1 2004-02-17 multipath-017
2 * remove the restrictive -f flag.
3 Introduce a more generic "-m iopolicy" one.
4 * remove useless "int with_sysfs" in env struct
5 2004-02-04 multipath-016
6 * add a GROUP_BY_SERIAL flag. This should be useful for
7 controlers that activate they spare paths on simple IO
8 submition with a penalty. The StorageWorks HW defaults to
9 this mode, even if the MULTIBUS mode is OK.
10 * remove unused sg_err.c
11 * big restructuring : split devinfo.c from main.c. Export :
12 * void basename (char *, char *);
13 * int get_serial (int, char *);
14 * int get_lun_strings (char *, char *, char *, char *);
15 * int get_evpd_wwid(char *, char *);
16 * long get_disk_size (char *);
17 * stop passing struct env as param
18 * add devmap_name proggy for udev to name devmaps as per their
19 internal DM name and not only by their sysfs enum name (dm-*)
20 The corresponding udev.rules line is :
21 KERNEL="dm-[0-9]*", PROGRAM="/sbin/devmap_name %M %m", NAME="%k", SYMLINK="%c"
22 * remove make_dm_node fn & call. Rely on udev for this.
23 * don't rely on the linux symlink in the udev/klibc dir since
24 udev build doesn't use it anymore. This corrects build breakage
25 2004-01-19 multipath-013
26 * update the DM target synthax to the 2.6.0-udm5 style
27 2003-12-29 multipath-012
28 * check hotplug event refers to a block device; if not exit early
30 * add the uninstall target in Makefile
31 2003-12-22 multipath-010
32 * tweak the install target in Makefile
33 * stop passing fds as argument : this change enable a strict
34 segregation of ugly 2.4 code
35 * sysfs version of get_lun_strings()
36 * be careful about the return of get_unique_id() since errors
37 formerly caught up by if(open()) in the caller fn are now returned
39 * send get_serial() in unused.c
40 * introduce dm-simplecmd for RESUME & SUSPEND requests
41 * split add_map() in setup_map() & dm-addmap()
42 * setup_map() correctly submits "SUSPEND-RELOAD-RESUME or CREATE"
43 sequences instead of the bogus "RELOAD or CREATE"
44 * don't print .sg_dev if equal to .dev (2.6) in print_path()
45 * since the kernel code handles defective paths, remove all
46 code to cope with them :
47 * move do_tur() to unused.c
48 * remove .state from path struct
49 * remove .state settings & conditionals
50 * add a cmdline switch to force maps to failover mode,
51 ie 1 path per priority group
52 * add default policies to the whitelist array (spread io ==
53 MULTIBUS / io forced to 1 path == FAILOVER)
54 * move get_disk_size() call out of add_map() to coalesce()
55 * comment tricky coalesce() fn
56 * bogus unsused.c file renamed to unused.c
57 2003-12-20 multipath-010
58 * big ChangeLog update
59 * start to give a little control over target params :
60 introduce cmdline arg -i to control polling interval
61 * cope with hotplug-style calling convention :
62 ie "multipath scsi $DEVPATH" ... to avoid messing with
63 online maps not concerned by an event
64 * example hotplug agent to drop in /etc/hotplug.d/scsi
65 * revert the run & resched patch : unless someone proves me
66 wrong, this was overdesigned
67 * move commented out functions in unused.c
68 * update multipath target params to "udm[23] style"
69 * mp target now supports nr_path == 1, so do we
70 * add gratuitous free()
71 * push version forward
72 2003-12-15 multipath-009
73 * Make the HW-specific get_unique_id switch pretty
74 * Prepare to field-test by whitelisting all known fibre array,
75 try to fetch WWID from the standard EVPD 0x83 off 8 for everyone
76 * configure the multipath target with round-robin path selector and
77 conservative default for a start (udm1 style) :
78 yes it makes this release the firstreally useful one.
79 * temporarily disable map creation for single path device
80 due to current restrictive defaults in the kernel target.
81 Sistina should work it out.
82 * correct the strncmp logic in blacklist function.
83 * update the Makefiles to autodetect libgcc.a & gcc includes
84 "ulibc-style". Factorisation of udevdirs & others niceties
85 * drop a hint about absent /dev/sd? on failed open()
86 * implement a reschedule flag in /var/run.
87 Last thing the prog do before exit is check if a call to multipath
88 was done (but canceled by /var/run/multipath.run check) during its
89 execution. If so restart themain loop.
90 * implement a blacklist of sysfs bdev to not bother with for now
91 (hd,md, dm, sr, scd, ram, raw).
92 This avoid sending SG_IO to unappropiate devices.
93 * Adds a /var/run/multipath.run handling to avoid simultaneous runs.
94 * Remove a commented-out "printf"
95 * drop a libdevmapper copy in extras/multipath;
96 maybe discussions w/Sistina folks will bring a better solution in the future.
97 * drop a putchar usage in libdevmapper to compile cleanly with klibc
98 * drop another such usage of my own in main.c
99 * massage the Makefile to compile libdevmapper against klibc
100 * use "ld" to produce the binary rather than "gcc -static"
101 * stop being stupid w/ uneeded major, minor & dev in main.c:dm_mk_node()
102 * reverse to creating striped target for now because the multipath target
103 is more hairy than expected initialy
104 * push the version code to 009 to be in synch w/ udev
105 2003-11-27 multipath-007
106 * removes sg_err.[ch] deps
107 * makes sure the core code play nice with klibc
108 * port the sysfs calls to dlist helpers
109 * links against udev's sysfs (need libsysfs.a & dlist.a)
110 * finally define DM_TARGET as "multipath" as Joe posted the code today (not tested yet)
111 * push version forward (do you want it in sync with udev version?)
113 * merged in udev-006 tree
114 2003-09-18 Christophe Varoqui <christophe.varoqui@free.fr>
115 * multipath 0.0.1 released.