[bug / patch ]patch for glibc 2.26 to avoid errors in compiling with CUDA(NVCC)
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
gcc-7 (Ubuntu) |
Fix Released
|
Undecided
|
Adam Conrad | ||
glibc (Ubuntu) |
Fix Released
|
Undecided
|
Adam Conrad |
Bug Description
DistroRelease: Ubuntu 17.10 (Proposed)
Package: glibc 2.26-0ubuntu1
Architecture: amd64
extrapackage (not incl. in UBUNTU) NVIDIA-DRIVER, CUDA
-------
Here is another proposal regarding bug report I posted (but in reject).
https:/
I reported that glibc 2.26-0ubuntu1 ( adopted in Artful-proposed on Sep 5.) causes compilation error by using NVIDIA's CUDA8.0 and 9.0RC.
Under glibc 2.24, they work.
It maybe comes from glibc226's new feature, i.e. "128-bit floating point as defined by ISO/IEC/IEEE 60559:2011 (IEEE 754-2008) and ISO/IEC TS 18661-3:2015" . (Evidence Scripted at bottom)
I proposed the patch to
/usr/
which lets NVCC (CUDA's compiler) avoid __float128 (NVCC does support neither 8.0 nor 9.0).
So can the patch be merged to glibc2.25 be adopted in Arful or later?
-------
*** floatn.h-dist 2017-09-04 16:34:21.000000000 +0900
--- floatn.h 2017-09-14 21:46:15.334033614 +0900
***************
*** 28,34 ****
support, for x86_64 and x86. */
#if (defined __x86_64__ \
? __GNUC_PREREQ (4, 3) \
! : (defined __GNU__ ? __GNUC_PREREQ (4, 5) : __GNUC_PREREQ (4, 4)))
# define __HAVE_FLOAT128 1
#else
# define __HAVE_FLOAT128 0
--- 28,35 ----
support, for x86_64 and x86. */
#if (defined __x86_64__ \
? __GNUC_PREREQ (4, 3) \
! : (defined __GNU__ ? __GNUC_PREREQ (4, 5) : __GNUC_PREREQ (4, 4))) \
! && !defined(
# define __HAVE_FLOAT128 1
#else
# define __HAVE_FLOAT128 0
-------
(evidence)
----------
1. Here is part of message during compiing Tensorflow with CUDA on UBUNTU17.10 beta with proposed-
--------------
typedef _Complex float __cfloat128 __attribute__ ((__mode__ (__TC__)));
INFO: From Compiling external/
/usr/include/
/usr/include/
--------------
2. Forums in INVIDIA
where issues the problems around glibc.2.26
3. THis bug has already discussed in Arch linux and the patch (tas same ) proposed
https:/
description: | updated |
description: | updated |
description: | updated |
description: | updated |
summary: |
- proposal of patch to avoid erros in compiping NVCC + proposal of patch for glibc to avoid erros in compiping NVCC |
summary: |
- proposal of patch for glibc to avoid erros in compiping NVCC + [bug / patch ]patch for glibc 2.26 to avoid errors in compiling with + CUDA(NVCC) |
Changed in glibc (Ubuntu): | |
assignee: | nobody → Adam Conrad (adconrad) |
Changed in gcc-7 (Ubuntu): | |
assignee: | nobody → Adam Conrad (adconrad) |
Changed in llvm-toolchain-3.8 (Ubuntu): | |
status: | Confirmed → Invalid |
Changed in gcc-7 (Ubuntu): | |
status: | New → Fix Committed |
no longer affects: | llvm-toolchain-3.8 (Ubuntu) |
Status changed to 'Confirmed' because the bug affects multiple users.