From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: by sourceware.org (Postfix, from userid 1005) id 12A6B3858C66; Thu, 12 Jan 2023 18:43:52 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 12A6B3858C66 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org; s=default; t=1673549032; bh=RaBp3KcGHQXKUp0/+BpPpYrw+1N8Suo6gTeQmFt14QQ=; h=From:To:Subject:Date:From; b=oKgF+3+S2wSWWveLvGFTqYF76NJquBBm5sD06TGGcSLL0QhOQKOG6kAiYw118q9kx kjLPtkY/AA34mBgvfFfCcoGDz+CG+/cHRgChyJ2ZQV0mCeKqKfrw0y1FrGfkYJhnT9 CV7595iJ3yqHytEp3/vAID/A4AcmC7H7cGrI3yLM= Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit From: Michael Meissner To: gcc-cvs@gcc.gnu.org Subject: [gcc(refs/users/meissner/heads/work105y)] Update ChangeLog.meissner X-Act-Checkin: gcc X-Git-Author: Michael Meissner X-Git-Refname: refs/users/meissner/heads/work105y X-Git-Oldrev: 66c1bf485041e18014ba85ebde082d6cf2b83b36 X-Git-Newrev: 2d259915721c6930847e9cbc5c05d10fc0ea7a7e Message-Id: <20230112184352.12A6B3858C66@sourceware.org> Date: Thu, 12 Jan 2023 18:43:52 +0000 (GMT) List-Id: https://gcc.gnu.org/g:2d259915721c6930847e9cbc5c05d10fc0ea7a7e commit 2d259915721c6930847e9cbc5c05d10fc0ea7a7e Author: Michael Meissner Date: Thu Jan 12 13:43:49 2023 -0500 Update ChangeLog.meissner Diff: --- gcc/ChangeLog.meissner | 71 ++++++++++++++++++++++++++++++++------------------ 1 file changed, 46 insertions(+), 25 deletions(-) diff --git a/gcc/ChangeLog.meissner b/gcc/ChangeLog.meissner index 266970c0eec..56517d4eca1 100644 --- a/gcc/ChangeLog.meissner +++ b/gcc/ChangeLog.meissner @@ -1,48 +1,69 @@ -==================== work105y, patch #2 ==================== +==================== work105y, patch #4 ==================== -PR target/107299: Use _Float128 and _Complex _Float128 types for PowerPC libgcc +PR target/107299: Update IEEE 128-bit types for PowerPC libgcc -This patch tells GCC to use the _Float128 and _Complex _Float128 types when -building the IEEE 128-bit support in libgcc. At the moment, we cannot build -GCC when the target uses IEEE 128-bit long doubles, such as building the -compiler for a native Fedora 36 system. The build dies when it is trying to -build the _mulkc3.c module. +This patch updates the IEEE 128-bit types in libgcc. -The _mulkc3 and _divkc3 modules were written before the _Float128 and _Complex -_Float128 types were available. These routines had to use the mode attributes -to get access to the __float128 complex type. Unfortunately, with the current -implementation of __float128, _Float128, and long double in the PowerPC -backend, this code now breaks if long double is IEEE 128-bit. +At the moment, we cannot build GCC when the target uses IEEE 128-bit long +doubles, such as building the compiler for a native Fedora 36 system. The +build dies when it is trying to build the _mulkc3.c and _divkc3 modules. -Originally I had hoped that we could fix the underlying problem with the -compiler. But that has become more complex than than I originally thought. I -believe the compiler should be fixed to avoid the breakage. +This patch changes the IEEE 128-bit type (TFtype) in the IEEE 128-bit support +to use either _Float128 or long double. Previously we would use the __float128 +type. In addition, the TCtype will use the same type along with _Complex. -This patch is a much simpler patch, and it avoids the problem by allowing the -machine depenendent parts of libgcc to override the definition of TFtype -(128-bit IEEE scalar) and TCtype (128-bit IEEE _Complex). This allows us to use -_Float128 and _Complex _Float128 when long double is IBM extended and plain long -double when IEEE 128-bit is used. We need to adjust the built-in functions used -as well. +I changed the libgcc build so that the IEEE 128-bit support is now compiled +with -mabi=ieeelongdouble. For the _mulkc3 and _divkc3, I explicitly use long +double and long double _Complex to avoid type mismatch errors. + +While it is desirable to ultimately have __float128 and _Float128 use the same +internal type and mode within GCC, at present if you use the option +-mabi=ieeelongdouble, the __float128 type will use the long double type. We +get an internal compiler error if we combine the signbitf128 built-in with a +long double type. + +I've gone through several iterations of trying to fix this within GCC, and +there are various problems that have come up. I developed this alternative +patch that changes libgcc so that it does not tickle the issue. I hope we can +fix the compiler at some point, but right now, this is preventing people on +Fedora 36 systems from building compilers where the default long double is IEEE +128-bit. 2023-01-12 Michael Meissner PR target/107299 - * config/rs6000/_divkc3.c (COPYSIGN): Use the correct built-in depending - on whether TFmode is _Float128 or long double. + * config/rs6000/_divkc3.c (toplevel): Add assertion for building with + -mabi=ieeelongdouble. Remove code supporting -mabi=ibmlongdouble. + (COPYSIGN): Use the long double built-in. (INFINITY): Likewise. (FABS): Likewise. - * config/rs6000/_mulkc3.c (COPYSIGN): Likewise. + * config/rs6000/_mulkc3.c (toplevel): Add assertion for building with + -mabi=ieeelongdouble. + (COPYSIGN): Use the long double built-in. (INFINITY): Likewise. + * config/rs6000/float128-ifunc.c (toplevel): Add assertion for building + with -mabi=ieeelongdouble. * config/rs6000/quad-float128.h (TF): Delete definition. (TFtype): Define to be _Float128 or long double. (TCtype): Define to be _Complex _Float128 or _Complex long double. - (__mulkc3): Delete definition since these are built-in functions. + (__mulkc3_sw): Only declare function if -mabi=ieeelongdouble. + (__divkc3_sw): Likewise. + (__mulkc3_hw): Likewise. + (__divkc3_hw): Likewise. + (__mulkc3): Likewise. (__divkc3): Likewise. + * config/rs6000/t-float128 (FP128_CFLAGS_SW): Add -mabi=ieeelongdouble. + * config/rs6000/t-float128-hw (FP128_CFLAGS_HW): Likewise. * libgcc2.h (TFtype): Allow MD code to override definition. (TCtype): Likewise. * soft-fp/quad.h (TFtype): Likewise. +==================== work105y, patch #3 was reverted ==================== + +==================== work105y, patch #2 was reverted ==================== + +==================== work105y, patch #1 was reverted ==================== + ==================== work105y branch point ==================== 2023-01-11 Michael Meissner