Skip to content

build system: add float_math feature to indicate support of floating point arithmetic #21727

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
wants to merge 8 commits into
base: master
Choose a base branch
from
Draft

build system: add float_math feature to indicate support of floating point arithmetic #21727

Changes from 1 commit
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
16e2506
build system: Add float_math feature
maribu Sep 17, 2025
1a9a192
drivers: require float_math feature where needed
maribu Sep 17, 2025
4b12599
sys: require float_math feature where needed
maribu Sep 17, 2025
fc3a860
tests: require float_math where needed
maribu Sep 17, 2025
b5d904e
sys/phydat: avoid use of pow()
maribu Sep 17, 2025
73f4160
pkg: require float_math where needed
maribu Sep 17, 2025
f2efe72
examples: require float_math where needed
maribu Sep 18, 2025
56be5ab
cpu/native: fix FPU/SSE register related context switching issues
maribu Sep 17, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 19 additions & 0 deletions cpu/native/Makefile.include
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,3 +37,22 @@ ifeq ($(USE_LIBUCONTEXT),1)
CFLAGS += $(pkg-config libucontext --cflags) -DUSE_LIBUCONTEXT=1
LINKFLAGS += $(shell pkg-config libucontext --libs)
endif

ifeq (x86_64,$(OS_ARCH))
Copy link
Contributor

@benpicco benpicco Sep 29, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Don't we also need that on 32 bit?

Copy link
Member Author

@maribu maribu Sep 29, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, but I wonder if someone actually still has an i585/i686 CPU in use these days.

It won't hurt to add the check, though.

Copy link
Contributor

@benpicco benpicco Oct 9, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's more about testing the code on a 32 bit target which is what most of our targets are.

Copy link
Member Author

@maribu maribu Oct 9, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

But the output of uname -m would still be x86_64, even with BOARD=native32.

But I honestly don't really see a path forward here with including C++ headers resulting in compilation errors when SSE is disabled.

Copy link
Contributor

@benpicco benpicco Oct 9, 2025 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ah I thought OS_ARCH was referring to RIOT 😅

# Disable use of FPU/SSE registers if FPU is not used. This prevents radom
# stack corruptions and floating point exceptions to not occur during
# context switching.
ifeq (,$(filter float_math,$(FEATURES_USED)))
# FPU/SSE registers are not reliably saved and restored in glibc's ucontext
# implementation (see https://github.com/RIOT-OS/RIOT/issues/495) and not at
# all in libucontext's (but there at least this is documented). So we force
# the compiler to only use registers known to be properly backed up and
# restored during context switching
CFLAGS += -mgeneral-regs-only

# Since we disabled the use of non-general registers (such as FPU registers),
# we need to use a soft FPU
CFLAGS += -msoft-float
LINKFLAGS += -msoft-float
endif
endif
Loading