This kernel version doesn't build with GCC 15:
In file included from include/uapi/linux/posix_types.h:5, from include/uapi/linux/types.h:14, from include/linux/types.h:6, from arch/x86/realmode/rm/wakeup.h:11, from arch/x86/realmode/rm/wakemain.c:2: include/linux/stddef.h:11:9: error: cannot use keyword 'false' as enumeration constant 11 | false = 0, | ^~~~~ include/linux/stddef.h:11:9: note: 'false' is a keyword with '-std=c23' onwards include/linux/types.h:30:33: error: 'bool' cannot be defined via 'typedef' 30 | typedef _Bool bool; | ^~~~ include/linux/types.h:30:33: note: 'bool' is a keyword with '-std=c23' onwards include/linux/types.h:30:1: warning: useless type name in empty declaration 30 | typedef _Bool bool; | ^~~~~~~
I initially fixed this by adding -std=gnu11 in arch/x86/Makefile, then I realised this fix was already done in an upstream commit, created before the GCC 15 release and not mentioning the error I had. This is the first patch.
When I was investigating my error, I noticed other commits were already backported to stable versions. They were all adding -std=gnu11 in different Makefiles. In their commit message, they were mentioning 'gnu11' was picked to use the same as the one from the main Makefile. But this is not the case in this kernel version. Patch 2 fixes that.
Finally, I noticed an extra warning that I didn't have in v5.15. Patch 3 fixes that.
Signed-off-by: Matthieu Baerts (NGI0) matttbe@kernel.org --- Alexey Dobriyan (1): x86/boot: Compile boot code with -std=gnu11 too
Matthieu Baerts (NGI0) (2): arch: back to -std=gnu89 in < v5.18 tracing: fix declaration-after-statement warning
arch/parisc/boot/compressed/Makefile | 2 +- arch/s390/Makefile | 2 +- arch/s390/purgatory/Makefile | 2 +- arch/x86/Makefile | 2 +- arch/x86/boot/compressed/Makefile | 2 +- drivers/firmware/efi/libstub/Makefile | 2 +- kernel/trace/trace_events_synth.c | 3 ++- 7 files changed, 8 insertions(+), 7 deletions(-) --- base-commit: a32db271d59d9f35f3a937ac27fcc2db1e029cdc change-id: 20251017-v5-10-gcc-15-ad2510f784f7
Best regards,
From: Alexey Dobriyan adobriyan@gmail.com
commit b3bee1e7c3f2b1b77182302c7b2131c804175870 upstream.
Use -std=gnu11 for consistency with main kernel code.
It doesn't seem to change anything in vmlinux.
Signed-off-by: Alexey Dobriyan adobriyan@gmail.com Signed-off-by: Ingo Molnar mingo@kernel.org Acked-by: H. Peter Anvin (Intel) hpa@zytor.com Link: https://lore.kernel.org/r/2058761e-12a4-4b2f-9690-3c3c1c9902a5@p183 [ This kernel version doesn't build with GCC 15:
In file included from include/uapi/linux/posix_types.h:5, from include/uapi/linux/types.h:14, from include/linux/types.h:6, from arch/x86/realmode/rm/wakeup.h:11, from arch/x86/realmode/rm/wakemain.c:2: include/linux/stddef.h:11:9: error: cannot use keyword 'false' as enumeration constant 11 | false = 0, | ^~~~~ include/linux/stddef.h:11:9: note: 'false' is a keyword with '-std=c23' onwards include/linux/types.h:30:33: error: 'bool' cannot be defined via 'typedef' 30 | typedef _Bool bool; | ^~~~ include/linux/types.h:30:33: note: 'bool' is a keyword with '-std=c23' onwards include/linux/types.h:30:1: warning: useless type name in empty declaration 30 | typedef _Bool bool; | ^~~~~~~
The fix is similar to commit ee2ab467bddf ("x86/boot: Use '-std=gnu11' to fix build with GCC 15") which has been backported to this kernel.
Note: In < 5.18 version, -std=gnu89 is used instead of -std=gnu11, see commit e8c07082a810 ("Kbuild: move to -std=gnu11"). I suggest not to modify that in this commit here as all the other similar fixes to support GCC 15 set -std=gnu11. This can be done in a dedicated commit if needed. There was a conflict, because commit 2838307b019d ("x86/build: Remove -m16 workaround for unsupported versions of GCC") is not in this version and change code in the context. -std=gnu11 can still be added at the same place. ] Signed-off-by: Matthieu Baerts (NGI0) matttbe@kernel.org --- Cc: Nathan Chancellor nathan@kernel.org Cc: Dave Hansen dave.hansen@linux.intel.com Cc: Ard Biesheuvel ardb@kernel.org --- arch/x86/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/x86/Makefile b/arch/x86/Makefile index 8b9fa777f513..f584d07095f1 100644 --- a/arch/x86/Makefile +++ b/arch/x86/Makefile @@ -31,7 +31,7 @@ endif CODE16GCC_CFLAGS := -m32 -Wa,$(srctree)/arch/x86/boot/code16gcc.h M16_CFLAGS := $(call cc-option, -m16, $(CODE16GCC_CFLAGS))
-REALMODE_CFLAGS := $(M16_CFLAGS) -g -Os -DDISABLE_BRANCH_PROFILING -D__DISABLE_EXPORTS \ +REALMODE_CFLAGS := -std=gnu11 $(M16_CFLAGS) -g -Os -DDISABLE_BRANCH_PROFILING -D__DISABLE_EXPORTS \ -Wall -Wstrict-prototypes -march=i386 -mregparm=3 \ -fno-strict-aliasing -fomit-frame-pointer -fno-pic \ -mno-mmx -mno-sse $(call cc-option,-fcf-protection=none)
Recent fixes have been backported to < v5.18 to fix build issues with GCC 5.15. They all force -std=gnu11 in the CFLAGS, "because [the kernel] requests the gnu11 standard via '-std=' in the main Makefile".
This is true for >= 5.18 versions, but not before. This switch to -std=gnu11 has been done in commit e8c07082a810 ("Kbuild: move to -std=gnu11").
For a question of uniformity, force -std=gnu89, similar to what is done in the main Makefile.
Note: the fixes tags below refers to upstream commits, but this fix is only for kernels not having commit e8c07082a810 ("Kbuild: move to -std=gnu11").
Fixes: 7cbb015e2d3d ("parisc: fix building with gcc-15") Fixes: 3b8b80e99376 ("s390: Add '-std=gnu11' to decompressor and purgatory CFLAGS") Fixes: b3bee1e7c3f2 ("x86/boot: Compile boot code with -std=gnu11 too") Fixes: ee2ab467bddf ("x86/boot: Use '-std=gnu11' to fix build with GCC 15") Fixes: 8ba14d9f490a ("efi: libstub: Use '-std=gnu11' to fix build with GCC 15") Signed-off-by: Matthieu Baerts (NGI0) matttbe@kernel.org --- Note: An alternative is to backport commit e8c07082a810 ("Kbuild: move to -std=gnu11"), but I guess we might not want to do that for stable, as it might introduce new warnings.
Cc: Nathan Chancellor nathan@kernel.org Cc: Ard Biesheuvel ardb@kernel.org Cc: Alexey Dobriyan adobriyan@gmail.com Cc: Arnd Bergmann arnd@arndb.de --- arch/parisc/boot/compressed/Makefile | 2 +- arch/s390/Makefile | 2 +- arch/s390/purgatory/Makefile | 2 +- arch/x86/Makefile | 2 +- arch/x86/boot/compressed/Makefile | 2 +- drivers/firmware/efi/libstub/Makefile | 2 +- 6 files changed, 6 insertions(+), 6 deletions(-)
diff --git a/arch/parisc/boot/compressed/Makefile b/arch/parisc/boot/compressed/Makefile index 4e5aecc263a2..f2efd55a0c81 100644 --- a/arch/parisc/boot/compressed/Makefile +++ b/arch/parisc/boot/compressed/Makefile @@ -22,7 +22,7 @@ KBUILD_CFLAGS += -fno-PIE -mno-space-regs -mdisable-fpregs -Os ifndef CONFIG_64BIT KBUILD_CFLAGS += -mfast-indirect-calls endif -KBUILD_CFLAGS += -std=gnu11 +KBUILD_CFLAGS += -std=gnu89
OBJECTS += $(obj)/head.o $(obj)/real2.o $(obj)/firmware.o $(obj)/misc.o $(obj)/piggy.o
diff --git a/arch/s390/Makefile b/arch/s390/Makefile index 92f2426d8797..9680c6c9a710 100644 --- a/arch/s390/Makefile +++ b/arch/s390/Makefile @@ -23,7 +23,7 @@ endif aflags_dwarf := -Wa,-gdwarf-2 KBUILD_AFLAGS_DECOMPRESSOR := $(CLANG_FLAGS) -m64 -D__ASSEMBLY__ KBUILD_AFLAGS_DECOMPRESSOR += $(if $(CONFIG_DEBUG_INFO),$(aflags_dwarf)) -KBUILD_CFLAGS_DECOMPRESSOR := $(CLANG_FLAGS) -m64 -O2 -std=gnu11 +KBUILD_CFLAGS_DECOMPRESSOR := $(CLANG_FLAGS) -m64 -O2 -std=gnu89 KBUILD_CFLAGS_DECOMPRESSOR += -DDISABLE_BRANCH_PROFILING -D__NO_FORTIFY KBUILD_CFLAGS_DECOMPRESSOR += -fno-delete-null-pointer-checks -msoft-float KBUILD_CFLAGS_DECOMPRESSOR += -fno-asynchronous-unwind-tables diff --git a/arch/s390/purgatory/Makefile b/arch/s390/purgatory/Makefile index 955f113cf320..e03f234fcfbe 100644 --- a/arch/s390/purgatory/Makefile +++ b/arch/s390/purgatory/Makefile @@ -20,7 +20,7 @@ GCOV_PROFILE := n UBSAN_SANITIZE := n KASAN_SANITIZE := n
-KBUILD_CFLAGS := -std=gnu11 -fno-strict-aliasing -Wall -Wstrict-prototypes +KBUILD_CFLAGS := -std=gnu89 -fno-strict-aliasing -Wall -Wstrict-prototypes KBUILD_CFLAGS += -Wno-pointer-sign -Wno-sign-compare KBUILD_CFLAGS += -fno-zero-initialized-in-bss -fno-builtin -ffreestanding KBUILD_CFLAGS += -c -MD -Os -m64 -msoft-float -fno-common diff --git a/arch/x86/Makefile b/arch/x86/Makefile index f584d07095f1..9c08f2cd399f 100644 --- a/arch/x86/Makefile +++ b/arch/x86/Makefile @@ -31,7 +31,7 @@ endif CODE16GCC_CFLAGS := -m32 -Wa,$(srctree)/arch/x86/boot/code16gcc.h M16_CFLAGS := $(call cc-option, -m16, $(CODE16GCC_CFLAGS))
-REALMODE_CFLAGS := -std=gnu11 $(M16_CFLAGS) -g -Os -DDISABLE_BRANCH_PROFILING -D__DISABLE_EXPORTS \ +REALMODE_CFLAGS := -std=gnu89 $(M16_CFLAGS) -g -Os -DDISABLE_BRANCH_PROFILING -D__DISABLE_EXPORTS \ -Wall -Wstrict-prototypes -march=i386 -mregparm=3 \ -fno-strict-aliasing -fomit-frame-pointer -fno-pic \ -mno-mmx -mno-sse $(call cc-option,-fcf-protection=none) diff --git a/arch/x86/boot/compressed/Makefile b/arch/x86/boot/compressed/Makefile index e1a750baf036..6dd03543235a 100644 --- a/arch/x86/boot/compressed/Makefile +++ b/arch/x86/boot/compressed/Makefile @@ -33,7 +33,7 @@ targets := vmlinux vmlinux.bin vmlinux.bin.gz vmlinux.bin.bz2 vmlinux.bin.lzma \ # avoid errors with '-march=i386', and future flags may depend on the target to # be valid. KBUILD_CFLAGS := -m$(BITS) -O2 $(CLANG_FLAGS) -KBUILD_CFLAGS += -std=gnu11 +KBUILD_CFLAGS += -std=gnu89 KBUILD_CFLAGS += -fno-strict-aliasing -fPIE KBUILD_CFLAGS += -DDISABLE_BRANCH_PROFILING cflags-$(CONFIG_X86_32) := -march=i386 diff --git a/drivers/firmware/efi/libstub/Makefile b/drivers/firmware/efi/libstub/Makefile index d7bcfe0d50d1..d719134e2d3a 100644 --- a/drivers/firmware/efi/libstub/Makefile +++ b/drivers/firmware/efi/libstub/Makefile @@ -7,7 +7,7 @@ # cflags-$(CONFIG_X86_32) := -march=i386 cflags-$(CONFIG_X86_64) := -mcmodel=small -cflags-$(CONFIG_X86) += -m$(BITS) -D__KERNEL__ -std=gnu11 \ +cflags-$(CONFIG_X86) += -m$(BITS) -D__KERNEL__ -std=gnu89 \ -fPIC -fno-strict-aliasing -mno-red-zone \ -mno-mmx -mno-sse -fshort-wchar \ -Wno-pointer-sign \
When building this kernel version this warning is visible:
kernel/trace/trace_events_synth.c: In function 'synth_event_reg': kernel/trace/trace_events_synth.c:847:9: warning: ISO C90 forbids mixed declarations and code [-Wdeclaration-after-statement] 847 | int ret = trace_event_reg(call, type, data); | ^~~
This can be easily fixed by declaring 'ret' earlier.
This issue is visible in < v5.18, because -std=gnu89 is used by default, see commit e8c07082a810 ("Kbuild: move to -std=gnu11").
Please note that in v5.15.y, the 'Fixes' commit has been modified during the backport, not to have this warning. See commit 72848b81b3dd ("tracing: Ensure module defining synth event cannot be unloaded while tracing") from v5.15.y.
Fixes: 21581dd4e7ff ("tracing: Ensure module defining synth event cannot be unloaded while tracing") Signed-off-by: Matthieu Baerts (NGI0) matttbe@kernel.org --- Cc: Douglas Raillard douglas.raillard@arm.com Cc: Masami Hiramatsu (Google) mhiramat@kernel.org Cc: Steven Rostedt (Google) rostedt@goodmis.org Cc: Mathieu Desnoyers mathieu.desnoyers@efficios.com --- kernel/trace/trace_events_synth.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/kernel/trace/trace_events_synth.c b/kernel/trace/trace_events_synth.c index 613d45e7b608..b126af59e61c 100644 --- a/kernel/trace/trace_events_synth.c +++ b/kernel/trace/trace_events_synth.c @@ -831,6 +831,7 @@ static int synth_event_reg(struct trace_event_call *call, enum trace_reg type, void *data) { struct synth_event *event = container_of(call, struct synth_event, call); + int ret;
switch (type) { #ifdef CONFIG_PERF_EVENTS @@ -844,7 +845,7 @@ static int synth_event_reg(struct trace_event_call *call, break; }
- int ret = trace_event_reg(call, type, data); + ret = trace_event_reg(call, type, data);
switch (type) { #ifdef CONFIG_PERF_EVENTS
On Fri, Oct 17, 2025 at 06:53:24PM +0200, Matthieu Baerts (NGI0) wrote:
This kernel version doesn't build with GCC 15:
In file included from include/uapi/linux/posix_types.h:5, from include/uapi/linux/types.h:14, from include/linux/types.h:6, from arch/x86/realmode/rm/wakeup.h:11, from arch/x86/realmode/rm/wakemain.c:2: include/linux/stddef.h:11:9: error: cannot use keyword 'false' as enumeration constant 11 | false = 0, | ^~~~~ include/linux/stddef.h:11:9: note: 'false' is a keyword with '-std=c23' onwards include/linux/types.h:30:33: error: 'bool' cannot be defined via 'typedef' 30 | typedef _Bool bool; | ^~~~ include/linux/types.h:30:33: note: 'bool' is a keyword with '-std=c23' onwards include/linux/types.h:30:1: warning: useless type name in empty declaration 30 | typedef _Bool bool; | ^~~~~~~
I initially fixed this by adding -std=gnu11 in arch/x86/Makefile, then I realised this fix was already done in an upstream commit, created before the GCC 15 release and not mentioning the error I had. This is the first patch.
When I was investigating my error, I noticed other commits were already backported to stable versions. They were all adding -std=gnu11 in different Makefiles. In their commit message, they were mentioning 'gnu11' was picked to use the same as the one from the main Makefile. But this is not the case in this kernel version. Patch 2 fixes that.
Finally, I noticed an extra warning that I didn't have in v5.15. Patch 3 fixes that.
As with 5.4.y, this kernel isn't going to be around all that much longer, and I strongly doubt anyone that relies on it is also using gcc15 with it. Normally near the end of the 6 year window of kernels, we are "stuck" with using older gcc releases with that, and not newer ones, and that's fine, as that's what the users of those kernels are also doing.
So I don't think the effort is worth it. gcc14 will be around in distros for the next year so all should be ok here. Just like gcc13 was around long enough for us to keep 4.19.y working just fine with that maximum compiler release as well.
thanks,
greg k-h
Hi Greg,
Thank you for your reply!
On 20/10/2025 15:32, Greg Kroah-Hartman wrote:
On Fri, Oct 17, 2025 at 06:53:24PM +0200, Matthieu Baerts (NGI0) wrote:
This kernel version doesn't build with GCC 15:
In file included from include/uapi/linux/posix_types.h:5, from include/uapi/linux/types.h:14, from include/linux/types.h:6, from arch/x86/realmode/rm/wakeup.h:11, from arch/x86/realmode/rm/wakemain.c:2: include/linux/stddef.h:11:9: error: cannot use keyword 'false' as enumeration constant 11 | false = 0, | ^~~~~ include/linux/stddef.h:11:9: note: 'false' is a keyword with '-std=c23' onwards include/linux/types.h:30:33: error: 'bool' cannot be defined via 'typedef' 30 | typedef _Bool bool; | ^~~~ include/linux/types.h:30:33: note: 'bool' is a keyword with '-std=c23' onwards include/linux/types.h:30:1: warning: useless type name in empty declaration 30 | typedef _Bool bool; | ^~~~~~~
I initially fixed this by adding -std=gnu11 in arch/x86/Makefile, then I realised this fix was already done in an upstream commit, created before the GCC 15 release and not mentioning the error I had. This is the first patch.
When I was investigating my error, I noticed other commits were already backported to stable versions. They were all adding -std=gnu11 in different Makefiles. In their commit message, they were mentioning 'gnu11' was picked to use the same as the one from the main Makefile. But this is not the case in this kernel version. Patch 2 fixes that.
Finally, I noticed an extra warning that I didn't have in v5.15. Patch 3 fixes that.
As with 5.4.y, this kernel isn't going to be around all that much longer, and I strongly doubt anyone that relies on it is also using gcc15 with it. Normally near the end of the 6 year window of kernels, we are "stuck" with using older gcc releases with that, and not newer ones, and that's fine, as that's what the users of those kernels are also doing.
So I don't think the effort is worth it. gcc14 will be around in distros for the next year so all should be ok here. Just like gcc13 was around long enough for us to keep 4.19.y working just fine with that maximum compiler release as well.
Fine by me, but just to avoid confusions, I would like to add:
- Only the first patch fixes the GCC 15 support, and it only adds '-std=gnu11' in x86/boot. All the other GCC 15 specific patches have already been backported, only that one didn't have a Fixes tag and was missing in <= v5.15.
- Patch 2/3 fixes the (accidental) switch to C11 standard for some arch. It should not be a problem in v5.10, because GCC minimal version is 4.9 which supports C11 if I'm not mistaken. It is just to be consistent with what is done in the main Makefile, but it can be drop if preferred.
- Patch 3/3 is not linked to GCC 15 support, it fixes an ISO C90 warning.
Cheers, Matt
On 10/17/25 18:53, Matthieu Baerts (NGI0) wrote:
This kernel version doesn't build with GCC 15:
In file included from include/uapi/linux/posix_types.h:5, from include/uapi/linux/types.h:14, from include/linux/types.h:6, from arch/x86/realmode/rm/wakeup.h:11, from arch/x86/realmode/rm/wakemain.c:2: include/linux/stddef.h:11:9: error: cannot use keyword 'false' as enumeration constant 11 | false = 0, | ^~~~~ include/linux/stddef.h:11:9: note: 'false' is a keyword with '-std=c23' onwards include/linux/types.h:30:33: error: 'bool' cannot be defined via 'typedef' 30 | typedef _Bool bool; | ^~~~ include/linux/types.h:30:33: note: 'bool' is a keyword with '-std=c23' onwards include/linux/types.h:30:1: warning: useless type name in empty declaration 30 | typedef _Bool bool; | ^~~~~~~
I initially fixed this by adding -std=gnu11 in arch/x86/Makefile, then I realised this fix was already done in an upstream commit, created before the GCC 15 release and not mentioning the error I had. This is the first patch.
When I was investigating my error, I noticed other commits were already backported to stable versions. They were all adding -std=gnu11 in different Makefiles. In their commit message, they were mentioning 'gnu11' was picked to use the same as the one from the main Makefile. But this is not the case in this kernel version. Patch 2 fixes that.
Finally, I noticed an extra warning that I didn't have in v5.15. Patch 3 fixes that.
Signed-off-by: Matthieu Baerts (NGI0) matttbe@kernel.org
Alexey Dobriyan (1): x86/boot: Compile boot code with -std=gnu11 too
Matthieu Baerts (NGI0) (2): arch: back to -std=gnu89 in < v5.18 tracing: fix declaration-after-statement warning
arch/parisc/boot/compressed/Makefile | 2 +- arch/s390/Makefile | 2 +- arch/s390/purgatory/Makefile | 2 +- arch/x86/Makefile | 2 +- arch/x86/boot/compressed/Makefile | 2 +- drivers/firmware/efi/libstub/Makefile | 2 +- kernel/trace/trace_events_synth.c | 3 ++- 7 files changed, 8 insertions(+), 7 deletions(-)
base-commit: a32db271d59d9f35f3a937ac27fcc2db1e029cdc change-id: 20251017-v5-10-gcc-15-ad2510f784f7
Best regards,
The last kernel I was able to build with GCC15 was 5.10.245. With 5.10.246 we have again the 'FAILED unresolved symbol filp_close' error. Attached you may find the full build log. 5.4.x and 5.15.x series still compile fine with GCC15. We already had the issue with 5.10.235 back in March: https://lore.kernel.org/stable/b39ca723-16a4-42a2-b8ca-b97d0e4bf7f5@manjaro....
linux-stable-mirror@lists.linaro.org
-
Greg Kroah-Hartman -
Matthieu Baerts -
Matthieu Baerts (NGI0)
-
Philip Müller