28781 commits

This branch

This branch

All branches

Author	SHA1	Message	Date
Charles Zhang	9412c7ab71	Remove some unnecessary checks. ... And declare a type.	2026-09-07 11:48:56 +02:00
Stas Boukarev	3e6f374427	More options for combination-match	2026-09-07 10:58:55 +03:00
Stas Boukarev	bb2895f75a	Transfer a couple more uses from combination-match to combination-match2	2026-09-07 10:03:49 +03:00
Stas Boukarev	3ca74b1072	Fix a logcount transform ... Fixes lp#2166634	2026-09-07 08:13:43 +03:00
Stas Boukarev	2880c35e95	Add aliases to combination-match	2026-09-07 05:17:39 +03:00
Douglas Katzman	0d46c3c373	Delete obsolete comment ... Linked lists of objects are not the right way to implement layout remsets	2026-09-06 21:22:09 -04:00
Douglas Katzman	9bc9b7ceaf	Add with-pinned-objects where needed for "documentation" ... This is kind of a no-op for x86-64 and doesn't actually improve the situation for arm64. When fixedobj space was changed to use software card marking instead of mprotect, I wanted to avoid creating more hand-written asm code, and I also thought at the time that the remembered set of modified layouts would be more than just setting a card mark. But it ended up as just that.	2026-09-06 21:21:19 -04:00
Douglas Katzman	c23e7716ae	Fix sb-vm::with-pseudo-atomic-foreign-calls ... The intent is to say that here is logic which it not machine-specific and so we're not going to put it inside of pseudo-atomic inside of a vop, but it needs to act as if it was pseudo-atomic. For sb-safepoint, even though as the comment said it "could" be controlled by an optimize decl, it was not getting done. For nonstop-foreign-call not only should we skip emitting instructions to save the stack pointer and possibly take a stop-for-gc trap, we must insert the pseudo-atomic barrier instructions, otherwise the foreign code would receive a stop-for-gc signal as if in lisp. And the implementation really should get finished for the architecturs other than x86-64, though it does not seem to be as critical and I'm not sure why. The uses guarded by #+immobile-space are clearly unimportant unless immobile space exists, but uses involving dlopen and dladdr seem important.	2026-09-06 20:36:25 -04:00
Stas Boukarev	53d0b3c4ff	Invent a new combination matching macro to do it all	2026-09-07 03:28:02 +03:00
Stas Boukarev	313fd89575	Fix the combination-match macro around :sole-dest ... Fixes lp#2166618 lp#2166617	2026-09-06 21:18:08 +03:00
Stas Boukarev	e8407e53ce	Don't change the number of values types in transforms ... Fixes lp#2166598	2026-09-06 17:11:59 +03:00
Stas Boukarev	935551f30a	Don't ash large integers in an integer-length transform ... Fixes lp#2166599	2026-09-06 17:11:59 +03:00
Stas Boukarev	7ec2408a9d	ltn-analyze-mv-bind: discard more unused variables ... So that default-unknown-values doesn't need to.	2026-09-06 10:01:56 +03:00
Stas Boukarev	db0339a5c4	Resolve a loop in an ash transform	2026-09-06 08:08:07 +03:00
Stas Boukarev	fce3a00f0e	x86-64, default-unknown-values: remove the last CMP	2026-09-06 08:08:07 +03:00
Stas Boukarev	fc5e832ae5	x86-64, default-unknown-values: use cmov for one value	2026-09-06 08:08:07 +03:00
Stas Boukarev	d6fb6072e5	x86-64: default-unknown-values ignore upper unused values completely ... Since the stack doesn't need to be adjusted just pretend they don't exist.	2026-09-06 08:08:07 +03:00
Douglas Katzman	7aadac2648	Revise some outdated comments	2026-09-06 04:48:43 +00:00
Stas Boukarev	790814f16e	Transform (< (abs rational) c) before abs is transformed	2026-09-05 23:40:09 +03:00
Stas Boukarev	783a2ab0e9	Transform (</> (integer-length x) constants)	2026-09-05 22:21:56 +03:00
Stas Boukarev	d3ff35b362	(logcount (* x power-of-two)) => (logcount x) ... It's not always transformed to ash.	2026-09-05 20:45:28 +03:00
Stas Boukarev	6925ac5a0c	(integer-length (lognot x)) => (integer-length x)	2026-09-05 20:07:22 +03:00
Stas Boukarev	67570e5c28	Add some logcount identities ... (logcount (logcount x)) => (logcount x) (logcount (ash x positive)) => (logcount x)	2026-09-05 19:52:25 +03:00
Stas Boukarev	961c8c87dd	Transform (= (logcount x) (integer-length x))	2026-09-05 19:52:19 +03:00
Stas Boukarev	120d5eab81	Add transforms for (= (logcount word-sized) 1)	2026-09-05 18:45:30 +03:00
Stas Boukarev	f8160932da	integer-type-length: work on union types	2026-09-05 18:45:30 +03:00
Stas Boukarev	2865cf89d7	(logxor n (ash n -63)) is (unsigned-byte 63) for (signed-byte 64) n	2026-09-05 18:45:30 +03:00
Douglas Katzman	42549b22cc	De-flake a test of sxhash hash quality ... I asked Gemini to do the math on this, and it said that it's no surprise a flake was seen with only 20 objects. Changing to 60 objects reduces the chance of spurious failure to 1 in 5e-16	2026-09-05 11:14:43 -04:00
Qiantan Hong	65f214fd09	add sb-simd-ssse3:u8.16-shuffle	2026-09-05 14:56:08 +03:00
Stas Boukarev	1fd7b4721e	Catch destination-less casts in leaf-refs-not-escape-elsewhere-p ... Fixes lp#2166183	2026-09-05 14:50:59 +03:00
Stas Boukarev	6e8c214ff1	Use word-sized-lvar-p more ... Addresses an issue reported by Qiantan Hong. Tests by Qiantan Hong.	2026-09-05 14:38:26 +03:00
Stas Boukarev	15c2369194	Resolve some stack pointer issues on nonstop-foreign-call on x86-64 ... scrub_thread_control_stack can't be called on the altstack. sigtrap_handler might be updating th->control_stack_pointer which might become stale.	2026-09-05 07:41:42 +03:00
Stas Boukarev	c100979210	#+ an optional function	2026-09-05 04:11:32 +03:00
Stas Boukarev	37ddc92c87	Stop stack-analyze interfering with direct mv-calls ... They don't push any values.	2026-09-05 04:05:56 +03:00
Stas Boukarev	b193d1f288	Fix calls	2026-09-05 03:11:17 +03:00
Stas Boukarev	9107d0306d	Add --interpret to parallel-exec ... [skip ci]	2026-09-04 22:09:31 +03:00
Stas Boukarev	ea4332cd1f	Disable pass-through-path-p for NLX ... nlx-entry-pass-through currently makes the whole function return, bypassing any other cleanups.	2026-09-04 20:07:50 +03:00
Douglas Katzman	07bb993d53	Disregard .lispwords when marking used labels ... Seems to have caused a problem for 32-bit x86	2026-09-04 16:27:45 +00:00
Stas Boukarev	a1918d4e0a	Include variant-vars when removing inherited ignores from define-vop	2026-09-04 18:41:52 +03:00
Stas Boukarev	82fdbae144	arm64: start implementing pass-through-call ... Add only the VOPs in call.lisp Unbreak the build where pass-through-call is disabled/unimplemented.	2026-09-04 17:46:30 +03:00
Douglas Katzman	f2e1794438	x86-64: Pass-through multiple values across harmless cleanups ... Optimize non-tail full calls followed by "easy" cleanups that do not execute arbitrary forms. "Easy" includes DX allocations, special var unbinding, popping the CATCH block stack. Then the unknown values can be directly returned as if by tail call, bypassing the RETURN-MULTIPLE helper routine. We annotate return LVAR as :pass-through, suppress register restoration after the call (:save-p :compute-only), and emit return as LEAVE+RET. Cleanup VOPs (unbind-n, %catch-breakup) must preserve the carry flag, and use only clobberable temporary registers. For CATCH, decouple the normal return path from the NLX path. The normal path eliminates use of the stack for buffering the mv-save area and then restoring it from the stack.	2026-09-04 04:44:09 +00:00
Stas Boukarev	f9db67bdfc	Actually invoke mv-call-direct-named	2026-09-04 07:37:14 +03:00
Stas Boukarev	136b501c85	assem: avoid emitting instructions with a lot of arguments ... Like lookup tables.	2026-09-04 07:23:40 +03:00
Douglas Katzman	6c4e7b5b72	Fix misplaced paren	2026-09-04 04:20:37 +00:00
Stas Boukarev	a05ed363f0	Allow direct calls for (mv-call x (values 1 2) (funcall y))	2026-09-04 06:58:24 +03:00
Stas Boukarev	d0a6ee5bba	Disable useless optimize notes from (truncate signed-word signed-word)	2026-09-04 06:29:32 +03:00
Douglas Katzman	a894522bb2	Add a slot to SB-ASSEM::STMT ... Separating the slot makes treatment of the prefix much more uniform, and the compiler trace file for x86-64 becomes easier to understand. Previously the trace didn't decode prefix bits in the IR2 dump of assembly instructions in each vop, while the symbolic asm listing showing vops as comments did. The peephole optimizer's instruction matcher is also a little simpler. arm64 could use a similar idea for its vector length prefixes.	2026-09-04 01:45:00 +00:00
Stas Boukarev	213c270561	Better (m-v-call f (g)) on tls-based-mv-return ... Copy the values directly from the return stack. If there are no other non-multiple arguments preceding the call then the registers stay in the same place.	2026-09-04 00:55:16 +03:00
Gabor Melis	289db46c16	sb-manual: do not document unbuilt contribs ... Also, test texinfo generation on github.	2026-09-03 14:09:06 +02:00
Gabor Melis	ccbf25c80d	Improve DEFPACKAGE docstring ... Add pointers to the documentation of non-standard options.	2026-09-03 11:16:52 +02:00