g++10.3 Segmentation fault when compiling CUDA code
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
gcc-10 (Ubuntu) |
Fix Released
|
Medium
|
Matthias Klose |
Bug Description
> lsb_release -rd
Description: Ubuntu 20.04.1 LTS
Release: 20.04
>apt-cache policy gcc-10
gcc-10:
Installed: 10.3.0-
Candidate: 10.3.0-
Version table:
*** 10.3.0-
500 http://
500 http://
100 /var/lib/
10-
500 http://
I expect to be able to compile CUDA code with gcc-10.
It was working with gcc 10.2 and failed with gcc-10.3
When compiling CUDA code g++-10.3 failed with following error message
/usr/include/
/usr/include/
/usr/include/
428 | _S_gcd(intmax_t __m, intmax_t __n) noexcept
| ^~~~~~
This bug has been corrected in gcc-10.4 :
https:/
Could it be possible to make Ubuntu 20.04 gcc-10 package use gcc-10.4 instead of 10.3 ?
description: | updated |
GCC 10.04 won't be release before early 2022. I'll see to get a backport into 20.04 LTS.