3 ;;;; tags which are set during the build process and which end up in
4 ;;;; CL:*FEATURES* in the target SBCL, plus some comments about other
5 ;;;; CL:*FEATURES* tags which have special meaning to SBCL or which
6 ;;;; have a special conventional meaning
8 ;;;; Note that the recommended way to customize the features of a
9 ;;;; local build of SBCL is not to edit this file, but instead to
10 ;;;; tweak customize-target-features.lisp. (You must create this file
11 ;;;; first; it is not in the SBCL distribution, and is in fact
12 ;;;; explicitly excluded from the distribution in places like
13 ;;;; .cvsignore.) If you define a function in
14 ;;;; customize-target-features.lisp, it will be used to transform the
15 ;;;; target features list after it's read and before it's used. E.g.,
16 ;;;; you can use code like this:
18 ;;;; (flet ((enable (x) (pushnew x list))
19 ;;;; (disable (x) (setf list (remove x list))))
20 ;;;; #+nil (enable :sb-show)
21 ;;;; (enable :sb-after-xc-core)
22 ;;;; #+nil (disable :sb-doc)
24 ;;;; By thus editing a local file (one which is not in the source
25 ;;;; distribution, and which is in .cvsignore) your customizations
26 ;;;; will remain local even if you do things like "cvs update",
27 ;;;; will not show up if you try to submit a patch with "cvs diff",
28 ;;;; and might even stay out of the way if you use other non-CVS-based
29 ;;;; methods to upgrade the files or store your configuration.
31 ;;;; This software is part of the SBCL system. See the README file for
32 ;;;; more information.
34 ;;;; This software is derived from the CMU CL system, which was
35 ;;;; written at Carnegie Mellon University and released into the
36 ;;;; public domain. The software is in the public domain and is
37 ;;;; provided with absolutely no warranty. See the COPYING and CREDITS
38 ;;;; files for more information.
42 ;; features present in all builds
47 ;; FIXME: Isn't there a :x3jsomething feature which we should set too?
48 ;; No. CLHS says ":x3j13 [...] A conforming implementation might or
49 ;; might not contain such a feature." -- CSR, 2002-02-21
54 ;; Douglas Thomas Crosher's conservative generational GC (the only one
55 ;; we currently support for X86).
56 ;; :gencgc used to be here; CSR moved it into
57 ;; local-target-features.lisp-expr via make-config.sh, as alpha,
58 ;; sparc and ppc ports don't currently support it. -- CSR, 2002-02-21
61 ;; features present in this particular build
64 ;; Setting this enables the compilation of documentation strings
65 ;; from the system sources into the target Lisp executable.
66 ;; Traditional Common Lisp folk will want this option set.
67 ;; I (WHN) made it optional because I came to Common Lisp from
68 ;; C++ through Scheme, so I'm accustomed to asking
69 ;; Emacs about things that I'm curious about instead of asking
70 ;; the executable I'm running.
73 ;; Make more debugging information available (for debugging SBCL
74 ;; itself). If you aren't hacking or troubleshooting SBCL itself,
75 ;; you probably don't want this set.
77 ;; At least two varieties of debugging information are enabled by this
79 ;; * SBCL is compiled with a higher level of OPTIMIZE DEBUG, so that
80 ;; the debugger can tell more about the state of the system.
81 ;; * Various code to print debugging messages, and similar debugging code,
82 ;; is compiled only when this feature is present.
84 ;; Note that the extra information recorded by the compiler at
85 ;; this higher level of OPTIMIZE DEBUG includes the source location
86 ;; forms. In order for the debugger to use this information, it has to
87 ;; re-READ the source file. In an ordinary installation of SBCL, this
88 ;; re-READing may not work very well, for either of two reasons:
89 ;; * The sources aren't present on the system in the same location that
90 ;; they were on the system where SBCL was compiled.
91 ;; * SBCL is using the standard readtable, without the added hackage
92 ;; which allows it to handle things like target features.
93 ;; If you want to be able to use the extra debugging information,
94 ;; therefore, be sure to keep the sources around, and run with the
95 ;; readtable configured so that the system sources can be read.
98 ;; Enable the low level debugger, "ldb", by default. In the ideal
99 ;; world you would not need this unless you are messing with SBCL at
100 ;; a very low level (e.g., trying to diagnose GC problems, or trying
101 ;; to debug assembly code for a port to a new CPU). However,
102 ;; experience shows that sooner or later everyone lose()'s, in which
103 ;; case SB-LDB can at least provide an informative backtrace.
106 ;; This isn't really a target Lisp feature at all, but controls
107 ;; whether the build process produces an after-xc.core file. This
108 ;; can be useful for shortening the edit/compile/debug cycle when
109 ;; you modify SBCL's own source code, as in slam.sh. Otherwise
110 ;; you don't need it.
113 ;; Enable extra debugging output in the assem.lisp assembler/scheduler
114 ;; code. (This is the feature which was called :DEBUG in the
115 ;; original CMU CL code.)
118 ;; Compile the C runtime with support for low-level debugging output
119 ;; through FSHOW and FSHOW_SIGNAL. If enabled, this feature allows
120 ;; users to turn on such debugging output using environment variables at
124 ;; Setting this makes SBCL more "fluid", i.e. more amenable to
125 ;; modification at runtime, by suppressing various INLINE declarations,
126 ;; compiler macro definitions, FREEZE-TYPE declarations; and by
127 ;; suppressing various burning-our-ships-behind-us actions after
128 ;; initialization is complete; and so forth. This tends to clobber the
129 ;; performance of the system, so unless you have some special need for
130 ;; this when hacking SBCL itself, you don't want this set.
133 ;; Enable code for collecting statistics on usage of various operations,
134 ;; useful for performance tuning of the SBCL system itself. This code
135 ;; is probably pretty stale (having not been tested since the fork from
136 ;; base CMU CL) but might nonetheless be a useful starting point for
137 ;; anyone who wants to collect such statistics in the future.
140 ;; Enable code for detecting concurrent accesses to the same hash-table
141 ;; in multiple threads. Note that this implementation is currently
142 ;; (2007-09-11) somewhat too eager: even though in the current implementation
143 ;; multiple readers are thread safe as long as there are no writers, this
144 ;; code will also trap multiple readers.
145 ; :sb-hash-table-debug
147 ;; Enabled automatically by make-config.sh for platforms which implement
148 ;; short vector SIMD intrinsics.
152 ;; Enabled automatically by make-config.sh for platforms which implement
153 ;; the %READ-CYCLE-COUNTER VOP. Can be disabled manually: affects TIME.
155 ;; FIXME: Should this be :SB-CYCLE-COUNTER instead? If so, then the same goes
156 ;; for :COMPARE-AND-SWAP-VOPS as well, and a bunch of others. Perhaps
157 ;; built-time convenience features like this should all live in eg. SB!INT
162 ;; Allow the main dynamic space to be allocated anywhere
163 ;; by the OS kernel regardless of the configuration-time value
164 ;; of dynamic-space-start.
165 ;; The downside is the potential need for startup-time relocation.
166 ;; Enabled by default if supported
169 ;; Build with support for an additional dynamic heap
170 ;; differing from the main dynamic heap in two ways:
171 ;; 1. it is guaranteed to reside below 4GB so that all pointers
172 ;; into it fit in 32 bits. (Only an issue for >32 bit address space)
173 ;; 2. all objects therein are immovable, and space is reclaimed
174 ;; by a mark-and-sweep collector.
175 ;; That combination of aspects potentially allows various efficiencies
176 ;; in code generation, especially for the x86-64 backend.
177 ;; The extra space has a fixed size which can only be changed by a rebuild,
178 ;; and out-of-space conditions are not easily preventable, so the space
179 ;; is sized rather generously to sidestep the issue.
180 ;; Additionally, it is assumed that for all objects in the immobile heap,
181 ;; speed of allocation of those objects is relatively unimportant.
182 ;; If unexpected performance regressions are observed,
183 ;; consider disabling this feature and reporting a bug.
186 ;; Allocate most functions in the immobile space.
187 ;; Enabled by default if supported.
188 ;; The down-side of this feature is that the allocator is significantly
189 ;; slower than the allocator for movable code. If a particular application
190 ;; is performance-constrained by speed of creation of compiled functions
191 ;; (not including closures), the feature can be disabled.
194 ;; Combine the layout pointer, instance-length, and widetag of INSTANCE
195 ;; into a single machine word. This represents a space savings of anywhere
196 ;; from 4% to 8% in typical applications. (Your mileage may vary).
197 ; :compact-instance-header
199 ;; Enabled automatically for platforms which implement complex arithmetic
200 ;; VOPs. Such platforms should implement real-complex, complex-real and
201 ;; complex-complex addition and subtractions (for complex-single-float
202 ;; and complex-double-float). They should also also implement complex-real
203 ;; and real-complex multiplication, complex-real division, and
204 ;; sb!vm::swap-complex, which swaps the real and imaginary parts.
205 ;; Finally, they should implement conjugate and complex-real, real-complex
206 ;; and complex-complex CL:= (complex-complex EQL would usually be a good
209 ; :complex-float-vops
211 ;; Enabled automatically for platforms which implement VOPs for EQL
212 ;; of single and double floats.
216 ;; Enabled automatically for platform that can implement inline constants.
218 ;; Such platform must implement 5 functions, in SB!VM:
219 ;; * canonicalize-inline-constant: converts a constant descriptor (list) into
220 ;; a canonical description, to be used as a key in an EQUAL hash table
221 ;; and to guide the generation of the constant itself.
222 ;; * inline-constant-value: given a canonical constant descriptor, computes
224 ;; 1. A label that will be used to emit the constant (usually a
226 ;; 2. A value that will be returned to code generators referring to
227 ;; the constant (on x86oids, an EA object)
228 ;; * sort-inline-constants: Receives a vector of unique constants;
229 ;; the car of each entry is the constant descriptor, and the cdr the
230 ;; corresponding label. Destructively returns a vector of constants
231 ;; sorted in emission order. It could actually perform arbitrary
232 ;; modifications to the vector, e.g. to fuse constants of different
234 ;; * emit-inline-constant: receives a constant descriptor and its associated
235 ;; label. Emits the constant.
237 ;; Implementing this features lets VOP generators use sb!c:register-inline-constant
238 ;; to get handles (as returned by sb!vm:inline-constant-value) from constant
243 ;; Peter Van Eynde's increase-bulletproofness code for CMU CL
245 ;; Some of the code which was #+high-security before the fork has now
246 ;; been either made unconditional, deleted, or rewritten into
247 ;; unrecognizability, but some remains. What remains is not maintained
248 ;; or tested in current SBCL, but I haven't gone out of my way to
252 ; :high-security-support
254 ;; low-level thread primitives support
256 ;; As of SBCL 1.0.33.26, threads are part of the default build on
257 ;; x86oid Linux. Other platforms that support them include
258 ;; x86oid Darwin, FreeBSD, and Solaris.
263 ;; While on linux we are able to use futexes for our locking
264 ;; primitive, on other platforms we don't have this luxury.
268 ;; On some operating systems the FS segment register (used for SBCL's
269 ;; thread local storage) is not reliably preserved in signal
270 ;; handlers, so we need to restore its value from the pthread thread
272 ; :restore-fs-segment-register-from-tls
274 ;; On some x86oid operating systems (darwin) SIGTRAP is not reliably
275 ;; delivered for the INT3 instruction, so we use the UD2 instruction
276 ;; which generates SIGILL instead.
279 ;; Support for detection of unportable code (when applied to the
280 ;; COMMON-LISP package, or SBCL-internal pacakges) or bad-neighbourly
281 ;; code (when applied to user-level packages), relating to material
282 ;; alteration to packages or to bindings in symbols in packages.
285 ;; Support for the entirety of the 21-bit character space defined by
286 ;; the Unicode consortium, rather than the classical 8-bit ISO-8859-1
290 ;; Support for a full evaluator that can execute all the CL special
291 ;; forms, as opposed to the traditional SBCL evaluator which called
292 ;; COMPILE for everything complicated.
294 ;; Support for a different evaluator (interpreter) with improved performance.
295 ;; You can't have both.
298 ;; Record source location information for variables, classes, conditions,
299 ;; packages, etc. Gives much better information on M-. in Slime, but
300 ;; increases core size by about 100kB.
303 ;; Record xref data for SBCL internals. This can be rather useful for
304 ;; people who want to develop on SBCL itself because it'll make M-?
305 ;; (slime-edit-uses) work which lists call/expansion/etc. sites.
306 ;; It'll increase the core size by major 5-6mB, though.
307 ; :sb-xref-for-internals
309 ;; We support package local nicknames. No :sb-prefix here as we vainly
310 ;; believe our API is worth copying to other implementations as well.
311 ;; This doesn't affect the build at all, merely declares how things are.
312 :package-local-nicknames
314 ;; This affects the definition of a lot of things in bignum.lisp. It
315 ;; doesn't seem to be documented anywhere what systems it might apply
316 ;; to. It doesn't seem to be needed for X86 systems anyway.
319 ;; This is set in classic CMU CL, and presumably there it means
320 ;; that the floating point arithmetic implementation
321 ;; conforms to IEEE's standard. Here it definitely means that the
322 ;; floating point arithmetic implementation conforms to IEEE's standard.
323 ;; I (WHN 19990702) haven't tried to verify
324 ;; that it does conform, but it should at least mostly conform (because
325 ;; the underlying x86 hardware tries).
328 ;; CMU CL had, and we inherited, code to support 80-bit LONG-FLOAT on the x86
329 ;; architecture. Nothing has been done to actively destroy the long float
330 ;; support, but it hasn't been thoroughly maintained, and needs at least
331 ;; some maintenance before it will work. (E.g. the LONG-FLOAT-only parts of
332 ;; genesis are still implemented in terms of unportable CMU CL functions
333 ;; which are not longer available at genesis time in SBCL.) A deeper
334 ;; problem is SBCL's bootstrap process implicitly assumes that the
335 ;; cross-compilation host will be able to make the same distinctions
336 ;; between floating point types that it does. This assumption is
337 ;; fundamentally sleazy, even though in practice it's unlikely to break down
338 ;; w.r.t. distinguishing SINGLE-FLOAT from DOUBLE-FLOAT; it's much more
339 ;; likely to break down w.r.t. distinguishing DOUBLE-FLOAT from LONG-FLOAT.
340 ;; Still it's likely to be quite doable to get LONG-FLOAT support working
341 ;; again, if anyone's sufficiently motivated.
344 ;; Some platforms don't use a 32-bit off_t by default, and thus can't
345 ;; handle files larger than 2GB. This feature will control whether
346 ;; we'll try to use platform-specific compilation options to enable a
347 ;; 64-bit off_t. The intent is for this feature to be automatically
348 ;; enabled by make-config.sh on platforms where it's needed and known
349 ;; to work, you shouldn't be enabling it manually. You might however
350 ;; want to disable it, if you need to pass file descriptors to
351 ;; foreign code that uses a 32-bit off_t.
354 ;; Enabled automatically on platforms that have VOPs to compute the
355 ;; high half of a full word-by-word multiplication. When disabled,
356 ;; SB-KERNEL:%MULTIPLY-HIGH is implemented in terms of
357 ;; SB-BIGNUM:%MULTIPLY.
358 ; :multiply-high-vops
360 ;; SBCL has optional support for zlib-based compressed core files. Enable
361 ;; this feature to compile it in. Obviously, doing so adds a dependency
363 ; :sb-core-compression
365 ;; On certain thread-enabled platforms, synchronization between threads
366 ;; for the purpose of stopping and starting the world around GC can be
367 ;; performed using safepoints instead of signals. Enable this feature
368 ;; to compile with safepoints and to use them for GC.
369 ;; (Replaces use of SIG_STOP_FOR_GC.)
372 ;; When compiling with safepoints, the INTERRUPT-THREAD mechanism can
373 ;; also use safepoints to roll the target thread to a point at which it
374 ;; can be interrupted safely, instead of using a signal for this
375 ;; purpose. Enable this feature in addition to :SB-SAFEPOINT to enable
377 ;; (Replaces use of SIGPIPE, except to wake up syscalls.)
380 ;; When compiling with safepoints and thruptions, the TIMER facility
381 ;; can replace its use of setitimer with a background thread.
382 ;; (Replaces use of SIGALRM.)
385 ;; This platform implements VOPs for %ash/right, variable-width shift right
389 ;; miscellaneous notes on other things which could have special significance
390 ;; in the *FEATURES* list
393 ;; Any target feature which affects binary compatibility of fasl files
394 ;; needs to be recorded in *FEATURES-POTENTIALLY-AFFECTING-FASL-FORMAT*
397 ;; notes on the :NIL and :IGNORE features:
399 ;; #+NIL is used to comment out forms. Occasionally #+IGNORE is used
400 ;; for this too. So don't use :NIL or :IGNORE as the names of features..
402 ;; notes on :SB-XC and :SB-XC-HOST features (which aren't controlled by this
403 ;; file, but are instead temporarily pushed onto *FEATURES* or
404 ;; *TARGET-FEATURES* during some phases of cross-compilation):
406 ;; :SB-XC-HOST stands for "cross-compilation host" and is in *FEATURES*
407 ;; during the first phase of cross-compilation bootstrapping, when the
408 ;; host Lisp is being used to compile the cross-compiler.
410 ;; :SB-XC stands for "cross compiler", and is in *FEATURES* during the second
411 ;; phase of cross-compilation bootstrapping, when the cross-compiler is
412 ;; being used to create the first target Lisp.
414 ;; notes on the :SB-ASSEMBLING feature (which isn't controlled by
417 ;; This is a flag for whether we're in the assembler. It's
418 ;; temporarily pushed onto the *FEATURES* list in the setup for
419 ;; the ASSEMBLE-FILE function. It would be a bad idea
420 ;; to use it as a name for a permanent feature.
422 ;; notes on local features (which are set automatically by the
423 ;; configuration script, and should not be set here unless you
424 ;; really, really know what you're doing):
426 ;; machine architecture features:
428 ;; any Intel 386 or better, or compatibles like the AMD K6 or K7
430 ;; any x86-64 CPU running in 64-bit mode
432 ;; DEC/Compaq Alpha CPU
434 ;; any Sun UltraSPARC (possibly also non-Ultras -- currently untested)
440 ;; any MIPS CPU (in little-endian mode with :little-endian)
442 ;; an ARM CPU (details yet to be determined)
444 ;; an ARMv8 AArch64 CPU
445 ;; (CMU CL also had a :pentium feature, which affected the definition
446 ;; of some floating point vops. It was present but not enabled or
447 ;; documented in the CMU CL code that SBCL is derived from, and has
448 ;; now been moved to the backend-subfeatures mechanism.)
450 ;; properties derived from the machine architecture
453 ;; means (= sb-vm:n-word-bits 64) currently true for x86-64 and arm64
456 ;; means (= sb-vm:n-machine-word-bits 64) currently true for alpha,
457 ;; arm64, and x86-64.
459 ;; :control-stack-grows-downward-not-upward
460 ;; On the X86, the Lisp control stack grows downward. On the
461 ;; other supported CPU architectures as of sbcl-0.7.1.40, the
462 ;; system stack grows upward.
463 ;; Note that there are other stack-related differences between the
464 ;; X86 port and the other ports. E.g. on the X86, the Lisp control
465 ;; stack coincides with the C stack, meaning that on the X86 there's
466 ;; stuff on the control stack that the Lisp-level debugger doesn't
467 ;; understand very well. As of sbcl-0.7.1.40 things like that are
468 ;; just parameterized by #!+X86, but it'd probably be better to
469 ;; use new flags like :CONTROL-STACK-CONTAINS-C-STACK.
471 ;; :stack-allocatable-closures
472 ;; The compiler can allocate dynamic-extent closures on stack.
475 ;; Alien callbacks have been implemented for this platform.
477 ;; :compare-and-swap-vops
478 ;; The backend implements compare-and-swap VOPs.
480 ;; :memory-barrier-vops
481 ;; Memory barriers (for multi-threaded synchronization) have been
482 ;; implemented for this platform.
484 ;; operating system features:
485 ;; :unix = We're intended to run under some Unix-like OS. (This is not
486 ;; exclusive with the features which indicate which particular
487 ;; Unix-like OS we're intended to run under.)
488 ;; :linux = We're intended to run under some version of Linux.
489 ;; :bsd = We're intended to run under some version of BSD Unix. (This
490 ;; is not exclusive with the features which indicate which
491 ;; particular version of BSD we're intended to run under.)
492 ;; :freebsd = We're intended to run under FreeBSD.
493 ;; :openbsd = We're intended to run under OpenBSD.
494 ;; :netbsd = We're intended to run under NetBSD.
495 ;; :dragonfly = We're intended to run under DragonFly BSD.
496 ;; :darwin = We're intended to run under Darwin (including MacOS X).
497 ;; :sunos = We're intended to run under Solaris user environment
498 ;; with the SunOS kernel.
499 ;; :hpux = We're intended to run under HP-UX 11.11 or later
500 ;; :osf1 = We're intended to run under Tru64 (aka Digital Unix
502 ;; :win32 = We're intended to under some version of Microsoft Windows.
503 ;; (No others are supported by SBCL as of 1.0.8, but :hpux or :irix
504 ;; support could be ported from CMU CL if anyone is sufficiently
505 ;; motivated to do so.)