llvm does does not understand -march=z9-109 and older target
specifiers, so disable the respective Kconfig settings and
the logic to make the boot code work on old systems when
building with clang.
Part of the early boot code is normally compiled with -march=z900
for maximum compatibility. This also has to get changed with
clang to the oldest supported ISA, which is -march=z10 here.
Signed-off-by: Arnd Bergmann <[email protected]>
---
arch/s390/Kconfig | 6 ++++++
arch/s390/boot/Makefile | 18 ++++++++++++------
2 files changed, 18 insertions(+), 6 deletions(-)
diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig
index 8cd860cba4d1..1a2eec61196d 100644
--- a/arch/s390/Kconfig
+++ b/arch/s390/Kconfig
@@ -240,6 +240,7 @@ choice
config MARCH_Z900
bool "IBM zSeries model z800 and z900"
+ depends on !CC_IS_CLANG
select HAVE_MARCH_Z900_FEATURES
help
Select this to enable optimizations for model z800/z900 (2064 and
@@ -248,6 +249,7 @@ config MARCH_Z900
config MARCH_Z990
bool "IBM zSeries model z890 and z990"
+ depends on !CC_IS_CLANG
select HAVE_MARCH_Z990_FEATURES
help
Select this to enable optimizations for model z890/z990 (2084 and
@@ -256,6 +258,7 @@ config MARCH_Z990
config MARCH_Z9_109
bool "IBM System z9"
+ depends on !CC_IS_CLANG
select HAVE_MARCH_Z9_109_FEATURES
help
Select this to enable optimizations for IBM System z9 (2094 and
@@ -347,12 +350,15 @@ config TUNE_DEFAULT
config TUNE_Z900
bool "IBM zSeries model z800 and z900"
+ depends on !CC_IS_CLANG
config TUNE_Z990
bool "IBM zSeries model z890 and z990"
+ depends on !CC_IS_CLANG
config TUNE_Z9_109
bool "IBM System z9"
+ depends on !CC_IS_CLANG
config TUNE_Z10
bool "IBM System z10"
diff --git a/arch/s390/boot/Makefile b/arch/s390/boot/Makefile
index c844eaf24ed7..c6a6b6dd3d52 100644
--- a/arch/s390/boot/Makefile
+++ b/arch/s390/boot/Makefile
@@ -12,18 +12,24 @@ KBUILD_AFLAGS := $(KBUILD_AFLAGS_DECOMPRESSOR)
KBUILD_CFLAGS := $(KBUILD_CFLAGS_DECOMPRESSOR)
#
-# Use -march=z900 for als.c to be able to print an error
+# Use minimum architecture for als.c to be able to print an error
# message if the kernel is started on a machine which is too old
#
-ifneq ($(CC_FLAGS_MARCH),-march=z900)
+ifndef CONFIG_CC_IS_CLANG
+CC_FLAGS_MARCH_MINIMUM := -march=z900
+else
+CC_FLAGS_MARCH_MINIMUM := -march=z10
+endif
+
+ifneq ($(CC_FLAGS_MARCH),$(CC_FLAGS_MARCH_MINIMUM))
AFLAGS_REMOVE_head.o += $(CC_FLAGS_MARCH)
-AFLAGS_head.o += -march=z900
+AFLAGS_head.o += $(CC_FLAGS_MARCH_MINIMUM)
AFLAGS_REMOVE_mem.o += $(CC_FLAGS_MARCH)
-AFLAGS_mem.o += -march=z900
+AFLAGS_mem.o += $(CC_FLAGS_MARCH_MINIMUM)
CFLAGS_REMOVE_als.o += $(CC_FLAGS_MARCH)
-CFLAGS_als.o += -march=z900
+CFLAGS_als.o += $(CC_FLAGS_MARCH_MINIMUM)
CFLAGS_REMOVE_sclp_early_core.o += $(CC_FLAGS_MARCH)
-CFLAGS_sclp_early_core.o += -march=z900
+CFLAGS_sclp_early_core.o += $(CC_FLAGS_MARCH_MINIMUM)
endif
CFLAGS_sclp_early_core.o += -I$(srctree)/drivers/s390/char
--
2.20.0
The purgatory and boot Makefiles do not inherit the original cflags,
so clang falls back to the default target architecture when building it,
typically this would be x86 when cross-compiling.
Add $(CLANG_FLAGS) everywhere so we pass the correct --target=s390x-linux
option when cross-compiling.
Signed-off-by: Arnd Bergmann <[email protected]>
---
arch/s390/Makefile | 4 ++--
arch/s390/purgatory/Makefile | 1 +
2 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/arch/s390/Makefile b/arch/s390/Makefile
index 9c079a506325..9a228786e34f 100644
--- a/arch/s390/Makefile
+++ b/arch/s390/Makefile
@@ -17,9 +17,9 @@ KBUILD_CFLAGS_MODULE += -fPIC
KBUILD_AFLAGS += -m64
KBUILD_CFLAGS += -m64
aflags_dwarf := -Wa,-gdwarf-2
-KBUILD_AFLAGS_DECOMPRESSOR := -m64 -D__ASSEMBLY__
+KBUILD_AFLAGS_DECOMPRESSOR := $(CLANG_FLAGS) -m64 -D__ASSEMBLY__
KBUILD_AFLAGS_DECOMPRESSOR += $(if $(CONFIG_DEBUG_INFO),$(aflags_dwarf))
-KBUILD_CFLAGS_DECOMPRESSOR := -m64 -O2
+KBUILD_CFLAGS_DECOMPRESSOR := $(CLANG_FLAGS) -m64 -O2
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 ce6a3f75065b..ecd0b3847fef 100644
--- a/arch/s390/purgatory/Makefile
+++ b/arch/s390/purgatory/Makefile
@@ -22,6 +22,7 @@ KBUILD_CFLAGS := -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
+KBUILD_CFLAGS += $(CLANG_FLAGS)
KBUILD_CFLAGS += $(call cc-option,-fno-PIE)
KBUILD_AFLAGS := $(filter-out -DCC_USING_EXPOLINE,$(KBUILD_AFLAGS))
--
2.20.0
VIRT_TO_BUS is only used for legacy device PCI and ISA drivers using
virt_to_bus() instead of the streaming DMA mapping API, and the
remaining drivers generally don't work on 64-bit architectures.
Two of these drivers also cause a build warning on s390, so instead
of trying to fix that, let's just disable the option as we do on
most architectures now.
Signed-off-by: Arnd Bergmann <[email protected]>
---
arch/s390/Kconfig | 1 -
1 file changed, 1 deletion(-)
diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig
index 1a2eec61196d..25e0e7375166 100644
--- a/arch/s390/Kconfig
+++ b/arch/s390/Kconfig
@@ -188,7 +188,6 @@ config S390
select TTY
select VIRT_CPU_ACCOUNTING
select ARCH_HAS_SCALED_CPUTIME
- select VIRT_TO_BUS
select HAVE_NMI
--
2.20.0
The arch/s390/boot directory is built with its own set of compiler
options that does not include -Wno-pointer-sign like the rest of
the kernel does, this causes a lot of harmess but correct warnings
when building with clang.
For the atomics, we can add type casts to avoid the warnings, for
everything else the easiest way is to slightly adapt the types
to be more consistent.
Signed-off-by: Arnd Bergmann <[email protected]>
---
arch/s390/boot/ipl_parm.c | 2 +-
arch/s390/include/asm/bitops.h | 12 ++++++------
arch/s390/include/asm/ebcdic.h | 2 +-
arch/s390/include/asm/lowcore.h | 2 +-
drivers/s390/char/sclp.h | 4 ++--
5 files changed, 11 insertions(+), 11 deletions(-)
diff --git a/arch/s390/boot/ipl_parm.c b/arch/s390/boot/ipl_parm.c
index 36beb56de021..3f2f0a9f1dad 100644
--- a/arch/s390/boot/ipl_parm.c
+++ b/arch/s390/boot/ipl_parm.c
@@ -51,7 +51,7 @@ void store_ipl_parmblock(void)
early_ipl_block_valid = 1;
}
-static size_t scpdata_length(const char *buf, size_t count)
+static size_t scpdata_length(const u8 *buf, size_t count)
{
while (count) {
if (buf[count - 1] != '\0' && buf[count - 1] != ' ')
diff --git a/arch/s390/include/asm/bitops.h b/arch/s390/include/asm/bitops.h
index d1f8a4d94cca..9900d655014c 100644
--- a/arch/s390/include/asm/bitops.h
+++ b/arch/s390/include/asm/bitops.h
@@ -73,7 +73,7 @@ static inline void set_bit(unsigned long nr, volatile unsigned long *ptr)
}
#endif
mask = 1UL << (nr & (BITS_PER_LONG - 1));
- __atomic64_or(mask, addr);
+ __atomic64_or(mask, (long *)addr);
}
static inline void clear_bit(unsigned long nr, volatile unsigned long *ptr)
@@ -94,7 +94,7 @@ static inline void clear_bit(unsigned long nr, volatile unsigned long *ptr)
}
#endif
mask = ~(1UL << (nr & (BITS_PER_LONG - 1)));
- __atomic64_and(mask, addr);
+ __atomic64_and(mask, (long *)addr);
}
static inline void change_bit(unsigned long nr, volatile unsigned long *ptr)
@@ -115,7 +115,7 @@ static inline void change_bit(unsigned long nr, volatile unsigned long *ptr)
}
#endif
mask = 1UL << (nr & (BITS_PER_LONG - 1));
- __atomic64_xor(mask, addr);
+ __atomic64_xor(mask, (long *)addr);
}
static inline int
@@ -125,7 +125,7 @@ test_and_set_bit(unsigned long nr, volatile unsigned long *ptr)
unsigned long old, mask;
mask = 1UL << (nr & (BITS_PER_LONG - 1));
- old = __atomic64_or_barrier(mask, addr);
+ old = __atomic64_or_barrier(mask, (long *)addr);
return (old & mask) != 0;
}
@@ -136,7 +136,7 @@ test_and_clear_bit(unsigned long nr, volatile unsigned long *ptr)
unsigned long old, mask;
mask = ~(1UL << (nr & (BITS_PER_LONG - 1)));
- old = __atomic64_and_barrier(mask, addr);
+ old = __atomic64_and_barrier(mask, (long *)addr);
return (old & ~mask) != 0;
}
@@ -147,7 +147,7 @@ test_and_change_bit(unsigned long nr, volatile unsigned long *ptr)
unsigned long old, mask;
mask = 1UL << (nr & (BITS_PER_LONG - 1));
- old = __atomic64_xor_barrier(mask, addr);
+ old = __atomic64_xor_barrier(mask, (long *)addr);
return (old & mask) != 0;
}
diff --git a/arch/s390/include/asm/ebcdic.h b/arch/s390/include/asm/ebcdic.h
index 29441beb92e6..d3ce6adacb10 100644
--- a/arch/s390/include/asm/ebcdic.h
+++ b/arch/s390/include/asm/ebcdic.h
@@ -20,7 +20,7 @@ extern __u8 _ebc_tolower[256]; /* EBCDIC -> lowercase */
extern __u8 _ebc_toupper[256]; /* EBCDIC -> uppercase */
static inline void
-codepage_convert(const __u8 *codepage, volatile __u8 * addr, unsigned long nr)
+codepage_convert(const __u8 *codepage, volatile char* addr, unsigned long nr)
{
if (nr-- <= 0)
return;
diff --git a/arch/s390/include/asm/lowcore.h b/arch/s390/include/asm/lowcore.h
index cc0947e08b6f..f3a637afd485 100644
--- a/arch/s390/include/asm/lowcore.h
+++ b/arch/s390/include/asm/lowcore.h
@@ -128,7 +128,7 @@ struct lowcore {
/* SMP info area */
__u32 cpu_nr; /* 0x0398 */
__u32 softirq_pending; /* 0x039c */
- __u32 preempt_count; /* 0x03a0 */
+ __s32 preempt_count; /* 0x03a0 */
__u32 spinlock_lockval; /* 0x03a4 */
__u32 spinlock_index; /* 0x03a8 */
__u32 fpu_flags; /* 0x03ac */
diff --git a/drivers/s390/char/sclp.h b/drivers/s390/char/sclp.h
index 367e9d384d85..5918479ffa0e 100644
--- a/drivers/s390/char/sclp.h
+++ b/drivers/s390/char/sclp.h
@@ -365,14 +365,14 @@ sclp_ascebc(unsigned char ch)
/* translate string from EBCDIC to ASCII */
static inline void
-sclp_ebcasc_str(unsigned char *str, int nr)
+sclp_ebcasc_str(char *str, int nr)
{
(MACHINE_IS_VM) ? EBCASC(str, nr) : EBCASC_500(str, nr);
}
/* translate string from ASCII to EBCDIC */
static inline void
-sclp_ascebc_str(unsigned char *str, int nr)
+sclp_ascebc_str(char *str, int nr)
{
(MACHINE_IS_VM) ? ASCEBC(str, nr) : ASCEBC_500(str, nr);
}
--
2.20.0
On Mon, Apr 15, 2019 at 10:35:52AM +0200, Arnd Bergmann wrote:
> The purgatory and boot Makefiles do not inherit the original cflags,
> so clang falls back to the default target architecture when building it,
> typically this would be x86 when cross-compiling.
>
> Add $(CLANG_FLAGS) everywhere so we pass the correct --target=s390x-linux
> option when cross-compiling.
>
> Signed-off-by: Arnd Bergmann <[email protected]>
Reviewed-by: Nathan Chancellor <[email protected]>
> ---
> arch/s390/Makefile | 4 ++--
> arch/s390/purgatory/Makefile | 1 +
> 2 files changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/arch/s390/Makefile b/arch/s390/Makefile
> index 9c079a506325..9a228786e34f 100644
> --- a/arch/s390/Makefile
> +++ b/arch/s390/Makefile
> @@ -17,9 +17,9 @@ KBUILD_CFLAGS_MODULE += -fPIC
> KBUILD_AFLAGS += -m64
> KBUILD_CFLAGS += -m64
> aflags_dwarf := -Wa,-gdwarf-2
> -KBUILD_AFLAGS_DECOMPRESSOR := -m64 -D__ASSEMBLY__
> +KBUILD_AFLAGS_DECOMPRESSOR := $(CLANG_FLAGS) -m64 -D__ASSEMBLY__
> KBUILD_AFLAGS_DECOMPRESSOR += $(if $(CONFIG_DEBUG_INFO),$(aflags_dwarf))
> -KBUILD_CFLAGS_DECOMPRESSOR := -m64 -O2
> +KBUILD_CFLAGS_DECOMPRESSOR := $(CLANG_FLAGS) -m64 -O2
> 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 ce6a3f75065b..ecd0b3847fef 100644
> --- a/arch/s390/purgatory/Makefile
> +++ b/arch/s390/purgatory/Makefile
> @@ -22,6 +22,7 @@ KBUILD_CFLAGS := -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
> +KBUILD_CFLAGS += $(CLANG_FLAGS)
> KBUILD_CFLAGS += $(call cc-option,-fno-PIE)
> KBUILD_AFLAGS := $(filter-out -DCC_USING_EXPOLINE,$(KBUILD_AFLAGS))
>
> --
> 2.20.0
>
On Mon, Apr 15, 2019 at 1:37 AM Arnd Bergmann <[email protected]> wrote:
>
> The arch/s390/boot directory is built with its own set of compiler
> options that does not include -Wno-pointer-sign like the rest of
> the kernel does, this causes a lot of harmess but correct warnings
s/harmess/harmless/
> when building with clang.
>
> For the atomics, we can add type casts to avoid the warnings, for
> everything else the easiest way is to slightly adapt the types
> to be more consistent.
>
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---
> arch/s390/boot/ipl_parm.c | 2 +-
> arch/s390/include/asm/bitops.h | 12 ++++++------
> arch/s390/include/asm/ebcdic.h | 2 +-
> arch/s390/include/asm/lowcore.h | 2 +-
> drivers/s390/char/sclp.h | 4 ++--
> 5 files changed, 11 insertions(+), 11 deletions(-)
>
> diff --git a/arch/s390/boot/ipl_parm.c b/arch/s390/boot/ipl_parm.c
> index 36beb56de021..3f2f0a9f1dad 100644
> --- a/arch/s390/boot/ipl_parm.c
> +++ b/arch/s390/boot/ipl_parm.c
> @@ -51,7 +51,7 @@ void store_ipl_parmblock(void)
> early_ipl_block_valid = 1;
> }
>
> -static size_t scpdata_length(const char *buf, size_t count)
> +static size_t scpdata_length(const u8 *buf, size_t count)
> {
> while (count) {
> if (buf[count - 1] != '\0' && buf[count - 1] != ' ')
> diff --git a/arch/s390/include/asm/bitops.h b/arch/s390/include/asm/bitops.h
> index d1f8a4d94cca..9900d655014c 100644
> --- a/arch/s390/include/asm/bitops.h
> +++ b/arch/s390/include/asm/bitops.h
> @@ -73,7 +73,7 @@ static inline void set_bit(unsigned long nr, volatile unsigned long *ptr)
> }
> #endif
> mask = 1UL << (nr & (BITS_PER_LONG - 1));
> - __atomic64_or(mask, addr);
> + __atomic64_or(mask, (long *)addr);
> }
>
> static inline void clear_bit(unsigned long nr, volatile unsigned long *ptr)
> @@ -94,7 +94,7 @@ static inline void clear_bit(unsigned long nr, volatile unsigned long *ptr)
> }
> #endif
> mask = ~(1UL << (nr & (BITS_PER_LONG - 1)));
> - __atomic64_and(mask, addr);
> + __atomic64_and(mask, (long *)addr);
> }
>
> static inline void change_bit(unsigned long nr, volatile unsigned long *ptr)
> @@ -115,7 +115,7 @@ static inline void change_bit(unsigned long nr, volatile unsigned long *ptr)
> }
> #endif
> mask = 1UL << (nr & (BITS_PER_LONG - 1));
> - __atomic64_xor(mask, addr);
> + __atomic64_xor(mask, (long *)addr);
> }
>
> static inline int
> @@ -125,7 +125,7 @@ test_and_set_bit(unsigned long nr, volatile unsigned long *ptr)
> unsigned long old, mask;
>
> mask = 1UL << (nr & (BITS_PER_LONG - 1));
> - old = __atomic64_or_barrier(mask, addr);
> + old = __atomic64_or_barrier(mask, (long *)addr);
> return (old & mask) != 0;
> }
>
> @@ -136,7 +136,7 @@ test_and_clear_bit(unsigned long nr, volatile unsigned long *ptr)
> unsigned long old, mask;
>
> mask = ~(1UL << (nr & (BITS_PER_LONG - 1)));
> - old = __atomic64_and_barrier(mask, addr);
> + old = __atomic64_and_barrier(mask, (long *)addr);
> return (old & ~mask) != 0;
> }
>
> @@ -147,7 +147,7 @@ test_and_change_bit(unsigned long nr, volatile unsigned long *ptr)
> unsigned long old, mask;
>
> mask = 1UL << (nr & (BITS_PER_LONG - 1));
> - old = __atomic64_xor_barrier(mask, addr);
> + old = __atomic64_xor_barrier(mask, (long *)addr);
Thanks for the follow up. This hunk and above all LGTM.
> return (old & mask) != 0;
> }
>
> diff --git a/arch/s390/include/asm/ebcdic.h b/arch/s390/include/asm/ebcdic.h
> index 29441beb92e6..d3ce6adacb10 100644
> --- a/arch/s390/include/asm/ebcdic.h
> +++ b/arch/s390/include/asm/ebcdic.h
> @@ -20,7 +20,7 @@ extern __u8 _ebc_tolower[256]; /* EBCDIC -> lowercase */
> extern __u8 _ebc_toupper[256]; /* EBCDIC -> uppercase */
>
> static inline void
> -codepage_convert(const __u8 *codepage, volatile __u8 * addr, unsigned long nr)
> +codepage_convert(const __u8 *codepage, volatile char* addr, unsigned long nr)
> {
> if (nr-- <= 0)
> return;
There are many call sites of ASCEBC which is defined in terms of this
function. Do they all use `char*`? grep shows an explicit cast to
`unsigned char*` in drivers/s390/char/tape_std.c for example.
> diff --git a/arch/s390/include/asm/lowcore.h b/arch/s390/include/asm/lowcore.h
> index cc0947e08b6f..f3a637afd485 100644
> --- a/arch/s390/include/asm/lowcore.h
> +++ b/arch/s390/include/asm/lowcore.h
> @@ -128,7 +128,7 @@ struct lowcore {
> /* SMP info area */
> __u32 cpu_nr; /* 0x0398 */
> __u32 softirq_pending; /* 0x039c */
> - __u32 preempt_count; /* 0x03a0 */
> + __s32 preempt_count; /* 0x03a0 */
This change is less obvious. Do you still have the warning for that
this hunk fixes?
> __u32 spinlock_lockval; /* 0x03a4 */
> __u32 spinlock_index; /* 0x03a8 */
> __u32 fpu_flags; /* 0x03ac */
> diff --git a/drivers/s390/char/sclp.h b/drivers/s390/char/sclp.h
> index 367e9d384d85..5918479ffa0e 100644
> --- a/drivers/s390/char/sclp.h
> +++ b/drivers/s390/char/sclp.h
> @@ -365,14 +365,14 @@ sclp_ascebc(unsigned char ch)
>
> /* translate string from EBCDIC to ASCII */
> static inline void
> -sclp_ebcasc_str(unsigned char *str, int nr)
> +sclp_ebcasc_str(char *str, int nr)
> {
> (MACHINE_IS_VM) ? EBCASC(str, nr) : EBCASC_500(str, nr);
> }
>
> /* translate string from ASCII to EBCDIC */
> static inline void
> -sclp_ascebc_str(unsigned char *str, int nr)
> +sclp_ascebc_str(char *str, int nr)
> {
> (MACHINE_IS_VM) ? ASCEBC(str, nr) : ASCEBC_500(str, nr);
> }
> --
> 2.20.0
>
--
Thanks,
~Nick Desaulniers
On Mon, Apr 22, 2019 at 7:52 PM 'Nick Desaulniers' via Clang Built
Linux <[email protected]> wrote:
> > @@ -20,7 +20,7 @@ extern __u8 _ebc_tolower[256]; /* EBCDIC -> lowercase */
> > extern __u8 _ebc_toupper[256]; /* EBCDIC -> uppercase */
> >
> > static inline void
> > -codepage_convert(const __u8 *codepage, volatile __u8 * addr, unsigned long nr)
> > +codepage_convert(const __u8 *codepage, volatile char* addr, unsigned long nr)
> > {
> > if (nr-- <= 0)
> > return;
>
> There are many call sites of ASCEBC which is defined in terms of this
> function. Do they all use `char*`? grep shows an explicit cast to
> `unsigned char*` in drivers/s390/char/tape_std.c for example.
Generally speaking, the kernel is full of Wpointer-sign warnings, that's why
this warning is disabled in the top-level Makefile by default. My patch fixes
the ones in the s390 boot code that is not built with those default flags, but
I made no attempt to fix the rest of the kernel.
Fun fact: on most architectures, 'char' is signed, but on s390 and 32-bit
arm it is unsigned. The compiler treats 'char', 'unsigned char' and
'signed char'
as three distinct types here for that reason.
> > diff --git a/arch/s390/include/asm/lowcore.h b/arch/s390/include/asm/lowcore.h
> > index cc0947e08b6f..f3a637afd485 100644
> > --- a/arch/s390/include/asm/lowcore.h
> > +++ b/arch/s390/include/asm/lowcore.h
> > @@ -128,7 +128,7 @@ struct lowcore {
> > /* SMP info area */
> > __u32 cpu_nr; /* 0x0398 */
> > __u32 softirq_pending; /* 0x039c */
> > - __u32 preempt_count; /* 0x03a0 */
> > + __s32 preempt_count; /* 0x03a0 */
>
> This change is less obvious. Do you still have the warning for that
> this hunk fixes?
I can't find it now, but there are multiple users that would cause
this, like:
static inline void preempt_count_set(int pc)
{
int old, new;
do {
old = READ_ONCE(S390_lowcore.preempt_count);
new = (old & PREEMPT_NEED_RESCHED) |
(pc & ~PREEMPT_NEED_RESCHED);
} while (__atomic_cmpxchg(&S390_lowcore.preempt_count,
old, new) != old);
}
Arnd
On Tue, Apr 23, 2019 at 1:06 AM Arnd Bergmann <[email protected]> wrote:
>
> On Mon, Apr 22, 2019 at 7:52 PM 'Nick Desaulniers' via Clang Built
> Linux <[email protected]> wrote:
>
> > > @@ -20,7 +20,7 @@ extern __u8 _ebc_tolower[256]; /* EBCDIC -> lowercase */
> > > extern __u8 _ebc_toupper[256]; /* EBCDIC -> uppercase */
> > >
> > > static inline void
> > > -codepage_convert(const __u8 *codepage, volatile __u8 * addr, unsigned long nr)
> > > +codepage_convert(const __u8 *codepage, volatile char* addr, unsigned long nr)
> > > {
> > > if (nr-- <= 0)
> > > return;
> >
> > There are many call sites of ASCEBC which is defined in terms of this
> > function. Do they all use `char*`? grep shows an explicit cast to
> > `unsigned char*` in drivers/s390/char/tape_std.c for example.
>
> Generally speaking, the kernel is full of Wpointer-sign warnings, that's why
> this warning is disabled in the top-level Makefile by default. My patch fixes
> the ones in the s390 boot code that is not built with those default flags, but
> I made no attempt to fix the rest of the kernel.
Right, sorry, I forgot about that. This patch looks good to me.
Reviewed-by: Nick Desaulniers <[email protected]>
>
> Fun fact: on most architectures, 'char' is signed, but on s390 and 32-bit
> arm it is unsigned. The compiler treats 'char', 'unsigned char' and
> 'signed char'
> as three distinct types here for that reason.
I think I recall reading about that in:
https://www.amazon.com/ARM-System-Developers-Guide-Architecture/dp/1558608745
(I'll try to dig it up and post what the explanation was).
--
Thanks,
~Nick Desaulniers
On Mon, Apr 15, 2019 at 10:35:51AM +0200, Arnd Bergmann wrote:
> llvm does does not understand -march=z9-109 and older target
> specifiers, so disable the respective Kconfig settings and
> the logic to make the boot code work on old systems when
> building with clang.
>
> Part of the early boot code is normally compiled with -march=z900
> for maximum compatibility. This also has to get changed with
> clang to the oldest supported ISA, which is -march=z10 here.
>
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---
> arch/s390/Kconfig | 6 ++++++
> arch/s390/boot/Makefile | 18 ++++++++++++------
> 2 files changed, 18 insertions(+), 6 deletions(-)
Applied, thanks.
On Mon, Apr 15, 2019 at 09:12:06AM -0700, Nathan Chancellor wrote:
> On Mon, Apr 15, 2019 at 10:35:52AM +0200, Arnd Bergmann wrote:
> > The purgatory and boot Makefiles do not inherit the original cflags,
> > so clang falls back to the default target architecture when building it,
> > typically this would be x86 when cross-compiling.
> >
> > Add $(CLANG_FLAGS) everywhere so we pass the correct --target=s390x-linux
> > option when cross-compiling.
> >
> > Signed-off-by: Arnd Bergmann <[email protected]>
>
> Reviewed-by: Nathan Chancellor <[email protected]>
>
> > ---
> > arch/s390/Makefile | 4 ++--
> > arch/s390/purgatory/Makefile | 1 +
> > 2 files changed, 3 insertions(+), 2 deletions(-)
Applied, thanks.
On Mon, Apr 15, 2019 at 10:35:53AM +0200, Arnd Bergmann wrote:
> VIRT_TO_BUS is only used for legacy device PCI and ISA drivers using
> virt_to_bus() instead of the streaming DMA mapping API, and the
> remaining drivers generally don't work on 64-bit architectures.
>
> Two of these drivers also cause a build warning on s390, so instead
> of trying to fix that, let's just disable the option as we do on
> most architectures now.
>
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---
> arch/s390/Kconfig | 1 -
> 1 file changed, 1 deletion(-)
Applied, thanks.
On Mon, Apr 15, 2019 at 10:35:54AM +0200, Arnd Bergmann wrote:
> The arch/s390/boot directory is built with its own set of compiler
> options that does not include -Wno-pointer-sign like the rest of
> the kernel does, this causes a lot of harmess but correct warnings
> when building with clang.
>
> For the atomics, we can add type casts to avoid the warnings, for
> everything else the easiest way is to slightly adapt the types
> to be more consistent.
>
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---
> arch/s390/boot/ipl_parm.c | 2 +-
> arch/s390/include/asm/bitops.h | 12 ++++++------
> arch/s390/include/asm/ebcdic.h | 2 +-
> arch/s390/include/asm/lowcore.h | 2 +-
> drivers/s390/char/sclp.h | 4 ++--
> 5 files changed, 11 insertions(+), 11 deletions(-)
Applied, with typo fix and a simple coding style issue.
Thanks!