projects / platform / upstream / glibc.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 5fe8e35) | patch
powerpc: Fix fsqrt build in libm [BZ#16576]
authorAdhemerval Zanella <azanella@linux.vnet.ibm.com>
Tue, 27 Jan 2015 18:16:39 +0000 (13:16 -0500)
committerAdhemerval Zanella <azanella@linux.vnet.ibm.com>
Wed, 28 Jan 2015 10:59:16 +0000 (05:59 -0500)
commit08cee2a464f614a6d4275b5af6c52481f1aa16e6
treee0a0b946ed40ac3c884db74de69b7cf291194808tree | snapshot
parent5fe8e3597562ac8e0e3df1399ebf804f72e7f661commit | diff
powerpc: Fix fsqrt build in libm [BZ#16576]

Some powerpc64 processors (e5500 core for instance) does not provide the
fsqrt instruction, however current check to use in math_private.h is
__WORDSIZE and _ARCH_PWR4 (ISA 2.02).  This is patch change it to use
the compiler flag _ARCH_PPCSQ (which is the same condition GCC uses to
decide whether to generate fsqrt instruction).

It fixes BZ#16576.
ChangeLog diff | blob | history
NEWS diff | blob | history
sysdeps/powerpc/fpu/e_sqrt.c diff | blob | history
sysdeps/powerpc/fpu/e_sqrtf.c diff | blob | history
sysdeps/powerpc/fpu/math_private.h diff | blob | history
sysdeps/powerpc/powerpc64/fpu/e_sqrt.c [deleted file] blob | history
sysdeps/powerpc/powerpc64/fpu/e_sqrtf.c [deleted file] blob | history
Domain: System / Toolchain;