- devel - example.com

[PATCH] sched_setaffinity: use the same style with glibc's
by Guo Ren 16 Aug '18

16 Aug '18

Call getpid() in INTERNAL_SYSCALL will break the argument regs, because gcc couldn't save destoryed regs for system call asm. Ref to glibc, we could just remove all the check code. Signed-off-by: Guo Ren <ren_guo(a)c-sky.com> --- libc/sysdeps/linux/common/sched_setaffinity.c | 34 --------------------------- 1 file changed, 34 deletions(-) diff --git a/libc/sysdeps/linux/common/sched_setaffinity.c b/libc/sysdeps/linux/common/sched_setaffinity.c index 3d513ef..39f6ec8 100644 --- a/libc/sysdeps/linux/common/sched_setaffinity.c +++ b/libc/sysdeps/linux/common/sched_setaffinity.c @@ -27,42 +27,8 @@ static __always_inline _syscall3(int, __syscall_sched_setaffinity, __kernel_pid_t, pid, size_t, cpusetsize, const cpu_set_t *, cpuset) -static size_t __kernel_cpumask_size; - int sched_setaffinity(pid_t pid, size_t cpusetsize, const cpu_set_t *cpuset) { - size_t cnt; - int res; - size_t psize = 128; - void *p = alloca (psize); - - if (unlikely (__kernel_cpumask_size == 0)) { - - INTERNAL_SYSCALL_DECL (err); - while (res = INTERNAL_SYSCALL (sched_getaffinity, err, 3, getpid (), - psize, p), - INTERNAL_SYSCALL_ERROR_P (res, err) - && INTERNAL_SYSCALL_ERRNO (res, err) == EINVAL) - p = extend_alloca (p, psize, 2 * psize); - - if (res == 0 || INTERNAL_SYSCALL_ERROR_P (res, err)) { - __set_errno (INTERNAL_SYSCALL_ERRNO (res, err)); - return -1; - } - - __kernel_cpumask_size = res; - } - - /* We now know the size of the kernel cpumask_t. Make sure the user - does not request to set a bit beyond that. */ - for (cnt = __kernel_cpumask_size; cnt < cpusetsize; ++cnt) - if (((char *) cpuset)[cnt] != '\0') { - /* Found a nonzero byte. This means the user request cannot be - fulfilled. */ - __set_errno (EINVAL); - return -1; - } - return __syscall_sched_setaffinity(pid, cpusetsize, cpuset); } #endif -- 2.7.4

2 1

[PATCH 00/32] FDPIC ABI for ARM
by Christophe Lyon 10 Aug '18

10 Aug '18

Hello, This patch series implements the uClibc-ng contribution of the FDPIC ABI for ARM targets. This ABI enables to run Linux on ARM MMU-less cores and supports shared libraries to reduce the memory footprint. Without MMU, text and data segments relative distances are different from one process to another, hence the need for a dedicated FDPIC register holding the start address of the data segment. One of the side effects is that function pointers require two words to be represented: the address of the code, and the data segment start address. These two words are designated as "Function Descriptor", hence the "FD PIC" name. On ARM, the FDPIC register is r9 [1], and the target name is arm-uclinuxfdpiceabi. Note that arm-uclinux exists, but uses another ABI and the BFLAT file format; it does not support code sharing. This work was developed some time ago by STMicroelectronics, and was presented during Linaro Connect SFO15 (September 2015). You can watch the discussion and read the slides [2]. This presentation was related to the toolchain published on github [3], which is based on binutils-2.22, gcc-4.7, uclibc-0.9.33.2, gdb-7.5.1 and qemu-2.3.0, and for which pre-built binaries are available [3]. The ABI itself is described in details in [1]. Our Linux kernel patches have been updated and committed by Nicolas Pitre (Linaro) in July 2017. They are required so that the loader is able to handle this new file type. Indeed, the ELF files are tagged with ELFOSABI_ARM_FDPIC. This new tag has been allocated by ARM, as well as the new relocations involved. The binutils and QEMU patch series have been merged recently. [4][5] The GCC patch series has been submitted and is under discussion. [6] This patch series consists in the original uClibc-based patches, which I have rebased and updated for uClibc-ng. Most of them are strictly related to FDPIC on ARM, while a few others address generic issues. Are the uClibc-ng patches OK for inclusion in master? Thanks, Christophe. [1] https://github.com/mickael-guene/fdpic_doc/blob/master/abi.txt [2] http://connect.linaro.org/resource/sfo15/sfo15-406-arm-fdpic-toolset-kernel… [3] https://github.com/mickael-guene/fdpic_manifest [4] https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commit;h=f1ac0af… [5] https://git.qemu.org/?p=qemu.git;a=commit;h=e8fa72957419c11984608062c7dcb20… [6] https://gcc.gnu.org/ml/gcc-patches/2018-05/msg01439.html Christophe Lyon (32): [ARM][FDPIC] Allow to select FDPIC ELF for arm architecture [FDPIC] rtld: Do not protect RELRO segments when we don't use an MMU. [ARM][FDPIC] rtld: Add #if defined (__FDPIC__) [ARM][FDPIC] rtld: Add FDPIC code for arm [ARM][FDPIC] rtld: Avoid FUNCDESC relocation on _start [ARM][FDPIC] rtld: Add startup code [ARM] Fix bug in _dl_pread when using pread64 syscall [ARM][FDPIC] Add application startup code for FDPIC [ARM][FDPIC] Add runtime support needed for C++ exceptions [ARM][FDPIC] Allow to generate PIE 'static' binary [ARM][FDPIC] rtld: Add lazy binding support [ARM][FDPIC] TLS: fix relocation computation [ARM][FDPIC] rtld: Compile with -fno-unwind-tables -fno-asynchronous-unwind-tables [ARM][FDPIC] arm/clone.S: Fix threading support [ARM][FDPIC] enable NPTL on TARGET_arm [ARM][FDPIC] rtld: Use ELF_RTYPE_CLASS_DLSYM [FDPIC] rtld: Initialize _dl_error_catch_tsd without FUNCDESC relocation [FDPIC] nptl: disable mprotect usage in stack protection [FDPIC] nptl: Disable fork and atfork on MMU-less systems. [FDPIC] nptl: Do not use madvise [ARM][FDPIC] nptl: Use linker-defined symbol to find start of .tdata section. [FDPIC] nptl: Allow sem_open to work on MMU-less systems [FDPIC] nptl: Add pthread_mutex_getprioceiling and pthread_mutex_setprioceiling support [FDPIC] nptl: Use vfork on MMU-less for system() nptl: Clear TLS area for static binaries. [FDPIC] nptl: Replace sbrk with mmap nptl threads: Fix bug in using a weak variable. Fix htab_delete loop counter isnan: Add isnan weak alias to __isnan mbtowc: Fix non compliant behavior for end of string Fix shm_open posix compliance error code [ARM] rtld: Avoid crash on R_ARM_NONE relocation extra/Configs/Config.in | 2 +- extra/Configs/Config.in.arch | 2 +- include/elf.h | 2 + include/link.h | 6 +- ldso/include/dl-elf.h | 2 +- ldso/include/dl-hash.h | 2 +- ldso/include/dl-string.h | 3 +- ldso/include/dl-syscall.h | 13 +- ldso/include/inline-hashtab.h | 2 +- ldso/ldso/Makefile.in | 2 - ldso/ldso/arm/dl-inlines.h | 1 + ldso/ldso/arm/dl-startup.h | 112 +++++++++++++- ldso/ldso/arm/dl-sysdep.h | 63 ++++++-- ldso/ldso/arm/elfinterp.c | 166 +++++++++++++++++---- ldso/ldso/arm/resolve.S | 23 +++ ldso/ldso/dl-elf.c | 6 +- ldso/ldso/dl-hash.c | 4 +- ldso/ldso/dl-tls.c | 1 + ldso/ldso/fdpic/dl-inlines.h | 2 + ldso/ldso/fdpic/dl-sysdep.h | 2 +- ldso/ldso/ldso.c | 23 ++- libc/misc/elf/dl-iterate-phdr.c | 2 +- libc/misc/internals/__uClibc_main.c | 47 +++++- libc/stdlib/stdlib.c | 6 +- libc/stdlib/system.c | 5 + libc/sysdeps/linux/arm/Makefile.arch | 2 +- libc/sysdeps/linux/arm/bits/elf-fdpic.h | 114 ++++++++++++++ libc/sysdeps/linux/arm/clone.S | 12 ++ libc/sysdeps/linux/arm/crt1.S | 66 ++++++++ libc/sysdeps/linux/arm/crtreloc.c | 144 ++++++++++++++++++ libc/sysdeps/linux/arm/find_exidx.c | 38 +++++ libc/sysdeps/linux/arm/find_got.c | 72 +++++++++ libm/s_isnan.c | 1 + libpthread/nptl/Makefile.in | 1 - libpthread/nptl/allocatestack.c | 6 + libpthread/nptl/linux_fsinfo.h | 5 +- libpthread/nptl/pthread_create.c | 4 + libpthread/nptl/sem_open.c | 8 +- libpthread/nptl/sysdeps/generic/libc-tls.c | 27 +++- libpthread/nptl/sysdeps/unix/sysv/linux/fork.c | 3 + .../sysdeps/unix/sysv/linux/libc_pthread_init.c | 4 + .../nptl/sysdeps/unix/sysv/linux/register-atfork.c | 14 +- .../sysdeps/unix/sysv/linux/unregister-atfork.c | 10 +- librt/shm.c | 3 + 44 files changed, 958 insertions(+), 75 deletions(-) create mode 100644 ldso/ldso/arm/dl-inlines.h create mode 100644 libc/sysdeps/linux/arm/bits/elf-fdpic.h create mode 100644 libc/sysdeps/linux/arm/crtreloc.c create mode 100644 libc/sysdeps/linux/arm/find_got.c -- 2.6.3

5 58

[PATCH] uclibcng-testrunner.sh: create /etc/ethers
by Nikita Sobolev 01 Aug '18

01 Aug '18

/etc/ethers file is required for running tst-ethers and tst-ethers-line tests. If it does not exist, tests fail. Signed-off-by: Nikita Sobolev <Nikita.Sobolev(a)synopsys.com> --- test/uclibcng-testrunner.sh | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/test/uclibcng-testrunner.sh b/test/uclibcng-testrunner.sh index 4fd1661..4a02458 100644 --- a/test/uclibcng-testrunner.sh +++ b/test/uclibcng-testrunner.sh @@ -17,6 +17,9 @@ # of dealing in the work, even if advised of the possibility of such # damage or existence of a defect, except proven that it results out # of said person's immediate fault when using the work as intended. +# +# prepare function is used for creating ethers file. It is +# necessary for running tst-ethers and tst-ethers-line tests. #- # Testsuite runner @@ -25,6 +28,16 @@ die() { exit 1 } + +prepare() { + if [ ! -s /etc/ethers ] + then + echo "00:11:22:33:44:55 teeth" > /etc/ethers + fi +} + +prepare + test -s uclibcng-testrunner.in || die uclibcng-testrunner.in not found nfail=0 -- 2.16.2

1 0

[uclibc-ng-test] Revert "arc: disable deadlocking test"
by Vineet Gupta 21 Jul '18

21 Jul '18

This seems to work fine on ARC ! This reverts commit f0567d67075ff7a27ada8ab5bce750fe8a739763. Signed-off-by: Vineet Gupta <vgupta(a)synopsys.com> --- test/nptl/Makefile.in | 5 ----- 1 file changed, 5 deletions(-) diff --git a/test/nptl/Makefile.in b/test/nptl/Makefile.in index 761f52db8366..35859f944911 100644 --- a/test/nptl/Makefile.in +++ b/test/nptl/Makefile.in @@ -1,11 +1,6 @@ # uClibc-ng NPTL tests # Licensed under the LGPL v2.1, see the file COPYING.LIB in this tarball. -# deadlock -ifeq ($(TARGET_ARCH),arc) -TESTS_DISABLED := tst-cond16 -endif - ifeq ($(TARGET_ARCH),metag) TESTS_DISABLED := tst-cleanup0 tst-cleanupx0 endif -- 2.7.4

2 1

[PATCH 30/32] mbtowc: Fix non compliant behavior for end of string
by Christophe Lyon 04 Jul '18

04 Jul '18

Match glibc behavior. * libc/stdlib/stdlib.c (mbtowc): Fix end of string behavior. Signed-off-by: Mickaël Guêné <mickael.guene(a)st.com> Signed-off-by: Christophe Lyon <christophe.lyon(a)st.com> diff --git a/libc/stdlib/stdlib.c b/libc/stdlib/stdlib.c index 075e6e5..f593663 100644 --- a/libc/stdlib/stdlib.c +++ b/libc/stdlib/stdlib.c @@ -895,9 +895,13 @@ int mbtowc(wchar_t *__restrict pwc, register const char *__restrict s, size_t n) return is_stateful(ENCODING); } - if (*s == '\0') + if (*s == '\0') { /* According to the ISO C 89 standard this is the expected behaviour. */ + /* Standard not very clear here, so do like glibc. */ + if (pwc != NULL) + *pwc = L'\0'; return 0; + } if ((r = mbrtowc(pwc, s, n, &state)) == (size_t) -2) { /* TODO: Should we set an error state? */ -- 2.6.3

1 2

statfs call corrupts memory struct statfs too small
by Geoff Levand 29 Jun '18

29 Jun '18

Hi, I experienced seg faults due to stack corruption when calling statfs() with an automatic struct statfs variable. It seems there is a mismatch between the struct statfs used by libc and that used by the arm64 kernel. The call to statfs() writes 120 bytes, but struct statfs is only 88 bytes. The attached test program shows this. Building for arm64. libuClibc-1.0.30 kernel: Linux-4.17.3 Comparing these files: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/tre… https://cgit.uclibc-ng.org/cgi/cgit/uclibc-ng.git/tree/libc/sysdeps/linux/c… I see for the kernel all fields of struct statfs except f_fsid are the same length at 64 bits for arm64, and that gives 120 bytes. That differs from the libc version which has some 32 bit fields. I filed this bug over at buildroot, but it was suggested to report it here: https://bugs.busybox.net/show_bug.cgi?id=11121 The output of statfs-test program: sizeof statfs: 88 sizeof packed: 138 1: 94 94 2: 19 19 3: 2 2 4: 1 1 5: 0 0 6: 0 0 7: 0 0 8: 0 0 9: 0 0 10: 10 10 11: 0 0 12: 0 0 13: 0 0 14: 0 0 15: 0 0 16: 0 0 17: 72 72 18: a5 a5 19: 7 7 20: 0 0 21: 0 0 22: 0 0 23: 0 0 24: 0 0 25: 83 83 26: 80 80 27: 7 7 28: 0 0 29: 0 0 30: 0 0 31: 0 0 32: 0 0 33: 83 83 34: 80 80 35: 7 7 36: 0 0 37: 0 0 38: 0 0 39: 0 0 40: 0 0 41: 72 72 42: a5 a5 43: 7 7 44: 0 0 45: 0 0 46: 0 0 47: 0 0 48: 0 0 49: d9 d9 50: a0 a0 51: 7 7 52: 0 0 53: 0 0 54: 0 0 55: 0 0 56: 0 0 57: 0 0 58: 0 0 59: 0 0 60: 0 0 61: 0 0 62: 0 0 63: 0 0 64: 0 0 65: ff ff 66: 0 0 67: 0 0 68: 0 0 69: 0 0 70: 0 0 71: 0 0 72: 0 0 73: 0 0 74: 10 10 75: 0 0 76: 0 0 77: 0 0 78: 0 0 79: 0 0 80: 0 0 81: 20 20 82: 0 0 83: 0 0 84: 0 0 85: 0 0 86: 0 0 87: 0 0 88: 0 0 89: 0 0 90: 0 0 91: 0 0 92: 0 0 93: 0 0 94: 0 0 95: 0 0 96: 0 0 97: 0 0 98: 0 0 99: 0 0 100: 0 0 101: 0 0 102: 0 0 103: 0 0 104: 0 0 105: 0 0 106: 0 0 107: 0 0 108: 0 0 109: 0 0 110: 0 0 111: 0 0 112: 0 0 113: 0 0 114: 0 0 115: 0 0 116: 0 0 117: 0 0 118: 0 0 119: 0 0 120: 0 0 121: ff cc 122: ff cc 123: ff cc 124: ff cc 125: ff cc 126: ff cc 127: ff cc 128: ff cc 129: ff cc 130: ff cc 131: ff cc 132: ff cc 133: ff cc 134: ff cc 135: ff cc 136: ff cc 137: ff cc 138: ff cc -Geoff

2 2

[PATCH v2] ldso/arc: fix LD_DEBUG segv when printing R_ARC_NONE
by Vineet Gupta 27 Jun '18

27 Jun '18

Signed-off-by: Vineet Gupta <vgupta(a)synopsys.com> --- v2: Also remove a bogus trailing l in prints --- ldso/ldso/arc/elfinterp.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/ldso/ldso/arc/elfinterp.c b/ldso/ldso/arc/elfinterp.c index 5a02bc2348da..a30c19bcb832 100644 --- a/ldso/ldso/arc/elfinterp.c +++ b/ldso/ldso/arc/elfinterp.c @@ -64,7 +64,7 @@ _dl_linux_resolver(struct elf_resolve *tpnt, unsigned int plt_pc) if (_dl_debug_bindings) { _dl_dprintf(_dl_debug_file, "\nresolve function: %s", symname); if (_dl_debug_detail) - _dl_dprintf(_dl_debug_file, "\n\tpatched %x ==> %pc @ %pl\n", + _dl_dprintf(_dl_debug_file, "\n\tpatched %x ==> %pc @ %p\n", *got_addr, new_addr, got_addr); } @@ -178,7 +178,7 @@ _dl_do_reloc(struct elf_resolve *tpnt, struct r_scope_elem *scope, log_entry: #if defined __SUPPORT_LD_DEBUG__ - if (_dl_debug_detail) + if (_dl_debug_detail && (reloc_type != R_ARC_NONE)) _dl_dprintf(_dl_debug_file,"\tpatched: %x ==> %x @ %x", old_val, *reloc_addr, reloc_addr); #endif @@ -214,7 +214,7 @@ _dl_do_lazy_reloc(struct elf_resolve *tpnt, struct r_scope_elem *scope, } #if defined __SUPPORT_LD_DEBUG__ - if (_dl_debug_reloc && _dl_debug_detail) + if (_dl_debug_reloc && _dl_debug_detail && (reloc_type != R_ARC_NONE)) _dl_dprintf(_dl_debug_file, "\tpatched: %x ==> %x @ %x\n", old_val, *reloc_addr, reloc_addr); #endif -- 2.7.4

1 0

[PATCH] ldso/arc: fix LD_DEBUG segv when printing R_ARC_NONE
by Vineet Gupta 27 Jun '18

27 Jun '18

Signed-off-by: Vineet Gupta <vgupta(a)synopsys.com> --- ldso/ldso/arc/elfinterp.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/ldso/ldso/arc/elfinterp.c b/ldso/ldso/arc/elfinterp.c index 5a02bc2348da..17d588899dfc 100644 --- a/ldso/ldso/arc/elfinterp.c +++ b/ldso/ldso/arc/elfinterp.c @@ -178,7 +178,7 @@ _dl_do_reloc(struct elf_resolve *tpnt, struct r_scope_elem *scope, log_entry: #if defined __SUPPORT_LD_DEBUG__ - if (_dl_debug_detail) + if (_dl_debug_detail && (reloc_type != R_ARC_NONE)) _dl_dprintf(_dl_debug_file,"\tpatched: %x ==> %x @ %x", old_val, *reloc_addr, reloc_addr); #endif @@ -214,7 +214,7 @@ _dl_do_lazy_reloc(struct elf_resolve *tpnt, struct r_scope_elem *scope, } #if defined __SUPPORT_LD_DEBUG__ - if (_dl_debug_reloc && _dl_debug_detail) + if (_dl_debug_reloc && _dl_debug_detail && (reloc_type != R_ARC_NONE)) _dl_dprintf(_dl_debug_file, "\tpatched: %x ==> %x @ %x\n", old_val, *reloc_addr, reloc_addr); #endif -- 2.7.4

1 0

write is not weak?
by Christophe Lyon 25 Jun '18

25 Jun '18

Hi, While working on FDPIC support for ARM in uclibc-ng, I've noticed that the write() function if defined as a strong global symbol, while the __GI_write alias is weak. 00000034 w F .text 0000005c .hidden __GI_write 00000034 g F .text 0000005c write my pre-processed write.i contains: ssize_t __attribute__ ((weak)) write (int fd, const void *buf, size_t count) { int oldtype; ssize_t result; if (__builtin_expect (__libc_multiple_threads == 0, 1)) return __write_nocancel (fd, buf, count); oldtype = __libc_enable_asynccancel (); result = __write_nocancel (fd, buf, count); __libc_disable_asynccancel (oldtype); return result; } extern __typeof (write) __EI_write __asm__("" "write"); extern __typeof (write) __EI_write __attribute__((alias ("" "__GI_write"))); and write.s has: .weak __GI_write .hidden __GI_write .type __GI_write, %function __GI_write: .fnstart [...] .size __GI_write, .-__GI_write .global write .set write,__GI_write >From the .i file, I'd expect the weak attribute to be set on write. FWIW, I'd like write to be weak, because currently it prevents re-defining it user code and linking statically with libc.a I've checked that glibc and newlib have weak write. Since this seems to depend on the threads model, my .config has: # HAS_NO_THREADS is not set UCLIBC_HAS_LINUXTHREADS=y UCLIBC_HAS_THREADS_NATIVE=y UCLIBC_HAS_THREADS=y PTHREADS_DEBUG_SUPPORT=y and having # UCLIBC_HAS_LINUXTHREADS is not set has no effect on write's weakness. What am I missing? Thanks, Christophe

2 3

__WIFSTOPPED macro for MIPS
by Sergey Korolev 14 Jun '18

14 Jun '18

Does anyone know why __WIFSTOPPED macro redefined for MIPS architecture as #if !defined(__mips__) #define __WIFSTOPPED(status) (((status) & 0xff) == 0x7f) #else #define __WIFSTOPPED(status) (((status) & 0xff) == 0x7f && ((status) & 0xff00)) #endif ? The problem with this definition is in that newer gdb versions (>= 7.0) expect that WIFSTOPPED(W_STOPCODE(0)) is true (see [1]), otherwise a debugger erroneously blocks a child process in a trace status until gdb exit. In the glibc [2] this macro expansion is the same for all architectures. Is this a gdb or uClibc-ng bug? 1. https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git; a=blob;f=gdb/linux-nat.c;h=445b59fa4adadbb2890a9e2debffb5 330f1e09e4;hb=HEAD#l497 2. https://sourceware.org/git/gitweb.cgi?p=glibc.git;a=blob; f=bits/waitstatus.h;h=a1c232612cd13e758f222ade6b27bc2c85ab8f8f;hb=HEAD#l44

3 2