Chris Hanson [Sun, 27 Oct 2019 05:50:52 +0000 (22:50 -0700)]
Automatically pre-register libraries in standard places.
Chris Hanson [Sun, 27 Oct 2019 05:50:11 +0000 (22:50 -0700)]
Add param:hide-notifications?.
Chris Hanson [Wed, 23 Oct 2019 07:29:52 +0000 (00:29 -0700)]
Create a new directory src/libraries, to hold R7RS libraries.
Initially populate this with (srfi 133) and (mit inline-testing), plus unit
tests for the former.
Chris Hanson [Tue, 22 Oct 2019 07:16:20 +0000 (00:16 -0700)]
Preserve the initial host-library-db.
Chris Hanson [Tue, 22 Oct 2019 06:00:09 +0000 (23:00 -0700)]
Create a "standard" library providing access to the library database.
This is useful for programs that want to work directly on libraries.
Chris Hanson [Tue, 22 Oct 2019 05:58:01 +0000 (22:58 -0700)]
Fix problems with reporting why an auto property isn't available.
The previous approach was not filtering the dependencies right.
Chris Hanson [Tue, 22 Oct 2019 05:13:46 +0000 (22:13 -0700)]
Eliminate duplicate binding.
Chris Hanson [Tue, 22 Oct 2019 05:12:36 +0000 (22:12 -0700)]
Don't rewrite library imports after parsing them.
It's more convenient to have the retain their syntax, so that they can be used
by programs using reflective access to the libraries.
Chris Hanson [Tue, 22 Oct 2019 03:43:19 +0000 (20:43 -0700)]
Be more careful when checking for unquote in repl eval.
Chris Hanson [Mon, 21 Oct 2019 05:37:00 +0000 (22:37 -0700)]
Furthur improve error reporting for libraries.
Chris Hanson [Sun, 20 Oct 2019 23:33:52 +0000 (16:33 -0700)]
Just have one legacy library: (mit legacy runtime)
Otherwise it's a pain to figure out where all the exports are, since they aren't
necessarily where you might think.
Also, this eliminates deprecated bindings.
Chris Hanson [Sun, 20 Oct 2019 20:07:17 +0000 (13:07 -0700)]
Improve error reporting for library imports.
Chris Hanson [Sun, 20 Oct 2019 05:41:51 +0000 (22:41 -0700)]
Fix problem with syntaxing of DO.
This wouldn't work correctly in some weird edge cases. Specifically, it was
trying to detect the 'else and '=> keywords, but comparing them in the syntactic
environment outside of the DO, not the one inside of it. Fixed by rewriting the
macro to defer the cond-clause processing until the interior environment was
available.
Chris Hanson [Sun, 20 Oct 2019 05:20:19 +0000 (22:20 -0700)]
Fix typo in earlier change.
Chris Hanson [Mon, 7 Oct 2019 05:10:28 +0000 (01:10 -0400)]
Add printer for bytevectors and limit printing of vectors.
Chris Hanson [Mon, 7 Oct 2019 04:34:47 +0000 (00:34 -0400)]
Change debug printer to print vectors correctly.
Chris Hanson [Mon, 7 Oct 2019 04:34:26 +0000 (00:34 -0400)]
Fix cold-load problems when cross-compiling from 10.1.10.
Taylor R Campbell [Thu, 3 Oct 2019 16:23:03 +0000 (16:23 +0000)]
A reader has alerted me to a caveat about IEEE 754 min/max.
Taylor R Campbell [Thu, 3 Oct 2019 15:39:13 +0000 (15:39 +0000)]
Fix bugs in min/max on signalling NaN inputs.
Taylor R Campbell [Thu, 3 Oct 2019 14:49:34 +0000 (14:49 +0000)]
Fix floating-point min/max tests, and mark them xfail.
I misread IEEE 754 semantics on min/max for _signalling_ NaN.
If given a number and a _quiet_ NaN, min/max are to return the
_number_. But if given a _signalling_ NaN, min/max are to return a
quiet version of the NaN.
References:
Sec. 5.3 `Homogeneous general-computational operations',
subsec. 5.3.1 `General operations',
on minNum/maxNum/minNumMag/maxNumMag reads:
`minNum(x, y) is...the canonicalized number if one operand is a
number and the other is a quiet NaN.'
`When either x or y is a signaling NaN, then the result is according
to 6.2.'
Sec. 6.2 `Operations with NaNs' reads:
`Under default exception handling, any operation signaling an invalid
operation exception and for which a floating-point result is to be
delivered shall deliver a quiet NaN.'
[Consequently, _if_ min/max raise invalid-operation and
nevertheless return a result, they shall return a quiet NaN.]
`Signaling NaNs shall be reserved operands that, under default
exception handling, signal the invalid operation exception (see 7.2)
for every general-computational and signaling-computational
operation except for the conversions described in 5.12.'
[min/max (i.e., minNum/maxNum/minNumMag/maxNumMag) are listed
under homogeneous general-computational, so this clause applies.]
Subsec. 6.2.3 `NaN propagation' further reads:
`An operation that propagates a NaN operand to its result and has a
single NaN as an input should produce a NaN with the payload of the
input NaN.'
`If two or more inputs are NaN, then the payload of the resulting NaN
should be identical to the payload of one of the input NaNs if
representable in the destination format.'
Chris Hanson [Tue, 24 Sep 2019 07:35:38 +0000 (00:35 -0700)]
Change record types to be immutable.
The primary reason for this is to make the predicates slightly faster,
eliminating the need to look up the type markers in the predicates.
Additionally, make-record-type now accepts these additional options using a
keyword list. The define-record-type macro has not been updated to support
these new options, but that will come soon. Consequently the files using these
options have been modified to use make-record-type directly.
Finally, a small tweak was needed so that multiple values are available earlier
in the cold load.
Chris Hanson [Mon, 23 Sep 2019 20:41:39 +0000 (13:41 -0700)]
Fix aliasing problem with sub-record predicates.
Add test to check that it works.
Chris Hanson [Sun, 22 Sep 2019 21:14:02 +0000 (14:14 -0700)]
Restore compatibility with SRFI 131.
Also add some tests for records, and fix several bugs revealed by them.
Chris Hanson [Sun, 22 Sep 2019 05:54:01 +0000 (22:54 -0700)]
Refactor record layout following a suggestion from Taylor.
New layout is identical for records that don't have a parent type. For those
that do have a parent type, the layout is like this:
root-marker
root-fields
sub1-marker
sub1-fields
sub2-marker
sub2-fields
...
The primary advantage of this layout is to make the record predicate be constant
time, as opposed to the previous design in which it could be linear in the depth
of the parent chain.
In addition, a number of record operations have been bummed for slightly better
performance, and the layout of record types has been altered to keep track of
the type information in a way that's better organized for generating the record
operations.
There are some behavioral changes:
* This implementation is slightly incompatible with SRFI 131, since it prohibits
a child from having a field name that's the same as one of its ancestors.
I'll probably change this for compatibility.
* Only a root record type can have an applicability method, and that method is
called for all sub-types of that root type. Arguably this is reasonable
behavior.
* Non-root fasdumpable records must have proxy markers for all of their
component types. Previously, only the record type stored in slot 0 needed to
have a fasdumpable proxy. This isn't an immediate issue since fasdumpable
records are used very sparingly at the moment and probably won't be supported
outside of the runtime system.
Chris Hanson [Sun, 22 Sep 2019 04:09:13 +0000 (21:09 -0700)]
Tweak my special hack.
Chris Hanson [Sat, 21 Sep 2019 07:02:15 +0000 (00:02 -0700)]
Implement any-duplicates?.
Chris Hanson [Sat, 21 Sep 2019 06:16:04 +0000 (23:16 -0700)]
Fix syntax bug that showed up in syntax-rules.
Taylor R Campbell [Wed, 18 Sep 2019 14:40:06 +0000 (14:40 +0000)]
Merge branch 'riastradh-
20190917-fixamd64uuojmp'
Taylor R Campbell [Wed, 18 Sep 2019 01:50:37 +0000 (01:50 +0000)]
Fix tospace/newspace confusion.
This did not manifest in my testing on NetBSD because it happened
that on NetBSD, the tospace and newspace are always separated by more
than 4 GB, so the bogus jmprel32_offset was never used during GC,
e.g.:
newspace=0x17dd7f8 tospace=0x7f7feeb00000 delta=
17523427132673
I suspect it is the cause of Chris's problems on macOS, so this
should fix Chris's issue on macOS and speed things up even more for
everyone.
Taylor R Campbell [Wed, 18 Sep 2019 01:44:15 +0000 (01:44 +0000)]
Revert "Workaround for inexplicable fatal error on macOS."
This reverts commit
689da8bf97534063f1800caf544a694de81d511e.
Will fix in a subsequent commit.
Chris Hanson [Tue, 17 Sep 2019 06:50:23 +0000 (23:50 -0700)]
Workaround for inexplicable fatal error on macOS.
Chris Hanson [Tue, 17 Sep 2019 06:16:14 +0000 (23:16 -0700)]
Update incorrect comments.
Chris Hanson [Tue, 17 Sep 2019 06:13:44 +0000 (23:13 -0700)]
Teach fasload about cc_return objects.
Otherwise fasload gets SIGSEGV when the memory addresses are different.
Chris Hanson [Sun, 8 Sep 2019 07:31:51 +0000 (00:31 -0700)]
Adapt fegetexcept etc. workarounds to work when debugging turned on.
Chris Hanson [Sun, 8 Sep 2019 07:30:51 +0000 (00:30 -0700)]
Don't use MAP_TRYFIXED unless is it defined; otherwise the assert fails.
Taylor R Campbell [Tue, 3 Sep 2019 13:37:25 +0000 (13:37 +0000)]
Mark floenv in use so exceptions stay cleared if preempted.
Otherwise, we might do:
T1: clear-float-exceptions
preempt and switch
T2: floating-point operation that raises exceptions
preempt and switch
T1: use-floating-point-environment!
and the exceptions may remain raised from T2.
Taylor R Campbell [Mon, 26 Aug 2019 04:50:32 +0000 (04:50 +0000)]
Open-code flonum-fma (fused multiply-add) on aarch64.
The fused multiply-subtract doesn't kick in right now for reasons I
don't understand in rcompr.scm; maybe someone who understands that
code better can help.
Taylor R Campbell [Mon, 26 Aug 2019 03:20:37 +0000 (03:20 +0000)]
Fix missing ELSE (error).
Taylor R Campbell [Mon, 26 Aug 2019 03:15:43 +0000 (03:15 +0000)]
Fix leak of floating-point environment in test-arith.scm.
Taylor R Campbell [Mon, 26 Aug 2019 03:14:22 +0000 (03:14 +0000)]
If enabling fp traps for first time, clear sticky exceptions.
Otherwise, the indeterminate state of the sticky exceptions --
possibly from another thread -- may cause the machine to trap merely
by enabling traps.
Also, don't bother marking the floating-point environment in use when
disabling traps.
Taylor R Campbell [Mon, 26 Aug 2019 02:59:12 +0000 (02:59 +0000)]
Cache cleared exceptions to prevent SIGFPE loop on trap.
fesetenv, as used by fixup_float_environment at the top of Interpret,
will trap any trapped and raised exceptions in the floating-point
environment it is restoring, which is bad news during a trap.
Not really sure how this managed to work in the past...
Taylor R Campbell [Mon, 26 Aug 2019 02:48:45 +0000 (02:48 +0000)]
Teach continuation parser about last return code offsets.
This fixes a thirty-year-old (!) bug with creating continuations that
return into compiled code with #f as the last return code offset for
reenter-compiled-code. Manifests only with debugging enabled.
Taylor R Campbell [Mon, 26 Aug 2019 00:21:27 +0000 (00:21 +0000)]
Fix store-value restarts for unassigned/unbound variable references.
Turns out these have been busted since 2007...
Taylor R Campbell [Sun, 25 Aug 2019 21:20:39 +0000 (21:20 +0000)]
New primitive PRIMITIVE-PROCEDURE-OPEN-CODED?.
Usage:
(define (foo x y)
(if ((ucode-primitive primitive-procedure-open-coded? 1)
(ucode-primitive xyz 2))
((ucode-primitive xyz 2) x y)
... Scheme alternative implementation ...))
Taylor R Campbell [Sun, 25 Aug 2019 20:27:53 +0000 (20:27 +0000)]
Use paxctl on NetBSD to allow PROT_WRITE|PROT_EXEC.
Taylor R Campbell [Sun, 25 Aug 2019 20:27:05 +0000 (20:27 +0000)]
Use proper m4 quotation for AC_CHECK_TOOL usage.
Chris Hanson [Mon, 26 Aug 2019 03:16:36 +0000 (20:16 -0700)]
Tweak workaround for floating exceptions on macOS.
Taylor R Campbell [Sun, 25 Aug 2019 19:49:44 +0000 (19:49 +0000)]
Merge branch 'riastradh-
20190113-aarch64-v6'
Taylor R Campbell [Sun, 25 Aug 2019 19:48:23 +0000 (19:48 +0000)]
Open-code floating-point arithmetic on aarch64.
Disabled by default for now due to limited testing.
Taylor R Campbell [Sun, 25 Aug 2019 19:25:34 +0000 (19:25 +0000)]
Reduce some extraneous MOVK instructions in immediate loads.
Taylor R Campbell [Sun, 25 Aug 2019 19:03:51 +0000 (19:03 +0000)]
Fix set of architectures with busted fp compare exceptions.
Taylor R Campbell [Sun, 25 Aug 2019 19:03:01 +0000 (19:03 +0000)]
Work around buggy libm on NetBSD/aarch64.
Will be fixed upstream but it doesn't hurt to add one more bit here.
Taylor R Campbell [Sun, 25 Aug 2019 18:35:37 +0000 (18:35 +0000)]
Show for which parameters the test failed.
Taylor R Campbell [Sun, 25 Aug 2019 18:32:03 +0000 (18:32 +0000)]
Simplify flo:-inf.0 and flo:+inf.0.
We have put out a release now whose reader understands +inf.0/-inf.0,
so we should be able to use the notation.
Fixes bugs with incorrectly raising the divide-by-zero exception flag
(even if no trap), e.g.:
;|log1mexp-invalid/(4.
9406564584124654e-324)| failed 1 sub-tests out of 1 in 0. seconds:
assertion 3: value was 3 but expected an object eqv? to 1
Here 3 = (invalid-operation divide-by-zero), 1 = (invalid-operation)
[aarch64].
Taylor R Campbell [Sun, 25 Aug 2019 18:25:43 +0000 (18:25 +0000)]
Fix missing #include.
Mistake failed silently because it only changed #ifdef outcome.
Taylor R Campbell [Sun, 25 Aug 2019 15:40:15 +0000 (15:40 +0000)]
Let caller specify how to get a temporary in ADD-IMMEDIATE.
Fixes
; Compiling file: "test-hack.bin" => "test-hack.com"...
;Unassigned variable: *needed-registers*
because of attempt to allocate a temporary in ADD-IMMEDIATE during
GENERATE/REMOTE-LINKS.
Taylor R Campbell [Sun, 25 Aug 2019 15:38:25 +0000 (15:38 +0000)]
New GENERAL-TEMPORARY! shorthand.
Taylor R Campbell [Sun, 25 Aug 2019 15:09:11 +0000 (15:09 +0000)]
Fix units for cc_entry_to_block_offset/cc_return_to_block_offset.
It would make more sense for the compiler to generate debug data
labels in instruction units, but this is a simpler change and is what
was done in the past on machines like mips with 32-bit aligned
instructions.
aarch64 now has functioning debug data!
Taylor R Campbell [Sat, 24 Aug 2019 07:10:39 +0000 (07:10 +0000)]
Fix comment about odd rounding mode.
Taylor R Campbell [Sat, 24 Aug 2019 07:09:39 +0000 (07:09 +0000)]
Add missing include guards.
Taylor R Campbell [Sat, 24 Aug 2019 07:07:11 +0000 (07:07 +0000)]
Update and clarify fp-env documentation.
Taylor R Campbell [Sat, 24 Aug 2019 04:32:50 +0000 (04:32 +0000)]
Simplify flo:update-environment!.
Caller is interested in exceptions afterward, so it is not sensible
to deregister interest in the floating-point environment afterward.
If you really want that, surround it in flo:preserving-environment.
Taylor R Campbell [Sat, 24 Aug 2019 04:26:31 +0000 (04:26 +0000)]
Fix two bugs in floating-point environment.
1. Don't cache it in the current thread.
We can't mark everywhere the cache needs to be invalidated --
i.e., every floating-point instruction -- and it's not clear
there's any performance benefit to the cache anyway. The main
performance cost, as I recall, was swapping environments on every
thread switch, which we avoid for all threads in the default
environment.
2. The default environment initialization left the machine in a wacky
state after reset-package!, which caused many spurious exception
traps once I undid the cache. There's no need to preserve the
machine environment here; we are setting up the default
environment, after all, so the environment we're in when done
should be the default one.
Taylor R Campbell [Sat, 24 Aug 2019 01:40:11 +0000 (01:40 +0000)]
Test busted flo:set-environment!.
Taylor R Campbell [Sat, 20 Apr 2019 19:48:07 +0000 (19:48 +0000)]
Simplify some logic by using dedicated scratch registers or targets.
No need to allocate temporaries except when we need more than two or
three of them.
Taylor R Campbell [Thu, 22 Aug 2019 14:23:08 +0000 (14:23 +0000)]
Use AC_CHECK_TOOL for as and allow configure parameter.
This way we auto-detect a cross-assembler more easily with, e.g.,
--build=aarch64--netbsd.
Taylor R Campbell [Mon, 22 Apr 2019 04:07:27 +0000 (04:07 +0000)]
Fix sign error in previous.
Taylor R Campbell [Sat, 20 Apr 2019 20:09:12 +0000 (20:09 +0000)]
Link the endian-independent cmpauxmd for aarch64.
Taylor R Campbell [Sat, 9 Feb 2019 17:40:30 +0000 (17:40 +0000)]
Update aarch64 release note.
Taylor R Campbell [Fri, 1 Feb 2019 05:30:38 +0000 (05:30 +0000)]
Validate signed-26 PC-relative offset in B instruction.
Taylor R Campbell [Fri, 1 Feb 2019 05:29:29 +0000 (05:29 +0000)]
Comment out bogus ADRP-ADD pseudo-instruction definition.
This will cause the compiler to fail noisily if it tries to assemble
code with sufficiently distant PC-relative addresses, which is better
than silently assembling garbage.
Need to find a proper fix.
Taylor R Campbell [Fri, 25 Jan 2019 04:35:12 +0000 (04:35 +0000)]
No need for a separate function here.
Taylor R Campbell [Fri, 25 Jan 2019 04:31:15 +0000 (04:31 +0000)]
Fix comment about ADRP to reflect reality.
Taylor R Campbell [Fri, 25 Jan 2019 04:30:38 +0000 (04:30 +0000)]
Add another step to the compiler tests README.
Taylor R Campbell [Fri, 25 Jan 2019 04:21:09 +0000 (04:21 +0000)]
Add note on ughbletchness of ADRP.
Taylor R Campbell [Fri, 25 Jan 2019 03:56:51 +0000 (03:56 +0000)]
Add some basic compiler tests.
For use when bringing up a compiler or back end that can't yet run
the normal test suite.
Taylor R Campbell [Fri, 25 Jan 2019 03:39:32 +0000 (03:39 +0000)]
Fix some bugs in logical immediate encoding.
Taylor R Campbell [Wed, 23 Jan 2019 08:17:57 +0000 (08:17 +0000)]
One more TODO item.
Taylor R Campbell [Wed, 23 Jan 2019 08:15:50 +0000 (08:15 +0000)]
Update TODO.
Taylor R Campbell [Wed, 23 Jan 2019 08:12:26 +0000 (08:12 +0000)]
Disable some tests when trapping is busted.
Taylor R Campbell [Wed, 23 Jan 2019 07:52:00 +0000 (07:52 +0000)]
Expect failure, not error, if trapping is not supported.
Makes these tests pass on my arm64 laptop.
Taylor R Campbell [Wed, 23 Jan 2019 07:50:38 +0000 (07:50 +0000)]
Check at runtime whether feenableexcept does anything.
The bits are defined on aarch64, but apparently some CPUs are
fabricated without support for them so they just read back as zers.
Bummer!
Taylor R Campbell [Wed, 23 Jan 2019 07:49:00 +0000 (07:49 +0000)]
Fix excessive sign bits in uuo link instruction.
Taylor R Campbell [Wed, 23 Jan 2019 05:33:39 +0000 (05:33 +0000)]
Fix displaced byte load rule.
The offset is not a machine register! Yikes.
Taylor R Campbell [Wed, 23 Jan 2019 05:21:52 +0000 (05:21 +0000)]
Fix FIXNUM-NOT rule: don't set the low bits.
Taylor R Campbell [Wed, 23 Jan 2019 05:21:34 +0000 (05:21 +0000)]
Implement aarch64 logical immediate encoding.
Taylor R Campbell [Wed, 23 Jan 2019 03:02:31 +0000 (03:02 +0000)]
Fix far uuo links.
Apparently ADRP really does do Rd <- (PC & ~0xfff) + (imm << 12), not
PC + (imm << 12), which means it's gonna cause some trouble for the
assembler in LIAR, since it means the code needs to know its own
offset within a page of memory and the target's offset within a page
of memory.
Taylor R Campbell [Wed, 23 Jan 2019 03:02:16 +0000 (03:02 +0000)]
Fix loading complemented immediates.
Taylor R Campbell [Wed, 23 Jan 2019 03:01:43 +0000 (03:01 +0000)]
Fix UNSIGNED-LESS-THAN-FIXNUM? branch condition.
Add some condition code aliases while here and clarify comments.
Taylor R Campbell [Tue, 22 Jan 2019 09:01:33 +0000 (09:01 +0000)]
Fix indexing of remote links.
Taylor R Campbell [Tue, 22 Jan 2019 03:47:47 +0000 (03:47 +0000)]
Fix byte ordering in GENERATE/NSECTS.
Taylor R Campbell [Tue, 22 Jan 2019 03:47:35 +0000 (03:47 +0000)]
Fix register choices in GENERATE/REMOTE-LINKS.
Taylor R Campbell [Tue, 22 Jan 2019 03:47:03 +0000 (03:47 +0000)]
Fix sense of INVOCATION-PREFIX:DYNAMIC-LINK choice.
Taylor R Campbell [Tue, 22 Jan 2019 01:55:18 +0000 (01:55 +0000)]
Fix reference to constant section in GENERATE/REMOTE-LINKS.
Taylor R Campbell [Mon, 21 Jan 2019 23:37:32 +0000 (23:37 +0000)]
Sign-extend PC-relative branch target.
Taylor R Campbell [Mon, 21 Jan 2019 22:39:29 +0000 (22:39 +0000)]
Fix indexing in MOVE-FRAME-UP code: objects, not bytes, here.
And with this, the cold load completes on aarch64!
Taylor R Campbell [Mon, 21 Jan 2019 22:39:11 +0000 (22:39 +0000)]
Fix large application setup.
Taylor R Campbell [Mon, 21 Jan 2019 20:59:02 +0000 (20:59 +0000)]
Teach cmpintmd to flush the instruction cache on aarch64.
Taylor R Campbell [Mon, 21 Jan 2019 20:53:14 +0000 (20:53 +0000)]
Fix argument to PUSH_D_CACHE_REGION.
Takes startptr/count, not startptr/endptr.
This was not an issue before because until aarch64, the only extant
port that even used this, i386, ignored the argument as a macro and
flushed the entire cache.
Taylor R Campbell [Mon, 21 Jan 2019 19:06:38 +0000 (19:06 +0000)]
Fix branch instruction in uuo link stub.