tools/nolibc: define time_t in terms of __kernel_old_time_t

Nolibc assumes that the kernel ABI is using a time values that are as
large as a long integer. For most ABIs this holds true.
But for x32

tools/nolibc: define time_t in terms of __kernel_old_time_t

Nolibc assumes that the kernel ABI is using a time values that are as
large as a long integer. For most ABIs this holds true.
But for x32 this is not correct, as it uses 32bit longs but 64bit times.

Also the 'struct stat' implementation of nolibc relies on timespec::tv_sec
and time_t being the same type. While timespec::tv_sec comes from the
kernel and is of type __kernel_old_time_t, time_t is defined within nolibc.

Switch to the __kernel_old_time_t to always get the correct type.

Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://lore.kernel.org/r/20250712-nolibc-x32-v1-1-6d81cb798710@weissschuh.net
Acked-by: Willy Tarreau <w@1wt.eu>

show more ...

tools/nolibc: drop s390 clang target override

tools/scripts/Makefile.include now has the same override,
removing the need for the one in the nolibc Makefile.

Drop the superfluous custom override.

tools/nolibc: drop s390 clang target override

tools/scripts/Makefile.include now has the same override,
removing the need for the one in the nolibc Makefile.

Drop the superfluous custom override.

Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Link: https://lore.kernel.org/r/20250620-tools-cross-s390-v2-2-ecda886e00e5@linutronix.de
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>

show more ...

tools/nolibc: avoid false-positive -Wmaybe-uninitialized through waitpid()

The compiler does not know that waitid() will only ever return 0 or -1.
If waitid() would return a positive value than wait

tools/nolibc: avoid false-positive -Wmaybe-uninitialized through waitpid()

The compiler does not know that waitid() will only ever return 0 or -1.
If waitid() would return a positive value than waitpid() would return that
same value and *status would not be initialized.
However users calling waitpid() know that the only possible return values
of it are 0 or -1. They therefore might check for errors with
'ret == -1' or 'ret < 0' and use *status otherwise. The compiler will then
warn about the usage of a potentially uninitialized variable.

Example:

$ cat test.c
#include <stdio.h>
#include <unistd.h>

int main(void)
{
int ret, status;

ret = waitpid(0, &status, 0);
if (ret == -1)
return 0;

printf("status %x\n", status);

return 0;
}

$ gcc --version
gcc (GCC) 15.1.1 20250425

$ gcc -Wall -Os -Werror -nostdlib -nostdinc -static -Iusr/include -Itools/include/nolibc/ -o /dev/null test.c
test.c: In function ‘main’:
test.c:12:9: error: ‘status’ may be used uninitialized [-Werror=maybe-uninitialized]
12 | printf("status %x\n", status);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
test.c:6:18: note: ‘status’ was declared here
6 | int ret, status;
| ^~~~~~
cc1: all warnings being treated as errors

Avoid the warning by normalizing waitid() errors to '-1' in waitpid().

Fixes: 0c89abf5ab3f ("tools/nolibc: implement waitpid() in terms of waitid()")
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://lore.kernel.org/r/20250707-nolibc-waitpid-uninitialized-v1-1-dcd4e70bcd8f@linutronix.de
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>

show more ...

selftests/nolibc: correctly report errors from printf() and friends

When an error is encountered by printf() it needs to be reported.
errno() is already set by the callback.

sprintf() is different,

selftests/nolibc: correctly report errors from printf() and friends

When an error is encountered by printf() it needs to be reported.
errno() is already set by the callback.

sprintf() is different, but that keeps working and is already tested.

Also add a new test.

Fixes: 7e4346f4a3a6 ("tools/nolibc/stdio: add a minimal [vf]printf() implementation")
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://lore.kernel.org/r/20250704-nolibc-printf-error-v1-2-74b7a092433b@linutronix.de
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>

show more ...

tools/nolibc: add support for clock_nanosleep() and nanosleep()

Also add some tests.

Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https

tools/nolibc: add support for clock_nanosleep() and nanosleep()

Also add some tests.

Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://lore.kernel.org/r/20250704-nolibc-nanosleep-v1-1-d79c19701952@linutronix.de
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>

show more ...

tools/nolibc: Provide vfork()

To allow testing of vfork() support in the arm64 basic-gcs test provide an
implementation for nolibc, using the vfork() syscall if one is available
and otherwise clone3

tools/nolibc: Provide vfork()

To allow testing of vfork() support in the arm64 basic-gcs test provide an
implementation for nolibc, using the vfork() syscall if one is available
and otherwise clone3(). We implement in terms of clone3() since the order
of the arguments for clone() varies between architectures.

As for fork() SPARC returns the parent PID rather than 0 in the child
for vfork() so needs custom handling.

Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20250703-arm64-gcs-vfork-exit-v3-2-1e9a9d2ddbbe@kernel.org
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>

show more ...

tools/nolibc: Replace ifdef with if defined() in sys.h

Thomas has requested that if defined() be used in place of ifdef but
currently ifdef is used consistently in sys.h. Update all the instances of

tools/nolibc: Replace ifdef with if defined() in sys.h

Thomas has requested that if defined() be used in place of ifdef but
currently ifdef is used consistently in sys.h. Update all the instances of
ifdef to if defined().

Suggested-by: Thomas Weißschuh <linux@weissschuh.net>
Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20250703-arm64-gcs-vfork-exit-v3-1-1e9a9d2ddbbe@kernel.org
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>

show more ...

tools/nolibc: add support for SuperH

Add support for SuperH/"sh" to nolibc.
Only sh4 is tested for now.

The startup code is special:

__nolibc_entrypoint_epilogue() calls __builtin_unreachable() wh

tools/nolibc: add support for SuperH

Add support for SuperH/"sh" to nolibc.
Only sh4 is tested for now.

The startup code is special:

__nolibc_entrypoint_epilogue() calls __builtin_unreachable() which emits
a call to abort(). To make this work a function prologue is generated to
set up a GOT pointer which corrupts "sp".
__builtin_unreachable() is necessary for __attribute__((noreturn)).
Also depending on compiler flags (for example -fPIC) even more prologue
is generated.

Work around this by defining a nested function in asm.

Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70216
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Acked-by: Rob Landley <rob@landley.net>
Acked-by: D. Jeff Dionne <jeff@coresemi.io>
Tested-by: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Link: https://lore.kernel.org/r/20250623-nolibc-sh-v2-3-0f5b4b303025@weissschuh.net

show more ...

tools/nolibc: MIPS: add support for N64 and N32 ABIs

Add support for the MIPS 64bit N64 and ILP32 N32 ABIs.

In addition to different byte orders and ABIs there are also different
releases of the MI

tools/nolibc: MIPS: add support for N64 and N32 ABIs

Add support for the MIPS 64bit N64 and ILP32 N32 ABIs.

In addition to different byte orders and ABIs there are also different
releases of the MIPS architecture. To avoid blowing up the test matrix,
only add a subset of all possible test combinations.

Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Tested-by: Sebastian Andrzej Siewior <sebastian@breakpoint.cc>
Link: https://lore.kernel.org/r/20250623-nolibc-mips-n32-v3-4-6ae2d89f4259@weissschuh.net

show more ...

tools/nolibc: MIPS: drop noreorder option

There are no more statements in the assembly code which would require
the usage of ".set noreorder".

Remove the option.

This also allows removal of the ma

tools/nolibc: MIPS: drop noreorder option

There are no more statements in the assembly code which would require
the usage of ".set noreorder".

Remove the option.

This also allows removal of the manual "nop" instruction in the
delay slot.

Suggested-by: Maciej W. Rozycki <macro@orcam.me.uk>
Link: https://lore.kernel.org/lkml/alpine.DEB.2.21.2502172208570.65342@angie.orcam.me.uk/
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Tested-by: Sebastian Andrzej Siewior <sebastian@breakpoint.cc>
Link: https://lore.kernel.org/r/20250623-nolibc-mips-n32-v3-3-6ae2d89f4259@weissschuh.net

show more ...

tools/nolibc: MIPS: drop manual stack pointer alignment

The stack pointer is already aligned by the kernel to a multiple of 16.
All modifications of the register have been removed from the entrypoin

tools/nolibc: MIPS: drop manual stack pointer alignment

The stack pointer is already aligned by the kernel to a multiple of 16.
All modifications of the register have been removed from the entrypoint,
so the manual realignment is unnecessary.

Drop the manual alignment.

Suggested-by: Maciej W. Rozycki <macro@orcam.me.uk>
Link: https://lore.kernel.org/lkml/alpine.DEB.2.21.2502161523290.65342@angie.orcam.me.uk/
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Tested-by: Sebastian Andrzej Siewior <sebastian@breakpoint.cc>
Link: https://lore.kernel.org/r/20250623-nolibc-mips-n32-v3-2-6ae2d89f4259@weissschuh.net

show more ...

tools/nolibc: MIPS: drop $gp setup

The setup of the global pointer "$gp" register was necessary when the C
entrypoint was called through "jal <symbol>".
However since commit 0daf8c86a451 ("tools/nol

tools/nolibc: MIPS: drop $gp setup

The setup of the global pointer "$gp" register was necessary when the C
entrypoint was called through "jal <symbol>".
However since commit 0daf8c86a451 ("tools/nolibc: mips: load current function to $t9")
"jalr" is used instead which does not require "$gp".

Remove the unnecessary $gp setup, simplifying the code and opening the
road for some other cleanups.

Suggested-by: Maciej W. Rozycki <macro@orcam.me.uk>
Link: https://lore.kernel.org/lkml/alpine.DEB.2.21.2502172208570.65342@angie.orcam.me.uk/
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Tested-by: Sebastian Andrzej Siewior <sebastian@breakpoint.cc>
Link: https://lore.kernel.org/r/20250623-nolibc-mips-n32-v3-1-6ae2d89f4259@weissschuh.net

show more ...

tools/nolibc: use arm64 name over aarch64

Nolibc generally uses the kernel's architecture names.
aarch64 is the only exception.

Remove the special case.
Nothing changes for the users.

Acked-by: Wi

tools/nolibc: use arm64 name over aarch64

Nolibc generally uses the kernel's architecture names.
aarch64 is the only exception.

Remove the special case.
Nothing changes for the users.

Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://lore.kernel.org/r/20250623-nolibc-aarch64-arm64-v1-1-a2892f1c1b27@weissschuh.net
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>

show more ...

tools/nolibc: hide headers_check command by default

If V=1 is not specified the executed commands should not be printed.

Hide the commands by default.

Fixes: a6a054c8ad32 ("tools/nolibc: add targe

tools/nolibc: hide headers_check command by default

If V=1 is not specified the executed commands should not be printed.

Hide the commands by default.

Fixes: a6a054c8ad32 ("tools/nolibc: add target to check header usability")
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://lore.kernel.org/r/20250623-nolibc-headers-silent-v1-1-f568facf014c@weissschuh.net
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>

show more ...

tools/nolibc: merge i386 and x86_64 into a single x86 arch

This remained the only exception to the kernel's architectures
organization and it's always a bit cumbersome to deal with. Let's merge
i386

tools/nolibc: merge i386 and x86_64 into a single x86 arch

This remained the only exception to the kernel's architectures
organization and it's always a bit cumbersome to deal with. Let's merge
i386 and x86_64 into x86. This will result in a single arch-x86.h file
by default, and we'll no longer need to merge the two manually during
installation. Requesting either i386 or x86_64 will also result in
installing x86.

Acked-by: Thomas Weißschuh <linux@weissschuh.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

show more ...

tools/nolibc: fix spelling of FD_SETBITMASK in FD_* macros

While nolibc-test does test syscalls, it doesn't test as much the rest
of the macros, and a wrong spelling of FD_SETBITMASK in commit
feaf7

tools/nolibc: fix spelling of FD_SETBITMASK in FD_* macros

While nolibc-test does test syscalls, it doesn't test as much the rest
of the macros, and a wrong spelling of FD_SETBITMASK in commit
feaf75658783a broke programs using either FD_SET() or FD_CLR() without
being noticed. Let's fix these macros.

Fixes: feaf75658783a ("nolibc: fix fd_set type")
Cc: stable@vger.kernel.org # v6.2+
Acked-by: Thomas Weißschuh <linux@weissschuh.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

show more ...