gdb/prologue-value.h - gdb

Data types defined

Macros defined

Source code

  1. /* Interface to prologue value handling for GDB.
  2.    Copyright (C) 2003-2015 Free Software Foundation, Inc.

  4.    This file is part of GDB.

  6.    This program is free software; you can redistribute it and/or modify
  7.    it under the terms of the GNU General Public License as published by
  8.    the Free Software Foundation; either version 3 of the License, or
  9.    (at your option) any later version.

  11.    This program is distributed in the hope that it will be useful,
  12.    but WITHOUT ANY WARRANTY; without even the implied warranty of
  13.    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14.    GNU General Public License for more details.

  16.    You should have received a copy of the GNU General Public License
  17.    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */

  19. #ifndef PROLOGUE_VALUE_H
  20. ‌#define PROLOGUE_VALUE_H

  22. /* When we analyze a prologue, we're really doing 'abstract
  23.    interpretation' or 'pseudo-evaluation': running the function's code
  24.    in simulation, but using conservative approximations of the values
  25.    it would have when it actually runs.  For example, if our function
  26.    starts with the instruction:

  28.       addi r1, 42     # add 42 to r1

  30.    we don't know exactly what value will be in r1 after executing this
  31.    instruction, but we do know it'll be 42 greater than its original
  32.    value.

  34.    If we then see an instruction like:

  36.       addi r1, 22     # add 22 to r1

  38.    we still don't know what r1's value is, but again, we can say it is
  39.    now 64 greater than its original value.

  41.    If the next instruction were:

  43.       mov r2, r1      # set r2 to r1's value

  45.    then we can say that r2's value is now the original value of r1
  46.    plus 64.

  48.    It's common for prologues to save registers on the stack, so we'll
  49.    need to track the values of stack frame slots, as well as the
  50.    registers.  So after an instruction like this:

  52.       mov (fp+4), r2

  54.    then we'd know that the stack slot four bytes above the frame
  55.    pointer holds the original value of r1 plus 64.

  57.    And so on.

  59.    Of course, this can only go so far before it gets unreasonable.  If
  60.    we wanted to be able to say anything about the value of r1 after
  61.    the instruction:

  63.       xor r1, r3      # exclusive-or r1 and r3, place result in r1

  65.    then things would get pretty complex.  But remember, we're just
  66.    doing a conservative approximation; if exclusive-or instructions
  67.    aren't relevant to prologues, we can just say r1's value is now
  68.    'unknown'.  We can ignore things that are too complex, if that loss
  69.    of information is acceptable for our application.

  71.    So when I say "conservative approximation" here, what I mean is an
  72.    approximation that is either accurate, or marked "unknown", but
  73.    never inaccurate.

  75.    Once you've reached the current PC, or an instruction that you
  76.    don't know how to simulate, you stop.  Now you can examine the
  77.    state of the registers and stack slots you've kept track of.

  79.    - To see how large your stack frame is, just check the value of the
  80.      stack pointer register; if it's the original value of the SP
  81.      minus a constant, then that constant is the stack frame's size.
  82.      If the SP's value has been marked as 'unknown', then that means
  83.      the prologue has done something too complex for us to track, and
  84.      we don't know the frame size.

  86.    - To see where we've saved the previous frame's registers, we just
  87.      search the values we've tracked --- stack slots, usually, but
  88.      registers, too, if you want --- for something equal to the
  89.      register's original value.  If the ABI suggests a standard place
  90.      to save a given register, then we can check there first, but
  91.      really, anything that will get us back the original value will
  92.      probably work.

  94.    Sure, this takes some work.  But prologue analyzers aren't
  95.    quick-and-simple pattern patching to recognize a few fixed prologue
  96.    forms any more; they're big, hairy functions.  Along with inferior
  97.    function calls, prologue analysis accounts for a substantial
  98.    portion of the time needed to stabilize a GDB port.  So I think
  99.    it's worthwhile to look for an approach that will be easier to
  100.    understand and maintain.  In the approach used here:

  102.    - It's easier to see that the analyzer is correct: you just see
  103.      whether the analyzer properly (albiet conservatively) simulates
  104.      the effect of each instruction.

  106.    - It's easier to extend the analyzer: you can add support for new
  107.      instructions, and know that you haven't broken anything that
  108.      wasn't already broken before.

  110.    - It's orthogonal: to gather new information, you don't need to
  111.      complicate the code for each instruction.  As long as your domain
  112.      of conservative values is already detailed enough to tell you
  113.      what you need, then all the existing instruction simulations are
  114.      already gathering the right data for you.

  116.    A 'struct prologue_value' is a conservative approximation of the
  117.    real value the register or stack slot will have.  */

  119. ‌struct prologue_value {

  121.   /* What sort of value is this?  This determines the interpretation
  122.      of subsequent fields.  */
  123.   enum {

  125.     /* We don't know anything about the value.  This is also used for
  126.        values we could have kept track of, when doing so would have
  127.        been too complex and we don't want to bother.  The bottom of
  128.        our lattice.  */
  129.     pvk_unknown,

  131.     /* A known constant.  K is its value.  */
  132.     pvk_constant,

  134.     /* The value that register REG originally had *UPON ENTRY TO THE
  135.        FUNCTION*, plus K.  If K is zero, this means, obviously, just
  136.        the value REG had upon entry to the function.  REG is a GDB
  137.        register number.  Before we start interpreting, we initialize
  138.        every register R to { pvk_register, R, 0 }.  */
  139.     pvk_register,

  141.   } kind;

  143.   /* The meanings of the following fields depend on 'kind'; see the
  144.      comments for the specific 'kind' values.  */
  145.   int reg;
  146.   CORE_ADDR k;
  147. };

  149. ‌typedef struct prologue_value pv_t;


  152. /* Return the unknown prologue value --- { pvk_unknown, ?, ? }.  */
  153. pv_t pv_unknown (void);

  155. /* Return the prologue value representing the constant K.  */
  156. pv_t pv_constant (CORE_ADDR k);

  158. /* Return the prologue value representing the original value of
  159.    register REG, plus the constant K.  */
  160. pv_t pv_register (int reg, CORE_ADDR k);


  163. /* Return conservative approximations of the results of the following
  164.    operations.  */
  165. pv_t pv_add (pv_t a, pv_t b);               /* a + b */
  166. pv_t pv_add_constant (pv_t v, CORE_ADDR k); /* a + k */
  167. pv_t pv_subtract (pv_t a, pv_t b);          /* a - b */
  168. pv_t pv_logical_and (pv_t a, pv_t b);       /* a & b */


  171. /* Return non-zero iff A and B are identical expressions.

  173.    This is not the same as asking if the two values are equal; the
  174.    result of such a comparison would have to be a pv_boolean, and
  175.    asking whether two 'unknown' values were equal would give you
  176.    pv_maybe.  Same for comparing, say, { pvk_register, R1, 0 } and {
  177.    pvk_register, R2, 0}.

  179.    Instead, this function asks whether the two representations are the
  180.    same.  */
  181. int pv_is_identical (pv_t a, pv_t b);


  184. /* Return non-zero if A is known to be a constant.  */
  185. int pv_is_constant (pv_t a);

  187. /* Return non-zero if A is the original value of register number R
  188.    plus some constant, zero otherwise.  */
  189. int pv_is_register (pv_t a, int r);


  192. /* Return non-zero if A is the original value of register R plus the
  193.    constant K.  */
  194. int pv_is_register_k (pv_t a, int r, CORE_ADDR k);

  196. /* A conservative boolean type, including "maybe", when we can't
  197.    figure out whether something is true or not.  */
  198. ‌enum pv_boolean {
  199.   pv_maybe,
  200.   pv_definite_yes,
  201.   pv_definite_no,
  202. };


  205. /* Decide whether a reference to SIZE bytes at ADDR refers exactly to
  206.    an element of an array.  The array starts at ARRAY_ADDR, and has
  207.    ARRAY_LEN values of ELT_SIZE bytes each.  If ADDR definitely does
  208.    refer to an array element, set *I to the index of the referenced
  209.    element in the array, and return pv_definite_yes.  If it definitely
  210.    doesn't, return pv_definite_no.  If we can't tell, return pv_maybe.

  212.    If the reference does touch the array, but doesn't fall exactly on
  213.    an element boundary, or doesn't refer to the whole element, return
  214.    pv_maybe.  */
  215. enum pv_boolean pv_is_array_ref (pv_t addr, CORE_ADDR size,
  216.                                  pv_t array_addr, CORE_ADDR array_len,
  217.                                  CORE_ADDR elt_size,
  218.                                  int *i);


  221. /* A 'struct pv_area' keeps track of values stored in a particular
  222.    region of memory.  */
  223. struct pv_area;

  225. /* Create a new area, tracking stores relative to the original value
  226.    of BASE_REG.  If BASE_REG is SP, then this effectively records the
  227.    contents of the stack frame: the original value of the SP is the
  228.    frame's CFA, or some constant offset from it.

  230.    Stores to constant addresses, unknown addresses, or to addresses
  231.    relative to registers other than BASE_REG will trash this area; see
  232.    pv_area_store_would_trash.

  234.    To check whether a pointer refers to this area, only the low
  235.    ADDR_BIT bits will be compared.  */
  236. struct pv_area *make_pv_area (int base_reg, int addr_bit);

  238. /* Free AREA.  */
  239. void free_pv_area (struct pv_area *area);


  242. /* Register a cleanup to free AREA.  */
  243. struct cleanup *make_cleanup_free_pv_area (struct pv_area *area);


  246. /* Store the SIZE-byte value VALUE at ADDR in AREA.

  248.    If ADDR is not relative to the same base register we used in
  249.    creating AREA, then we can't tell which values here the stored
  250.    value might overlap, and we'll have to mark everything as
  251.    unknown.  */
  252. void pv_area_store (struct pv_area *area,
  253.                     pv_t addr,
  254.                     CORE_ADDR size,
  255.                     pv_t value);

  257. /* Return the SIZE-byte value at ADDR in AREA.  This may return
  258.    pv_unknown ().  */
  259. pv_t pv_area_fetch (struct pv_area *area, pv_t addr, CORE_ADDR size);

  261. /* Return true if storing to address ADDR in AREA would force us to
  262.    mark the contents of the entire area as unknown.  This could happen
  263.    if, say, ADDR is unknown, since we could be storing anywhere.  Or,
  264.    it could happen if ADDR is relative to a different register than
  265.    the other stores base register, since we don't know the relative
  266.    values of the two registers.

  268.    If you've reached such a store, it may be better to simply stop the
  269.    prologue analysis, and return the information you've gathered,
  270.    instead of losing all that information, most of which is probably
  271.    okay.  */
  272. int pv_area_store_would_trash (struct pv_area *area, pv_t addr);


  275. /* Search AREA for the original value of REGISTER.  If we can't find
  276.    it, return zero; if we can find it, return a non-zero value, and if
  277.    OFFSET_P is non-zero, set *OFFSET_P to the register's offset within
  278.    AREA.  GDBARCH is the architecture of which REGISTER is a member.

  280.    In the worst case, this takes time proportional to the number of
  281.    items stored in AREA.  If you plan to gather a lot of information
  282.    about registers saved in AREA, consider calling pv_area_scan
  283.    instead, and collecting all your information in one pass.  */
  284. int pv_area_find_reg (struct pv_area *area,
  285.                       struct gdbarch *gdbarch,
  286.                       int reg,
  287.                       CORE_ADDR *offset_p);


  290. /* For every part of AREA whose value we know, apply FUNC to CLOSURE,
  291.    the value's address, its size, and the value itself.  */
  292. void pv_area_scan (struct pv_area *area,
  293.                    void (*func) (void *closure,
  294.                                  pv_t addr,
  295.                                  CORE_ADDR size,
  296.                                  pv_t value),
  297.                    void *closure);


  300. #endif /* PROLOGUE_VALUE_H */