-
Notifications
You must be signed in to change notification settings - Fork 1.3k
/
Copy pathChangeLog.opaque.txt
221 lines (169 loc) · 9.54 KB
/
ChangeLog.opaque.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
----- * ----- * ----- * ----- * ----- * ----- * ----- * ----- * ----- * -----
Changes 2002.12.20
1. Bug fixes
1.1 When an opaque LSA is being removed from (or added to) the LSDB,
it does not mean a change in network topology. Therefore, SPF
recalculation should not be triggered in that case.
There was an assertion failure problem "assert (rn && rn->info)"
inside the function "ospf_ase_incremental_update()", because
the upper function "ospf_lsa_maxage_walker_remover()" called it
when a type-11 opaque LSA is removed due to MaxAge.
1.2 Type-9 LSA is defined to have "link-local" flooding scope.
In the Database exchange procedure with a new neighbor, a type-9
LSA was added in the database summary of a DD message, even if
the link is different from the one that have bound to.
2. Feature enhancements
2.1 Though a "wildcard" concept to handle type-9/10/11 LSAs altogether
has introduced about a year ago, it was only a symbol definition
and actual handling mechanism was not implemented. Now it works.
----- * ----- * ----- * ----- * ----- * ----- * ----- * ----- * ----- * -----
Changes 2002.7.8
1. Bug fixes
1.1 When "ospf_delete_opaque_functab()" is called, internal structure
"oipt" remain unfreed. If register/delete functab is repeated,
illegal memory access happens due to this "oipt".
1.2 In "free_opaque_info_per_id()", there was a crucial typo which
ignores a condition test.
"if (oipi->lsa != NULL);" <-- semicolon!
2. Feature enhancements
None.
----- * ----- * ----- * ----- * ----- * ----- * ----- * ----- * ----- * -----
Changes 2001.12.03
1. Bug fixes
1.1 Though a new member "oi" has added to "struct ospf_lsa" to control
flooding scope of type-9 Opaque-LSAs, the value was always NULL
because no one set it.
1.2 In the function "show_ip_ospf_database_summary()" and "show_lsa_
detail_adv_router()", VTY output for type-11 Opaque-LSAs did not
work properly.
1.3 URL for the opaque-type assignment reference has changed.
1.4 In the file "ospf_mpls_te.c", printf formats have changed to
avoid compiler warning messages; "%lu" -> "%u", "%lx" -> "%x".
Note that this hack depends on OS, compiler and their versions.
1.5 One of attached documentation "opaque_lsa.txt" has changed to
reflect the latest coding.
2. Feature enhancements
2.1 Knowing that it is an ugly hack, an "officially unallocated"
opaque-type value 0 has newly introduced as a "wildcard",
which matches to all opaque-type.
This value must not be flooded to the network, of course.
2.2 The Opaque-core module makes use of newly introduced hooks to
dispatch every LSDB change (LSA installation and deletion) to
preregistered opaque users.
Therefore, by providing appropriate callback functions as new
parameters of "ospf_register_opaque_functab()", an opaque user
can refer to every LSA instance to be installed into, or to be
deleted from, the LSDB.
----- * ----- * ----- * ----- * ----- * ----- * ----- * ----- * ----- * -----
Changes 2001.10.31
1. Bug fixes
1.1 Since each LSA has their own lifetime, they will remain in a
routing domain (being stored in LSDB of each router), until their
age naturally reach to MaxAge or explicitly being flushed by the
originated router. Therefore, if a router restarted with a short
downtime, it is possible that previously flooded self-originated
LSAs might received if the NSM status is not less than Exchange.
There were some problems in the way of handling self-originated
Opaque-LSAs if they are contained in a received LSUpd message,
but not installed to the local LSDB yet.
Regardless of some conditions to start originating Opaque-LSAs
(there should be at least one opaque-capable full-state neighbor),
the function "ospf_flood()" will be called to flood and install
this brand-new looking LSA.
As the result, when the NSM of an opaque-capable neighbor gets
full, internal state inconsistency happens; a user of Opaque-LSA
such as MPLS-TE can refer to self-originated LSAs in the local
LSDB, but cannot modify their contents...
Above problems have fixed with a policy "flush it from the whole
routing domain and keep silent until the flushing completed".
By using this sweeping technique, we can be free from confusion
caused by self-originated LSAs received via network.
1.2 The function "ospf_opaque_type_name()" contained massive ifdefs
corresponding to each "opaque-type".
These unnecessary ifdefs are removed completely.
1.3 In the function "ospf_delete_opaque_functab()", there was an
improper loop control that causes illegal memory access.
Original coding was "next = nextnode (node)".
1.4 The function "ospf_mpls_te_ism_change()" could not handle the
case when the ISM changes from Waiting to DR/BDR/Other.
So, there was a case that even if one of an ISM become
operational and MPLS-TE module has started, the corresponding
Opaque-LSA cannot be originated.
1.5 The function "ospf_opaque_lsa_reoriginate_schedule()" did not
allow to be called multiple times, simply because handling
module for the given "lsa-type & opaque-type" already exists.
But this assumption seems to be wrong.
Change the policy to allow this function to be called multiple
times and let the caller to decide what should do when the
corresponding callback function "(* functab->lsa_originator)()"
is called.
2. Feature enhancements
2.1 The global bitmap "opaque" has introduced instead of former flag
"OpaqueCapable", to store complex conditions to handle Opaque-LSAs.
2.2 The MPLS-TE module now referes to "draft-katz-yeung-ospf-traffic
-06.txt", no significant changes with 05 version, though.
----- * ----- * ----- * ----- * ----- * ----- * ----- * ----- * ----- * -----
Changes 2001.08.03
1. Bug fixes
1.1 Even if the ospfd started with opaque capability enabled, when
the ospfd receives an unknown opaque-type (unregistered by the
function "ospf_register_opaque_functab()" beforehand), the LSA
was discarded. As the result, only the opaque-LSAs that have
commonly registered by opaque-capable ospf routers can be
flooded in a routing domain.
This behavior has fixed so that arbitrary opaque-type LSAs can
be flooded among opaque-capable ospf routers.
If the ospfd has opaque-LSA capability but disabled at runtime,
received opaque-LSAs can be accepted and registered to LSDB as
is, but not be flooded to the network; those opaque LSAs will
remain in LSDB until explicitly flushed by incoming LSUpd
messages with MaxAge, or their age naturally reaches to MaxAge.
1.2 The function "ospf_register_opaque_functab()" did not check
if the entry corresponding to the given "lsa-type, opaque-type"
combination already exists or not.
This problem has fixed not to allow multiple registration.
1.3 Since type-11 (AS external) LSAs will be flooded beyond areas,
there is little relationship between "struct lsa" and "struct
area". More specifically, the pointer address "lsa->area" can
be NULL if the lsa-type is 11, thus an illegal memory access
will happen. This problem has fixed.
1.4 When self-originated opaque-LSAs are received via network and
if the corresponding opaque-type functions are not available
(they have already deleted) at that time, those LSAs were
dropped due to "unknown opaque-type" error.
After the problem 1.1 has fixed, those "self-originated" LSAs
were registered to LSDB and then flooded to the network, even
if the processing functions did not exist...
After all, this problem has fixed so that those LSAs should
explicitly be flushed from the routing domain immediately, if
the processing functions cannot find at that time.
1.5 Some typo have fixed.
--- EXAMPLE ---
static int
opaque_lsa_originate_callback (list funclist, void *lsa_type_dependent)
^^^^^
--- EXAMPLE ---
2. Feature enhancements
2.1 According to the description of rfc2328 in section 10.8, any
change in the router's optional capabilities should trigger
the option re-negotiation procedures with neighbors.
--- EXCERPT ---
If for some reason the router's optional
capabilities change, the Database Exchange procedure should be
restarted by reverting to neighbor state ExStart.
--- EXCERPT ---
For the opaque-capability changes, this feature has implemented.
More specifically, if "ospf opaque-lsa" or "no ospf opaque-lsa"
VTY command is given at runtime, all self-originated LSAs will
be flushed immediately and then all neighbor status will be
forced to ExStart by generating SeqNumberMismatch events.
2.1 When we change opaque-capability dynamically (ON -> OFF -> ON),
there was no trigger at "OFF->ON" timing to reactivate opaque
LSA handling modules (such as MPLS-TE) that have once forcibly
stopped at "ON->OFF" timing.
Now this dynamic reactivation feature has added.
2.2 The MPLS-TE module now referes to "draft-katz-yeung-ospf-traffic
-05.txt", no significant changes with 04 version, though.
----- * ----- * ----- * ----- * ----- * ----- * ----- * ----- * ----- * -----
Changes 2001.03.28
Initial release of Opaque-LSA/MPLS-TE extensions for the zebra/ospfd.