[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
support newer MIPS CPUs
From: |
Bruno Haible |
Subject: |
support newer MIPS CPUs |
Date: |
Fri, 19 Apr 2024 02:20:29 +0200 |
On all MIPS machines that I had tested Gnulib on, the nan-mips.m4 configure
test reported:
checking whether the NaN float encoding is IEEE 754-2008 compliant... no
checking whether the NaN double encoding is IEEE 754-2008 compliant... no
checking whether the NaN long double encoding is IEEE 754-2008 compliant... no
On the OpenBSD/mips machine of the GCC compilefarm, however, it reports:
checking whether the NaN float encoding is IEEE 754-2008 compliant... no
checking whether the NaN double encoding is IEEE 754-2008 compliant... no
checking whether the NaN long double encoding is IEEE 754-2008 compliant...
yes
Which means that the "NaN is signalling" bit on 'long double' values is like
on most other CPU types. And the code that inverts that bit, conditionalized
with 'defined __mips__', causes test failures on this machine.
This set of patches fixes it.
2024-04-18 Bruno Haible <bruno@clisp.org>
setpayloadsig*: Support newer MIPS CPUs.
* lib/setpayloadsig.c (setpayloadsig): On MIPS CPUs with
MIPS_NAN2008_DOUBLE,
reject a zero payload.
* lib/setpayloadsigf.c (setpayloadsigf): On MIPS CPUs with
MIPS_NAN2008_FLOAT,
reject a zero payload.
* lib/setpayloadsigl.c (setpayloadsigl): On MIPS CPUs with
MIPS_NAN2008_LONG_DOUBLE, reject a zero payload.
2024-04-18 Bruno Haible <bruno@clisp.org>
setpayload*: Support newer MIPS CPUs.
* lib/setpayload.c (setpayload): On MIPS CPUs with MIPS_NAN2008_DOUBLE,
accept a zero payload.
* lib/setpayloadf.c (setpayloadf): On MIPS CPUs with MIPS_NAN2008_FLOAT,
accept a zero payload.
* lib/setpayloadl.c (setpayloadl): On MIPS CPUs with
MIPS_NAN2008_LONG_DOUBLE, accept a zero payload.
* m4/setpayload.m4 (gl_FUNC_SETPAYLOADF, gl_FUNC_SETPAYLOAD,
gl_FUNC_SETPAYLOADL): Invoke gl_NAN_MIPS.
* modules/setpayload (Files): Add m4/nan-mips.m4.
* modules/setpayloadf (Files): Likewise.
* modules/setpayloadl (Files): Likewise.
2024-04-18 Bruno Haible <bruno@clisp.org>
totalorder*: Support newer MIPS CPUs.
* lib/totalorder.c (totalorder): On MIPS CPUs with MIPS_NAN2008_DOUBLE,
don't invert the most significant bit of the mantissa field.
* lib/totalorderf.c (totalorderf): On MIPS CPUs with MIPS_NAN2008_FLOAT,
don't invert the most significant bit of the mantissa field.
* lib/totalorderl.c (totalorderl): On MIPS CPUs with
MIPS_NAN2008_LONG_DOUBLE, don't invert the most significant bit of the
mantissa field.
* m4/totalorder.m4 (gl_FUNC_TOTALORDERF, gl_FUNC_TOTALORDER,
gl_FUNC_TOTALORDERL): Invoke gl_NAN_MIPS.
* modules/totalorder (Files): Add m4/nan-mips.m4.
* modules/totalorderf (Files): Likewise.
* modules/totalorderl (Files): Likewise.
0001-totalorder-Support-newer-MIPS-CPUs.patch
Description: Text Data
0002-setpayload-Support-newer-MIPS-CPUs.patch
Description: Text Data
0003-setpayloadsig-Support-newer-MIPS-CPUs.patch
Description: Text Data
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- support newer MIPS CPUs,
Bruno Haible <=