 	* lpfc_fdmi_tmo() and lpfc_qthrottle_up() now use system
1652 	* lpfc_disc_timeout() and lpfc_establish_link_tmo() now use
1655 	* Remove use of PAN (pseudo adapter number).
1659 	* Fix no-unlock-before return in lpfc_els_retry_delay which was
1663 	  elx_stop_timer().  All timeout routines now use these common
1687 	* Rewrite of Makefile. Fixes breakages with make -j4 during
1747 	  lpfc_linkdown_timeout to use the system timer instead of
1765 	  OS-provided #defines. Also added linux/pci.h to *.c files.
1772 	* Change use of 2.4 SCSI typedef Scsi_Host_Template to	struct
1774 	* Change use of 2.4 SCSI typedefs (Scsi_Device, Scsi_Cmnd,
1793 	* Use lpfc_is_LC_HBA() macro that tests pcidev->device directly
1800 	* Use dev_warn instead of printk for 2.6 kernels
1805 	* Removed memory pool for buf/bpl buffers and use kmalloc/kfree
1807 	* Move PPC check for DMA address 0 in scatter-gather list, into
1809 	* Always use pci_unmap_single() instead of pci_unmap_page()
1820 	* Change elx_printf_log to use vsprintf.
1824 	* Removed memory pool for nlp/bind buffers and use kmalloc/kfree
1826 	* Removed memory pool for iocb buffers and use kmalloc/kfree
1828 	* Removed memory pool for mailbox buffers and use kmalloc/kfree
1841 	* Use kmalloc/kfree for ELX_SCSI_BUF_t
1842 	* Use linux pci_pools for SCSI_DMA_EXT
1843 	* Use linux pci_pools for BPLs.
1848 	* Use strcpy, strncmp, isdigit, strlen instead of abstractions