https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100293
--- Comment #9 from Brecht Sanders
---
Any update on this?
Issue still exists today (in GCC 11.2.0 and in latest snapshot
11.2.1-20210814).
Both when building gcc on Windows for nvptx as well as the offload engine for
nvptx there is an error l
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100293
--- Comment #8 from Brecht Sanders
---
Additional test:
Running this manually (in MSYS2 shell) also fails:
R:/winlibs32_stage/gcc-offload-nvptx-11.1.0/gcc-11.1.0/build_win_offload-nvptx/gcc/as
-v -m sm_35 -o 'R:\winlibs32_stage\_TMP_\ccYJWYIt.o
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100293
--- Comment #7 from Brecht Sanders
---
I ran the following commands based on what was in config.log
cat > conftest.c << EOF
/* confdefs.h */
#define PACKAGE_NAME "GNU Atomic Library"
#define PACKAGE_TARNAME "libatomic"
#define PACKAGE_VERSION
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100293
--- Comment #6 from Brecht Sanders
---
Yes, that folder exists and that's where my TMP and TEMP environment variables
point to.
I also tried to point them to a folder on the C: drive, as R: is a RAM drive
and I wanted to exclude that that was
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100293
--- Comment #5 from Jakub Jelinek ---
The log file contains:
error reading R:\winlibs64_stage\_TMP_\ccYEwkZW.o
collect2.exe: error: ld returned 1 exit status
which I bet is the problem, but it is unclear if the _TMP_ subdir exists...
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100293
Tom de Vries changed:
What|Removed |Added
CC||vries at gcc dot gnu.org
--- Comment #4
