- cmucl-cvs - mailman3.common-lisp.net

[Git][cmucl/cmucl] Pushed new branch issue-381-cmucl-unix-depends-on-os
by Raymond Toy (＠rtoy) 16 Sep '25

16 Sep '25

Raymond Toy pushed new branch issue-381-cmucl-unix-depends-on-os at cmucl / cmucl -- View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/tree/issue-381-cmucl-unix-depe… You're receiving this email because of your account on gitlab.common-lisp.net.

1 0

[Git][cmucl/cmucl] Pushed new branch issue-438-call-out-sets-x87-precision
by Raymond Toy (＠rtoy) 05 Sep '25

05 Sep '25

Raymond Toy pushed new branch issue-438-call-out-sets-x87-precision at cmucl / cmucl -- View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/tree/issue-438-call-out-sets-x… You're receiving this email because of your account on gitlab.common-lisp.net.

1 0

[Git][cmucl/cmucl][master] 2 commits: Fix #431: Fix setting of x87 modes in set-floating-point-modes
by Raymond Toy (＠rtoy) 05 Sep '25

05 Sep '25

Raymond Toy pushed to branch master at cmucl / cmucl Commits: 633e64a4 by Raymond Toy at 2025-09-05T07:55:14-07:00 Fix #431: Fix setting of x87 modes in set-floating-point-modes - - - - - 7ab74e13 by Raymond Toy at 2025-09-05T07:55:15-07:00 Merge branch 'issue-431-fix-set-floating-point-modes' into 'master' Fix #431: Fix setting of x87 modes in set-floating-point-modes Closes #431 See merge request cmucl/cmucl!314 - - - - - 10 changed files: - .gitlab-ci.yml - + src/bootfiles/21e/boot-2025-07.lisp - src/code/float-trap.lisp - src/code/x86-vm.lisp - src/compiler/x86/float-sse2.lisp - src/compiler/x86/parms.lisp - src/general-info/release-21f.md - src/i18n/locale/cmucl-x86-vm.pot - + tests/float-x86.lisp - tests/float.lisp Changes: ===================================== .gitlab-ci.yml ===================================== @@ -4,7 +4,7 @@ variables: download_url: "https://common-lisp.net/project/cmucl/downloads/snapshots/$year/$month" version: "$year-$month-x86" tar_ext: "xz" - bootstrap: "" + bootstrap: "-B boot-2025-07" # Default install configuration to download the cmucl tarballs to use # for building. ===================================== src/bootfiles/21e/boot-2025-07.lisp ===================================== @@ -0,0 +1,25 @@ +;; Bootstrap file to define some constants we use early in the build. + +(in-package :x86) + +(defconstant x87-float-infinity-control-byte + (byte 1 (+ 12 16)) + "The bit in the x87 FPU control word that controls the infinity mode.") +(defconstant x87-float-rounding-mode + (byte 2 (+ 10 16)) + "The bits in the x87 FPU control word for the rounding mode.") +(defconstant x87-float-precision-control-byte + (byte 2 (+ 8 16)) + "The bits in the x87 FPU contol word for the FP operation precision.") +(defconstant x87-float-traps-byte + (byte 6 16) + "The bits in the x87 FPU control word indicating the exceptions that + are enabled.") +(defconstant x87-float-precision-control-alist + `((:24-bits . 0) + (:reserved . 1) + (:53-bits . 2) + (:64-bits . 3)) + "Alist for the x87 precison control. The car is the symbolic + precision and the cdr is the value for the precision control field.") + ===================================== src/code/float-trap.lisp ===================================== @@ -114,21 +114,24 @@ ;; Set the floating point modes for both X87 and SSE2. This ;; include the rounding control bits. (let* ((rc (ldb float-rounding-mode new-mode)) - (new-exceptions (logand #x3f new-mode)) - (new-enables (logand #x3f (ash new-mode -7))) - (x87-modes - (logior new-exceptions - (ash rc 10) - (ash new-enables 16) - ;; Set precision control to be 64-bit, always. We - ;; don't use the x87 registers with sse2, so this - ;; is ok and would be the correct setting if we - ;; ever support long-floats. - (ash 3 (+ 8 16))))) - (setf (vm::sse2-floating-point-modes) (ldb (byte 24 0) new-mode)) - (setf (vm::x87-floating-point-modes) (ldb (byte 24 0) x87-modes))) - new-mode) - ) + (new-exceptions (ldb float-exceptions-byte new-mode)) + (new-enables (ldb float-traps-byte new-mode)) + (x87-modes 0)) + ;; From the new mode, update the x87 FPU mode with the same set + ;; of accrued exceptions, enabled exceptions, and rounding mode. + (setf x87-modes (dpb new-exceptions float-exceptions-byte x87-modes)) + (setf x87-modes (dpb new-enables x87-float-traps-byte x87-modes)) + (setf x87-modes (dpb rc x87-float-rounding-mode x87-modes)) + + ;; Set precision to 64-bits (double-extended). + (setf x87-modes + (dpb (cdr (assoc :64-bits x87-float-precision-control-alist)) + x87-float-precision-control-byte + x87-modes)) + + (setf (vm::sse2-floating-point-modes) (ldb (byte 16 0) new-mode)) + (setf (vm::x87-floating-point-modes) (ldb (byte 30 0) x87-modes))) + new-mode)) #+(and sse2 darwin) (progn ===================================== src/code/x86-vm.lisp ===================================== @@ -733,3 +733,88 @@ (multiple-value-bind (fop dst src) (get-fp-operation scp) (values fop (list dst src)))) + +;; See src/compiler/x86/float-sse2.lisp for a description of the x87 +;; control and status words. + +(defconstant x87-float-infinity-control-byte + (byte 1 (+ 12 16)) + "The bit in the x87 FPU control word that controls the infinity mode.") + +(defconstant x87-float-rounding-mode + (byte 2 (+ 10 16)) + "The bits in the x87 FPU control word for the rounding mode.") + +(defconstant x87-float-precision-control-byte + (byte 2 (+ 8 16)) + "The bits in the x87 FPU contol word for the FP operation precision.") + +(defconstant x87-float-traps-byte + (byte 6 16) + "The bits in the x87 FPU control word indicating the exceptions that + are enabled.") + +(defconstant x87-float-precision-control-alist + `((:24-bits . 0) + (:reserved . 1) + (:53-bits . 2) + (:64-bits . 3)) + "Alist for the x87 precison control. The car is the symbolic + precision and the cdr is the value for the precision control field.") + +(defun print-fp-exceptions-enabled (enabled) + (format t "Precision enable: ~30T~A~%" (ldb (byte 1 5) enabled)) + (format t "Underflow enable: ~30T~A~%" (ldb (byte 1 4) enabled)) + (format t "Overflow enable: ~30T~A~%" (ldb (byte 1 3) enabled)) + (format t "Divide-by-zero enable: ~30T~A~%" (ldb (byte 1 2) enabled)) + (format t "Denormal op enable: ~30T~A~%" (ldb (byte 1 1) enabled)) + (format t "Invalid op enable: ~30T~A~%" (ldb (byte 1 0) enabled))) + +(defun print-fp-current-exceptions (current) + (format t "Precision flag: ~30T~A~%" (ldb (byte 1 5) current)) + (format t "Underflow flag: ~30T~A~%" (ldb (byte 1 4) current)) + (format t "Overflow flag: ~30T~A~%" (ldb (byte 1 3) current)) + (format t "Divide-by-zero flag: ~30T~A~%" (ldb (byte 1 2) current)) + (format t "Denormal op flag: ~30T~A~%" (ldb (byte 1 1) current)) + (format t "Invalid op flag: ~30T~A~%" (ldb (byte 1 0) current))) + +(defun print-sse2-fp-modes (&optional (sse-mode (sse2-floating-point-modes))) + "Print SSE2 floating modes word in a human-readable fashion." + ;; Note that Intel uses masks to disable the exception, but to match + ;; the rest of cmucl, these bits are represented as enable bits. + (format t "Flush-to-zero: ~30T~A~%" (ldb (byte 1 15) sse-mode)) + (let ((rc (ldb float-rounding-mode sse-mode))) + (format t "Rounding control: ~30T#b~2,'0b ~S~%" + rc + (car (rassoc rc rounding-mode-alist)))) + (print-fp-exceptions-enabled (ldb float-traps-byte sse-mode)) + (print-fp-current-exceptions (ldb float-exceptions-byte sse-mode))) + +(defun print-x87-fp-modes (&optional (x87-mode (x87-floating-point-modes))) + "Print X87 floating modes word in a human-readable fashion." + ;; Note that Intel uses masks to disable the exception, but to match + ;; the rest of cmucl, these bits are represented as enable bits. + (format t "Status word:~%") + (format t "FPU busy: ~30T~A~%" (ldb (byte 1 15) x87-mode)) + (format t "Condition code C3: ~30T~A~%" (ldb (byte 1 14) x87-mode)) + (format t "Top of stack: ~30T~D~%" (ldb (byte 3 11) x87-mode)) + (format t "Condition code C2: ~30T~A~%" (ldb (byte 1 10) x87-mode)) + (format t "Condition code C1: ~30T~A~%" (ldb (byte 1 9) x87-mode)) + (format t "Condition code C0: ~30T~A~%" (ldb (byte 1 8) x87-mode)) + (format t "Error summary: ~30T~A~%" (ldb (byte 1 7) x87-mode)) + (format t "Stack fault: ~30T~A~%" (ldb (byte 1 6) x87-mode)) + (print-fp-current-exceptions (ldb float-exceptions-byte x87-mode)) + (format t "~%Control word:~%") + (format t "Reserved: ~30T#b~2,'0b~%" (ldb (byte 2 (+ 13 16)) x87-mode)) + (format t "Infinity control: ~30T~A~%" (ldb x87-float-infinity-control-byte x87-mode)) + (let ((rc (ldb x87-float-rounding-mode x87-mode))) + (format t "Rounding control: ~30T#b~2,'0b ~S~%" + rc + (car (rassoc rc rounding-mode-alist)))) + (let ((pc (ldb x87-float-precision-control-byte x87-mode))) + (format t "Precision control: ~30T#b~2,'0b ~S~%" + pc + (car (rassoc pc x87-float-precision-control-alist)))) + (format t "Reserved: ~30T#b~2,'0b~%" (ldb (byte 2 (+ 6 16)) x87-mode)) + (print-fp-exceptions-enabled (ldb x87-float-traps-byte x87-mode))) + ===================================== src/compiler/x86/float-sse2.lisp ===================================== @@ -1288,6 +1288,10 @@ ;; 0 invalid operation flag ;; ;; See below for rounding control +;; +;; Also see float-rounding-mode, float-sticky-bits, float-traps-byte, +;; and float-exceptions-byte in compiler/x86/parms.lisp. + (defknown sse2-floating-point-modes () float-modes (flushable)) (defknown ((setf sse2-floating-point-modes)) (float-modes) float-modes) ===================================== src/compiler/x86/parms.lisp ===================================== @@ -205,9 +205,15 @@ ;; need to do it this way because the interface assumes the modes are ;; in the same order as the MXCSR register. (defconstant float-rounding-mode (byte 2 13)) -(defconstant float-sticky-bits (byte 6 0)) -(defconstant float-traps-byte (byte 6 7)) -(defconstant float-exceptions-byte (byte 6 0)) +(defconstant float-sticky-bits (byte 6 0) + "The bits in the FP mode that hold the accrued exceptions that have + occurred since the bits were reset.") +(defconstant float-traps-byte (byte 6 7) + "The bits in the FP mode that hold FP exceptions that are enabled.") +(defconstant float-exceptions-byte (byte 6 0) + "The bits in the FP mode that hold the current exception. However + for x86, there aren't separate bits for this, so we use the stick + bits from the accrued exceptions") (progn ;; SSE2 has a flush-to-zero flag, which we use as the fast bit. Some ===================================== src/general-info/release-21f.md ===================================== @@ -139,6 +139,7 @@ public domain. package-local-nicknames) * #424 Use fdlibm `hypot` to fix bug in snapshot 2025-07 * #426 Define float-modes type correctly for `(setf (x87-set-floating-point-modes))` + * $431 Fix setting of x87 FP modes in `set-floating-point-modes` * Other changes: * Improvements to the PCL implementation of CLOS: * Changes to building procedure: ===================================== src/i18n/locale/cmucl-x86-vm.pot ===================================== @@ -77,6 +77,38 @@ msgstr "" msgid "Thread safe push of val onto the list in the vector element." msgstr "" +#: src/code/x86-vm.lisp +msgid "The bit in the x87 FPU control word that controls the infinity mode." +msgstr "" + +#: src/code/x86-vm.lisp +msgid "The bits in the x87 FPU control word for the rounding mode." +msgstr "" + +#: src/code/x86-vm.lisp +msgid "The bits in the x87 FPU contol word for the FP operation precision." +msgstr "" + +#: src/code/x86-vm.lisp +msgid "" +"The bits in the x87 FPU control word indicating the exceptions that\n" +" are enabled." +msgstr "" + +#: src/code/x86-vm.lisp +msgid "" +"Alist for the x87 precison control. The car is the symbolic\n" +" precision and the cdr is the value for the precision control field." +msgstr "" + +#: src/code/x86-vm.lisp +msgid "Print SSE2 floating modes word in a human-readable fashion." +msgstr "" + +#: src/code/x86-vm.lisp +msgid "Print X87 floating modes word in a human-readable fashion." +msgstr "" + #: src/code/load.lisp msgid "Top-Level Form" msgstr "" @@ -136,6 +168,23 @@ msgstr "" msgid "Maximum number of bits in a positive fixnum" msgstr "" +#: src/compiler/x86/parms.lisp +msgid "" +"The bits in the FP mode that hold the accrued exceptions that have\n" +" occurred since the bits were reset." +msgstr "" + +#: src/compiler/x86/parms.lisp +msgid "The bits in the FP mode that hold FP exceptions that are enabled." +msgstr "" + +#: src/compiler/x86/parms.lisp +msgid "" +"The bits in the FP mode that hold the current exception. However\n" +" for x86, there aren't separate bits for this, so we use the stick\n" +" bits from the accrued exceptions" +msgstr "" + #: src/compiler/x86/vm.lisp msgid "Redefining SC number ~D from ~S to ~S." msgstr "" ===================================== tests/float-x86.lisp ===================================== @@ -0,0 +1,59 @@ +;; Tests of float functions + +(defpackage :float-x86-tests + (:use :cl :lisp-unit)) + +(in-package "FLOAT-X86-TESTS") + +(define-test set-floating-point-modes + (let ((old-x87-modes (x86::x87-floating-point-modes)) + (old-sse2-modes (x86::sse2-floating-point-modes)) + x87-modes sse2-modes) + (unwind-protect + (progn + ;; Set some new traps and rounding mode. + (ext:set-floating-point-modes :traps '(:underflow + :overflow + :invalid + :divide-by-zero) + :rounding-mode :zero) + ;; Save these new FP modes + (setf x87-modes (x86::x87-floating-point-modes)) + (setf sse2-modes (x86::sse2-floating-point-modes))) + + (setf (x86::x87-floating-point-modes) old-x87-modes) + (setf (x86::sse2-floating-point-modes) old-sse2-modes)) + + (let* ((x87-exceptions-enabled (ldb x86::x87-float-traps-byte x87-modes)) + (x87-rc (ldb x86::x87-float-rounding-mode x87-modes)) + (sse2-exceptions-enabled (ldb x86::float-traps-byte sse2-modes)) + (sse2-rc (ldb x86::float-rounding-mode sse2-modes))) + (format t "*X87 FP mode words:~%") + (x86::print-x87-fp-modes x87-modes) + (format t "~%*SSE2 FP mode words:~%") + (x86::print-sse2-fp-modes sse2-modes) + + ;; Verify that we set the enabled exceptions + ;; correctly. First for sse2, then for x87. + (assert-false (logbitp 5 sse2-exceptions-enabled)) ; precision + (assert-true (logbitp 4 sse2-exceptions-enabled)) ; underflow + (assert-true (logbitp 3 sse2-exceptions-enabled)) ; overflow + (assert-true (logbitp 2 sse2-exceptions-enabled)) ; divide-by-zero + (assert-false (logbitp 1 sse2-exceptions-enabled)) ; denormal + (assert-true (logbitp 0 sse2-exceptions-enabled)) ; invalid + + (assert-false (logbitp 5 x87-exceptions-enabled)) ; precision + (assert-true (logbitp 4 x87-exceptions-enabled)) ; underflow + (assert-true (logbitp 3 x87-exceptions-enabled)) ; overflow + (assert-true (logbitp 2 x87-exceptions-enabled)) ; divide-by-zero + (assert-false (logbitp 1 x87-exceptions-enabled)) ; denormal + (assert-true (logbitp 0 x87-exceptions-enabled)) ; invalid + + ;; Verify the rounding mode is set to zero + (assert-eql :zero (car (rassoc sse2-rc x86::rounding-mode-alist))) + (assert-eql :zero (car (rassoc x87-rc x86::rounding-mode-alist))) + + ;; Verify precision for x87 + (assert-eql :64-bits + (car (rassoc (ldb x86::x87-float-precision-control-byte x87-modes) + x86::x87-float-precision-control-alist)))))) ===================================== tests/float.lisp ===================================== @@ -342,4 +342,3 @@ (x86::x87-floating-point-modes))))) (assert-true (typep new-mode 'x86::float-modes)) (assert-equal new-mode (setf (x86::x87-floating-point-modes) new-mode)))) - View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/compare/c65554a544fc86c05e0f0f… -- View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/compare/c65554a544fc86c05e0f0f… You're receiving this email because of your account on gitlab.common-lisp.net.

1 0

[Git][cmucl/cmucl][issue-431-fix-set-floating-point-modes] Oops. Done.
by Raymond Toy (＠rtoy) 05 Sep '25

05 Sep '25

Raymond Toy pushed to branch issue-431-fix-set-floating-point-modes at cmucl / cmucl Commits: 70e174a5 by Raymond Toy at 2025-09-05T06:48:57-07:00 Oops. Done. - - - - - 1 changed file: - tests/float-x86.lisp Changes: ===================================== tests/float-x86.lisp ===================================== @@ -56,6 +56,4 @@ ;; Verify precision for x87 (assert-eql :64-bits (car (rassoc (ldb x86::x87-float-precision-control-byte x87-modes) - x86::x87-float-precision-control-alist)))) - ) - ) + x86::x87-float-precision-control-alist)))))) View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/commit/70e174a5eb8f8bd7b1abb19… -- View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/commit/70e174a5eb8f8bd7b1abb19… You're receiving this email because of your account on gitlab.common-lisp.net.

1 0

[Git][cmucl/cmucl][issue-425-correctly-rounded-math-functions-single-float] Include core-math 32-bit functions in lisp.a
by Raymond Toy (＠rtoy) 03 Sep '25

03 Sep '25

Raymond Toy pushed to branch issue-425-correctly-rounded-math-functions-single-float at cmucl / cmucl Commits: 8c957e76 by Raymond Toy at 2025-09-03T12:26:49-07:00 Include core-math 32-bit functions in lisp.a When building lisp.a, we were only adding CORE64_OBJS and forgot to include CORE32_OBJS. To do this, lisp.a needs to include CORE_MATH_OBJS. - - - - - 1 changed file: - src/lisp/GNUmakefile Changes: ===================================== src/lisp/GNUmakefile ===================================== @@ -60,7 +60,7 @@ lisp: ${OBJS} ${CORE_MATH_OBJS} version.o # Create a library out of all the object files so we can build an # executable. However, we need to remove exec-init.o from the library lisp.a: version.o ${OBJS} ${CORE_MATH_OBJS} ${EXEC_FINAL_OBJ} - $(AR) crs lisp.a ${OBJS} ${CORE64_OBJS} version.o + $(AR) crs lisp.a ${OBJS} ${CORE_MATH_OBJS} version.o ifneq (${EXEC_FINAL_OBJ},) $(AR) d lisp.a exec-init.o endif View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/commit/8c957e766463e0bac0d2d14… -- View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/commit/8c957e766463e0bac0d2d14… You're receiving this email because of your account on gitlab.common-lisp.net.

1 0

[Git][cmucl/cmucl][issue-431-fix-set-floating-point-modes] 2 commits: Address review comments
by Raymond Toy (＠rtoy) 02 Sep '25

02 Sep '25

Raymond Toy pushed to branch issue-431-fix-set-floating-point-modes at cmucl / cmucl Commits: 67432557 by Raymond Toy at 2025-09-02T15:52:58-07:00 Address review comments Move new x87 constants and FP mode word printers from compiler/x86/float-sse2.lisp to code/x86-vm.lisp. Move new test from tests/float.lisp to the new file tests/float-x86.lisp. - - - - - 5002a779 by Raymond Toy at 2025-09-02T16:12:27-07:00 Address review comments Reduce scope of `unwind-protect` in tests/float-x86.lisp. Also print out the sse2 FP mode word, just like we do for the x87 FP mode word. - - - - - 6 changed files: - src/code/x86-vm.lisp - src/compiler/x86/float-sse2.lisp - src/i18n/locale/cmucl-sse2.pot - src/i18n/locale/cmucl-x86-vm.pot - + tests/float-x86.lisp - tests/float.lisp Changes: ===================================== src/code/x86-vm.lisp ===================================== @@ -733,3 +733,88 @@ (multiple-value-bind (fop dst src) (get-fp-operation scp) (values fop (list dst src)))) + +;; See src/compiler/x86/float-sse2.lisp for a description of the x87 +;; control and status words. + +(defconstant x87-float-infinity-control-byte + (byte 1 (+ 12 16)) + "The bit in the x87 FPU control word that controls the infinity mode.") + +(defconstant x87-float-rounding-mode + (byte 2 (+ 10 16)) + "The bits in the x87 FPU control word for the rounding mode.") + +(defconstant x87-float-precision-control-byte + (byte 2 (+ 8 16)) + "The bits in the x87 FPU contol word for the FP operation precision.") + +(defconstant x87-float-traps-byte + (byte 6 16) + "The bits in the x87 FPU control word indicating the exceptions that + are enabled.") + +(defconstant x87-float-precision-control-alist + `((:24-bits . 0) + (:reserved . 1) + (:53-bits . 2) + (:64-bits . 3)) + "Alist for the x87 precison control. The car is the symbolic + precision and the cdr is the value for the precision control field.") + +(defun print-fp-exceptions-enabled (enabled) + (format t "Precision enable: ~30T~A~%" (ldb (byte 1 5) enabled)) + (format t "Underflow enable: ~30T~A~%" (ldb (byte 1 4) enabled)) + (format t "Overflow enable: ~30T~A~%" (ldb (byte 1 3) enabled)) + (format t "Divide-by-zero enable: ~30T~A~%" (ldb (byte 1 2) enabled)) + (format t "Denormal op enable: ~30T~A~%" (ldb (byte 1 1) enabled)) + (format t "Invalid op enable: ~30T~A~%" (ldb (byte 1 0) enabled))) + +(defun print-fp-current-exceptions (current) + (format t "Precision flag: ~30T~A~%" (ldb (byte 1 5) current)) + (format t "Underflow flag: ~30T~A~%" (ldb (byte 1 4) current)) + (format t "Overflow flag: ~30T~A~%" (ldb (byte 1 3) current)) + (format t "Divide-by-zero flag: ~30T~A~%" (ldb (byte 1 2) current)) + (format t "Denormal op flag: ~30T~A~%" (ldb (byte 1 1) current)) + (format t "Invalid op flag: ~30T~A~%" (ldb (byte 1 0) current))) + +(defun print-sse2-fp-modes (&optional (sse-mode (sse2-floating-point-modes))) + "Print SSE2 floating modes word in a human-readable fashion." + ;; Note that Intel uses masks to disable the exception, but to match + ;; the rest of cmucl, these bits are represented as enable bits. + (format t "Flush-to-zero: ~30T~A~%" (ldb (byte 1 15) sse-mode)) + (let ((rc (ldb float-rounding-mode sse-mode))) + (format t "Rounding control: ~30T#b~2,'0b ~S~%" + rc + (car (rassoc rc rounding-mode-alist)))) + (print-fp-exceptions-enabled (ldb float-traps-byte sse-mode)) + (print-fp-current-exceptions (ldb float-exceptions-byte sse-mode))) + +(defun print-x87-fp-modes (&optional (x87-mode (x87-floating-point-modes))) + "Print X87 floating modes word in a human-readable fashion." + ;; Note that Intel uses masks to disable the exception, but to match + ;; the rest of cmucl, these bits are represented as enable bits. + (format t "Status word:~%") + (format t "FPU busy: ~30T~A~%" (ldb (byte 1 15) x87-mode)) + (format t "Condition code C3: ~30T~A~%" (ldb (byte 1 14) x87-mode)) + (format t "Top of stack: ~30T~D~%" (ldb (byte 3 11) x87-mode)) + (format t "Condition code C2: ~30T~A~%" (ldb (byte 1 10) x87-mode)) + (format t "Condition code C1: ~30T~A~%" (ldb (byte 1 9) x87-mode)) + (format t "Condition code C0: ~30T~A~%" (ldb (byte 1 8) x87-mode)) + (format t "Error summary: ~30T~A~%" (ldb (byte 1 7) x87-mode)) + (format t "Stack fault: ~30T~A~%" (ldb (byte 1 6) x87-mode)) + (print-fp-current-exceptions (ldb float-exceptions-byte x87-mode)) + (format t "~%Control word:~%") + (format t "Reserved: ~30T#b~2,'0b~%" (ldb (byte 2 (+ 13 16)) x87-mode)) + (format t "Infinity control: ~30T~A~%" (ldb x87-float-infinity-control-byte x87-mode)) + (let ((rc (ldb x87-float-rounding-mode x87-mode))) + (format t "Rounding control: ~30T#b~2,'0b ~S~%" + rc + (car (rassoc rc rounding-mode-alist)))) + (let ((pc (ldb x87-float-precision-control-byte x87-mode))) + (format t "Precision control: ~30T#b~2,'0b ~S~%" + pc + (car (rassoc pc x87-float-precision-control-alist)))) + (format t "Reserved: ~30T#b~2,'0b~%" (ldb (byte 2 (+ 6 16)) x87-mode)) + (print-fp-exceptions-enabled (ldb x87-float-traps-byte x87-mode))) + ===================================== src/compiler/x86/float-sse2.lisp ===================================== @@ -1379,83 +1379,6 @@ ;; 10 double precision (53 bits) ;; 11 double extended precision (64 bits) -(defconstant x87-float-infinity-control-byte - (byte 1 (+ 12 16)) - "The bit in the x87 FPU control word that controls the infinity mode.") -(defconstant x87-float-rounding-mode - (byte 2 (+ 10 16)) - "The bits in the x87 FPU control word for the rounding mode.") -(defconstant x87-float-precision-control-byte - (byte 2 (+ 8 16)) - "The bits in the x87 FPU contol word for the FP operation precision.") -(defconstant x87-float-traps-byte - (byte 6 16) - "The bits in the x87 FPU control word indicating the exceptions that - are enabled.") -(defconstant x87-float-precision-control-alist - `((:24-bits . 0) - (:reserved . 1) - (:53-bits . 2) - (:64-bits . 3)) - "Alist for the x87 precison control. The car is the symbolic - precision and the cdr is the value for the precision control field.") - -(defun print-fp-exceptions-enabled (enabled) - (format t "Precision enable: ~30T~A~%" (ldb (byte 1 5) enabled)) - (format t "Underflow enable: ~30T~A~%" (ldb (byte 1 4) enabled)) - (format t "Overflow enable: ~30T~A~%" (ldb (byte 1 3) enabled)) - (format t "Divide-by-zero enable: ~30T~A~%" (ldb (byte 1 2) enabled)) - (format t "Denormal op enable: ~30T~A~%" (ldb (byte 1 1) enabled)) - (format t "Invalid op enable: ~30T~A~%" (ldb (byte 1 0) enabled))) - -(defun print-fp-current-exceptions (current) - (format t "Precision flag: ~30T~A~%" (ldb (byte 1 5) current)) - (format t "Underflow flag: ~30T~A~%" (ldb (byte 1 4) current)) - (format t "Overflow flag: ~30T~A~%" (ldb (byte 1 3) current)) - (format t "Divide-by-zero flag: ~30T~A~%" (ldb (byte 1 2) current)) - (format t "Denormal op flag: ~30T~A~%" (ldb (byte 1 1) current)) - (format t "Invalid op flag: ~30T~A~%" (ldb (byte 1 0) current))) - -(defun print-sse2-fp-modes (&optional (sse-mode (sse2-floating-point-modes))) - "Print SSE2 modes word in a readable fashion." - ;; Note that Intel uses masks to disable the exception, but to match - ;; the rest of cmucl, these bits are represented as enable bits. - (format t "Flush-to-zero: ~30T~A~%" (ldb (byte 1 15) sse-mode)) - (let ((rc (ldb float-rounding-mode sse-mode))) - (format t "Rounding control: ~30T#b~2,'0b ~S~%" - rc - (car (rassoc rc rounding-mode-alist)))) - (print-fp-exceptions-enabled (ldb float-traps-byte sse-mode)) - (print-fp-current-exceptions (ldb float-exceptions-byte sse-mode))) - -(defun print-x87-fp-modes (&optional (x87-mode (x87-floating-point-modes))) - "Print X87 floating modes word in a readable fashion." - ;; Note that Intel uses masks to disable the exception, but to match - ;; the rest of cmucl, these bits are represented as enable bits. - (format t "Status word:~%") - (format t "FPU busy: ~30T~A~%" (ldb (byte 1 15) x87-mode)) - (format t "Condition code C3: ~30T~A~%" (ldb (byte 1 14) x87-mode)) - (format t "Top of stack: ~30T~D~%" (ldb (byte 3 11) x87-mode)) - (format t "Condition code C2: ~30T~A~%" (ldb (byte 1 10) x87-mode)) - (format t "Condition code C1: ~30T~A~%" (ldb (byte 1 9) x87-mode)) - (format t "Condition code C0: ~30T~A~%" (ldb (byte 1 8) x87-mode)) - (format t "Error summary: ~30T~A~%" (ldb (byte 1 7) x87-mode)) - (format t "Stack fault: ~30T~A~%" (ldb (byte 1 6) x87-mode)) - (print-fp-current-exceptions (ldb float-exceptions-byte x87-mode)) - (format t "~%Control word:~%") - (format t "Reserved: ~30T#b~2,'0b~%" (ldb (byte 2 (+ 13 16)) x87-mode)) - (format t "Infinity control: ~30T~A~%" (ldb x87-float-infinity-control-byte x87-mode)) - (let ((rc (ldb x87-float-rounding-mode x87-mode))) - (format t "Rounding control: ~30T#b~2,'0b ~S~%" - rc - (car (rassoc rc rounding-mode-alist)))) - (let ((pc (ldb x87-float-precision-control-byte x87-mode))) - (format t "Precision control: ~30T#b~2,'0b ~S~%" - pc - (car (rassoc pc x87-float-precision-control-alist)))) - (format t "Reserved: ~30T#b~2,'0b~%" (ldb (byte 2 (+ 6 16)) x87-mode)) - (print-fp-exceptions-enabled (ldb x87-float-traps-byte x87-mode))) - (defknown x87-floating-point-modes () float-modes (flushable)) (defknown ((setf x87-floating-point-modes)) (float-modes) float-modes) ===================================== src/i18n/locale/cmucl-sse2.pot ===================================== @@ -86,38 +86,6 @@ msgstr "" msgid "inline ftruncate" msgstr "" -#: src/compiler/x86/float-sse2.lisp -msgid "The bit in the x87 FPU control word that controls the infinity mode." -msgstr "" - -#: src/compiler/x86/float-sse2.lisp -msgid "The bits in the x87 FPU control word for the rounding mode." -msgstr "" - -#: src/compiler/x86/float-sse2.lisp -msgid "The bits in the x87 FPU contol word for the FP operation precision." -msgstr "" - -#: src/compiler/x86/float-sse2.lisp -msgid "" -"The bits in the x87 FPU control word indicating the exceptions that\n" -" are enabled." -msgstr "" - -#: src/compiler/x86/float-sse2.lisp -msgid "" -"Alist for the x87 precison control. The car is the symbolic\n" -" precision and the cdr is the value for the precision control field." -msgstr "" - -#: src/compiler/x86/float-sse2.lisp -msgid "Print SSE2 modes word in a readable fashion." -msgstr "" - -#: src/compiler/x86/float-sse2.lisp -msgid "Print X87 floating modes word in a readable fashion." -msgstr "" - #: src/compiler/x86/float-sse2.lisp msgid "inline complex single-float creation" msgstr "" ===================================== src/i18n/locale/cmucl-x86-vm.pot ===================================== @@ -77,6 +77,38 @@ msgstr "" msgid "Thread safe push of val onto the list in the vector element." msgstr "" +#: src/code/x86-vm.lisp +msgid "The bit in the x87 FPU control word that controls the infinity mode." +msgstr "" + +#: src/code/x86-vm.lisp +msgid "The bits in the x87 FPU control word for the rounding mode." +msgstr "" + +#: src/code/x86-vm.lisp +msgid "The bits in the x87 FPU contol word for the FP operation precision." +msgstr "" + +#: src/code/x86-vm.lisp +msgid "" +"The bits in the x87 FPU control word indicating the exceptions that\n" +" are enabled." +msgstr "" + +#: src/code/x86-vm.lisp +msgid "" +"Alist for the x87 precison control. The car is the symbolic\n" +" precision and the cdr is the value for the precision control field." +msgstr "" + +#: src/code/x86-vm.lisp +msgid "Print SSE2 floating modes word in a human-readable fashion." +msgstr "" + +#: src/code/x86-vm.lisp +msgid "Print X87 floating modes word in a human-readable fashion." +msgstr "" + #: src/code/load.lisp msgid "Top-Level Form" msgstr "" ===================================== tests/float-x86.lisp ===================================== @@ -0,0 +1,61 @@ +;; Tests of float functions + +(defpackage :float-x86-tests + (:use :cl :lisp-unit)) + +(in-package "FLOAT-X86-TESTS") + +(define-test set-floating-point-modes + (let ((old-x87-modes (x86::x87-floating-point-modes)) + (old-sse2-modes (x86::sse2-floating-point-modes)) + x87-modes sse2-modes) + (unwind-protect + (progn + ;; Set some new traps and rounding mode. + (ext:set-floating-point-modes :traps '(:underflow + :overflow + :invalid + :divide-by-zero) + :rounding-mode :zero) + ;; Save these new FP modes + (setf x87-modes (x86::x87-floating-point-modes)) + (setf sse2-modes (x86::sse2-floating-point-modes))) + + (setf (x86::x87-floating-point-modes) old-x87-modes) + (setf (x86::sse2-floating-point-modes) old-sse2-modes)) + + (let* ((x87-exceptions-enabled (ldb x86::x87-float-traps-byte x87-modes)) + (x87-rc (ldb x86::x87-float-rounding-mode x87-modes)) + (sse2-exceptions-enabled (ldb x86::float-traps-byte sse2-modes)) + (sse2-rc (ldb x86::float-rounding-mode sse2-modes))) + (format t "*X87 FP mode words:~%") + (x86::print-x87-fp-modes x87-modes) + (format t "~%*SSE2 FP mode words:~%") + (x86::print-sse2-fp-modes sse2-modes) + + ;; Verify that we set the enabled exceptions + ;; correctly. First for sse2, then for x87. + (assert-false (logbitp 5 sse2-exceptions-enabled)) ; precision + (assert-true (logbitp 4 sse2-exceptions-enabled)) ; underflow + (assert-true (logbitp 3 sse2-exceptions-enabled)) ; overflow + (assert-true (logbitp 2 sse2-exceptions-enabled)) ; divide-by-zero + (assert-false (logbitp 1 sse2-exceptions-enabled)) ; denormal + (assert-true (logbitp 0 sse2-exceptions-enabled)) ; invalid + + (assert-false (logbitp 5 x87-exceptions-enabled)) ; precision + (assert-true (logbitp 4 x87-exceptions-enabled)) ; underflow + (assert-true (logbitp 3 x87-exceptions-enabled)) ; overflow + (assert-true (logbitp 2 x87-exceptions-enabled)) ; divide-by-zero + (assert-false (logbitp 1 x87-exceptions-enabled)) ; denormal + (assert-true (logbitp 0 x87-exceptions-enabled)) ; invalid + + ;; Verify the rounding mode is set to zero + (assert-eql :zero (car (rassoc sse2-rc x86::rounding-mode-alist))) + (assert-eql :zero (car (rassoc x87-rc x86::rounding-mode-alist))) + + ;; Verify precision for x87 + (assert-eql :64-bits + (car (rassoc (ldb x86::x87-float-precision-control-byte x87-modes) + x86::x87-float-precision-control-alist)))) + ) + ) ===================================== tests/float.lisp ===================================== @@ -342,49 +342,3 @@ (x86::x87-floating-point-modes))))) (assert-true (typep new-mode 'x86::float-modes)) (assert-equal new-mode (setf (x86::x87-floating-point-modes) new-mode)))) - -(define-test set-floating-point-modes - (let ((old-x87-modes (x86::x87-floating-point-modes)) - (old-sse2-modes (x86::sse2-floating-point-modes))) - (unwind-protect - (progn - ;; Set some new traps and rounding mode - (ext:set-floating-point-modes :traps '(:underflow - :overflow - :invalid - :divide-by-zero) - :rounding-mode :zero) - (let* ((x87-modes (x86::x87-floating-point-modes)) - (x87-exceptions-enabled (ldb x86::x87-float-traps-byte x87-modes)) - (x87-rc (ldb x86::x87-float-rounding-mode x87-modes)) - (sse2-modes (x86::sse2-floating-point-modes)) - (sse2-exceptions-enabled (ldb x86::float-traps-byte sse2-modes)) - (sse2-rc (ldb x86::float-rounding-mode sse2-modes))) - (x86::print-x87-fp-modes x87-modes) - ;; Verify that we set the enabled exceptions - ;; correctly. First for sse2, then for x87. - (assert-false (logbitp 5 sse2-exceptions-enabled)) ; precision - (assert-true (logbitp 4 sse2-exceptions-enabled)) ; underflow - (assert-true (logbitp 3 sse2-exceptions-enabled)) ; overflow - (assert-true (logbitp 2 sse2-exceptions-enabled)) ; divide-by-zero - (assert-false (logbitp 1 sse2-exceptions-enabled)) ; denormal - (assert-true (logbitp 0 sse2-exceptions-enabled)) ; invalid - - (assert-false (logbitp 5 x87-exceptions-enabled)) ; precision - (assert-true (logbitp 4 x87-exceptions-enabled)) ; underflow - (assert-true (logbitp 3 x87-exceptions-enabled)) ; overflow - (assert-true (logbitp 2 x87-exceptions-enabled)) ; divide-by-zero - (assert-false (logbitp 1 x87-exceptions-enabled)) ; denormal - (assert-true (logbitp 0 x87-exceptions-enabled)) ; invalid - - ;; Verify the rounding mode is set to zero - (assert-eql :zero (car (rassoc sse2-rc x86::rounding-mode-alist))) - (assert-eql :zero (car (rassoc x87-rc x86::rounding-mode-alist))) - - ;; Verify precision for x87 - (assert-eql :64-bits - (car (rassoc (ldb x86::x87-float-precision-control-byte x87-modes) - x86::x87-float-precision-control-alist))))) - - (setf (x86::x87-floating-point-modes) old-x87-modes) - (setf (x86::sse2-floating-point-modes) old-sse2-modes)))) View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/compare/12cae745fdb002782b06b4… -- View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/compare/12cae745fdb002782b06b4… You're receiving this email because of your account on gitlab.common-lisp.net.

1 0

[Git][cmucl/cmucl][issue-431-fix-set-floating-point-modes] Address review comments
by Raymond Toy (＠rtoy) 21 Aug '25

21 Aug '25

Raymond Toy pushed to branch issue-431-fix-set-floating-point-modes at cmucl / cmucl Commits: 12cae745 by Raymond Toy at 2025-08-21T07:30:12-07:00 Address review comments - - - - - 1 changed file: - src/code/float-trap.lisp Changes: ===================================== src/code/float-trap.lisp ===================================== @@ -109,7 +109,6 @@ (ash x87-enables 7)))) final-mode)) - (defun (setf floating-point-modes) (new-mode) (declare (type (unsigned-byte 32) new-mode)) ;; Set the floating point modes for both X87 and SSE2. This View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/commit/12cae745fdb002782b06b40… -- View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/commit/12cae745fdb002782b06b40… You're receiving this email because of your account on gitlab.common-lisp.net.

1 0

[Git][cmucl/cmucl][issue-425-correctly-rounded-math-functions-single-float] Add coshf.c
by Raymond Toy (＠rtoy) 21 Aug '25

21 Aug '25

Raymond Toy pushed to branch issue-425-correctly-rounded-math-functions-single-float at cmucl / cmucl Commits: f42458c7 by Raymond Toy at 2025-08-20T18:49:14-07:00 Add coshf.c Forgot to add this file - - - - - 1 changed file: - + src/lisp/core-math/src/binary32/cosh/coshf.c Changes: ===================================== src/lisp/core-math/src/binary32/cosh/coshf.c ===================================== @@ -0,0 +1,133 @@ +/* Correctly-rounded hyperbolic cosine function for binary32 value. + +Copyright (c) 2022-2023 Alexei Sibidanov. + +This file is part of the CORE-MATH project +(https://core-math.gitlabpages.inria.fr/). + +Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), to deal +in the Software without restriction, including without limitation the rights +to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +copies of the Software, and to permit persons to whom the Software is +furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in all +copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE +SOFTWARE. +*/ + +#include <stdint.h> +#include <errno.h> + +// Warning: clang also defines __GNUC__ +#if defined(__GNUC__) && !defined(__clang__) +#pragma GCC diagnostic ignored "-Wunknown-pragmas" +#endif + +#pragma STDC FENV_ACCESS ON + +/* __builtin_roundeven was introduced in gcc 10: + https://gcc.gnu.org/gcc-10/changes.html, + and in clang 17 */ +#if ((defined(__GNUC__) && __GNUC__ >= 10) || (defined(__clang__) && __clang_major__ >= 17)) && (defined(__aarch64__) || defined(__x86_64__) || defined(__i386__)) +# define roundeven_finite(x) __builtin_roundeven (x) +#else +/* round x to nearest integer, breaking ties to even */ +static double +roundeven_finite (double x) +{ + double ix; +# if (defined(__GNUC__) || defined(__clang__)) && (defined(__AVX__) || defined(__SSE4_1__) || (__ARM_ARCH >= 8)) +# if defined __AVX__ + __asm__("vroundsd $0x8,%1,%1,%0":"=x"(ix):"x"(x)); +# elif __ARM_ARCH >= 8 + __asm__ ("frintn %d0, %d1":"=w"(ix):"w"(x)); +# else /* __SSE4_1__ */ + __asm__("roundsd $0x8,%1,%0":"=x"(ix):"x"(x)); +# endif +# else + ix = __builtin_round (x); /* nearest, away from 0 */ + if (__builtin_fabs (ix - x) == 0.5) + { + /* if ix is odd, we should return ix-1 if x>0, and ix+1 if x<0 */ + union { double f; uint64_t n; } u, v; + u.f = ix; + v.f = ix - __builtin_copysign (1.0, x); + if (__builtin_ctz (v.n) > __builtin_ctz (u.n)) + ix = v.f; + } +# endif + return ix; +} +#endif + +typedef union {float f; uint32_t u;} b32u32_u; +typedef union {double f; uint64_t u;} b64u64_u; + +float cr_coshf(float x){ + static const double c[] = + {1, 0x1.62e42fef4c4e7p-6, 0x1.ebfd1b232f475p-13, 0x1.c6b19384ecd93p-20}; + static const double ch[] = + {1, 0x1.62e42fefa39efp-6, 0x1.ebfbdff82c58fp-13, 0x1.c6b08d702e0edp-20, 0x1.3b2ab6fb92e5ep-27, + 0x1.5d886e6d54203p-35, 0x1.430976b8ce6efp-43}; + static const uint64_t tb[] = + {0x3fe0000000000000, 0x3fe059b0d3158574, 0x3fe0b5586cf9890f, 0x3fe11301d0125b51, + 0x3fe172b83c7d517b, 0x3fe1d4873168b9aa, 0x3fe2387a6e756238, 0x3fe29e9df51fdee1, + 0x3fe306fe0a31b715, 0x3fe371a7373aa9cb, 0x3fe3dea64c123422, 0x3fe44e086061892d, + 0x3fe4bfdad5362a27, 0x3fe5342b569d4f82, 0x3fe5ab07dd485429, 0x3fe6247eb03a5585, + 0x3fe6a09e667f3bcd, 0x3fe71f75e8ec5f74, 0x3fe7a11473eb0187, 0x3fe82589994cce13, + 0x3fe8ace5422aa0db, 0x3fe93737b0cdc5e5, 0x3fe9c49182a3f090, 0x3fea5503b23e255d, + 0x3feae89f995ad3ad, 0x3feb7f76f2fb5e47, 0x3fec199bdd85529c, 0x3fecb720dcef9069, + 0x3fed5818dcfba487, 0x3fedfc97337b9b5f, 0x3feea4afa2a490da, 0x3fef50765b6e4540}; + const double iln2 = 0x1.71547652b82fep+5; + b32u32_u t = {.f = x}; + double z = x; + uint32_t ax = t.u<<1; + if(__builtin_expect(ax>0x8565a9f8u, 0)){ // |x| >~ 89.4 + if(ax>=0xff000000u) { + if(ax<<8) return x + x; // nan + return 1.0f/0.0f; // +-inf + } + float r = 2.0f*0x1.fffffep127f; +#ifdef CORE_MATH_SUPPORT_ERRNO + errno = ERANGE; +#endif + return r; + } + if(__builtin_expect(ax<0x7c000000u, 0)){ // |x| < 0.125 + if(__builtin_expect(ax<0x74000000u, 0)){ // |x| < 0x1p-11 + if(__builtin_expect(ax<0x66000000u, 0)) // |x| < 0x1p-24 + return __builtin_fmaf(__builtin_fabsf(x), 0x1p-25, 1.0f); + return (0.5f*x)*x + 1.0f; + } + static const double cp[] = + {0x1.fffffffffffe3p-2, 0x1.55555555723cfp-5, 0x1.6c16bee4a5986p-10, 0x1.a0483fc0328f7p-16}; + double z2 = z*z, z4 = z2*z2; + return 1.0 + z2*((cp[0] + z2*cp[1]) + z4*(cp[2] + z2*(cp[3]))); + } + double a = iln2*z, ia = roundeven_finite(a), h = a - ia, h2 = h*h; + b64u64_u ja = {.f = ia + 0x1.8p52}; + int64_t jp = ja.u, jm = -jp; + b64u64_u sp = {.u = tb[jp&31] + ((uint64_t)(jp>>5)<<52)}, sm = {.u = tb[jm&31] + ((uint64_t)(jm>>5)<<52)}; + double te = c[0] + h2*c[2], to = (c[1] + h2*c[3]); + double rp = sp.f*(te + h*to), rm = sm.f*(te - h*to), r = rp + rm; + float ub = r, lb = r - 1.45e-10*r; + if(__builtin_expect(ub != lb, 0)){ + const double iln2h = 0x1.7154765p+5, iln2l = 0x1.5c17f0bbbe88p-26; + h = (iln2h*z - ia) + iln2l*z; + h2 = h*h; + te = ch[0] + h2*ch[2] + (h2*h2)*(ch[4] + h2*ch[6]); + to = ch[1] + h2*(ch[3] + h2*ch[5]); + r = sp.f*(te + h*to) + sm.f*(te - h*to); + ub = r; + } + return ub; +} View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/commit/f42458c790dd69af8f94202… -- View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/commit/f42458c790dd69af8f94202… You're receiving this email because of your account on gitlab.common-lisp.net.

1 0

[Git][cmucl/cmucl] Pushed new branch issue-425-correctly-rounded-math-functions-single-float
by Raymond Toy (＠rtoy) 21 Aug '25

21 Aug '25

Raymond Toy pushed new branch issue-425-correctly-rounded-math-functions-single-float at cmucl / cmucl -- View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/tree/issue-425-correctly-round… You're receiving this email because of your account on gitlab.common-lisp.net.

1 0

[Git][cmucl/cmucl][issue-380-p2-clean-up-shell-scripts] Address comments and update disable shellcheck notes
by Raymond Toy (＠rtoy) 20 Aug '25

20 Aug '25

Raymond Toy pushed to branch issue-380-p2-clean-up-shell-scripts at cmucl / cmucl Commits: 9b796002 by Raymond Toy at 2025-08-20T13:04:37-07:00 Address comments and update disable shellcheck notes Don't want to use :lower: and :upper: for tr. Disable some the warnings/notes that shellcheck produces for the :lower: note and also for the suggesiong on using shell variable pattern matching and replacement. - - - - - 1 changed file: - bin/make-dist.sh Changes: ===================================== bin/make-dist.sh ===================================== @@ -63,6 +63,8 @@ def_arch_os () { GTAR=gtar ;; esac uname_r=$(uname -r) + # shellcheck disable=SC2001 + # sed is ok here instead of shell variable pattern matching case $uname_r in 5.*) rel=$(echo "$uname_r" | sed 's/5\.//') ;; *) rel=$uname_r;; @@ -86,7 +88,9 @@ def_arch_os () { ;; FreeBSD) ARCH=x86 - OS=freebsd_$(uname -r | tr '[:upper:]' '[:lower:]') + # shellcheck disable=SC2018,SC2019 + # Don't want to use :lower: and :upper here since that depends on the locale. + OS=freebsd_$(uname -r | tr 'A-Z' 'a-z') ;; esac } @@ -159,6 +163,7 @@ if [ -z "$INSTALL_DIR" ]; then fi fi +# shellcheck disable=SC2001 TARGET="$(echo "$1" | sed 's:/*$::')" # Choose a version based on the git hash as the default version. We View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/commit/9b79600285eaf879208c57c… -- View it on GitLab: https://gitlab.common-lisp.net/cmucl/cmucl/-/commit/9b79600285eaf879208c57c… You're receiving this email because of your account on gitlab.common-lisp.net.

1 0