1 /* Interface to prologue value handling for GDB.
2 Copyright 2003, 2004, 2005, 2007, 2008, 2009, 2010
3 Free Software Foundation, Inc.
5 This file is part of GDB.
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 3 of the License, or
10 (at your option) any later version.
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with this program. If not, see <http://www.gnu.org/licenses/>. */
20 #ifndef PROLOGUE_VALUE_H
21 #define PROLOGUE_VALUE_H
23 /* When we analyze a prologue, we're really doing 'abstract
24 interpretation' or 'pseudo-evaluation': running the function's code
25 in simulation, but using conservative approximations of the values
26 it would have when it actually runs. For example, if our function
27 starts with the instruction:
29 addi r1, 42 # add 42 to r1
31 we don't know exactly what value will be in r1 after executing this
32 instruction, but we do know it'll be 42 greater than its original
35 If we then see an instruction like:
37 addi r1, 22 # add 22 to r1
39 we still don't know what r1's value is, but again, we can say it is
40 now 64 greater than its original value.
42 If the next instruction were:
44 mov r2, r1 # set r2 to r1's value
46 then we can say that r2's value is now the original value of r1
49 It's common for prologues to save registers on the stack, so we'll
50 need to track the values of stack frame slots, as well as the
51 registers. So after an instruction like this:
55 then we'd know that the stack slot four bytes above the frame
56 pointer holds the original value of r1 plus 64.
60 Of course, this can only go so far before it gets unreasonable. If
61 we wanted to be able to say anything about the value of r1 after
64 xor r1, r3 # exclusive-or r1 and r3, place result in r1
66 then things would get pretty complex. But remember, we're just
67 doing a conservative approximation; if exclusive-or instructions
68 aren't relevant to prologues, we can just say r1's value is now
69 'unknown'. We can ignore things that are too complex, if that loss
70 of information is acceptable for our application.
72 So when I say "conservative approximation" here, what I mean is an
73 approximation that is either accurate, or marked "unknown", but
76 Once you've reached the current PC, or an instruction that you
77 don't know how to simulate, you stop. Now you can examine the
78 state of the registers and stack slots you've kept track of.
80 - To see how large your stack frame is, just check the value of the
81 stack pointer register; if it's the original value of the SP
82 minus a constant, then that constant is the stack frame's size.
83 If the SP's value has been marked as 'unknown', then that means
84 the prologue has done something too complex for us to track, and
85 we don't know the frame size.
87 - To see where we've saved the previous frame's registers, we just
88 search the values we've tracked --- stack slots, usually, but
89 registers, too, if you want --- for something equal to the
90 register's original value. If the ABI suggests a standard place
91 to save a given register, then we can check there first, but
92 really, anything that will get us back the original value will
95 Sure, this takes some work. But prologue analyzers aren't
96 quick-and-simple pattern patching to recognize a few fixed prologue
97 forms any more; they're big, hairy functions. Along with inferior
98 function calls, prologue analysis accounts for a substantial
99 portion of the time needed to stabilize a GDB port. So I think
100 it's worthwhile to look for an approach that will be easier to
101 understand and maintain. In the approach used here:
103 - It's easier to see that the analyzer is correct: you just see
104 whether the analyzer properly (albiet conservatively) simulates
105 the effect of each instruction.
107 - It's easier to extend the analyzer: you can add support for new
108 instructions, and know that you haven't broken anything that
109 wasn't already broken before.
111 - It's orthogonal: to gather new information, you don't need to
112 complicate the code for each instruction. As long as your domain
113 of conservative values is already detailed enough to tell you
114 what you need, then all the existing instruction simulations are
115 already gathering the right data for you.
117 A 'struct prologue_value' is a conservative approximation of the
118 real value the register or stack slot will have. */
120 struct prologue_value {
122 /* What sort of value is this? This determines the interpretation
123 of subsequent fields. */
126 /* We don't know anything about the value. This is also used for
127 values we could have kept track of, when doing so would have
128 been too complex and we don't want to bother. The bottom of
132 /* A known constant. K is its value. */
135 /* The value that register REG originally had *UPON ENTRY TO THE
136 FUNCTION*, plus K. If K is zero, this means, obviously, just
137 the value REG had upon entry to the function. REG is a GDB
138 register number. Before we start interpreting, we initialize
139 every register R to { pvk_register, R, 0 }. */
144 /* The meanings of the following fields depend on 'kind'; see the
145 comments for the specific 'kind' values. */
150 typedef struct prologue_value pv_t;
153 /* Return the unknown prologue value --- { pvk_unknown, ?, ? }. */
154 pv_t pv_unknown (void);
156 /* Return the prologue value representing the constant K. */
157 pv_t pv_constant (CORE_ADDR k);
159 /* Return the prologue value representing the original value of
160 register REG, plus the constant K. */
161 pv_t pv_register (int reg, CORE_ADDR k);
164 /* Return conservative approximations of the results of the following
166 pv_t pv_add (pv_t a, pv_t b); /* a + b */
167 pv_t pv_add_constant (pv_t v, CORE_ADDR k); /* a + k */
168 pv_t pv_subtract (pv_t a, pv_t b); /* a - b */
169 pv_t pv_logical_and (pv_t a, pv_t b); /* a & b */
172 /* Return non-zero iff A and B are identical expressions.
174 This is not the same as asking if the two values are equal; the
175 result of such a comparison would have to be a pv_boolean, and
176 asking whether two 'unknown' values were equal would give you
177 pv_maybe. Same for comparing, say, { pvk_register, R1, 0 } and {
178 pvk_register, R2, 0}.
180 Instead, this function asks whether the two representations are the
182 int pv_is_identical (pv_t a, pv_t b);
185 /* Return non-zero if A is known to be a constant. */
186 int pv_is_constant (pv_t a);
188 /* Return non-zero if A is the original value of register number R
189 plus some constant, zero otherwise. */
190 int pv_is_register (pv_t a, int r);
193 /* Return non-zero if A is the original value of register R plus the
195 int pv_is_register_k (pv_t a, int r, CORE_ADDR k);
197 /* A conservative boolean type, including "maybe", when we can't
198 figure out whether something is true or not. */
206 /* Decide whether a reference to SIZE bytes at ADDR refers exactly to
207 an element of an array. The array starts at ARRAY_ADDR, and has
208 ARRAY_LEN values of ELT_SIZE bytes each. If ADDR definitely does
209 refer to an array element, set *I to the index of the referenced
210 element in the array, and return pv_definite_yes. If it definitely
211 doesn't, return pv_definite_no. If we can't tell, return pv_maybe.
213 If the reference does touch the array, but doesn't fall exactly on
214 an element boundary, or doesn't refer to the whole element, return
216 enum pv_boolean pv_is_array_ref (pv_t addr, CORE_ADDR size,
217 pv_t array_addr, CORE_ADDR array_len,
222 /* A 'struct pv_area' keeps track of values stored in a particular
226 /* Create a new area, tracking stores relative to the original value
227 of BASE_REG. If BASE_REG is SP, then this effectively records the
228 contents of the stack frame: the original value of the SP is the
229 frame's CFA, or some constant offset from it.
231 Stores to constant addresses, unknown addresses, or to addresses
232 relative to registers other than BASE_REG will trash this area; see
233 pv_area_store_would_trash.
235 To check whether a pointer refers to this area, only the low
236 ADDR_BIT bits will be compared. */
237 struct pv_area *make_pv_area (int base_reg, int addr_bit);
240 void free_pv_area (struct pv_area *area);
243 /* Register a cleanup to free AREA. */
244 struct cleanup *make_cleanup_free_pv_area (struct pv_area *area);
247 /* Store the SIZE-byte value VALUE at ADDR in AREA.
249 If ADDR is not relative to the same base register we used in
250 creating AREA, then we can't tell which values here the stored
251 value might overlap, and we'll have to mark everything as
253 void pv_area_store (struct pv_area *area,
258 /* Return the SIZE-byte value at ADDR in AREA. This may return
260 pv_t pv_area_fetch (struct pv_area *area, pv_t addr, CORE_ADDR size);
262 /* Return true if storing to address ADDR in AREA would force us to
263 mark the contents of the entire area as unknown. This could happen
264 if, say, ADDR is unknown, since we could be storing anywhere. Or,
265 it could happen if ADDR is relative to a different register than
266 the other stores base register, since we don't know the relative
267 values of the two registers.
269 If you've reached such a store, it may be better to simply stop the
270 prologue analysis, and return the information you've gathered,
271 instead of losing all that information, most of which is probably
273 int pv_area_store_would_trash (struct pv_area *area, pv_t addr);
276 /* Search AREA for the original value of REGISTER. If we can't find
277 it, return zero; if we can find it, return a non-zero value, and if
278 OFFSET_P is non-zero, set *OFFSET_P to the register's offset within
279 AREA. GDBARCH is the architecture of which REGISTER is a member.
281 In the worst case, this takes time proportional to the number of
282 items stored in AREA. If you plan to gather a lot of information
283 about registers saved in AREA, consider calling pv_area_scan
284 instead, and collecting all your information in one pass. */
285 int pv_area_find_reg (struct pv_area *area,
286 struct gdbarch *gdbarch,
288 CORE_ADDR *offset_p);
291 /* For every part of AREA whose value we know, apply FUNC to CLOSURE,
292 the value's address, its size, and the value itself. */
293 void pv_area_scan (struct pv_area *area,
294 void (*func) (void *closure,
301 #endif /* PROLOGUE_VALUE_H */