Matt Birkholz [Thu, 8 May 2014 04:22:58 +0000 (21:22 -0700)]
ffi: Punt AUXDIR from shim-config.scm.
Procedures like compile-shim need to use the first directory in
MITSCHEME_LIBRARY_PATH (not the installation directory) so that the
FFI test can override it, getting the not-yet-installed configuration
from the build directory.
Symlink shim-config.scm to src/lib/ and generate it during `make all',
not `make install', so that it is available to the test FFI (during
`make check').
Fix FFI test NOT to use with-notification around `make', which cannot
use fresh-line. Just write whole lines to notification-output-port.
Matt Birkholz [Wed, 7 May 2014 06:19:42 +0000 (23:19 -0700)]
configure: Move check for __x86_64__.
Move AC_CHECK_DECL(__x86_64__) before AC_MSG_CHECKING([for native-code
support]) so that the former's progress note does not arrive in the
middle of the latter's, like this:
checking for native-code support... checking whether __x86_64__ is declared... no
yes, for i386
Matt Birkholz [Wed, 7 May 2014 04:11:25 +0000 (21:11 -0700)]
ffi: Add install-html using documentation build parameters.
Install parameters in infodir/mit-scheme-doc-config.scm.
Matt Birkholz [Wed, 7 May 2014 03:36:12 +0000 (20:36 -0700)]
ffi: Use Makefile to save shim build parameters.
Install them in AUXDIR/shim-config.scm and use them in the
compile-shim, link-shim, install-shim and install-load-option
procedures.
Matt Birkholz [Tue, 6 May 2014 02:26:44 +0000 (19:26 -0700)]
Fix continue_from_trap to NOT open a FREE_PARANOIA_MARGIN hole.
Just reset Free to heap_alloc_limit if its sanity check fails.
Matt Birkholz [Tue, 6 May 2014 02:23:06 +0000 (19:23 -0700)]
ffi: Update Makefile for prhello example.
Use the generate-shim, compile-shim, link-shim and install-shim
procedures. Move to a separate file included via @verbatiminclude,
for more convenient testing.
Matt Birkholz [Tue, 6 May 2014 01:51:40 +0000 (18:51 -0700)]
Add FFI to doc/info-dir and reformat.
Renamed the dir entries for consistency, and so that they sort
together in a large category like "Programming Languages". Updated
the @direntry in each toplevel .texinfo to match.
Taylor R Campbell [Sun, 27 Apr 2014 03:17:43 +0000 (03:17 +0000)]
Add portable IEEE 754 format utilities.
Not hooked up to anything, but these have been floating around my
disk for months and were likely to get lost. Feel free to hook these
up and start using them. If you do, you should start by writing some
automatic tests.
Taylor R Campbell [Sat, 26 Apr 2014 21:00:01 +0000 (21:00 +0000)]
Correct types in BIGNUM_REDUCE_LENGTH and use output parameter as such.
Taylor R Campbell [Sat, 26 Apr 2014 20:58:24 +0000 (20:58 +0000)]
Make definition of VERIFY-HEAP primitive unconditional.
Define a verify_heap stub that always returns true if
!ENABLE_DEBUGGING_TOOLS.
Matt Birkholz [Sat, 26 Apr 2014 15:00:13 +0000 (08:00 -0700)]
Fix string-head! to hide the unused (non-marked) words.
This quiets 3 of the 17 complaints from verify_heap during `make
check'. The remaining 14 appear to be caused by SIGFPE recovery.
Matt Birkholz [Sat, 26 Apr 2014 04:10:43 +0000 (21:10 -0700)]
Fix BIGNUM_REDUCE_LENGTH to hide the unused (non-marked) words.
This quiets 7093 of the 7110 complaints from verify_heap during the
tests.
Matt Birkholz [Fri, 25 Apr 2014 23:47:06 +0000 (16:47 -0700)]
Call verify_heap before GC to scan for bogosities.
Trap recovery (on Unix) occasionally fails the assert in
linkage_section_type when its scan of the heap encounters non-marked
words that are not "hidden" by a non-marked vector header.
The new verify_heap function scans the heap in the same way, and
complains on stderr about any invalid linkage section types, object
types, etc. If the GC_Debug flag is set, verify_heap is called by the
garbage-collect primitive to scan maximal heap.
The new dump_heap_at function can be used to inspect the heap around a
reported bogosity.
Taylor R Campbell [Sun, 6 Apr 2014 22:10:20 +0000 (22:10 +0000)]
Fix technically incorrect malloc size in re_match. Omit needless cast.
Taylor R Campbell [Sun, 6 Apr 2014 22:09:47 +0000 (22:09 +0000)]
Nix dead initialization in scan_ephemerons.
Taylor R Campbell [Sun, 6 Apr 2014 22:09:32 +0000 (22:09 +0000)]
Mark scan unused in non-debug builds.
Taylor R Campbell [Sun, 6 Apr 2014 22:08:53 +0000 (22:08 +0000)]
Make (cd "/nonexistent") fail.
Kinda kludgey, but it'll do for interactive use.
See XXX comment for what a proper solution should look like.
Taylor R Campbell [Sun, 16 Mar 2014 17:30:55 +0000 (17:30 +0000)]
Remove vestigial unused variable in FAST-DIVISION.
Taylor R Campbell [Sun, 16 Mar 2014 15:21:38 +0000 (15:21 +0000)]
Revert "Assemble (JMP (@PCR FOO)) (LABEL FOO) to zero bits on amd64."
This reverts commit
f4dda7bf4a6f9f529fd2b295a0f51ee9ed6dc22b.
Apparently this doesn't work.
Taylor R Campbell [Sun, 16 Mar 2014 14:56:10 +0000 (14:56 +0000)]
Clarify comment over amd64 fixnum_shift hook.
Taylor R Campbell [Sun, 16 Mar 2014 14:50:30 +0000 (14:50 +0000)]
Open-code non-2^n quotients and remainders by multiplication on amd64.
Taylor R Campbell [Sun, 16 Mar 2014 14:49:26 +0000 (14:49 +0000)]
Assemble (JMP (@PCR FOO)) (LABEL FOO) to zero bits on amd64.
Taylor R Campbell [Sun, 16 Mar 2014 14:48:35 +0000 (14:48 +0000)]
Reorder branches in FIX:QUOTIENT/FIX:REMAINDER code on amd64.
Make the negative numerator case a forward branch so it will be
statically predicted not-taken.
Taylor R Campbell [Sat, 15 Mar 2014 20:32:07 +0000 (20:32 +0000)]
Avoid modifying rcx in amd64 fixnum_shift assembly hook.
Taylor R Campbell [Sat, 15 Mar 2014 20:30:09 +0000 (20:30 +0000)]
Fix sf/cgen so it doesn't just drop declarations for the compiler.
Taylor R Campbell [Sun, 23 Feb 2014 21:19:03 +0000 (21:19 +0000)]
Attempt to fix an undefined variable reference in svm disassembler.
Guessing here what the right way is; I don't know this code well
enough to say for sure.
Taylor R Campbell [Sun, 23 Feb 2014 17:57:35 +0000 (17:57 +0000)]
Generate all OS constructors for compiler.
Taylor R Campbell [Sun, 23 Feb 2014 17:30:43 +0000 (17:30 +0000)]
Remove obsolete AC_SYS_RESTARTABLE_SYSCALLS.
We don't use the result anyway, and it always causes an annoying
delay in configure.
Taylor R Campbell [Sun, 23 Feb 2014 17:10:24 +0000 (17:10 +0000)]
Use FLO:NEGATE, not (FLO:- 0. ...), to apply sign in NUMBER->STRING.
Taylor R Campbell [Sun, 23 Feb 2014 17:08:55 +0000 (17:08 +0000)]
Don't rewrite (flo:- 0. x) as (flo:negate x) on amd64.
This is wrong because (flo:- 0. 0.) is 0. but (flo:negate 0.) is -0.
This caused some broken code to be miscompiled and yield the correct
result accidentally!
Taylor R Campbell [Sun, 23 Feb 2014 16:42:02 +0000 (16:42 +0000)]
More hacks to get liarc to build.
Taylor R Campbell [Sun, 23 Feb 2014 16:40:46 +0000 (16:40 +0000)]
Try builtin objects before fasl files in liarc FASLOAD.
Taylor R Campbell [Sun, 23 Feb 2014 16:39:24 +0000 (16:39 +0000)]
Generate code for -0. constants correctly in liarc.
Taylor R Campbell [Sun, 23 Feb 2014 03:43:01 +0000 (03:43 +0000)]
compile-liarc-bundle rules depend on some tools in microcode/.
Taylor R Campbell [Sun, 23 Feb 2014 01:17:57 +0000 (01:17 +0000)]
Fix up fni->bci files in crsend.scm.
Taylor R Campbell [Sat, 22 Feb 2014 22:31:17 +0000 (22:31 +0000)]
Temporary kludge: bake a fresh runtime into the toolchain.
This is practically necessary for now because of the SEQUENCE-2 mess.
It may continue to be necessary in the future until we have proper
cross-compilation tools that can handle macros and fasdump format
changes sanely.
Taylor R Campbell [Sat, 22 Feb 2014 22:30:09 +0000 (22:30 +0000)]
Put back some temporary SEQUENCE-2 scaffolding.
Taylor R Campbell [Sat, 22 Feb 2014 22:29:31 +0000 (22:29 +0000)]
Simplify x86 fenv structures.
Avoid needless typedefs and unions.
Taylor R Campbell [Sat, 22 Feb 2014 22:28:39 +0000 (22:28 +0000)]
On i386, avoid FCOM[P] and FTST in favour of FUCOM[P].
The FCOM[P] and FTST instructions trap even on quiet NaN; the
FUCOM[P] instructions don't.
Taylor R Campbell [Sat, 22 Feb 2014 20:59:52 +0000 (20:59 +0000)]
Avoid using new-style SEQUENCE stuff in LIAR.
Taylor R Campbell [Sat, 22 Feb 2014 19:20:29 +0000 (19:20 +0000)]
Fix default (ISO-8859-1) char image strings.
We previously used `ANSI' char image strings, which really meant
Windows-1252 for some reason. Now we use ISO-8859-1 so things get
displayed correctly in X at least. Some characters may not display
*nicely* on Windows, but they will display *correctly* (with an octal
escape).
Matt Birkholz [Fri, 29 Nov 2013 18:26:29 +0000 (11:26 -0700)]
svm: Fix the disassembler.
Re-wrote it with a "cursor" that wraps up the block and offset and
symbol-table previously passed around and/or fluid-bound to *block,
*current-offset, *symbol-table, etc.
Matt Birkholz [Tue, 26 Nov 2013 17:08:59 +0000 (10:08 -0700)]
svm: Convert program_counter from a static variable to an argument.
This seems to speed the machine up 20%. Before this patch, the
compile-everything phase ("Use the new machine and compiler to
re-compile everything.") took ~67 minutes on a 3.2GHz Core i3-550.
After applying this patch, that phase took ~53 minutes (21% less).
This conversion was incomplete in the original implementation (in
6ae556b). I backed it out in order to get a simpler machine working.
Matt Birkholz [Tue, 26 Nov 2013 17:07:29 +0000 (10:07 -0700)]
ffi: Avoid longjmps in C-CALL primitive (callout_continue).
Rather than abort after every callout (in callout_continue), call the
second trampoline directly, after unsealing the Scheme stack.
Matt Birkholz [Mon, 18 Nov 2013 18:43:29 +0000 (11:43 -0700)]
make.so is supposed to be found in the compiler bundle.
It must have been Chaos that slipped in this bogus fix.
Matt Birkholz [Mon, 18 Nov 2013 18:41:26 +0000 (11:41 -0700)]
src/Makefile.tools.in: Missing dependency made cross-compiler crawl.
Matt Birkholz [Thu, 31 Oct 2013 19:23:20 +0000 (12:23 -0700)]
ffi: Add flovec_* functions and abort_to_c for shims.
Add flovec_length, flovec_loc and flovec_ref. The abort_to_c function
allows the Gtk shim's run_gtk() to stop Scheme and return to the
toolkit.
Matt Birkholz [Thu, 31 Oct 2013 19:05:52 +0000 (12:05 -0700)]
ffi/build.scm.in: Install compile-file:sf-only?ed options.
Copy the .bin if there is no .com. Copy .ext too.
Matt Birkholz [Thu, 31 Oct 2013 19:03:12 +0000 (12:03 -0700)]
doc/ffi: typo
Matt Birkholz [Thu, 31 Oct 2013 19:02:10 +0000 (12:02 -0700)]
mhash: Simplify configure.ac.
Matt Birkholz [Wed, 16 Oct 2013 00:01:35 +0000 (17:01 -0700)]
doc/user-manual (Unix Installation): Added plugin building.
Matt Birkholz [Tue, 15 Oct 2013 18:05:32 +0000 (11:05 -0700)]
src/Makefile.in: Make xml/xml-unx.pkd depend on cref-star-parser.
Matt Birkholz [Tue, 15 Oct 2013 17:55:39 +0000 (10:55 -0700)]
dist/make-src-files: Fix $plugin-config log file names.
Matt Birkholz [Tue, 15 Oct 2013 17:19:42 +0000 (10:19 -0700)]
Revert
13b5bca. Use the host's runtime, not a cross runtime.
Revert
41c0070 and
5cf6bfc too, and the "training" in cref/make.scm
(for the compiler) and runtime.sf (for a cross runtime).
Matt Birkholz [Tue, 15 Oct 2013 16:18:50 +0000 (09:18 -0700)]
dist/make-src-files: Autoconf plugins (FFI wrappers).
Fix a confusing error message too.
Matt Birkholz [Tue, 15 Oct 2013 16:15:54 +0000 (09:15 -0700)]
ffi: Add DESTDIR support.
Matt Birkholz [Sun, 13 Oct 2013 08:03:15 +0000 (01:03 -0700)]
Fix LIAR/svm build to cross from 9.1.1.
As for LIAR/C, do not cross compile host .bins containing sequence-3s.
Chuck them to STAGEX/ and re-syntax with cross runtime. And syntax
compiler/base/crsend AFTER said chucking.
Matt Birkholz [Sun, 13 Oct 2013 07:48:19 +0000 (00:48 -0700)]
Make compile-xml dependent on cref-star-parser.
xml.pkg wants (global-definitions "../star-parser/parser") else
;Warning: Could not find global definitions: ...
Matt Birkholz [Sun, 13 Oct 2013 07:44:16 +0000 (00:44 -0700)]
Fix LIAR/C build to cross from 9.1.1.
Toss the .bins generated by the host runtime (which include
sequence-3). Only cross-compile .bins generated by the cross runtime.
Matt Birkholz [Sun, 13 Oct 2013 02:58:57 +0000 (19:58 -0700)]
More FFI cleanups: no need to include microcode/achost.ac.
Matt Birkholz [Sun, 13 Oct 2013 01:17:44 +0000 (18:17 -0700)]
Fix new build to cross from 9.1.1.
On a 9.1.1 host, the new build dies while compiling the compiler.
; Generating SCode for file: "fggen/canon.scm" => "fggen/canon.bin"...
;MICROCODE-TYPE: Unknown name sequence
The usual remedies did not suffice. In fact now the temporary code in
cref/make.scm is not needed. The compiler is compiled with a cross
runtime that provides the necessary patch via its utabs.scm, where
access to type-aliases and returns-aliases is easy, though
insufficient to train the host. 9.1.1 still needs the training in
runtime.sf.
Unlike a cross compile, a cross runtime was not used by this build(?).
Use a new compile-cross-runtime target and --fasl make.com to make
tools/syntaxer.com. Make syntax-cross-cref and syntax-cross-sf
dependent on syntax-cross-runtime for its runtime.pkd (pinching off a
little parallelism, but allowing them to proceed while the runtime
compiles). Did not try to syntax anything but the new compiler with
the new star-parser.
Matt Birkholz [Wed, 9 Oct 2013 18:59:26 +0000 (11:59 -0700)]
Eliminate compile-system.
Taylor R Campbell [Tue, 8 Oct 2013 02:56:57 +0000 (02:56 +0000)]
Automatically clean tree and build tools if necessary.
Taylor R Campbell [Tue, 8 Oct 2013 02:55:57 +0000 (02:55 +0000)]
Turn some phony rules into explicit file rules to reduce rebuilds. [retry]
Unfinished conversion; some of the explicit file rules still depend
on phony rules. Finishing this is left as an exercise for the
reader.
Taylor R Campbell [Tue, 8 Oct 2013 02:54:21 +0000 (02:54 +0000)]
Revert "Turn some phony rules into explicit file rules to reduce rebuilds."
This reverts commit
54e7f433ab16f65cd12779b7a672edf713c72427.
That commit wasn't supposed to revert an earlier change to SVM1; not
sure how that snuck in there.
Taylor R Campbell [Mon, 7 Oct 2013 22:57:53 +0000 (22:57 +0000)]
Turn some phony rules into explicit file rules to reduce rebuilds.
Unfinished conversion; some of the explicit file rules still depend
on phony rules. Finishing this is left as an exercise for the
reader.
Taylor R Campbell [Mon, 7 Oct 2013 22:35:16 +0000 (22:35 +0000)]
Reduce SVM1 address decoding indirection.
This must have hammered horrendously on the CPU's branch target
predictor. No measurements yet but I'm sure SVM1 should be a lot
faster now.
Taylor R Campbell [Mon, 7 Oct 2013 22:07:19 +0000 (22:07 +0000)]
Tweak svm1-interp a little.
- Mark several routines inline.
- Add branch predictions.
- Fail gracefully, don't abort, on illegal instructions.
Taylor R Campbell [Mon, 7 Oct 2013 21:07:26 +0000 (21:07 +0000)]
Add a few more file types for save/restore targets.
Taylor R Campbell [Mon, 7 Oct 2013 20:54:45 +0000 (20:54 +0000)]
In COMPILE-SYSTEM, load the bin if cross-compiling too.
XXX COMPILE-SYSTEM is completely wrong; it needs to be eliminated in
favour of a principled way to identify macro dependencies, like in
Scheme48 and Racket.
Taylor R Campbell [Mon, 7 Oct 2013 20:52:57 +0000 (20:52 +0000)]
Call it `saved-objects', not `obj', to avoid confusing BSD make.
Taylor R Campbell [Mon, 7 Oct 2013 20:46:08 +0000 (20:46 +0000)]
Don't list compiler twice in SUBDIRS.
Taylor R Campbell [Mon, 7 Oct 2013 20:36:55 +0000 (20:36 +0000)]
Load ffimacro.bin in win32.sf, not ffimacro.
Avoids accidentally loading cross-compiled gobbledegook which the
toolchain barfs on.
Taylor R Campbell [Mon, 7 Oct 2013 19:16:33 +0000 (19:16 +0000)]
Explain what the `cross-target' target does.
Taylor R Campbell [Mon, 7 Oct 2013 18:43:19 +0000 (18:43 +0000)]
toolchain target should really depend on tools/syntaxer.com too.
Taylor R Campbell [Mon, 7 Oct 2013 18:37:10 +0000 (18:37 +0000)]
Remove unfinished kludge to build IMAIL with LIARC.
Taylor R Campbell [Mon, 7 Oct 2013 18:32:32 +0000 (18:32 +0000)]
Recompile edwin/edwin.bld only if necessary.
Taylor R Campbell [Mon, 7 Oct 2013 17:30:55 +0000 (17:30 +0000)]
Recompile *.pkd files for liarc only if necessary.
Taylor R Campbell [Mon, 7 Oct 2013 17:30:11 +0000 (17:30 +0000)]
Fasload errors should fail gracefully, not terminate Scheme.
Taylor R Campbell [Mon, 7 Oct 2013 17:17:47 +0000 (17:17 +0000)]
Lots of changes toward making liarc build use new structure.
Doesn't fully work yet, but it's progress!
Old build system is still all there; you must use `make all-native'
to try the new one.
Taylor R Campbell [Mon, 7 Oct 2013 17:14:23 +0000 (17:14 +0000)]
Use $@ and $< for liarc .c.o rule, not $*.o and $*.c.
Otherwise, it won't work in subdirectories.
Taylor R Campbell [Mon, 7 Oct 2013 14:25:20 +0000 (14:25 +0000)]
Update links created by Setup.sh to reflect installed lib directory.
Taylor R Campbell [Mon, 7 Oct 2013 02:26:29 +0000 (02:26 +0000)]
Compile IMAIL in --batch-mode.
Taylor R Campbell [Sun, 6 Oct 2013 23:07:37 +0000 (23:07 +0000)]
OOPS! Don't forget to build compiler/rtlopt...
Taylor R Campbell [Sun, 6 Oct 2013 22:23:48 +0000 (22:23 +0000)]
Dependents of cross-target must be conditionalized on @IF_CROSS@.
Taylor R Campbell [Sun, 6 Oct 2013 22:22:09 +0000 (22:22 +0000)]
First whack at making new build system work for liarc.
Taylor R Campbell [Sun, 6 Oct 2013 22:21:43 +0000 (22:21 +0000)]
Rename cross-barrier/cross-end to cross-host/cross-target.
Taylor R Campbell [Sun, 6 Oct 2013 22:21:13 +0000 (22:21 +0000)]
Add cast to avoid negative arguments to isalnum.
Taylor R Campbell [Sun, 6 Oct 2013 21:46:08 +0000 (21:46 +0000)]
compile-imail needs cross-end, not just cross-barrier.
Taylor R Campbell [Sun, 6 Oct 2013 21:14:04 +0000 (21:14 +0000)]
Just cast to long; don't check whether %td will work for ptrdiff_t.
Taylor R Campbell [Sun, 6 Oct 2013 18:15:08 +0000 (18:15 +0000)]
Don't reset AS_FLAGS and M4_FLAGS to empty in microcode/configure.
Taylor R Campbell [Sun, 6 Oct 2013 18:14:53 +0000 (18:14 +0000)]
Use a compile-test, not a run-test, for __ELF__.
Taylor R Campbell [Sun, 6 Oct 2013 18:14:28 +0000 (18:14 +0000)]
Use CAREFUL-OBJECT-DATUM in i386 MULTIPLY-OBJECT-BY-2.
Taylor R Campbell [Sun, 6 Oct 2013 18:11:00 +0000 (18:11 +0000)]
Merge branch 'make-cross-tools'
Taylor R Campbell [Sun, 6 Oct 2013 18:10:33 +0000 (18:10 +0000)]
Make the `save' makefile targets fail noisily.
Taylor R Campbell [Sun, 6 Oct 2013 16:17:19 +0000 (16:17 +0000)]
Fix missing close paren in Makefile.in.
Taylor R Campbell [Sun, 6 Oct 2013 16:07:06 +0000 (16:07 +0000)]
Call it the toolchain, not the bootstrap toolchain.
Rename Makefile.boot -> Makefile.tools, boot/ -> tools/.
`Bootstrap toolchain' sounds like something that is useful only to
bootstrap MIT Scheme, when in fact the object of the exercise is to
make it a general-purpose cross-compilation toolchain application
that you can use to cross-compile anything.
Taylor R Campbell [Sat, 5 Oct 2013 21:55:06 +0000 (21:55 +0000)]
Tweak native rules so they work for svm too.
Taylor R Campbell [Sat, 5 Oct 2013 21:54:52 +0000 (21:54 +0000)]
Rename @CROSS@ to @IF_CROSS@.
Taylor R Campbell [Sat, 5 Oct 2013 21:42:03 +0000 (21:42 +0000)]
Load crsend.scm into the right environment.