25
|
1 |
(* Author: Andrew Boyton, 2012
|
|
2 |
Maintainers: Gerwin Klein <kleing at cse.unsw.edu.au>
|
|
3 |
Rafal Kolanski <rafal.kolanski at nicta.com.au>
|
|
4 |
*)
|
|
5 |
|
|
6 |
header "Instantiating capDL as a separation algebra."
|
|
7 |
|
|
8 |
theory Abstract_Separation_D
|
|
9 |
imports "../../Sep_Tactics" Types_D "../../Map_Extra"
|
|
10 |
begin
|
|
11 |
|
|
12 |
(**************************************
|
|
13 |
* Start of lemmas to move elsewhere. *
|
|
14 |
**************************************)
|
|
15 |
|
|
16 |
lemma inter_empty_not_both:
|
|
17 |
"\<lbrakk>x \<in> A; A \<inter> B = {}\<rbrakk> \<Longrightarrow> x \<notin> B"
|
|
18 |
by fastforce
|
|
19 |
|
|
20 |
lemma union_intersection:
|
|
21 |
"A \<inter> (A \<union> B) = A"
|
|
22 |
"B \<inter> (A \<union> B) = B"
|
|
23 |
"(A \<union> B) \<inter> A = A"
|
|
24 |
"(A \<union> B) \<inter> B = B"
|
|
25 |
by fastforce+
|
|
26 |
|
|
27 |
lemma union_intersection1: "A \<inter> (A \<union> B) = A"
|
|
28 |
by (rule inf_sup_absorb)
|
|
29 |
lemma union_intersection2: "B \<inter> (A \<union> B) = B"
|
|
30 |
by fastforce
|
|
31 |
|
|
32 |
(* This lemma is strictly weaker than restrict_map_disj. *)
|
|
33 |
lemma restrict_map_disj':
|
|
34 |
"S \<inter> T = {} \<Longrightarrow> h |` S \<bottom> h' |` T"
|
|
35 |
by (auto simp: map_disj_def restrict_map_def dom_def)
|
|
36 |
|
|
37 |
lemma map_add_restrict_comm:
|
|
38 |
"S \<inter> T = {} \<Longrightarrow> h |` S ++ h' |` T = h' |` T ++ h |` S"
|
|
39 |
apply (drule restrict_map_disj')
|
|
40 |
apply (erule map_add_com)
|
|
41 |
done
|
|
42 |
|
|
43 |
(************************************
|
|
44 |
* End of lemmas to move elsewhere. *
|
|
45 |
************************************)
|
|
46 |
|
|
47 |
|
|
48 |
|
|
49 |
(* The state for separation logic has:
|
|
50 |
* The memory heap.
|
|
51 |
* A function for which objects own which fields.
|
|
52 |
In capDL, we say that an object either owns all of its fields, or none of them.
|
|
53 |
These are both taken from the cdl_state.
|
|
54 |
*)
|
|
55 |
|
|
56 |
datatype sep_state = SepState cdl_heap cdl_ghost_state
|
|
57 |
|
|
58 |
(* Functions to get the heap and the ghost_state from the sep_state. *)
|
|
59 |
primrec sep_heap :: "sep_state \<Rightarrow> cdl_heap"
|
|
60 |
where "sep_heap (SepState h gs) = h"
|
|
61 |
|
|
62 |
primrec sep_ghost_state :: "sep_state \<Rightarrow> cdl_ghost_state"
|
|
63 |
where "sep_ghost_state (SepState h gs) = gs"
|
|
64 |
|
|
65 |
definition
|
|
66 |
the_set :: "'a option set \<Rightarrow> 'a set"
|
|
67 |
where
|
|
68 |
"the_set xs = {x. Some x \<in> xs}"
|
|
69 |
|
|
70 |
lemma the_set_union [simp]:
|
|
71 |
"the_set (A \<union> B) = the_set A \<union> the_set B"
|
|
72 |
by (fastforce simp: the_set_def)
|
|
73 |
|
|
74 |
lemma the_set_inter [simp]:
|
|
75 |
"the_set (A \<inter> B) = the_set A \<inter> the_set B"
|
|
76 |
by (fastforce simp: the_set_def)
|
|
77 |
|
|
78 |
lemma the_set_inter_empty:
|
|
79 |
"A \<inter> B = {} \<Longrightarrow> the_set A \<inter> the_set B = {}"
|
|
80 |
by (fastforce simp: the_set_def)
|
|
81 |
|
|
82 |
|
|
83 |
(* As the capDL operations mostly take the state (rather than the heap)
|
|
84 |
* we need to redefine some of them again to take just the heap.
|
|
85 |
*)
|
|
86 |
definition
|
|
87 |
slots_of_heap :: "cdl_heap \<Rightarrow> cdl_object_id \<Rightarrow> cdl_cap_map"
|
|
88 |
where
|
|
89 |
"slots_of_heap h \<equiv> \<lambda>obj_id.
|
|
90 |
case h obj_id of
|
|
91 |
None \<Rightarrow> empty
|
|
92 |
| Some obj \<Rightarrow> object_slots obj"
|
|
93 |
|
|
94 |
(* Adds new caps to an object. It won't overwrite on a collision. *)
|
|
95 |
definition
|
|
96 |
add_to_slots :: "cdl_cap_map \<Rightarrow> cdl_object \<Rightarrow> cdl_object"
|
|
97 |
where
|
|
98 |
"add_to_slots new_val obj \<equiv> update_slots (new_val ++ (object_slots obj)) obj"
|
|
99 |
|
|
100 |
lemma add_to_slots_assoc:
|
|
101 |
"add_to_slots x (add_to_slots (y ++ z) obj) =
|
|
102 |
add_to_slots (x ++ y) (add_to_slots z obj)"
|
|
103 |
apply (clarsimp simp: add_to_slots_def update_slots_def object_slots_def)
|
|
104 |
apply (fastforce simp: cdl_tcb.splits cdl_cnode.splits
|
|
105 |
split: cdl_object.splits)
|
|
106 |
done
|
|
107 |
|
|
108 |
(* Lemmas about add_to_slots, update_slots and object_slots. *)
|
|
109 |
lemma add_to_slots_twice [simp]:
|
|
110 |
"add_to_slots x (add_to_slots y a) = add_to_slots (x ++ y) a"
|
|
111 |
by (fastforce simp: add_to_slots_def update_slots_def object_slots_def
|
|
112 |
split: cdl_object.splits)
|
|
113 |
|
|
114 |
lemma slots_of_heap_empty [simp]: "slots_of_heap empty object_id = empty"
|
|
115 |
by (simp add: slots_of_heap_def)
|
|
116 |
|
|
117 |
lemma slots_of_heap_empty2 [simp]:
|
|
118 |
"h obj_id = None \<Longrightarrow> slots_of_heap h obj_id = empty"
|
|
119 |
by (simp add: slots_of_heap_def)
|
|
120 |
|
|
121 |
lemma update_slots_add_to_slots_empty [simp]:
|
|
122 |
"update_slots empty (add_to_slots new obj) = update_slots empty obj"
|
|
123 |
by (clarsimp simp: update_slots_def add_to_slots_def split:cdl_object.splits)
|
|
124 |
|
|
125 |
lemma update_object_slots_id [simp]: "update_slots (object_slots a) a = a"
|
|
126 |
by (clarsimp simp: update_slots_def object_slots_def
|
|
127 |
split: cdl_object.splits)
|
|
128 |
|
|
129 |
lemma update_slots_of_heap_id [simp]:
|
|
130 |
"h obj_id = Some obj \<Longrightarrow> update_slots (slots_of_heap h obj_id) obj = obj"
|
|
131 |
by (clarsimp simp: update_slots_def slots_of_heap_def object_slots_def
|
|
132 |
split: cdl_object.splits)
|
|
133 |
|
|
134 |
lemma add_to_slots_empty [simp]: "add_to_slots empty h = h"
|
|
135 |
by (simp add: add_to_slots_def)
|
|
136 |
|
|
137 |
lemma update_slots_eq:
|
|
138 |
"update_slots a o1 = update_slots a o2 \<Longrightarrow> update_slots b o1 = update_slots b o2"
|
|
139 |
by (fastforce simp: update_slots_def cdl_tcb.splits cdl_cnode.splits
|
|
140 |
split: cdl_object.splits)
|
|
141 |
|
|
142 |
|
|
143 |
|
|
144 |
(* If there are not two conflicting objects at a position in two states.
|
|
145 |
* Objects conflict if their types are different or their ghost_states collide.
|
|
146 |
*)
|
|
147 |
definition
|
|
148 |
not_conflicting_objects :: "sep_state \<Rightarrow> sep_state \<Rightarrow> cdl_object_id \<Rightarrow> bool"
|
|
149 |
where
|
|
150 |
"not_conflicting_objects state_a state_b = (\<lambda>obj_id.
|
|
151 |
let heap_a = sep_heap state_a;
|
|
152 |
heap_b = sep_heap state_b;
|
|
153 |
gs_a = sep_ghost_state state_a;
|
|
154 |
gs_b = sep_ghost_state state_b
|
|
155 |
in case (heap_a obj_id, heap_b obj_id) of
|
|
156 |
(Some o1, Some o2) \<Rightarrow> object_type o1 = object_type o2 \<and> gs_a obj_id \<inter> gs_b obj_id = {}
|
|
157 |
| _ \<Rightarrow> True)"
|
|
158 |
|
|
159 |
|
|
160 |
(* "Cleans" slots to conform with the compontents. *)
|
|
161 |
definition
|
|
162 |
clean_slots :: "cdl_cap_map \<Rightarrow> cdl_components \<Rightarrow> cdl_cap_map"
|
|
163 |
where
|
|
164 |
"clean_slots slots cmp \<equiv> slots |` the_set cmp"
|
|
165 |
|
|
166 |
(* Sets the fields of an object to a "clean" state.
|
|
167 |
Because a frame's size is part of it's type, we don't reset it. *)
|
|
168 |
definition
|
|
169 |
object_clean_fields :: "cdl_object \<Rightarrow> cdl_components \<Rightarrow> cdl_object"
|
|
170 |
where
|
|
171 |
"object_clean_fields obj cmp \<equiv> if None \<in> cmp then obj else case obj of
|
|
172 |
Tcb x \<Rightarrow> Tcb (x\<lparr>cdl_tcb_fault_endpoint := undefined\<rparr>)
|
|
173 |
| CNode x \<Rightarrow> CNode (x\<lparr>cdl_cnode_size_bits := undefined \<rparr>)
|
|
174 |
| _ \<Rightarrow> obj"
|
|
175 |
|
|
176 |
(* Sets the slots of an object to a "clean" state. *)
|
|
177 |
definition
|
|
178 |
object_clean_slots :: "cdl_object \<Rightarrow> cdl_components \<Rightarrow> cdl_object"
|
|
179 |
where
|
|
180 |
"object_clean_slots obj cmp \<equiv> update_slots (clean_slots (object_slots obj) cmp) obj"
|
|
181 |
|
|
182 |
(* Sets an object to a "clean" state. *)
|
|
183 |
definition
|
|
184 |
object_clean :: "cdl_object \<Rightarrow> cdl_components \<Rightarrow> cdl_object"
|
|
185 |
where
|
|
186 |
"object_clean obj gs \<equiv> object_clean_slots (object_clean_fields obj gs) gs"
|
|
187 |
|
|
188 |
(* Overrides the left object with the attributes of the right, as specified by the ghost state.
|
|
189 |
If the components for an object are empty, then this object is treated as empty, and thus ignored.
|
|
190 |
*)
|
|
191 |
definition
|
|
192 |
object_add :: "cdl_object \<Rightarrow> cdl_object \<Rightarrow> cdl_components \<Rightarrow> cdl_components \<Rightarrow> cdl_object"
|
|
193 |
where
|
|
194 |
"object_add obj_a obj_b cmps_a cmps_b \<equiv>
|
|
195 |
let clean_obj_a = object_clean obj_a cmps_a;
|
|
196 |
clean_obj_b = object_clean obj_b cmps_b
|
|
197 |
in if (cmps_a = {})
|
|
198 |
then clean_obj_b
|
|
199 |
else if (cmps_b = {})
|
|
200 |
then clean_obj_a
|
|
201 |
else if (None \<in> cmps_b)
|
|
202 |
then (update_slots (object_slots clean_obj_a ++ object_slots clean_obj_b) clean_obj_b)
|
|
203 |
else (update_slots (object_slots clean_obj_a ++ object_slots clean_obj_b) clean_obj_a)"
|
|
204 |
|
|
205 |
(* Heaps are added by adding their repsective objects.
|
|
206 |
* The ghost state tells us which object's fields should be taken.
|
|
207 |
* Adding objects of the same type adds their caps
|
|
208 |
* (overwrites the left with the right).
|
|
209 |
*)
|
|
210 |
definition
|
|
211 |
cdl_heap_add :: "sep_state \<Rightarrow> sep_state \<Rightarrow> cdl_heap"
|
|
212 |
where
|
|
213 |
"cdl_heap_add state_a state_b \<equiv> \<lambda>obj_id.
|
|
214 |
let heap_a = sep_heap state_a;
|
|
215 |
heap_b = sep_heap state_b;
|
|
216 |
gs_a = sep_ghost_state state_a;
|
|
217 |
gs_b = sep_ghost_state state_b
|
|
218 |
in case heap_b obj_id of
|
|
219 |
None \<Rightarrow> heap_a obj_id
|
|
220 |
| Some obj_b \<Rightarrow> case heap_a obj_id of
|
|
221 |
None \<Rightarrow> heap_b obj_id
|
|
222 |
| Some obj_a \<Rightarrow> Some (object_add obj_a obj_b (gs_a obj_id) (gs_b obj_id))"
|
|
223 |
|
|
224 |
(* Heaps are added by adding their repsective objects.
|
|
225 |
* The ghost state tells us which object's fields should be taken.
|
|
226 |
* Adding objects of the same type adds their caps
|
|
227 |
* (overwrites the left with the right).
|
|
228 |
*)
|
|
229 |
definition
|
|
230 |
cdl_ghost_state_add :: "sep_state \<Rightarrow> sep_state \<Rightarrow> cdl_ghost_state"
|
|
231 |
where
|
|
232 |
"cdl_ghost_state_add state_a state_b \<equiv> \<lambda>obj_id.
|
|
233 |
let heap_a = sep_heap state_a;
|
|
234 |
heap_b = sep_heap state_b;
|
|
235 |
gs_a = sep_ghost_state state_a;
|
|
236 |
gs_b = sep_ghost_state state_b
|
|
237 |
in if heap_a obj_id = None \<and> heap_b obj_id \<noteq> None then gs_b obj_id
|
|
238 |
else if heap_b obj_id = None \<and> heap_a obj_id \<noteq> None then gs_a obj_id
|
|
239 |
else gs_a obj_id \<union> gs_b obj_id"
|
|
240 |
|
|
241 |
|
|
242 |
(* Adding states adds their heaps,
|
|
243 |
* and each objects owns whichever fields it owned in either heap.
|
|
244 |
*)
|
|
245 |
definition
|
|
246 |
sep_state_add :: "sep_state \<Rightarrow> sep_state \<Rightarrow> sep_state"
|
|
247 |
where
|
|
248 |
"sep_state_add state_a state_b \<equiv>
|
|
249 |
let
|
|
250 |
heap_a = sep_heap state_a;
|
|
251 |
heap_b = sep_heap state_b;
|
|
252 |
gs_a = sep_ghost_state state_a;
|
|
253 |
gs_b = sep_ghost_state state_b
|
|
254 |
in
|
|
255 |
SepState (cdl_heap_add state_a state_b) (cdl_ghost_state_add state_a state_b)"
|
|
256 |
|
|
257 |
|
|
258 |
(* Heaps are disjoint if for all of their objects:
|
|
259 |
* the caps of their respective objects are disjoint,
|
|
260 |
* their respective objects don't conflict,
|
|
261 |
* they don't both own any of the same fields.
|
|
262 |
*)
|
|
263 |
definition
|
|
264 |
sep_state_disj :: "sep_state \<Rightarrow> sep_state \<Rightarrow> bool"
|
|
265 |
where
|
|
266 |
"sep_state_disj state_a state_b \<equiv>
|
|
267 |
let
|
|
268 |
heap_a = sep_heap state_a;
|
|
269 |
heap_b = sep_heap state_b;
|
|
270 |
gs_a = sep_ghost_state state_a;
|
|
271 |
gs_b = sep_ghost_state state_b
|
|
272 |
in
|
|
273 |
\<forall>obj_id. not_conflicting_objects state_a state_b obj_id"
|
|
274 |
|
|
275 |
lemma not_conflicting_objects_comm:
|
|
276 |
"not_conflicting_objects h1 h2 obj = not_conflicting_objects h2 h1 obj"
|
|
277 |
apply (clarsimp simp: not_conflicting_objects_def split:option.splits)
|
|
278 |
apply (fastforce simp: update_slots_def cdl_tcb.splits cdl_cnode.splits
|
|
279 |
split: cdl_object.splits)
|
|
280 |
done
|
|
281 |
|
|
282 |
lemma object_clean_comm:
|
|
283 |
"\<lbrakk>object_type obj_a = object_type obj_b;
|
|
284 |
object_slots obj_a ++ object_slots obj_b = object_slots obj_b ++ object_slots obj_a; None \<notin> cmp\<rbrakk>
|
|
285 |
\<Longrightarrow> object_clean (add_to_slots (object_slots obj_a) obj_b) cmp =
|
|
286 |
object_clean (add_to_slots (object_slots obj_b) obj_a) cmp"
|
|
287 |
apply (clarsimp simp: object_type_def split: cdl_object.splits)
|
|
288 |
apply (clarsimp simp: object_clean_def object_clean_slots_def object_clean_fields_def
|
|
289 |
add_to_slots_def object_slots_def update_slots_def
|
|
290 |
cdl_tcb.splits cdl_cnode.splits
|
|
291 |
split: cdl_object.splits)+
|
|
292 |
done
|
|
293 |
|
|
294 |
lemma add_to_slots_object_slots:
|
|
295 |
"object_type y = object_type z
|
|
296 |
\<Longrightarrow> add_to_slots (object_slots (add_to_slots (x) y)) z =
|
|
297 |
add_to_slots (x ++ object_slots y) z"
|
|
298 |
apply (clarsimp simp: add_to_slots_def update_slots_def object_slots_def)
|
|
299 |
apply (fastforce simp: object_type_def cdl_tcb.splits cdl_cnode.splits
|
|
300 |
split: cdl_object.splits)
|
|
301 |
done
|
|
302 |
|
|
303 |
lemma not_conflicting_objects_empty [simp]:
|
|
304 |
"not_conflicting_objects s (SepState empty (\<lambda>obj_id. {})) obj_id"
|
|
305 |
by (clarsimp simp: not_conflicting_objects_def split:option.splits)
|
|
306 |
|
|
307 |
lemma empty_not_conflicting_objects [simp]:
|
|
308 |
"not_conflicting_objects (SepState empty (\<lambda>obj_id. {})) s obj_id"
|
|
309 |
by (clarsimp simp: not_conflicting_objects_def split:option.splits)
|
|
310 |
|
|
311 |
lemma not_conflicting_objects_empty_object [elim!]:
|
|
312 |
"(sep_heap x) obj_id = None \<Longrightarrow> not_conflicting_objects x y obj_id"
|
|
313 |
by (clarsimp simp: not_conflicting_objects_def)
|
|
314 |
|
|
315 |
lemma empty_object_not_conflicting_objects [elim!]:
|
|
316 |
"(sep_heap y) obj_id = None \<Longrightarrow> not_conflicting_objects x y obj_id"
|
|
317 |
apply (drule not_conflicting_objects_empty_object [where y=x])
|
|
318 |
apply (clarsimp simp: not_conflicting_objects_comm)
|
|
319 |
done
|
|
320 |
|
|
321 |
lemma cdl_heap_add_empty [simp]:
|
|
322 |
"cdl_heap_add (SepState h gs) (SepState empty (\<lambda>obj_id. {})) = h"
|
|
323 |
by (simp add: cdl_heap_add_def)
|
|
324 |
|
|
325 |
lemma empty_cdl_heap_add [simp]:
|
|
326 |
"cdl_heap_add (SepState empty (\<lambda>obj_id. {})) (SepState h gs)= h"
|
|
327 |
apply (simp add: cdl_heap_add_def)
|
|
328 |
apply (rule ext)
|
|
329 |
apply (clarsimp split: option.splits)
|
|
330 |
done
|
|
331 |
|
|
332 |
lemma map_add_result_empty1: "a ++ b = empty \<Longrightarrow> a = empty"
|
|
333 |
apply (subgoal_tac "dom (a++b) = {}")
|
|
334 |
apply (subgoal_tac "dom (a) = {}")
|
|
335 |
apply clarsimp
|
|
336 |
apply (unfold dom_map_add)[1]
|
|
337 |
apply clarsimp
|
|
338 |
apply clarsimp
|
|
339 |
done
|
|
340 |
|
|
341 |
lemma map_add_result_empty2: "a ++ b = empty \<Longrightarrow> b = empty"
|
|
342 |
apply (subgoal_tac "dom (a++b) = {}")
|
|
343 |
apply (subgoal_tac "dom (a) = {}")
|
|
344 |
apply clarsimp
|
|
345 |
apply (unfold dom_map_add)[1]
|
|
346 |
apply clarsimp
|
|
347 |
apply clarsimp
|
|
348 |
done
|
|
349 |
|
|
350 |
lemma map_add_emptyE [elim!]: "\<lbrakk>a ++ b = empty; \<lbrakk>a = empty; b = empty\<rbrakk> \<Longrightarrow> R\<rbrakk> \<Longrightarrow> R"
|
|
351 |
apply (frule map_add_result_empty1)
|
|
352 |
apply (frule map_add_result_empty2)
|
|
353 |
apply clarsimp
|
|
354 |
done
|
|
355 |
|
|
356 |
lemma clean_slots_empty [simp]:
|
|
357 |
"clean_slots empty cmp = empty"
|
|
358 |
by (clarsimp simp: clean_slots_def)
|
|
359 |
|
|
360 |
lemma object_type_update_slots [simp]:
|
|
361 |
"object_type (update_slots slots x) = object_type x"
|
|
362 |
by (clarsimp simp: object_type_def update_slots_def split: cdl_object.splits)
|
|
363 |
|
|
364 |
lemma object_type_object_clean_slots [simp]:
|
|
365 |
"object_type (object_clean_slots x cmp) = object_type x"
|
|
366 |
by (clarsimp simp: object_clean_slots_def)
|
|
367 |
|
|
368 |
lemma object_type_object_clean_fields [simp]:
|
|
369 |
"object_type (object_clean_fields x cmp) = object_type x"
|
|
370 |
by (clarsimp simp: object_clean_fields_def object_type_def split: cdl_object.splits)
|
|
371 |
|
|
372 |
lemma object_type_object_clean [simp]:
|
|
373 |
"object_type (object_clean x cmp) = object_type x"
|
|
374 |
by (clarsimp simp: object_clean_def)
|
|
375 |
|
|
376 |
lemma object_type_add_to_slots [simp]:
|
|
377 |
"object_type (add_to_slots slots x) = object_type x"
|
|
378 |
by (clarsimp simp: object_type_def add_to_slots_def update_slots_def split: cdl_object.splits)
|
|
379 |
|
|
380 |
lemma object_slots_update_slots [simp]:
|
|
381 |
"has_slots obj \<Longrightarrow> object_slots (update_slots slots obj) = slots"
|
|
382 |
by (clarsimp simp: object_slots_def update_slots_def has_slots_def
|
|
383 |
split: cdl_object.splits)
|
|
384 |
|
|
385 |
lemma object_slots_update_slots_empty [simp]:
|
|
386 |
"\<not>has_slots obj \<Longrightarrow> object_slots (update_slots slots obj) = empty"
|
|
387 |
by (clarsimp simp: object_slots_def update_slots_def has_slots_def
|
|
388 |
split: cdl_object.splits)
|
|
389 |
|
|
390 |
lemma update_slots_no_slots [simp]:
|
|
391 |
"\<not>has_slots obj \<Longrightarrow> update_slots slots obj = obj"
|
|
392 |
by (clarsimp simp: update_slots_def has_slots_def split: cdl_object.splits)
|
|
393 |
|
|
394 |
lemma update_slots_update_slots [simp]:
|
|
395 |
"update_slots slots (update_slots slots' obj) = update_slots slots obj"
|
|
396 |
by (clarsimp simp: update_slots_def split: cdl_object.splits)
|
|
397 |
|
|
398 |
lemma update_slots_same_object:
|
|
399 |
"a = b \<Longrightarrow> update_slots a obj = update_slots b obj"
|
|
400 |
by (erule arg_cong)
|
|
401 |
|
|
402 |
lemma object_type_has_slots:
|
|
403 |
"\<lbrakk>has_slots x; object_type x = object_type y\<rbrakk> \<Longrightarrow> has_slots y"
|
|
404 |
by (clarsimp simp: object_type_def has_slots_def split: cdl_object.splits)
|
|
405 |
|
|
406 |
lemma object_slots_object_clean_fields [simp]:
|
|
407 |
"object_slots (object_clean_fields obj cmp) = object_slots obj"
|
|
408 |
by (clarsimp simp: object_slots_def object_clean_fields_def split: cdl_object.splits)
|
|
409 |
|
|
410 |
lemma object_slots_object_clean_slots [simp]:
|
|
411 |
"object_slots (object_clean_slots obj cmp) = clean_slots (object_slots obj) cmp"
|
|
412 |
by (clarsimp simp: object_clean_slots_def object_slots_def update_slots_def split: cdl_object.splits)
|
|
413 |
|
|
414 |
lemma object_slots_object_clean [simp]:
|
|
415 |
"object_slots (object_clean obj cmp) = clean_slots (object_slots obj) cmp"
|
|
416 |
by (clarsimp simp: object_clean_def)
|
|
417 |
|
|
418 |
lemma object_slots_add_to_slots [simp]:
|
|
419 |
"object_type y = object_type z \<Longrightarrow> object_slots (add_to_slots (object_slots y) z) = object_slots y ++ object_slots z"
|
|
420 |
by (clarsimp simp: object_slots_def add_to_slots_def update_slots_def object_type_def split: cdl_object.splits)
|
|
421 |
|
|
422 |
lemma update_slots_object_clean_slots [simp]:
|
|
423 |
"update_slots slots (object_clean_slots obj cmp) = update_slots slots obj"
|
|
424 |
by (clarsimp simp: object_clean_slots_def)
|
|
425 |
|
|
426 |
lemma object_clean_fields_idem [simp]:
|
|
427 |
"object_clean_fields (object_clean_fields obj cmp) cmp = object_clean_fields obj cmp"
|
|
428 |
by (clarsimp simp: object_clean_fields_def split: cdl_object.splits)
|
|
429 |
|
|
430 |
lemma object_clean_slots_idem [simp]:
|
|
431 |
"object_clean_slots (object_clean_slots obj cmp) cmp = object_clean_slots obj cmp"
|
|
432 |
apply (case_tac "has_slots obj")
|
|
433 |
apply (clarsimp simp: object_clean_slots_def clean_slots_def)+
|
|
434 |
done
|
|
435 |
|
|
436 |
lemma object_clean_fields_object_clean_slots [simp]:
|
|
437 |
"object_clean_fields (object_clean_slots obj gs) gs = object_clean_slots (object_clean_fields obj gs) gs"
|
|
438 |
by (clarsimp simp: object_clean_fields_def object_clean_slots_def
|
|
439 |
clean_slots_def object_slots_def update_slots_def
|
|
440 |
split: cdl_object.splits)
|
|
441 |
|
|
442 |
lemma object_clean_idem [simp]:
|
|
443 |
"object_clean (object_clean obj cmp) cmp = object_clean obj cmp"
|
|
444 |
by (clarsimp simp: object_clean_def)
|
|
445 |
|
|
446 |
lemma has_slots_object_clean_slots:
|
|
447 |
"has_slots (object_clean_slots obj cmp) = has_slots obj"
|
|
448 |
by (clarsimp simp: has_slots_def object_clean_slots_def update_slots_def split: cdl_object.splits)
|
|
449 |
|
|
450 |
lemma has_slots_object_clean_fields:
|
|
451 |
"has_slots (object_clean_fields obj cmp) = has_slots obj"
|
|
452 |
by (clarsimp simp: has_slots_def object_clean_fields_def split: cdl_object.splits)
|
|
453 |
|
|
454 |
lemma has_slots_object_clean:
|
|
455 |
"has_slots (object_clean obj cmp) = has_slots obj"
|
|
456 |
by (clarsimp simp: object_clean_def has_slots_object_clean_slots has_slots_object_clean_fields)
|
|
457 |
|
|
458 |
lemma object_slots_update_slots_object_clean_fields [simp]:
|
|
459 |
"object_slots (update_slots slots (object_clean_fields obj cmp)) = object_slots (update_slots slots obj)"
|
|
460 |
apply (case_tac "has_slots obj")
|
|
461 |
apply (clarsimp simp: has_slots_object_clean_fields)+
|
|
462 |
done
|
|
463 |
|
|
464 |
lemma object_clean_fields_update_slots [simp]:
|
|
465 |
"object_clean_fields (update_slots slots obj) cmp = update_slots slots (object_clean_fields obj cmp)"
|
|
466 |
by (clarsimp simp: object_clean_fields_def update_slots_def split: cdl_object.splits)
|
|
467 |
|
|
468 |
lemma object_clean_fields_twice [simp]:
|
|
469 |
"(object_clean_fields (object_clean_fields obj cmp') cmp) = object_clean_fields obj (cmp \<inter> cmp')"
|
|
470 |
by (clarsimp simp: object_clean_fields_def split: cdl_object.splits)
|
|
471 |
|
|
472 |
lemma update_slots_object_clean_fields:
|
|
473 |
"\<lbrakk>None \<notin> cmps; None \<notin> cmps'; object_type obj = object_type obj'\<rbrakk>
|
|
474 |
\<Longrightarrow> update_slots slots (object_clean_fields obj cmps) =
|
|
475 |
update_slots slots (object_clean_fields obj' cmps')"
|
|
476 |
by (fastforce simp: update_slots_def object_clean_fields_def object_type_def split: cdl_object.splits)
|
|
477 |
|
|
478 |
lemma object_clean_fields_no_slots:
|
|
479 |
"\<lbrakk>None \<notin> cmps; None \<notin> cmps'; object_type obj = object_type obj'; \<not> has_slots obj; \<not> has_slots obj'\<rbrakk>
|
|
480 |
\<Longrightarrow> object_clean_fields obj cmps = object_clean_fields obj' cmps'"
|
|
481 |
by (fastforce simp: object_clean_fields_def object_type_def has_slots_def split: cdl_object.splits)
|
|
482 |
|
|
483 |
lemma update_slots_object_clean:
|
|
484 |
"\<lbrakk>None \<notin> cmps; None \<notin> cmps'; object_type obj = object_type obj'\<rbrakk>
|
|
485 |
\<Longrightarrow> update_slots slots (object_clean obj cmps) = update_slots slots (object_clean obj' cmps')"
|
|
486 |
apply (clarsimp simp: object_clean_def object_clean_slots_def)
|
|
487 |
apply (erule (2) update_slots_object_clean_fields)
|
|
488 |
done
|
|
489 |
|
|
490 |
lemma cdl_heap_add_assoc':
|
|
491 |
"\<forall>obj_id. not_conflicting_objects x z obj_id \<and>
|
|
492 |
not_conflicting_objects y z obj_id \<and>
|
|
493 |
not_conflicting_objects x z obj_id \<Longrightarrow>
|
|
494 |
cdl_heap_add (SepState (cdl_heap_add x y) (cdl_ghost_state_add x y)) z =
|
|
495 |
cdl_heap_add x (SepState (cdl_heap_add y z) (cdl_ghost_state_add y z))"
|
|
496 |
apply (rule ext)
|
|
497 |
apply (rename_tac obj_id)
|
|
498 |
apply (erule_tac x=obj_id in allE)
|
|
499 |
apply (clarsimp simp: cdl_heap_add_def cdl_ghost_state_add_def not_conflicting_objects_def)
|
|
500 |
apply (simp add: Let_unfold split: option.splits)
|
|
501 |
apply (rename_tac obj_y obj_x obj_z)
|
|
502 |
apply (clarsimp simp: object_add_def clean_slots_def object_clean_def object_clean_slots_def Let_unfold)
|
|
503 |
apply (case_tac "has_slots obj_z")
|
|
504 |
apply (subgoal_tac "has_slots obj_y")
|
|
505 |
apply (subgoal_tac "has_slots obj_x")
|
|
506 |
apply ((clarsimp simp: has_slots_object_clean_fields has_slots_object_clean_slots has_slots_object_clean
|
|
507 |
map_add_restrict union_intersection |
|
|
508 |
drule inter_empty_not_both |
|
|
509 |
erule update_slots_object_clean_fields |
|
|
510 |
erule object_type_has_slots, simp |
|
|
511 |
simp | safe)+)[3]
|
|
512 |
apply (subgoal_tac "\<not> has_slots obj_y")
|
|
513 |
apply (subgoal_tac "\<not> has_slots obj_x")
|
|
514 |
apply ((clarsimp simp: has_slots_object_clean_fields has_slots_object_clean_slots has_slots_object_clean
|
|
515 |
map_add_restrict union_intersection |
|
|
516 |
drule inter_empty_not_both |
|
|
517 |
erule object_clean_fields_no_slots |
|
|
518 |
erule object_type_has_slots, simp |
|
|
519 |
simp | safe)+)
|
|
520 |
apply (fastforce simp: object_type_has_slots)+
|
|
521 |
done
|
|
522 |
|
|
523 |
lemma cdl_heap_add_assoc:
|
|
524 |
"\<lbrakk>sep_state_disj x y; sep_state_disj y z; sep_state_disj x z\<rbrakk>
|
|
525 |
\<Longrightarrow> cdl_heap_add (SepState (cdl_heap_add x y) (cdl_ghost_state_add x y)) z =
|
|
526 |
cdl_heap_add x (SepState (cdl_heap_add y z) (cdl_ghost_state_add y z))"
|
|
527 |
apply (clarsimp simp: sep_state_disj_def)
|
|
528 |
apply (cut_tac cdl_heap_add_assoc')
|
|
529 |
apply fast
|
|
530 |
apply fastforce
|
|
531 |
done
|
|
532 |
|
|
533 |
lemma cdl_ghost_state_add_assoc:
|
|
534 |
"cdl_ghost_state_add (SepState (cdl_heap_add x y) (cdl_ghost_state_add x y)) z =
|
|
535 |
cdl_ghost_state_add x (SepState (cdl_heap_add y z) (cdl_ghost_state_add y z))"
|
|
536 |
apply (rule ext)
|
|
537 |
apply (fastforce simp: cdl_heap_add_def cdl_ghost_state_add_def Let_unfold)
|
|
538 |
done
|
|
539 |
|
|
540 |
lemma clean_slots_map_add_comm:
|
|
541 |
"cmps_a \<inter> cmps_b = {}
|
|
542 |
\<Longrightarrow> clean_slots slots_a cmps_a ++ clean_slots slots_b cmps_b =
|
|
543 |
clean_slots slots_b cmps_b ++ clean_slots slots_a cmps_a"
|
|
544 |
apply (clarsimp simp: clean_slots_def)
|
|
545 |
apply (drule the_set_inter_empty)
|
|
546 |
apply (erule map_add_restrict_comm)
|
|
547 |
done
|
|
548 |
|
|
549 |
lemma object_clean_all:
|
|
550 |
"object_type obj_a = object_type obj_b \<Longrightarrow> object_clean obj_b {} = object_clean obj_a {}"
|
|
551 |
apply (clarsimp simp: object_clean_def object_clean_slots_def clean_slots_def the_set_def)
|
|
552 |
apply (rule_tac cmps'1="{}" and obj'1="obj_a" in trans [OF update_slots_object_clean_fields], fastforce+)
|
|
553 |
done
|
|
554 |
|
|
555 |
lemma object_add_comm:
|
|
556 |
"\<lbrakk>object_type obj_a = object_type obj_b; cmps_a \<inter> cmps_b = {}\<rbrakk>
|
|
557 |
\<Longrightarrow> object_add obj_a obj_b cmps_a cmps_b = object_add obj_b obj_a cmps_b cmps_a"
|
|
558 |
apply (clarsimp simp: object_add_def Let_unfold)
|
|
559 |
apply (rule conjI | clarsimp)+
|
|
560 |
apply fastforce
|
|
561 |
apply (rule conjI | clarsimp)+
|
|
562 |
apply (drule_tac slots_a = "object_slots obj_a" and slots_b = "object_slots obj_b" in clean_slots_map_add_comm)
|
|
563 |
apply fastforce
|
|
564 |
apply (rule conjI | clarsimp)+
|
|
565 |
apply (drule_tac slots_a = "object_slots obj_a" and slots_b = "object_slots obj_b" in clean_slots_map_add_comm)
|
|
566 |
apply fastforce
|
|
567 |
apply (rule conjI | clarsimp)+
|
|
568 |
apply (erule object_clean_all)
|
|
569 |
apply (clarsimp)
|
|
570 |
apply (rule_tac cmps'1=cmps_b and obj'1=obj_b in trans [OF update_slots_object_clean], assumption+)
|
|
571 |
apply (drule_tac slots_a = "object_slots obj_a" and slots_b = "object_slots obj_b" in clean_slots_map_add_comm)
|
|
572 |
apply fastforce
|
|
573 |
done
|
|
574 |
|
|
575 |
lemma sep_state_add_comm:
|
|
576 |
"sep_state_disj x y \<Longrightarrow> sep_state_add x y = sep_state_add y x"
|
|
577 |
apply (clarsimp simp: sep_state_add_def sep_state_disj_def)
|
|
578 |
apply (rule conjI)
|
|
579 |
apply (case_tac x, case_tac y, clarsimp)
|
|
580 |
apply (rename_tac heap_a gs_a heap_b gs_b)
|
|
581 |
apply (clarsimp simp: cdl_heap_add_def Let_unfold)
|
|
582 |
apply (rule ext)
|
|
583 |
apply (case_tac "heap_a obj_id")
|
|
584 |
apply (case_tac "heap_b obj_id", simp_all add: slots_of_heap_def)
|
|
585 |
apply (case_tac "heap_b obj_id", simp_all add: slots_of_heap_def)
|
|
586 |
apply (rename_tac obj_a obj_b)
|
|
587 |
apply (erule_tac x=obj_id in allE)
|
|
588 |
apply (rule object_add_comm)
|
|
589 |
apply (clarsimp simp: not_conflicting_objects_def)
|
|
590 |
apply (clarsimp simp: not_conflicting_objects_def)
|
|
591 |
apply (rule ext, fastforce simp: cdl_ghost_state_add_def Let_unfold Un_commute)
|
|
592 |
done
|
|
593 |
|
|
594 |
lemma add_to_slots_comm:
|
|
595 |
"\<lbrakk>object_slots y_obj \<bottom> object_slots z_obj; update_slots empty y_obj = update_slots empty z_obj \<rbrakk>
|
|
596 |
\<Longrightarrow> add_to_slots (object_slots z_obj) y_obj = add_to_slots (object_slots y_obj) z_obj"
|
|
597 |
by (fastforce simp: add_to_slots_def update_slots_def object_slots_def
|
|
598 |
cdl_tcb.splits cdl_cnode.splits
|
|
599 |
dest!: map_add_com
|
|
600 |
split: cdl_object.splits)
|
|
601 |
|
|
602 |
lemma cdl_heap_add_none1:
|
|
603 |
"cdl_heap_add x y obj_id = None \<Longrightarrow> (sep_heap x) obj_id = None"
|
|
604 |
by (clarsimp simp: cdl_heap_add_def Let_unfold split:option.splits split_if_asm)
|
|
605 |
|
|
606 |
lemma cdl_heap_add_none2:
|
|
607 |
"cdl_heap_add x y obj_id = None \<Longrightarrow> (sep_heap y) obj_id = None"
|
|
608 |
by (clarsimp simp: cdl_heap_add_def Let_unfold split:option.splits split_if_asm)
|
|
609 |
|
|
610 |
lemma object_type_object_addL:
|
|
611 |
"object_type obj = object_type obj'
|
|
612 |
\<Longrightarrow> object_type (object_add obj obj' cmp cmp') = object_type obj"
|
|
613 |
by (clarsimp simp: object_add_def Let_unfold)
|
|
614 |
|
|
615 |
lemma object_type_object_addR:
|
|
616 |
"object_type obj = object_type obj'
|
|
617 |
\<Longrightarrow> object_type (object_add obj obj' cmp cmp') = object_type obj'"
|
|
618 |
by (clarsimp simp: object_add_def Let_unfold)
|
|
619 |
|
|
620 |
lemma sep_state_add_disjL:
|
|
621 |
"\<lbrakk>sep_state_disj y z; sep_state_disj x (sep_state_add y z)\<rbrakk> \<Longrightarrow> sep_state_disj x y"
|
|
622 |
apply (clarsimp simp: sep_state_disj_def sep_state_add_def)
|
|
623 |
apply (rename_tac obj_id)
|
|
624 |
apply (clarsimp simp: not_conflicting_objects_def)
|
|
625 |
apply (erule_tac x=obj_id in allE)+
|
|
626 |
apply (fastforce simp: cdl_heap_add_def cdl_ghost_state_add_def object_type_object_addR
|
|
627 |
split: option.splits)
|
|
628 |
done
|
|
629 |
|
|
630 |
lemma sep_state_add_disjR:
|
|
631 |
"\<lbrakk>sep_state_disj y z; sep_state_disj x (sep_state_add y z)\<rbrakk> \<Longrightarrow> sep_state_disj x z"
|
|
632 |
apply (clarsimp simp: sep_state_disj_def sep_state_add_def)
|
|
633 |
apply (rename_tac obj_id)
|
|
634 |
apply (clarsimp simp: not_conflicting_objects_def)
|
|
635 |
apply (erule_tac x=obj_id in allE)+
|
|
636 |
apply (fastforce simp: cdl_heap_add_def cdl_ghost_state_add_def object_type_object_addR
|
|
637 |
split: option.splits)
|
|
638 |
done
|
|
639 |
|
|
640 |
lemma sep_state_add_disj:
|
|
641 |
"\<lbrakk>sep_state_disj y z; sep_state_disj x y; sep_state_disj x z\<rbrakk> \<Longrightarrow> sep_state_disj x (sep_state_add y z)"
|
|
642 |
apply (clarsimp simp: sep_state_disj_def sep_state_add_def)
|
|
643 |
apply (rename_tac obj_id)
|
|
644 |
apply (clarsimp simp: not_conflicting_objects_def)
|
|
645 |
apply (erule_tac x=obj_id in allE)+
|
|
646 |
apply (fastforce simp: cdl_heap_add_def cdl_ghost_state_add_def object_type_object_addR
|
|
647 |
split: option.splits)
|
|
648 |
done
|
|
649 |
|
|
650 |
|
|
651 |
|
|
652 |
|
|
653 |
(*********************************************)
|
|
654 |
(* Definition of separation logic for capDL. *)
|
|
655 |
(*********************************************)
|
|
656 |
|
|
657 |
instantiation "sep_state" :: zero
|
|
658 |
begin
|
|
659 |
definition "0 \<equiv> SepState empty (\<lambda>obj_id. {})"
|
|
660 |
instance ..
|
|
661 |
end
|
|
662 |
|
|
663 |
instantiation "sep_state" :: stronger_sep_algebra
|
|
664 |
begin
|
|
665 |
|
|
666 |
definition "(op ##) \<equiv> sep_state_disj"
|
|
667 |
definition "(op +) \<equiv> sep_state_add"
|
|
668 |
|
|
669 |
|
|
670 |
|
|
671 |
(**********************************************
|
|
672 |
* The proof that this is a separation logic. *
|
|
673 |
**********************************************)
|
|
674 |
|
|
675 |
instance
|
|
676 |
apply default
|
|
677 |
(* x ## 0 *)
|
|
678 |
apply (simp add: sep_disj_sep_state_def sep_state_disj_def zero_sep_state_def)
|
|
679 |
(* x ## y \<Longrightarrow> y ## x *)
|
|
680 |
apply (clarsimp simp: not_conflicting_objects_comm sep_disj_sep_state_def sep_state_disj_def Let_unfold
|
|
681 |
map_disj_com not_conflicting_objects_comm Int_commute)
|
|
682 |
(* x + 0 = x *)
|
|
683 |
apply (simp add: plus_sep_state_def sep_state_add_def zero_sep_state_def)
|
|
684 |
apply (case_tac x)
|
|
685 |
apply (clarsimp simp: cdl_heap_add_def)
|
|
686 |
apply (rule ext)
|
|
687 |
apply (clarsimp simp: cdl_ghost_state_add_def split:split_if_asm)
|
|
688 |
(* x ## y \<Longrightarrow> x + y = y + x *)
|
|
689 |
apply (clarsimp simp: plus_sep_state_def sep_disj_sep_state_def)
|
|
690 |
apply (erule sep_state_add_comm)
|
|
691 |
(* (x + y) + z = x + (y + z) *)
|
|
692 |
apply (simp add: plus_sep_state_def sep_state_add_def)
|
|
693 |
apply (rule conjI)
|
|
694 |
apply (clarsimp simp: sep_disj_sep_state_def)
|
|
695 |
apply (erule (2) cdl_heap_add_assoc)
|
|
696 |
apply (rule cdl_ghost_state_add_assoc)
|
|
697 |
(* x ## y + z = (x ## y \<and> x ## z) *)
|
|
698 |
apply (clarsimp simp: plus_sep_state_def sep_disj_sep_state_def)
|
|
699 |
apply (rule iffI)
|
|
700 |
(* x ## y + z \<Longrightarrow> (x ## y \<and> x ## z) *)
|
|
701 |
apply (rule conjI)
|
|
702 |
(* x ## y + z \<Longrightarrow> (x ## y) *)
|
|
703 |
apply (erule (1) sep_state_add_disjL)
|
|
704 |
(* x ## y + z \<Longrightarrow> (x ## z) *)
|
|
705 |
apply (erule (1) sep_state_add_disjR)
|
|
706 |
(* x ## y + z \<Longleftarrow> (x ## y \<and> x ## z) *)
|
|
707 |
apply clarsimp
|
|
708 |
apply (erule (2) sep_state_add_disj)
|
|
709 |
done
|
|
710 |
|
|
711 |
end
|
|
712 |
|
|
713 |
|
|
714 |
end
|