tools/testing/selftests/kselftest_harness.h | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)
Rename is_signed_type() to is_signed_var() to avoid colliding with a macro
of the same name defined by linux/overflow.h. Note, overflow.h's version
takes a type as the input, whereas the harness's version takes a variable!
This fixes warnings (and presumably potential test failures) in tests
that utilize the selftests harness and happen to (indirectly) include
overflow.h.
In file included from tools/include/linux/bits.h:34,
from tools/include/linux/bitops.h:14,
from tools/include/linux/hashtable.h:13,
from include/kvm_util.h:11,
from x86/userspace_msr_exit_test.c:11:
tools/include/linux/overflow.h:31:9: error: "is_signed_type" redefined [-Werror]
31 | #define is_signed_type(type) (((type)(-1)) < (type)1)
| ^~~~~~~~~~~~~~
In file included from include/kvm_test_harness.h:11,
from x86/userspace_msr_exit_test.c:9:
../kselftest_harness.h:754:9: note: this is the location of the previous definition
754 | #define is_signed_type(var) (!!(((__typeof__(var))(-1)) < (__typeof__(var))1))
| ^~~~~~~~~~~~~~
Opportunistically use is_signed_type() to implement is_signed_var() so
that the relationship and differences are obvious.
Fixes: fc92099902fb ("tools headers: Synchronize linux/bits.h with the kernel sources")
Cc: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
---
This is probably compile-tested only, I don't think any of the KVM selftests
utilize the harness's EXPECT macros.
tools/testing/selftests/kselftest_harness.h | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/tools/testing/selftests/kselftest_harness.h b/tools/testing/selftests/kselftest_harness.h
index 2925e47db995..f3e7a46345db 100644
--- a/tools/testing/selftests/kselftest_harness.h
+++ b/tools/testing/selftests/kselftest_harness.h
@@ -56,6 +56,7 @@
#include <asm/types.h>
#include <ctype.h>
#include <errno.h>
+#include <linux/overflow.h>
#include <linux/unistd.h>
#include <poll.h>
#include <stdbool.h>
@@ -751,7 +752,7 @@
for (; _metadata->trigger; _metadata->trigger = \
__bail(_assert, _metadata))
-#define is_signed_type(var) (!!(((__typeof__(var))(-1)) < (__typeof__(var))1))
+#define is_signed_var(var) is_signed_type(__typeof__(var))
#define __EXPECT(_expected, _expected_str, _seen, _seen_str, _t, _assert) do { \
/* Avoid multiple evaluation of the cases */ \
@@ -759,7 +760,7 @@
__typeof__(_seen) __seen = (_seen); \
if (!(__exp _t __seen)) { \
/* Report with actual signedness to avoid weird output. */ \
- switch (is_signed_type(__exp) * 2 + is_signed_type(__seen)) { \
+ switch (is_signed_var(__exp) * 2 + is_signed_var(__seen)) { \
case 0: { \
uintmax_t __exp_print = (uintmax_t)__exp; \
uintmax_t __seen_print = (uintmax_t)__seen; \
base-commit: 78f4e737a53e1163ded2687a922fce138aee73f5
--
2.50.0.714.g196bf9f422-googOn Tue, Jun 24, 2025 at 04:19:30PM -0700, Sean Christopherson wrote:
> Rename is_signed_type() to is_signed_var() to avoid colliding with a macro
> of the same name defined by linux/overflow.h. Note, overflow.h's version
> takes a type as the input, whereas the harness's version takes a variable!
This patch is in -next and is causing widespread breakage in the
selftests -next on at least arm and arm64 due to:
make --silent --keep-going --jobs=15 O=/build/stage/build-work INSTALL_PATH=/build/stage/build-work/kselftest_install ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- CROSS_COMPILE_COMPAT=arm-linux-gnueabihf- kselftest-install
...
In file included from test-pcmtest-driver.c:10:
../kselftest_harness.h:59:10: fatal error: linux/overflow.h: No such file or directory
59 | #include <linux/overflow.h>
| ^~~~~~~~~~~~~~~~~~
compilation terminated.
Sample build log:
https://builds.sirena.org.uk/5303936d609e09665deda94eaedf26a0e5c3a087/arm64/defconfig/build.log
On Wed, Aug 20, 2025, Mark Brown wrote:
> On Tue, Jun 24, 2025 at 04:19:30PM -0700, Sean Christopherson wrote:
> > Rename is_signed_type() to is_signed_var() to avoid colliding with a macro
> > of the same name defined by linux/overflow.h. Note, overflow.h's version
> > takes a type as the input, whereas the harness's version takes a variable!
>
> This patch is in -next and is causing widespread breakage in the
> selftests -next on at least arm and arm64 due to:
>
> make --silent --keep-going --jobs=15 O=/build/stage/build-work INSTALL_PATH=/build/stage/build-work/kselftest_install ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- CROSS_COMPILE_COMPAT=arm-linux-gnueabihf- kselftest-install
>
> ...
>
> In file included from test-pcmtest-driver.c:10:
> ../kselftest_harness.h:59:10: fatal error: linux/overflow.h: No such file or directory
> 59 | #include <linux/overflow.h>
> | ^~~~~~~~~~~~~~~~~~
> compilation terminated.
Argh, many selftests don't add tools/include to their include path. The least
awful thing I can think of is to go with a minimal fix to avoid the collision.
AFAICT, nothing outside of kselftest_harness.h uses is_signed_type(), so this
shouldn't cause a different flavor of breakage?
--
From: Sean Christopherson <seanjc@google.com>
Date: Tue, 24 Jun 2025 16:19:30 -0700
Subject: [PATCH] selftests: harness: Rework is_signed_type() to avoid
collision with overflow.h
Rename is_signed_type() to is_signed_var() to avoid colliding with a macro
of the same name defined by tools' linux/overflow.h. This fixes warnings
(and presumably potential test failures) in tests that utilize the
selftests harness and happen to (indirectly) include overflow.h.
In file included from tools/include/linux/bits.h:34,
from tools/include/linux/bitops.h:14,
from tools/include/linux/hashtable.h:13,
from include/kvm_util.h:11,
from x86/userspace_msr_exit_test.c:11:
tools/include/linux/overflow.h:31:9: error: "is_signed_type" redefined [-Werror]
31 | #define is_signed_type(type) (((type)(-1)) < (type)1)
| ^~~~~~~~~~~~~~
In file included from include/kvm_test_harness.h:11,
from x86/userspace_msr_exit_test.c:9:
../kselftest_harness.h:754:9: note: this is the location of the previous definition
754 | #define is_signed_type(var) (!!(((__typeof__(var))(-1)) < (__typeof__(var))1))
| ^~~~~~~~~~~~~~
Use a separate definition, at least for now, as many selftests build
without tools/include in their include path.
Fixes: fc92099902fb ("tools headers: Synchronize linux/bits.h with the kernel sources")
Cc: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Link: https://lore.kernel.org/r/20250624231930.583689-1-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
---
tools/testing/selftests/kselftest_harness.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/tools/testing/selftests/kselftest_harness.h b/tools/testing/selftests/kselftest_harness.h
index 2925e47db995..8516e8434bc4 100644
--- a/tools/testing/selftests/kselftest_harness.h
+++ b/tools/testing/selftests/kselftest_harness.h
@@ -751,7 +751,7 @@
for (; _metadata->trigger; _metadata->trigger = \
__bail(_assert, _metadata))
-#define is_signed_type(var) (!!(((__typeof__(var))(-1)) < (__typeof__(var))1))
+#define is_signed_var(var) (!!(((__typeof__(var))(-1)) < (__typeof__(var))1))
#define __EXPECT(_expected, _expected_str, _seen, _seen_str, _t, _assert) do { \
/* Avoid multiple evaluation of the cases */ \
@@ -759,7 +759,7 @@
__typeof__(_seen) __seen = (_seen); \
if (!(__exp _t __seen)) { \
/* Report with actual signedness to avoid weird output. */ \
- switch (is_signed_type(__exp) * 2 + is_signed_type(__seen)) { \
+ switch (is_signed_var(__exp) * 2 + is_signed_var(__seen)) { \
case 0: { \
uintmax_t __exp_print = (uintmax_t)__exp; \
uintmax_t __seen_print = (uintmax_t)__seen; \
base-commit: 923fcb3dbc0246fc5207093c0049af4c56f20e41
--
On Wed, Aug 20, 2025, Sean Christopherson wrote:
> On Wed, Aug 20, 2025, Mark Brown wrote:
> > On Tue, Jun 24, 2025 at 04:19:30PM -0700, Sean Christopherson wrote:
> > > Rename is_signed_type() to is_signed_var() to avoid colliding with a macro
> > > of the same name defined by linux/overflow.h. Note, overflow.h's version
> > > takes a type as the input, whereas the harness's version takes a variable!
> >
> > This patch is in -next and is causing widespread breakage in the
> > selftests -next on at least arm and arm64 due to:
> >
> > make --silent --keep-going --jobs=15 O=/build/stage/build-work INSTALL_PATH=/build/stage/build-work/kselftest_install ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- CROSS_COMPILE_COMPAT=arm-linux-gnueabihf- kselftest-install
> >
> > ...
> >
> > In file included from test-pcmtest-driver.c:10:
> > ../kselftest_harness.h:59:10: fatal error: linux/overflow.h: No such file or directory
> > 59 | #include <linux/overflow.h>
> > | ^~~~~~~~~~~~~~~~~~
> > compilation terminated.
>
> Argh, many selftests don't add tools/include to their include path. The least
> awful thing I can think of is to go with a minimal fix to avoid the collision.
> AFAICT, nothing outside of kselftest_harness.h uses is_signed_type(), so this
> shouldn't cause a different flavor of breakage?
I force-pushed a straight rename. Please holler if that somehow still breaks
builds.
[1/1] selftests: harness: Rename is_signed_type() to avoid collision with overflow.h
https://github.com/kvm-x86/linux/commit/dce1b33ed743
On Tue, 24 Jun 2025 16:19:30 -0700, Sean Christopherson wrote:
> Rename is_signed_type() to is_signed_var() to avoid colliding with a macro
> of the same name defined by linux/overflow.h. Note, overflow.h's version
> takes a type as the input, whereas the harness's version takes a variable!
>
> This fixes warnings (and presumably potential test failures) in tests
> that utilize the selftests harness and happen to (indirectly) include
> overflow.h.
>
> [...]
Applied to kvm-x86 fixes, as this is still a big thorn in my side.
[1/1] selftests: harness: Rework is_signed_type() to avoid collision with overflow.h
https://github.com/kvm-x86/linux/commit/40d5b2ab0d72
--
https://github.com/kvm-x86/linux/tree/next
On Tue, Jun 24, 2025 at 04:19:30PM -0700, Sean Christopherson wrote: > Rename is_signed_type() to is_signed_var() to avoid colliding with a macro > of the same name defined by linux/overflow.h. Note, overflow.h's version > takes a type as the input, whereas the harness's version takes a variable! Can we just update compiler.h to use typeof() and drop duplicates? (typeof() a type is a pass-thru). Totally untested: diff --git a/include/linux/compiler.h b/include/linux/compiler.h index 6f04a1d8c720..cb925b883806 100644 --- a/include/linux/compiler.h +++ b/include/linux/compiler.h @@ -347,7 +347,7 @@ static inline void *offset_to_ptr(const int *off) * Whether 'type' is a signed type or an unsigned type. Supports scalar types, * bool and also pointer types. */ -#define is_signed_type(type) (((type)(-1)) < (__force type)1) +#define is_signed_type(type) (((typeof(type))(-1)) < (__force typeof(type))1) #define is_unsigned_type(type) (!is_signed_type(type)) /* -- Kees Cook
On Fri, Jun 27, 2025, Kees Cook wrote: > On Tue, Jun 24, 2025 at 04:19:30PM -0700, Sean Christopherson wrote: > > Rename is_signed_type() to is_signed_var() to avoid colliding with a macro > > of the same name defined by linux/overflow.h. Note, overflow.h's version > > takes a type as the input, whereas the harness's version takes a variable! > > Can we just update compiler.h to use typeof() and drop duplicates? I have no objection to doing so, but I'd prefer any such change go on top. I'd like to get this fixed in 6.16, and as your diff alludes to, I think tools' version of compiler.h should follow the kernel's version (and presumably the below isn't an rc6+ candidate :-) ). > (typeof() a type is a pass-thru). Totally untested: > > diff --git a/include/linux/compiler.h b/include/linux/compiler.h > index 6f04a1d8c720..cb925b883806 100644 > --- a/include/linux/compiler.h > +++ b/include/linux/compiler.h > @@ -347,7 +347,7 @@ static inline void *offset_to_ptr(const int *off) > * Whether 'type' is a signed type or an unsigned type. Supports scalar types, > * bool and also pointer types. > */ > -#define is_signed_type(type) (((type)(-1)) < (__force type)1) > +#define is_signed_type(type) (((typeof(type))(-1)) < (__force typeof(type))1) > #define is_unsigned_type(type) (!is_signed_type(type)) > > /* > > > -- > Kees Cook
On 25/06/2025 at 08:19, Sean Christopherson wrote:
> Rename is_signed_type() to is_signed_var() to avoid colliding with a macro
> of the same name defined by linux/overflow.h. Note, overflow.h's version
> takes a type as the input, whereas the harness's version takes a variable!
>
> This fixes warnings (and presumably potential test failures) in tests
> that utilize the selftests harness and happen to (indirectly) include
> overflow.h.
>
> In file included from tools/include/linux/bits.h:34,
> from tools/include/linux/bitops.h:14,
> from tools/include/linux/hashtable.h:13,
> from include/kvm_util.h:11,
> from x86/userspace_msr_exit_test.c:11:
> tools/include/linux/overflow.h:31:9: error: "is_signed_type" redefined [-Werror]
> 31 | #define is_signed_type(type) (((type)(-1)) < (type)1)
> | ^~~~~~~~~~~~~~
> In file included from include/kvm_test_harness.h:11,
> from x86/userspace_msr_exit_test.c:9:
> ../kselftest_harness.h:754:9: note: this is the location of the previous definition
> 754 | #define is_signed_type(var) (!!(((__typeof__(var))(-1)) < (__typeof__(var))1))
> | ^~~~~~~~~~~~~~
>
> Opportunistically use is_signed_type() to implement is_signed_var() so
> that the relationship and differences are obvious.
>
> Fixes: fc92099902fb ("tools headers: Synchronize linux/bits.h with the kernel sources")
> Cc: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
> Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
> Signed-off-by: Sean Christopherson <seanjc@google.com>
> ---
>
> This is probably compile-tested only, I don't think any of the KVM selftests
> utilize the harness's EXPECT macros.
>
> tools/testing/selftests/kselftest_harness.h | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/tools/testing/selftests/kselftest_harness.h b/tools/testing/selftests/kselftest_harness.h
> index 2925e47db995..f3e7a46345db 100644
> --- a/tools/testing/selftests/kselftest_harness.h
> +++ b/tools/testing/selftests/kselftest_harness.h
> @@ -56,6 +56,7 @@
> #include <asm/types.h>
> #include <ctype.h>
> #include <errno.h>
> +#include <linux/overflow.h>
> #include <linux/unistd.h>
> #include <poll.h>
> #include <stdbool.h>
> @@ -751,7 +752,7 @@
> for (; _metadata->trigger; _metadata->trigger = \
> __bail(_assert, _metadata))
>
> -#define is_signed_type(var) (!!(((__typeof__(var))(-1)) < (__typeof__(var))1))
> +#define is_signed_var(var) is_signed_type(__typeof__(var))
It would potentially make sense to add the is_signed_var() to overflow.h. I see
a couple places doing some is_signed_type(typeof(x)):
$ git grep "is_signed_type(typeof" | wc -l
5
And I can imagine this number growing in the future.
But this is by no way a reason strong enough to block this patch. It is just a
random idea I am sharing here.
So that said:
Reviewed-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
> #define __EXPECT(_expected, _expected_str, _seen, _seen_str, _t, _assert) do { \
> /* Avoid multiple evaluation of the cases */ \
> @@ -759,7 +760,7 @@
> __typeof__(_seen) __seen = (_seen); \
> if (!(__exp _t __seen)) { \
> /* Report with actual signedness to avoid weird output. */ \
> - switch (is_signed_type(__exp) * 2 + is_signed_type(__seen)) { \
> + switch (is_signed_var(__exp) * 2 + is_signed_var(__seen)) { \
> case 0: { \
> uintmax_t __exp_print = (uintmax_t)__exp; \
> uintmax_t __seen_print = (uintmax_t)__seen; \
>
> base-commit: 78f4e737a53e1163ded2687a922fce138aee73f5
Yours sincerely,
Vincent Mailhol
© 2016 - 2025 Red Hat, Inc.