Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 1 | This file describes some special Python build types enabled via |
| 2 | compile-time preprocessor defines. |
| 3 | |
| 4 | --------------------------------------------------------------------------- |
Tim Peters | 62fc52e | 2002-07-11 00:23:58 +0000 | [diff] [blame] | 5 | Py_REF_DEBUG introduced in 1.4 |
| 6 | named REF_DEBUG before 1.4 |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 7 | |
| 8 | Turn on aggregate reference counting. This arranges that extern |
| 9 | _Py_RefTotal hold a count of all references, the sum of ob_refcnt across |
| 10 | all objects. In a debug-mode build, this is where the "8288" comes from |
| 11 | in |
| 12 | |
| 13 | >>> 23 |
| 14 | 23 |
| 15 | [8288 refs] |
| 16 | >>> |
| 17 | |
| 18 | Note that if this count increases when you're not storing away new objects, |
| 19 | there's probably a leak. Remember, though, that in interactive mode the |
| 20 | special name "_" holds a reference to the last result displayed! |
| 21 | |
| 22 | Py_REF_DEBUG also checks after every decref to verify that the refcount |
| 23 | hasn't gone negative, and causes an immediate fatal error if it has. |
| 24 | |
| 25 | Special gimmicks: |
| 26 | |
| 27 | sys.gettotalrefcount() |
| 28 | Return current total of all refcounts. |
| 29 | Available under Py_REF_DEBUG in Python 2.3. |
| 30 | Before 2.3, Py_TRACE_REFS was required to enable this function. |
| 31 | --------------------------------------------------------------------------- |
Tim Peters | 62fc52e | 2002-07-11 00:23:58 +0000 | [diff] [blame] | 32 | Py_TRACE_REFS introduced in 1.4 |
| 33 | named TRACE_REFS before 1.4 |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 34 | |
| 35 | Turn on heavy reference debugging. This is major surgery. Every PyObject |
| 36 | grows two more pointers, to maintain a doubly-linked list of all live |
| 37 | heap-allocated objects (note that, e.g., most builtin type objects are not |
| 38 | in this list, as they're statically allocated). Note that because the |
| 39 | fundamental PyObject layout changes, Python modules compiled with |
| 40 | Py_TRACE_REFS are incompatible with modules compiled without it. |
| 41 | |
| 42 | Py_TRACE_REFS implies Py_REF_DEBUG. |
| 43 | |
| 44 | Special gimmicks: |
| 45 | |
| 46 | sys.getobjects(max[, type]) |
Tim Peters | a788f5e | 2002-07-10 18:47:03 +0000 | [diff] [blame] | 47 | Return list of the (no more than) max most-recently allocated objects, |
| 48 | most recently allocated first in the list, least-recently allocated |
| 49 | last in the list. max=0 means no limit on list length. |
| 50 | If an optional type object is passed, the list is also restricted to |
| 51 | objects of that type. |
| 52 | The return list itself, and some temp objects created just to call |
| 53 | sys.getobjects(), are excluded from the return list. Note that the |
| 54 | list returned is just another object, though, so may appear in the |
| 55 | return list the next time you call getobjects(); note that every |
| 56 | object in the list is kept alive too, simply by virtue of being in |
| 57 | the list. |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 58 | |
| 59 | envar PYTHONDUMPREFS |
| 60 | If this envar exists, Py_Finalize() arranges to print a list of |
| 61 | all still-live heap objects. |
| 62 | --------------------------------------------------------------------------- |
Tim Peters | 62fc52e | 2002-07-11 00:23:58 +0000 | [diff] [blame] | 63 | PYMALLOC_DEBUG introduced in 2.3 |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 64 | |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 65 | When pymalloc is enabled (WITH_PYMALLOC is defined), calls to the PyObject_ |
| 66 | memory routines are handled by Python's own small-object allocator, while |
| 67 | calls to the PyMem_ memory routines are directed to the system malloc/ |
| 68 | realloc/free. If PYMALLOC_DEBUG is also defined, calls to both PyObject_ |
| 69 | and PyMem_ memory routines are directed to a special debugging mode of |
| 70 | Python's small-object allocator. |
| 71 | |
| 72 | This mode fills dynamically allocated memory blocks with special, |
| 73 | recognizable bit patterns, and adds debugging info on each end of |
| 74 | dynamically allocated memory blocks. The special bit patterns are: |
| 75 | |
| 76 | #define CLEANBYTE 0xCB /* clean (newly allocated) memory */ |
| 77 | #define DEADBYTE 0xDB /* dead (newly freed) memory */ |
| 78 | #define FORBIDDENBYTE 0xFB /* fordidden -- untouchable bytes */ |
| 79 | |
| 80 | Strings of these bytes are unlikely to be valid addresses, floats, or 7-bit |
| 81 | ASCII strings. |
| 82 | |
| 83 | 8 bytes are added at each end of each block of N bytes requested. The |
| 84 | memory layout is like so, where p represents the address returned by a |
Tim Peters | 20c8a04 | 2002-07-11 00:02:52 +0000 | [diff] [blame] | 85 | malloc-like or realloc-like function (p[i:j] means the slice of bytes |
| 86 | from *(p+i) inclusive up to *(p+j) exclusive; note that the treatment |
| 87 | of negative indices differs from a Python slice): |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 88 | |
| 89 | p[-8:-4] |
| 90 | Number of bytes originally asked for. 4-byte unsigned integer, |
| 91 | big-endian (easier to read in a memory dump). |
| 92 | p[-4:0] |
| 93 | Copies of FORBIDDENBYTE. Used to catch under- writes and reads. |
| 94 | p[0:N] |
Tim Peters | 62fc52e | 2002-07-11 00:23:58 +0000 | [diff] [blame] | 95 | The requested memory, filled with copies of CLEANBYTE, used to catch |
| 96 | reference to uninitialized memory. |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 97 | When a realloc-like function is called requesting a larger memory |
| 98 | block, the new excess bytes are also filled with CLEANBYTE. |
| 99 | When a free-like function is called, these are overwritten with |
Tim Peters | 62fc52e | 2002-07-11 00:23:58 +0000 | [diff] [blame] | 100 | DEADBYTE, to catch reference to freed memory. When a realloc- |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 101 | like function is called requesting a smaller memory block, the excess |
| 102 | old bytes are also filled with DEADBYTE. |
| 103 | p[N:N+4] |
| 104 | Copies of FORBIDDENBYTE. Used to catch over- writes and reads. |
| 105 | p[N+4:N+8] |
| 106 | A serial number, incremented by 1 on each call to a malloc-like or |
| 107 | realloc-like function. |
| 108 | 4-byte unsigned integer, big-endian. |
| 109 | If "bad memory" is detected later, the serial number gives an |
| 110 | excellent way to set a breakpoint on the next run, to capture the |
Tim Peters | 20c8a04 | 2002-07-11 00:02:52 +0000 | [diff] [blame] | 111 | instant at which this block was passed out. The static function |
| 112 | bumpserialno() in obmalloc.c is the only place the serial number |
| 113 | is incremented, and exists so you can set such a breakpoint easily. |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 114 | |
Tim Peters | 62fc52e | 2002-07-11 00:23:58 +0000 | [diff] [blame] | 115 | A realloc-like or free-like function first checks that the FORBIDDENBYTEs |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 116 | at each end are intact. If they've been altered, diagnostic output is |
Tim Peters | 62fc52e | 2002-07-11 00:23:58 +0000 | [diff] [blame] | 117 | written to stderr, and the program is aborted via Py_FatalError(). The |
| 118 | other main failure mode is provoking a memory error when a program |
| 119 | reads up one of the special bit patterns and tries to use it as an address. |
| 120 | If you get in a debugger then and look at the object, you're likely |
| 121 | to see that it's entirely filled with 0xDB (meaning freed memory is |
| 122 | getting used) or 0xCB (meaning uninitialized memory is getting used). |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 123 | |
| 124 | Note that PYMALLOC_DEBUG requires WITH_PYMALLOC. |
| 125 | |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 126 | Special gimmicks: |
| 127 | |
| 128 | envar PYTHONMALLOCSTATS |
| 129 | If this envar exists, a report of pymalloc summary statistics is |
| 130 | printed to stderr whenever a new arena is allocated, and also |
| 131 | by Py_Finalize(). |
| 132 | --------------------------------------------------------------------------- |
Tim Peters | 62fc52e | 2002-07-11 00:23:58 +0000 | [diff] [blame] | 133 | Py_DEBUG introduced in 1.5 |
| 134 | named DEBUG before 1.5 |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 135 | |
| 136 | This is what is generally meant by "a debug build" of Python. |
| 137 | |
Michael W. Hudson | a625523 | 2002-07-30 09:49:29 +0000 | [diff] [blame] | 138 | Py_DEBUG implies LLTRACE, Py_REF_DEBUG, Py_TRACE_REFS, and |
| 139 | PYMALLOC_DEBUG (if WITH_PYMALLOC is enabled). In addition, C |
| 140 | assert()s are enabled (via the C way: by not defining NDEBUG), and |
| 141 | some routines do additional sanity checks inside "#ifdef Py_DEBUG" |
| 142 | blocks. |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 143 | --------------------------------------------------------------------------- |
Michael W. Hudson | 202a4b6 | 2002-07-30 15:25:57 +0000 | [diff] [blame] | 144 | COUNT_ALLOCS introduced in 0.9.9 |
| 145 | partly broken in 2.2 and 2.2.1 |
Tim Peters | 48ba649 | 2002-07-09 19:24:54 +0000 | [diff] [blame] | 146 | |
| 147 | Each type object grows three new members: |
| 148 | |
| 149 | /* Number of times an object of this type was allocated. */ |
Guido van Rossum | 0c08864 | 2002-07-11 01:04:32 +0000 | [diff] [blame] | 150 | int tp_allocs; |
Tim Peters | 48ba649 | 2002-07-09 19:24:54 +0000 | [diff] [blame] | 151 | |
| 152 | /* Number of times an object of this type was deallocated. */ |
Guido van Rossum | 0c08864 | 2002-07-11 01:04:32 +0000 | [diff] [blame] | 153 | int tp_frees; |
Tim Peters | 48ba649 | 2002-07-09 19:24:54 +0000 | [diff] [blame] | 154 | |
Guido van Rossum | 0c08864 | 2002-07-11 01:04:32 +0000 | [diff] [blame] | 155 | /* Highwater mark: the maximum value of tp_allocs - tp_frees so |
| 156 | * far; or, IOW, the largest number of objects of this type alive at |
| 157 | * the same time. |
| 158 | */ |
| 159 | int tp_maxalloc; |
Tim Peters | 48ba649 | 2002-07-09 19:24:54 +0000 | [diff] [blame] | 160 | |
| 161 | Allocation and deallocation code keeps these counts up to date. |
| 162 | Py_Finalize() displays a summary of the info returned by sys.getcounts() |
| 163 | (see below), along with assorted other special allocation counts (like |
| 164 | the number of tuple allocations satisfied by a tuple free-list, the number |
| 165 | of 1-character strings allocated, etc). |
| 166 | |
| 167 | Before Python 2.2, type objects were immortal, and the COUNT_ALLOCS |
| 168 | implementation relies on that. As of Python 2.2, heap-allocated type/ |
| 169 | class objects can go away. COUNT_ALLOCS can blow up in 2.2 and 2.2.1 |
| 170 | because of this; this was fixed in 2.2.2. Use of COUNT_ALLOCS makes |
| 171 | all heap-allocated type objects immortal, except for those for which no |
| 172 | object of that type is ever allocated. |
| 173 | |
| 174 | Special gimmicks: |
| 175 | |
| 176 | sys.getcounts() |
| 177 | Return a list of 4-tuples, one entry for each type object for which |
| 178 | at least one object of that type was allocated. Each tuple is of |
| 179 | the form: |
| 180 | |
| 181 | (tp_name, tp_allocs, tp_frees, tp_maxalloc) |
| 182 | |
Tim Peters | 44c1a7b | 2002-07-09 19:27:20 +0000 | [diff] [blame] | 183 | Each distinct type object gets a distinct entry in this list, even |
Tim Peters | 48ba649 | 2002-07-09 19:24:54 +0000 | [diff] [blame] | 184 | if two or more type objects have the same tp_name (in which case |
| 185 | there's no way to distinguish them by looking at this list). The |
| 186 | list is ordered by time of first object allocation: the type object |
| 187 | for which the first allocation of an object of that type occurred |
| 188 | most recently is at the front of the list. |
| 189 | --------------------------------------------------------------------------- |
Michael W. Hudson | 202a4b6 | 2002-07-30 15:25:57 +0000 | [diff] [blame] | 190 | LLTRACE introduced well before 1.0 |
Michael W. Hudson | a625523 | 2002-07-30 09:49:29 +0000 | [diff] [blame] | 191 | |
Michael W. Hudson | 202a4b6 | 2002-07-30 15:25:57 +0000 | [diff] [blame] | 192 | Compile in support of Low Level TRACE-ing of the main interpreter loop. |
Michael W. Hudson | a625523 | 2002-07-30 09:49:29 +0000 | [diff] [blame] | 193 | |
| 194 | When this preprocessor symbol is defined, before eval_frame |
Michael W. Hudson | 202a4b6 | 2002-07-30 15:25:57 +0000 | [diff] [blame] | 195 | (eval_code2 before 2.2) executes a frame's code it checks the frame's |
| 196 | global namespace for a variable "__lltrace__". If such a variable is |
| 197 | found, mounds of information about what the interpreter is doing are |
| 198 | sprayed to stdout, such as every opcode and opcode argument and values |
| 199 | pushed onto and popped off the value stack. |
Michael W. Hudson | a625523 | 2002-07-30 09:49:29 +0000 | [diff] [blame] | 200 | |
| 201 | Not useful very often, but very useful when needed. |
Jeremy Hylton | 985eba5 | 2003-02-05 23:13:00 +0000 | [diff] [blame] | 202 | |
| 203 | --------------------------------------------------------------------------- |
| 204 | CALL_PROFILE introduced for Python 2.3 |
| 205 | |
| 206 | Count the number of function calls executed. |
| 207 | |
| 208 | When this symbol is defined, the ceval mainloop and helper functions |
| 209 | count the number of function calls made. It keeps detailed statistics |
| 210 | about what kind of object was called and whether the call hit any of |
| 211 | the special fast paths in the code. |