Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 1 | This file describes some special Python build types enabled via compile-time |
Paul Ganssle | 5c403b2 | 2019-04-27 14:14:35 -0400 | [diff] [blame] | 2 | preprocessor directives. |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 3 | |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 4 | IMPORTANT: if you want to build a debug-enabled Python, it is recommended that |
| 5 | you use ``./configure --with-pydebug``, rather than the options listed here. |
Antoine Pitrou | 3cbb24d | 2010-12-15 15:47:34 +0000 | [diff] [blame] | 6 | |
| 7 | However, if you wish to define some of these options individually, it is best |
| 8 | to define them in the EXTRA_CFLAGS make variable; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 9 | ``make EXTRA_CFLAGS="-DPy_REF_DEBUG"``. |
Brett Cannon | a267563 | 2005-04-19 20:28:09 +0000 | [diff] [blame] | 10 | |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 11 | |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 12 | Py_REF_DEBUG |
| 13 | ------------ |
| 14 | |
| 15 | Turn on aggregate reference counting. This arranges that extern _Py_RefTotal |
Benjamin Peterson | ff57aef | 2014-03-30 17:16:09 -0400 | [diff] [blame] | 16 | hold a count of all references, the sum of ob_refcnt across all objects. |
| 17 | Passing ``-X showrefcount`` on the command line causes the interactive |
| 18 | interpreter to print the reference count total as well the number of memory |
| 19 | blocks allocated after each statement: |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 20 | |
| 21 | >>> 23 |
| 22 | 23 |
Benjamin Peterson | ff57aef | 2014-03-30 17:16:09 -0400 | [diff] [blame] | 23 | [8288 refs, 14332 blocks] |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 24 | >>> |
| 25 | |
| 26 | Note that if this count increases when you're not storing away new objects, |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 27 | there's probably a leak. Remember, though, that in interactive mode the special |
| 28 | name "_" holds a reference to the last result displayed! |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 29 | |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 30 | Py_REF_DEBUG also checks after every decref to verify that the refcount hasn't |
| 31 | gone negative, and causes an immediate fatal error if it has. |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 32 | |
Victor Stinner | 235e7b2 | 2019-04-25 01:46:31 +0200 | [diff] [blame] | 33 | Py_DEBUG implies Py_REF_DEBUG. |
| 34 | |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 35 | Special gimmicks: |
| 36 | |
| 37 | sys.gettotalrefcount() |
| 38 | Return current total of all refcounts. |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 39 | |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 40 | |
| 41 | Py_TRACE_REFS |
| 42 | ------------- |
| 43 | |
Victor Stinner | 235e7b2 | 2019-04-25 01:46:31 +0200 | [diff] [blame] | 44 | Build option: ``./configure --with-trace-refs``. |
| 45 | |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 46 | Turn on heavy reference debugging. This is major surgery. Every PyObject grows |
| 47 | two more pointers, to maintain a doubly-linked list of all live heap-allocated |
| 48 | objects. Most built-in type objects are not in this list, as they're statically |
Victor Stinner | c6e5c11 | 2020-02-03 15:17:15 +0100 | [diff] [blame] | 49 | allocated. |
Tim Peters | 78be799 | 2003-03-23 02:51:01 +0000 | [diff] [blame] | 50 | |
| 51 | Note that because the fundamental PyObject layout changes, Python modules |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 52 | compiled with Py_TRACE_REFS are incompatible with modules compiled without it. |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 53 | |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 54 | Special gimmicks: |
| 55 | |
| 56 | sys.getobjects(max[, type]) |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 57 | Return list of the (no more than) max most-recently allocated objects, most |
| 58 | recently allocated first in the list, least-recently allocated last in the |
| 59 | list. max=0 means no limit on list length. If an optional type object is |
| 60 | passed, the list is also restricted to objects of that type. The return |
| 61 | list itself, and some temp objects created just to call sys.getobjects(), |
| 62 | are excluded from the return list. Note that the list returned is just |
| 63 | another object, though, so may appear in the return list the next time you |
| 64 | call getobjects(); note that every object in the list is kept alive too, |
| 65 | simply by virtue of being in the list. |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 66 | |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 67 | envvar PYTHONDUMPREFS |
Martin Panter | b4ce1fc | 2015-11-30 03:18:29 +0000 | [diff] [blame] | 68 | If this envvar exists, Py_FinalizeEx() arranges to print a list of all |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 69 | still-live heap objects. This is printed twice, in different formats, |
Martin Panter | b4ce1fc | 2015-11-30 03:18:29 +0000 | [diff] [blame] | 70 | before and after Py_FinalizeEx has cleaned up everything it can clean up. The |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 71 | first output block produces the repr() of each object so is more |
| 72 | informative; however, a lot of stuff destined to die is still alive then. |
| 73 | The second output block is much harder to work with (repr() can't be invoked |
| 74 | anymore -- the interpreter has been torn down too far), but doesn't list any |
| 75 | objects that will die. The tool script combinerefs.py can be run over this |
| 76 | to combine the info from both output blocks. The second output block, and |
Tim Peters | 21d7d4d | 2003-04-18 00:45:59 +0000 | [diff] [blame] | 77 | combinerefs.py, were new in Python 2.3b1. |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 78 | |
| 79 | |
| 80 | PYMALLOC_DEBUG |
| 81 | -------------- |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 82 | |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 83 | When pymalloc is enabled (WITH_PYMALLOC is defined), calls to the PyObject_ |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 84 | memory routines are handled by Python's own small-object allocator, while calls |
| 85 | to the PyMem_ memory routines are directed to the system malloc/ realloc/free. |
| 86 | If PYMALLOC_DEBUG is also defined, calls to both PyObject_ and PyMem_ memory |
| 87 | routines are directed to a special debugging mode of Python's small-object |
| 88 | allocator. |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 89 | |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 90 | This mode fills dynamically allocated memory blocks with special, recognizable |
| 91 | bit patterns, and adds debugging info on each end of dynamically allocated |
| 92 | memory blocks. The special bit patterns are: |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 93 | |
| 94 | #define CLEANBYTE 0xCB /* clean (newly allocated) memory */ |
| 95 | #define DEADBYTE 0xDB /* dead (newly freed) memory */ |
Thomas Wouters | 89f507f | 2006-12-13 04:49:30 +0000 | [diff] [blame] | 96 | #define FORBIDDENBYTE 0xFB /* forbidden -- untouchable bytes */ |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 97 | |
| 98 | Strings of these bytes are unlikely to be valid addresses, floats, or 7-bit |
| 99 | ASCII strings. |
| 100 | |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 101 | Let S = sizeof(size_t). 2*S bytes are added at each end of each block of N bytes |
| 102 | requested. The memory layout is like so, where p represents the address |
| 103 | returned by a malloc-like or realloc-like function (p[i:j] means the slice of |
| 104 | bytes from *(p+i) inclusive up to *(p+j) exclusive; note that the treatment of |
| 105 | negative indices differs from a Python slice): |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 106 | |
Thomas Wouters | 73e5a5b | 2006-06-08 15:35:45 +0000 | [diff] [blame] | 107 | p[-2*S:-S] |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 108 | Number of bytes originally asked for. This is a size_t, big-endian (easier |
| 109 | to read in a memory dump). |
Tim Peters | df099f5 | 2013-09-19 21:06:37 -0500 | [diff] [blame] | 110 | p[-S] |
| 111 | API ID. See PEP 445. This is a character, but seems undocumented. |
| 112 | p[-S+1:0] |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 113 | Copies of FORBIDDENBYTE. Used to catch under- writes and reads. |
| 114 | p[0:N] |
Tim Peters | 62fc52e | 2002-07-11 00:23:58 +0000 | [diff] [blame] | 115 | The requested memory, filled with copies of CLEANBYTE, used to catch |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 116 | reference to uninitialized memory. When a realloc-like function is called |
| 117 | requesting a larger memory block, the new excess bytes are also filled with |
| 118 | CLEANBYTE. When a free-like function is called, these are overwritten with |
| 119 | DEADBYTE, to catch reference to freed memory. When a realloc- like function |
| 120 | is called requesting a smaller memory block, the excess old bytes are also |
| 121 | filled with DEADBYTE. |
Thomas Wouters | 73e5a5b | 2006-06-08 15:35:45 +0000 | [diff] [blame] | 122 | p[N:N+S] |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 123 | Copies of FORBIDDENBYTE. Used to catch over- writes and reads. |
Thomas Wouters | 73e5a5b | 2006-06-08 15:35:45 +0000 | [diff] [blame] | 124 | p[N+S:N+2*S] |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 125 | A serial number, incremented by 1 on each call to a malloc-like or |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 126 | realloc-like function. Big-endian size_t. If "bad memory" is detected |
| 127 | later, the serial number gives an excellent way to set a breakpoint on the |
| 128 | next run, to capture the instant at which this block was passed out. The |
| 129 | static function bumpserialno() in obmalloc.c is the only place the serial |
| 130 | number is incremented, and exists so you can set such a breakpoint easily. |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 131 | |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 132 | A realloc-like or free-like function first checks that the FORBIDDENBYTEs at |
| 133 | each end are intact. If they've been altered, diagnostic output is written to |
| 134 | stderr, and the program is aborted via Py_FatalError(). The other main failure |
| 135 | mode is provoking a memory error when a program reads up one of the special bit |
| 136 | patterns and tries to use it as an address. If you get in a debugger then and |
| 137 | look at the object, you're likely to see that it's entirely filled with 0xDB |
| 138 | (meaning freed memory is getting used) or 0xCB (meaning uninitialized memory is |
| 139 | getting used). |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 140 | |
Victor Stinner | 235e7b2 | 2019-04-25 01:46:31 +0200 | [diff] [blame] | 141 | Note that PYMALLOC_DEBUG requires WITH_PYMALLOC. Py_DEBUG implies |
| 142 | PYMALLOC_DEBUG (if WITH_PYMALLOC is enabled). |
Tim Peters | 889f61d | 2002-07-10 19:29:49 +0000 | [diff] [blame] | 143 | |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 144 | Special gimmicks: |
| 145 | |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 146 | envvar PYTHONMALLOCSTATS |
| 147 | If this envvar exists, a report of pymalloc summary statistics is printed to |
Martin Panter | b4ce1fc | 2015-11-30 03:18:29 +0000 | [diff] [blame] | 148 | stderr whenever a new arena is allocated, and also by Py_FinalizeEx(). |
Thomas Wouters | 73e5a5b | 2006-06-08 15:35:45 +0000 | [diff] [blame] | 149 | |
| 150 | Changed in 2.5: The number of extra bytes allocated is 4*sizeof(size_t). |
| 151 | Before it was 16 on all boxes, reflecting that Python couldn't make use of |
| 152 | allocations >= 2**32 bytes even on 64-bit boxes before 2.5. |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 153 | |
| 154 | |
| 155 | Py_DEBUG |
| 156 | -------- |
Tim Peters | 6045d48 | 2002-07-09 18:35:34 +0000 | [diff] [blame] | 157 | |
| 158 | This is what is generally meant by "a debug build" of Python. |
| 159 | |
Victor Stinner | 235e7b2 | 2019-04-25 01:46:31 +0200 | [diff] [blame] | 160 | Py_DEBUG implies LLTRACE, Py_REF_DEBUG, and PYMALLOC_DEBUG (if |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 161 | WITH_PYMALLOC is enabled). In addition, C assert()s are enabled (via the C way: |
| 162 | by not defining NDEBUG), and some routines do additional sanity checks inside |
| 163 | "#ifdef Py_DEBUG" blocks. |
| 164 | |
| 165 | |
Georg Brandl | c166076 | 2010-12-28 11:38:12 +0000 | [diff] [blame] | 166 | LLTRACE |
| 167 | ------- |
Michael W. Hudson | a625523 | 2002-07-30 09:49:29 +0000 | [diff] [blame] | 168 | |
Michael W. Hudson | 46e6d92 | 2005-01-18 15:53:59 +0000 | [diff] [blame] | 169 | Compile in support for Low Level TRACE-ing of the main interpreter loop. |
Michael W. Hudson | a625523 | 2002-07-30 09:49:29 +0000 | [diff] [blame] | 170 | |
Benjamin Peterson | 6e820c0 | 2015-03-01 20:59:22 -0500 | [diff] [blame] | 171 | When this preprocessor symbol is defined, before PyEval_EvalFrame executes a |
| 172 | frame's code it checks the frame's global namespace for a variable |
Danish Prakash | 09efe49 | 2018-08-25 07:30:25 +0530 | [diff] [blame] | 173 | "__ltrace__". If such a variable is found, mounds of information about what |
Benjamin Peterson | 6e820c0 | 2015-03-01 20:59:22 -0500 | [diff] [blame] | 174 | the interpreter is doing are sprayed to stdout, such as every opcode and opcode |
| 175 | argument and values pushed onto and popped off the value stack. |
Michael W. Hudson | a625523 | 2002-07-30 09:49:29 +0000 | [diff] [blame] | 176 | |
| 177 | Not useful very often, but very useful when needed. |
Victor Stinner | 235e7b2 | 2019-04-25 01:46:31 +0200 | [diff] [blame] | 178 | |
| 179 | Py_DEBUG implies LLTRACE. |