This branch is out-of-date with the base branch
I missed this, it actually fails on rebase because I added tagaspi in bench6. I will need to drop it from the inputs as well.
Managed to fix all packages and tests (ignoring GPI-2 and TAGASPI).
hut% nix build -vL .#bsc-ci.all --print-out-paths
/nix/store/0ifyfdldpbrwkzi5asnkfgk68f8qrr7r-ci-all
It looks the gcc version no has extra components in some cases:
hut$ ls /nix/store/6i862vz60awrlsila8vw18rg4d4l66iy-gcc-14.2.1.20250322/lib/gcc/x86_64-unknown-linux-gnu/14.2.1.20250322
ls:…
Patching path_backup fixes it for C, but C++ fails:
hut$ cat hello.cpp
#include <cstdio>
int main()
{
printf("Hello world!\n");
return 0;
}
hut$ icpc hello.cpp -o…
Do you want to try luck upstream so they support the newer boost? Nothing here https://github.com/bsc-performance-tools/paraver-kernel/commits/devel/ or here https://github.com/bsc-performance-…
So... we depended on a bug as they seem to not export PATH, so it was being drived by our wrapper:
This also…
PATH="$path_backup" <--- WTF
# Old bash workaround, see above.
if (( "${NIX_CC_USE_RESPONSE_FILE:-0}" >= 1 )); then
responseFile=$(/nix/store/mp7ba85zcqdj2sqwa29pql02s6nqpcxy-coreutil…
Something is causing our $PATH to change between the cc-wrapper-hook and the execve syscall:
write(1, "cc-wrapper-hook: starting...\n", 29cc-wrapper-hook: starting...
) = 29
write(1,…
# Need the gcc in the path |
|
echo 'export "PATH=${mygcc}/bin:$PATH"' >> $out/nix-support/cc-wrapper-hook |
Using
boost186should solve itI have added the fix in #6, but you can include it in #5 instead
Thanks, I have cherry picked it and it builds fine. Do you want to try luck upstream so…