- devel - example.com

[PATCH] Clang support for gnu_inline attribute
by Marcus Haehnel 06 Jul '24

06 Jul '24

From: Marcus Hähnel <marcus.haehnel(a)kernkonzept.com> Clang also supports the gnu_inline attribute and the __GNUC_STDC_INLINE__ macro (C99 semantics). However, it reports as GCC 4.2 compatible (__GNUC_MINOR__ / __GNUC__) and thus the current defines do not think it can support this. Add clang as an alternative for this support. Documentation shows that this attribute is supported since at least Clang 8. --- Requires the features.h change that introduces the __CLANG_PREREQ macro include/sys/cdefs.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/include/sys/cdefs.h b/include/sys/cdefs.h index 9b3a02177..656548c52 100644 --- a/include/sys/cdefs.h +++ b/include/sys/cdefs.h @@ -320,7 +320,7 @@ inline semantics, unless -fgnu89-inline is used. For -std=gnu99, forcing gnu_inline attribute does not change behavior, but may silence spurious warnings (such as in GCC 4.2). */ -#if !defined __cplusplus || __GNUC_PREREQ (4,3) +#if !defined __cplusplus || __GNUC_PREREQ (4,3) || __CLANG_PREREQ(8,0) # if defined __GNUC_STDC_INLINE__ || defined __GNUC_GNU_INLINE__ || defined __cplusplus # define __extern_inline extern __inline __attribute__ ((__gnu_inline__)) # if __GNUC_PREREQ (4,3) -- 2.43.0

3 2

[PATCH v2] Clang support for gnu_inline attribute
by Marcus Haehnel 06 Jul '24

06 Jul '24

From: Marcus Hähnel <marcus.haehnel(a)kernkonzept.com> Clang also supports the gnu_inline attribute and the __GNUC_STDC_INLINE__ macro (C99 semantics). However, it reports as GCC 4.2 compatible (__GNUC_MINOR__ / __GNUC__) and thus the current defines do not think it can support this. Add clang as an alternative for this support. Documentation shows that this attribute is supported since at least Clang 8. --- include/sys/cdefs.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/include/sys/cdefs.h b/include/sys/cdefs.h index 9b3a02177..656548c52 100644 --- a/include/sys/cdefs.h +++ b/include/sys/cdefs.h @@ -320,7 +320,7 @@ inline semantics, unless -fgnu89-inline is used. For -std=gnu99, forcing gnu_inline attribute does not change behavior, but may silence spurious warnings (such as in GCC 4.2). */ -#if !defined __cplusplus || __GNUC_PREREQ (4,3) +#if !defined __cplusplus || __GNUC_PREREQ (4,3) || __CLANG_PREREQ(8,0) # if defined __GNUC_STDC_INLINE__ || defined __GNUC_GNU_INLINE__ || defined __cplusplus # define __extern_inline extern __inline __attribute__ ((__gnu_inline__)) # if __GNUC_PREREQ (4,3) -- 2.43.0

1 0

[PATCH] features.h: add clang prerequisite macro
by Marcus Haehnel 05 Jul '24

05 Jul '24

This can be used to guard fatures for specific clang versions. --- include/features.h | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/include/features.h b/include/features.h index b5d4e79f2..1a4efb9db 100644 --- a/include/features.h +++ b/include/features.h @@ -140,6 +140,18 @@ # define __GNUC_PREREQ(maj, min) 0 #endif +/* Convenience macro to test the version of clang. + Use like this: + #if __CLANG_PREREQ(3,2) + ... code requiring clang 3.2 or later ... + #endif */ +#if defined __clang__ +# define __CLANG_PREREQ(maj, min) \ + ((__clang_major__ << 16) + __clang_minor__ >= ((maj) << 16) + (min)) +#else +# define __CLANG_PREREQ(maj, min) 0 +#endif + /* Whether to use feature set F. */ #define __GLIBC_USE(F) __GLIBC_USE_ ## F -- 2.43.0

1 0

[PATCH] c++: Remove deprecated dynamic exception specification
by Marcus Haehnel 05 Jul '24

05 Jul '24

From: Sven Linker <sven.linker(a)kernkonzept.com> Dynamic exception specification using `throw()` is deprecated since C++11 and removed in C++17. This change replaces `throw()` with `noexcept` when compiling for newer standards. Co-authored-by: Sven Linker <sven.linker(a)kernkonzept.com> --- include/sys/cdefs.h | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/include/sys/cdefs.h b/include/sys/cdefs.h index 5c4daebcd..9b3a02177 100644 --- a/include/sys/cdefs.h +++ b/include/sys/cdefs.h @@ -58,9 +58,15 @@ # define __NTH(fct) __attribute__ ((__nothrow__ __LEAF)) fct # else # if defined __cplusplus && __GNUC_PREREQ (2,8) -# define __THROW throw () -# define __THROWNL throw () -# define __NTH(fct) __LEAF_ATTR fct throw () +/* Dynamic exception specification is deprecated since C++11, so + we only use it when compiling for an earlier standard. */ +# if __cplusplus < 201103UL +# define __THROW throw () +# else +# define __THROW noexcept +# endif +# define __THROWNL __THROW +# define __NTH(fct) __LEAF_ATTR fct __THROW # else # define __THROW # define __THROWNL -- 2.43.0

1 0

[PATCH v2] uclibc: Fix double promotion warning
by Marcus Haehnel 05 Jul '24

05 Jul '24

Add casts where necessary to convince clang that the promotion of float to double is intentional. Signed-off-by: Sven Linker <sven.linker(a)kernkonzept.com> --- v2: Missed some changed files in the first version. libc/stdio/_fpmaxtostr.c | 4 ++-- libc/stdlib/_strtod.c | 10 +++++----- libm/cexp.c | 6 +++--- libm/float_wrappers.c | 2 +- libm/nan.c | 4 ++-- libm/w_cabs.c | 2 +- 6 files changed, 14 insertions(+), 14 deletions(-) diff --git a/libc/stdio/_fpmaxtostr.c b/libc/stdio/_fpmaxtostr.c index b06b25aa0..1f268fdaa 100644 --- a/libc/stdio/_fpmaxtostr.c +++ b/libc/stdio/_fpmaxtostr.c @@ -318,8 +318,8 @@ ssize_t _fpmaxtostr(FILE * fp, __fpmax_t x, struct printf_info *info, #else /* __UCLIBC_HAS_HEXADECIMAL_FLOATS__ */ -#define lower_bnd 1e8 -#define upper_bnd 1e9 +#define lower_bnd (__fpmax_t)1e8 +#define upper_bnd (__fpmax_t)1e9 #define power_table exp10_table #define dpb DIGITS_PER_BLOCK #define base 10 diff --git a/libc/stdlib/_strtod.c b/libc/stdlib/_strtod.c index c4c79e511..483551e64 100644 --- a/libc/stdlib/_strtod.c +++ b/libc/stdlib/_strtod.c @@ -256,7 +256,7 @@ __fpmax_t attribute_hidden __XL_NPP(__strtofpmax)(const Wchar *str, Wchar **endp } #endif - number = 0.; + number = (__fpmax_t)0; #ifdef _STRTOD_NEED_NUM_DIGITS num_digits = -1; #endif @@ -339,7 +339,7 @@ __fpmax_t attribute_hidden __XL_NPP(__strtofpmax)(const Wchar *str, Wchar **endp while ((pos[j] | 0x20) == nan_inf_str[i+1+j]) { ++j; if (!nan_inf_str[i+1+j]) { - number = i / 0.; + number = i / (__fpmax_t)0.; if (negative) { /* Correct for sign. */ number = -number; } @@ -414,7 +414,7 @@ __fpmax_t attribute_hidden __XL_NPP(__strtofpmax)(const Wchar *str, Wchar **endp } #ifdef _STRTOD_ZERO_CHECK - if (number == 0.) { + if (number == (__fpmax_t)0.) { goto DONE; } #endif @@ -515,7 +515,7 @@ float __XL_NPP(strtof)(const Wchar *str, Wchar **endptr __LOCALE_PARAM ) x = __XL_NPP(__strtofpmax)(str, endptr, 0 __LOCALE_ARG ); y = (float) x; - __fp_range_check(y, x); + __fp_range_check((__fpmax_t)y, x); return y; #endif @@ -549,7 +549,7 @@ double __XL_NPP(strtod)(const Wchar *__restrict str, x = __XL_NPP(__strtofpmax)(str, endptr, 0 __LOCALE_ARG ); y = (double) x; - __fp_range_check(y, x); + __fp_range_check((__fpmax_t)y, x); return y; #endif diff --git a/libm/cexp.c b/libm/cexp.c index 87512b7c5..a08d12d4e 100644 --- a/libm/cexp.c +++ b/libm/cexp.c @@ -38,10 +38,10 @@ libm_hidden_proto(cexpf) __complex__ float cexpf(__complex__ float z) { __complex__ float ret; - double r_exponent = exp(__real__ z); + double r_exponent = exp((double)__real__ z); - __real__ ret = r_exponent * cosf(__imag__ z); - __imag__ ret = r_exponent * sinf(__imag__ z); + __real__ ret = r_exponent * (double) cosf(__imag__ z); + __imag__ ret = r_exponent * (double) sinf(__imag__ z); return ret; } diff --git a/libm/float_wrappers.c b/libm/float_wrappers.c index 948f6bc14..35887dde2 100644 --- a/libm/float_wrappers.c +++ b/libm/float_wrappers.c @@ -230,7 +230,7 @@ long_WRAPPER1(lround) float modff (float x, float *iptr) { double y, result; - result = modf( x, &y ); + result = modf( (double)x, &y ); *iptr = (float)y; return (float) result; } diff --git a/libm/nan.c b/libm/nan.c index 454734b6f..9c2cd6e90 100644 --- a/libm/nan.c +++ b/libm/nan.c @@ -28,7 +28,7 @@ double nan (const char *tagp) sprintf (buf, "NAN(%s)", tagp); return strtod (buf, NULL); } - return NAN; + return (double)NAN; } libm_hidden_def(nan) @@ -53,7 +53,7 @@ long double nanl (const char *tagp) sprintf (buf, "NAN(%s)", tagp); return strtold (buf, NULL); } - return NAN; + return (long double)NAN; } libm_hidden_def(nanl) #endif diff --git a/libm/w_cabs.c b/libm/w_cabs.c index b2592484c..546b6affa 100644 --- a/libm/w_cabs.c +++ b/libm/w_cabs.c @@ -17,7 +17,7 @@ libm_hidden_def(cabs) libm_hidden_proto(cabsf) float cabsf(float _Complex z) { - return (float) hypot(__real__ z, __imag__ z); + return (float) hypot((double)__real__ z, (double)__imag__ z); } libm_hidden_def(cabsf) -- 2.43.0

1 0

[PATCH] libm: Fix float conversion compiler warning
by Marcus Haehnel 05 Jul '24

05 Jul '24

Make two implicit casts from double to int explicit to silence compiler warnings about them. The casts are required during the computation of exponentiation. Signed-off-by: Sven Linker <sven.linker(a)kernkonzept.com> --- libm/e_exp.c | 2 +- libm/s_expm1.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/libm/e_exp.c b/libm/e_exp.c index ffa556120..f694f67d1 100644 --- a/libm/e_exp.c +++ b/libm/e_exp.c @@ -126,7 +126,7 @@ double __ieee754_exp(double x) /* default IEEE double exp */ if(hx < 0x3FF0A2B2) { /* and |x| < 1.5 ln2 */ hi = x-ln2HI[xsb]; lo=ln2LO[xsb]; k = 1-xsb-xsb; } else { - k = invln2*x+halF[xsb]; + k = (int32_t)(invln2*x+halF[xsb]); t = k; hi = x - t*ln2HI[0]; /* t*ln2HI is exact here */ lo = t*ln2LO[0]; diff --git a/libm/s_expm1.c b/libm/s_expm1.c index 8e51ae748..85defefa4 100644 --- a/libm/s_expm1.c +++ b/libm/s_expm1.c @@ -159,7 +159,7 @@ double expm1(double x) else {hi = x + ln2_hi; lo = -ln2_lo; k = -1;} } else { - k = invln2*x+((xsb==0)?0.5:-0.5); + k = (int32_t)(invln2*x+((xsb==0)?0.5:-0.5)); t = k; hi = x - t*ln2_hi; /* t*ln2_hi is exact here */ lo = t*ln2_lo; -- 2.43.0

1 0

[PATCH] uclibc: Fix double promotion warning
by Marcus Haehnel 05 Jul '24

05 Jul '24

Add casts where necessary to convince clang that the promotion of float to double is intentional. Signed-off-by: Sven Linker <sven.linker(a)kernkonzept.com> --- libm/cexp.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/libm/cexp.c b/libm/cexp.c index 87512b7c5..a08d12d4e 100644 --- a/libm/cexp.c +++ b/libm/cexp.c @@ -38,10 +38,10 @@ libm_hidden_proto(cexpf) __complex__ float cexpf(__complex__ float z) { __complex__ float ret; - double r_exponent = exp(__real__ z); + double r_exponent = exp((double)__real__ z); - __real__ ret = r_exponent * cosf(__imag__ z); - __imag__ ret = r_exponent * sinf(__imag__ z); + __real__ ret = r_exponent * (double) cosf(__imag__ z); + __imag__ ret = r_exponent * (double) sinf(__imag__ z); return ret; } -- 2.43.0

1 0

[PATCH] ldso/dl-vdso: avoid compiler warning if configured without vdso
by Marcus Haehnel 04 Jul '24

04 Jul '24

The parameters are not used in the function variant without vdso support, so tell this to the compiler to avoid a warning. Also fix the indentation to be in-line with other functions in the file. --- ldso/ldso/dl-vdso.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/ldso/ldso/dl-vdso.c b/ldso/ldso/dl-vdso.c index 196cbbb3b..01309011d 100755 --- a/ldso/ldso/dl-vdso.c +++ b/ldso/ldso/dl-vdso.c @@ -28,14 +28,14 @@ #ifndef __VDSO_SUPPORT__ - void load_vdso(void *sys_info_ehdr, char **envp ){ +void load_vdso( void *sys_info_ehdr attribute_unused, + char **envp attribute_unused ){ #ifdef __SUPPORT_LD_DEBUG__ - if ( _dl_debug_vdso != 0 ){ - _dl_dprintf(2,"_dl_vdso support not enabled\n" ); - } - -#endif + if ( _dl_debug_vdso != 0 ){ + _dl_dprintf(2,"_dl_vdso support not enabled\n" ); } +#endif +} #else void *_dl__vdso_gettimeofday = 0; -- 2.43.0

1 0

1.0.49 release
by Waldemar Brodkorb 17 Jun '24

17 Jun '24

Hi, I released uClibc-ng 1.0.49 on Saturday. Many gcc 14.1.0 fixes were added thanks to Dmitry and Max. Xtensa FDPIC support was added by Max. git shortlog v1.0.49...v1.0.48 Dmitry Chestnykh (6): Correct uClibc compilation. Fix uClibc build for nds32 with gcc-14. Fix riscv32 build with gcc-14. riscv32: Fix `struct ucontext` definition. Fix wrong `struct ucontext_t` typedef for all arches. Cast to proper types inside atomic macroses. Fabrice Fontaine (1): libc/sysdeps/linux/common/utime.c: fix riscv32 build Joe Damato (1): epoll.h: Add epoll ioctls Max Filippov (9): iconv: fix type mismatches ldso: FDPIC: fix type mismatches ldso: arm: fix build with gcc-14 malloc/memalign: avoid integer overflow fix kernel_stat64 definition linuxthreads: enable unwinding over signal frames xtensa: use compiler-provided XCHAL macros xtensa: make _init and _fini hidden xtensa: add FDPIC support Waldemar Brodkorb (9): futimesat: add missing header sparc64: Fix incorrect sigreturn stub function implementation csky: allow time64 m68k: fix noMMU ELF compile with gcc 14.x Revert "nds32: sync with binutils 2.37, gcc 11.2 and linux 5.10.93 changes" m68k: fix for m68000 cpu config: make ctor/dtor visible again config: enable TIME64 by default bump version for 1.0.49 release Yuriy Kolerov (1): libc: cast free() argument to void * in wchar.c best regards Waldemar

3 2

[PATCH] libc: cast free() argument to void * in wchar.c
by kolerov93＠gmail.com 17 Jun '24

17 Jun '24

From: Yuriy Kolerov <ykolerov(a)synopsys.com> iconv_close() accepts iconv_t type (which is void *) and passes it to free() which accepts void *. However, GCC 14 raises a -Wint-conversion warning if it is not casted to void * because GCC cannot unwind typedef of iconv_t. Signed-off-by: Yuriy Kolerov <ykolerov(a)synopsys.com> --- libc/misc/wchar/wchar.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libc/misc/wchar/wchar.c b/libc/misc/wchar/wchar.c index 2714d47d7..782b67f93 100644 --- a/libc/misc/wchar/wchar.c +++ b/libc/misc/wchar/wchar.c @@ -1298,7 +1298,7 @@ iconv_t weak_function iconv_open(const char *tocode, const char *fromcode) int weak_function iconv_close(iconv_t cd) { - free(cd); + free((void *) cd); return 0; } -- 2.39.2

2 1