pool/glibc
SHA256
2
0
Fork 1
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Accepting request 536907 from Base:System

Browse Source 
- spawni-assert.patch: Fix improper assert in Linux posix_spawn (BZ
  #22273) (forwarded request 536906 from Andreas_Schwab)

OBS-URL: https://build.opensuse.org/request/show/536907
OBS-URL: https://build.opensuse.org/package/show/openSUSE:Factory/glibc?expand=0&rev=216
This commit is contained in:
Dominique Leuenberger 2017-11-15 15:48:09 +00:00 committed by Git OBS Bridge
commit 5017ae7da4
16 changed files with 5158 additions and 32 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
aarch64-cpu-features.patch Normal file
View File

@ -0,0 +1,18 @@
2017-10-10 Steve Ellcey <sellcey@cavium.com>
* sysdeps/unix/sysv/linux/aarch64/cpu-features.c (get_midr_from_mcpu):
Use strcmp instead of tunable_is_name.
Index: glibc-2.26/sysdeps/unix/sysv/linux/aarch64/cpu-features.c
===================================================================
--- glibc-2.26.orig/sysdeps/unix/sysv/linux/aarch64/cpu-features.c
+++ glibc-2.26/sysdeps/unix/sysv/linux/aarch64/cpu-features.c
@@ -37,7 +37,7 @@ static uint64_t
get_midr_from_mcpu (const char *mcpu)
{
for (int i = 0; i < sizeof (cpu_list) / sizeof (struct cpu_list); i++)
- if (tunable_is_name (mcpu, cpu_list[i].name) == 0)
+ if (strcmp (mcpu, cpu_list[i].name) == 0)
return cpu_list[i].midr;
return UINT64_MAX;

851
dl-runtime-resolve-xsave.patch Normal file
View File

@ -0,0 +1,851 @@
2017-10-22 H.J. Lu <hongjiu.lu@intel.com>
[BZ #21265]
* sysdeps/x86/cpu-features-offsets.sym (XSAVE_STATE_SIZE_OFFSET):
New.
* sysdeps/x86/cpu-features.c: Include <libc-pointer-arith.h>.
(get_common_indeces): Set xsave_state_size, xsave_state_full_size
and bit_arch_XSAVEC_Usable if needed.
(init_cpu_features): Remove bit_arch_Use_dl_runtime_resolve_slow
and bit_arch_Use_dl_runtime_resolve_opt.
* sysdeps/x86/cpu-features.h (bit_arch_Use_dl_runtime_resolve_opt):
Removed.
(bit_arch_Use_dl_runtime_resolve_slow): Likewise.
(bit_arch_Prefer_No_AVX512): Updated.
(bit_arch_MathVec_Prefer_No_AVX512): Likewise.
(bit_arch_XSAVEC_Usable): New.
(STATE_SAVE_OFFSET): Likewise.
(STATE_SAVE_MASK): Likewise.
[__ASSEMBLER__]: Include <cpu-features-offsets.h>.
(cpu_features): Add xsave_state_size and xsave_state_full_size.
(index_arch_Use_dl_runtime_resolve_opt): Removed.
(index_arch_Use_dl_runtime_resolve_slow): Likewise.
(index_arch_XSAVEC_Usable): New.
* sysdeps/x86/cpu-tunables.c (TUNABLE_CALLBACK (set_hwcaps)):
Support XSAVEC_Usable. Remove Use_dl_runtime_resolve_slow.
* sysdeps/x86_64/Makefile (tst-x86_64-1-ENV): New if tunables
is enabled.
* sysdeps/x86_64/dl-machine.h (elf_machine_runtime_setup):
Replace _dl_runtime_resolve_sse, _dl_runtime_resolve_avx,
_dl_runtime_resolve_avx_slow, _dl_runtime_resolve_avx_opt,
_dl_runtime_resolve_avx512 and _dl_runtime_resolve_avx512_opt
with _dl_runtime_resolve_fxsave, _dl_runtime_resolve_xsave and
_dl_runtime_resolve_xsavec.
* sysdeps/x86_64/dl-trampoline.S (DL_RUNTIME_UNALIGNED_VEC_SIZE):
Removed.
(DL_RUNTIME_RESOLVE_REALIGN_STACK): Check STATE_SAVE_ALIGNMENT
instead of VEC_SIZE.
(REGISTER_SAVE_BND0): Removed.
(REGISTER_SAVE_BND1): Likewise.
(REGISTER_SAVE_BND3): Likewise.
(REGISTER_SAVE_RAX): Always defined to 0.
(VMOV): Removed.
(_dl_runtime_resolve_avx): Likewise.
(_dl_runtime_resolve_avx_slow): Likewise.
(_dl_runtime_resolve_avx_opt): Likewise.
(_dl_runtime_resolve_avx512): Likewise.
(_dl_runtime_resolve_avx512_opt): Likewise.
(_dl_runtime_resolve_sse): Likewise.
(_dl_runtime_resolve_sse_vex): Likewise.
(USE_FXSAVE): New.
(_dl_runtime_resolve_fxsave): Likewise.
(USE_XSAVE): Likewise.
(_dl_runtime_resolve_xsave): Likewise.
(USE_XSAVEC): Likewise.
(_dl_runtime_resolve_xsavec): Likewise.
* sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve_avx512):
Removed.
(_dl_runtime_resolve_avx512_opt): Likewise.
(_dl_runtime_resolve_avx): Likewise.
(_dl_runtime_resolve_avx_opt): Likewise.
(_dl_runtime_resolve_sse): Likewise.
(_dl_runtime_resolve_sse_vex): Likewise.
(_dl_runtime_resolve_fxsave): New.
(_dl_runtime_resolve_xsave): Likewise.
(_dl_runtime_resolve_xsavec): Likewise.
Index: glibc-2.26/sysdeps/x86/cpu-features-offsets.sym
===================================================================
--- glibc-2.26.orig/sysdeps/x86/cpu-features-offsets.sym
+++ glibc-2.26/sysdeps/x86/cpu-features-offsets.sym
@@ -15,6 +15,7 @@ CPUID_ECX_OFFSET offsetof (struct cpuid_
CPUID_EDX_OFFSET offsetof (struct cpuid_registers, edx)
FAMILY_OFFSET offsetof (struct cpu_features, family)
MODEL_OFFSET offsetof (struct cpu_features, model)
+XSAVE_STATE_SIZE_OFFSET offsetof (struct cpu_features, xsave_state_size)
FEATURE_OFFSET offsetof (struct cpu_features, feature)
FEATURE_SIZE sizeof (unsigned int)
Index: glibc-2.26/sysdeps/x86/cpu-features.c
===================================================================
--- glibc-2.26.orig/sysdeps/x86/cpu-features.c
+++ glibc-2.26/sysdeps/x86/cpu-features.c
@@ -19,6 +19,7 @@
#include <cpuid.h>
#include <cpu-features.h>
#include <dl-hwcap.h>
+#include <libc-pointer-arith.h>
#if HAVE_TUNABLES
# define TUNABLE_NAMESPACE tune
@@ -103,6 +104,76 @@ get_common_indeces (struct cpu_features
}
}
}
+
+ /* For _dl_runtime_resolve, set xsave_state_size to xsave area
+ size + integer register save size and align it to 64 bytes. */
+ if (cpu_features->max_cpuid >= 0xd)
+ {
+ unsigned int eax, ebx, ecx, edx;
+
+ __cpuid_count (0xd, 0, eax, ebx, ecx, edx);
+ if (ebx != 0)
+ {
+ unsigned int xsave_state_full_size
+ = ALIGN_UP (ebx + STATE_SAVE_OFFSET, 64);
+
+ cpu_features->xsave_state_size
+ = xsave_state_full_size;
+ cpu_features->xsave_state_full_size
+ = xsave_state_full_size;
+
+ __cpuid_count (0xd, 1, eax, ebx, ecx, edx);
+
+ /* Check if XSAVEC is available. */
+ if ((eax & (1 << 1)) != 0)
+ {
+ unsigned int xstate_comp_offsets[32];
+ unsigned int xstate_comp_sizes[32];
+ unsigned int i;
+
+ xstate_comp_offsets[0] = 0;
+ xstate_comp_offsets[1] = 160;
+ xstate_comp_offsets[2] = 576;
+ xstate_comp_sizes[0] = 160;
+ xstate_comp_sizes[1] = 256;
+
+ for (i = 2; i < 32; i++)
+ {
+ if ((STATE_SAVE_MASK & (1 << i)) != 0)
+ {
+ __cpuid_count (0xd, i, eax, ebx, ecx, edx);
+ xstate_comp_sizes[i] = eax;
+ }
+ else
+ {
+ ecx = 0;
+ xstate_comp_sizes[i] = 0;
+ }
+
+ if (i > 2)
+ {
+ xstate_comp_offsets[i]
+ = (xstate_comp_offsets[i - 1]
+ + xstate_comp_sizes[i -1]);
+ if ((ecx & (1 << 1)) != 0)
+ xstate_comp_offsets[i]
+ = ALIGN_UP (xstate_comp_offsets[i], 64);
+ }
+ }
+
+ /* Use XSAVEC. */
+ unsigned int size
+ = xstate_comp_offsets[31] + xstate_comp_sizes[31];
+ if (size)
+ {
+ cpu_features->xsave_state_size
+ = ALIGN_UP (size + STATE_SAVE_OFFSET, 64);
+ cpu_features->feature[index_arch_XSAVEC_Usable]
+ |= bit_arch_XSAVEC_Usable;
+ }
+ }
+ }
+ }
}
}
@@ -242,23 +313,6 @@ init_cpu_features (struct cpu_features *
else
cpu_features->feature[index_arch_Prefer_No_AVX512]
|= bit_arch_Prefer_No_AVX512;
-
- /* To avoid SSE transition penalty, use _dl_runtime_resolve_slow.
- If XGETBV suports ECX == 1, use _dl_runtime_resolve_opt.
- Use _dl_runtime_resolve_opt only with AVX512F since it is
- slower than _dl_runtime_resolve_slow with AVX. */
- cpu_features->feature[index_arch_Use_dl_runtime_resolve_slow]
- |= bit_arch_Use_dl_runtime_resolve_slow;
- if (CPU_FEATURES_ARCH_P (cpu_features, AVX512F_Usable)
- && cpu_features->max_cpuid >= 0xd)
- {
- unsigned int eax;
-
- __cpuid_count (0xd, 1, eax, ebx, ecx, edx);
- if ((eax & (1 << 2)) != 0)
- cpu_features->feature[index_arch_Use_dl_runtime_resolve_opt]
- |= bit_arch_Use_dl_runtime_resolve_opt;
- }
}
/* This spells out "AuthenticAMD". */
else if (ebx == 0x68747541 && ecx == 0x444d4163 && edx == 0x69746e65)
Index: glibc-2.26/sysdeps/x86/cpu-features.h
===================================================================
--- glibc-2.26.orig/sysdeps/x86/cpu-features.h
+++ glibc-2.26/sysdeps/x86/cpu-features.h
@@ -37,9 +37,8 @@
#define bit_arch_Prefer_No_VZEROUPPER (1 << 17)
#define bit_arch_Fast_Unaligned_Copy (1 << 18)
#define bit_arch_Prefer_ERMS (1 << 19)
-#define bit_arch_Use_dl_runtime_resolve_opt (1 << 20)
-#define bit_arch_Use_dl_runtime_resolve_slow (1 << 21)
-#define bit_arch_Prefer_No_AVX512 (1 << 22)
+#define bit_arch_Prefer_No_AVX512 (1 << 20)
+#define bit_arch_XSAVEC_Usable (1 << 21)
/* CPUID Feature flags. */
@@ -88,6 +87,15 @@
/* The current maximum size of the feature integer bit array. */
#define FEATURE_INDEX_MAX 1
+/* Offset for fxsave/xsave area used by _dl_runtime_resolve. Also need
+ space to preserve RCX, RDX, RSI, RDI, R8, R9 and RAX. It must be
+ aligned to 16 bytes for fxsave and 64 bytes for xsave. */
+#define STATE_SAVE_OFFSET (8 * 7 + 8)
+
+/* Save SSE, AVX, AVX512, mask and bound registers. */
+#define STATE_SAVE_MASK \
+ ((1 << 1) | (1 << 2) | (1 << 3) | (1 << 5) | (1 << 6) | (1 << 7))
+
#ifdef __ASSEMBLER__
# include <cpu-features-offsets.h>
@@ -123,8 +131,6 @@
# define index_arch_Prefer_No_VZEROUPPER FEATURE_INDEX_1*FEATURE_SIZE
# define index_arch_Fast_Unaligned_Copy FEATURE_INDEX_1*FEATURE_SIZE
# define index_arch_Prefer_ERMS FEATURE_INDEX_1*FEATURE_SIZE
-# define index_arch_Use_dl_runtime_resolve_opt FEATURE_INDEX_1*FEATURE_SIZE
-# define index_arch_Use_dl_runtime_resolve_slow FEATURE_INDEX_1*FEATURE_SIZE
# define index_arch_Prefer_No_AVX512 FEATURE_INDEX_1*FEATURE_SIZE
@@ -214,6 +220,18 @@ struct cpu_features
} cpuid[COMMON_CPUID_INDEX_MAX];
unsigned int family;
unsigned int model;
+ /* The state size for XSAVEC or XSAVE. The type must be unsigned long
+ int so that we use
+
+ sub xsave_state_size_offset(%rip) %RSP_LP
+
+ in _dl_runtime_resolve. */
+ unsigned long int xsave_state_size;
+ /* The full state size for XSAVE when XSAVEC is disabled by
+
+ GLIBC_TUNABLES=glibc.tune.hwcaps=-XSAVEC_Usable
+ */
+ unsigned int xsave_state_full_size;
unsigned int feature[FEATURE_INDEX_MAX];
/* Data cache size for use in memory and string routines, typically
L1 size. */
@@ -326,9 +344,8 @@ extern const struct cpu_features *__get_
# define index_arch_Prefer_No_VZEROUPPER FEATURE_INDEX_1
# define index_arch_Fast_Unaligned_Copy FEATURE_INDEX_1
# define index_arch_Prefer_ERMS FEATURE_INDEX_1
-# define index_arch_Use_dl_runtime_resolve_opt FEATURE_INDEX_1
-# define index_arch_Use_dl_runtime_resolve_slow FEATURE_INDEX_1
# define index_arch_Prefer_No_AVX512 FEATURE_INDEX_1
+# define index_arch_XSAVEC_Usable FEATURE_INDEX_1
#endif /* !__ASSEMBLER__ */
Index: glibc-2.26/sysdeps/x86/cpu-tunables.c
===================================================================
--- glibc-2.26.orig/sysdeps/x86/cpu-tunables.c
+++ glibc-2.26/sysdeps/x86/cpu-tunables.c
@@ -240,6 +240,16 @@ TUNABLE_CALLBACK (set_hwcaps) (tunable_v
Slow_SSE4_2, SSE4_2,
disable, 11);
break;
+ case 13:
+ if (disable)
+ {
+ /* Update xsave_state_size to XSAVE state size. */
+ cpu_features->xsave_state_size
+ = cpu_features->xsave_state_full_size;
+ CHECK_GLIBC_IFUNC_ARCH_OFF (n, cpu_features,
+ XSAVEC_Usable, 13);
+ }
+ break;
case 14:
if (disable)
{
@@ -308,13 +318,6 @@ TUNABLE_CALLBACK (set_hwcaps) (tunable_v
disable, 26);
}
break;
- case 27:
- {
- CHECK_GLIBC_IFUNC_ARCH_BOTH (n, cpu_features,
- Use_dl_runtime_resolve_slow,
- disable, 27);
- }
- break;
}
p += len + 1;
}
Index: glibc-2.26/sysdeps/x86_64/Makefile
===================================================================
--- glibc-2.26.orig/sysdeps/x86_64/Makefile
+++ glibc-2.26/sysdeps/x86_64/Makefile
@@ -55,6 +55,10 @@ CFLAGS-tst-quad2pie.c = $(PIE-ccflag)
tests += tst-x86_64-1
modules-names += x86_64/tst-x86_64mod-1
LDFLAGS-tst-x86_64mod-1.so = -Wl,-soname,tst-x86_64mod-1.so
+ifneq (no,$(have-tunables))
+# Test the state size for XSAVE when XSAVEC is disabled.
+tst-x86_64-1-ENV = GLIBC_TUNABLES=glibc.tune.hwcaps=-XSAVEC_Usable
+endif
$(objpfx)tst-x86_64-1: $(objpfx)x86_64/tst-x86_64mod-1.so
Index: glibc-2.26/sysdeps/x86_64/dl-machine.h
===================================================================
--- glibc-2.26.orig/sysdeps/x86_64/dl-machine.h
+++ glibc-2.26/sysdeps/x86_64/dl-machine.h
@@ -66,12 +66,9 @@ static inline int __attribute__ ((unused
elf_machine_runtime_setup (struct link_map *l, int lazy, int profile)
{
Elf64_Addr *got;
- extern void _dl_runtime_resolve_sse (ElfW(Word)) attribute_hidden;
- extern void _dl_runtime_resolve_avx (ElfW(Word)) attribute_hidden;
- extern void _dl_runtime_resolve_avx_slow (ElfW(Word)) attribute_hidden;
- extern void _dl_runtime_resolve_avx_opt (ElfW(Word)) attribute_hidden;
- extern void _dl_runtime_resolve_avx512 (ElfW(Word)) attribute_hidden;
- extern void _dl_runtime_resolve_avx512_opt (ElfW(Word)) attribute_hidden;
+ extern void _dl_runtime_resolve_fxsave (ElfW(Word)) attribute_hidden;
+ extern void _dl_runtime_resolve_xsave (ElfW(Word)) attribute_hidden;
+ extern void _dl_runtime_resolve_xsavec (ElfW(Word)) attribute_hidden;
extern void _dl_runtime_profile_sse (ElfW(Word)) attribute_hidden;
extern void _dl_runtime_profile_avx (ElfW(Word)) attribute_hidden;
extern void _dl_runtime_profile_avx512 (ElfW(Word)) attribute_hidden;
@@ -120,29 +117,14 @@ elf_machine_runtime_setup (struct link_m
/* This function will get called to fix up the GOT entry
indicated by the offset on the stack, and then jump to
the resolved address. */
- if (HAS_ARCH_FEATURE (AVX512F_Usable))
- {
- if (HAS_ARCH_FEATURE (Use_dl_runtime_resolve_opt))
- *(ElfW(Addr) *) (got + 2)
- = (ElfW(Addr)) &_dl_runtime_resolve_avx512_opt;
- else
- *(ElfW(Addr) *) (got + 2)
- = (ElfW(Addr)) &_dl_runtime_resolve_avx512;
- }
- else if (HAS_ARCH_FEATURE (AVX_Usable))
- {
- if (HAS_ARCH_FEATURE (Use_dl_runtime_resolve_opt))
- *(ElfW(Addr) *) (got + 2)
- = (ElfW(Addr)) &_dl_runtime_resolve_avx_opt;
- else if (HAS_ARCH_FEATURE (Use_dl_runtime_resolve_slow))
- *(ElfW(Addr) *) (got + 2)
- = (ElfW(Addr)) &_dl_runtime_resolve_avx_slow;
- else
- *(ElfW(Addr) *) (got + 2)
- = (ElfW(Addr)) &_dl_runtime_resolve_avx;
- }
+ if (GLRO(dl_x86_cpu_features).xsave_state_size != 0)
+ *(ElfW(Addr) *) (got + 2)
+ = (HAS_ARCH_FEATURE (XSAVEC_Usable)
+ ? (ElfW(Addr)) &_dl_runtime_resolve_xsavec
+ : (ElfW(Addr)) &_dl_runtime_resolve_xsave);
else
- *(ElfW(Addr) *) (got + 2) = (ElfW(Addr)) &_dl_runtime_resolve_sse;
+ *(ElfW(Addr) *) (got + 2)
+ = (ElfW(Addr)) &_dl_runtime_resolve_fxsave;
}
}
Index: glibc-2.26/sysdeps/x86_64/dl-trampoline.S
===================================================================
--- glibc-2.26.orig/sysdeps/x86_64/dl-trampoline.S
+++ glibc-2.26/sysdeps/x86_64/dl-trampoline.S
@@ -34,41 +34,24 @@
# define DL_STACK_ALIGNMENT 8
#endif
-#ifndef DL_RUNTIME_UNALIGNED_VEC_SIZE
-/* The maximum size in bytes of unaligned vector load and store in the
- dynamic linker. Since SSE optimized memory/string functions with
- aligned SSE register load and store are used in the dynamic linker,
- we must set this to 8 so that _dl_runtime_resolve_sse will align the
- stack before calling _dl_fixup. */
-# define DL_RUNTIME_UNALIGNED_VEC_SIZE 8
-#endif
-
-/* True if _dl_runtime_resolve should align stack to VEC_SIZE bytes. */
+/* True if _dl_runtime_resolve should align stack for STATE_SAVE or align
+ stack to 16 bytes before calling _dl_fixup. */
#define DL_RUNTIME_RESOLVE_REALIGN_STACK \
- (VEC_SIZE > DL_STACK_ALIGNMENT \
- && VEC_SIZE > DL_RUNTIME_UNALIGNED_VEC_SIZE)
-
-/* Align vector register save area to 16 bytes. */
-#define REGISTER_SAVE_VEC_OFF 0
+ (STATE_SAVE_ALIGNMENT > DL_STACK_ALIGNMENT \
+ || 16 > DL_STACK_ALIGNMENT)
/* Area on stack to save and restore registers used for parameter
passing when calling _dl_fixup. */
#ifdef __ILP32__
-# define REGISTER_SAVE_RAX (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 8)
# define PRESERVE_BND_REGS_PREFIX
#else
-/* Align bound register save area to 16 bytes. */
-# define REGISTER_SAVE_BND0 (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 8)
-# define REGISTER_SAVE_BND1 (REGISTER_SAVE_BND0 + 16)
-# define REGISTER_SAVE_BND2 (REGISTER_SAVE_BND1 + 16)
-# define REGISTER_SAVE_BND3 (REGISTER_SAVE_BND2 + 16)
-# define REGISTER_SAVE_RAX (REGISTER_SAVE_BND3 + 16)
# ifdef HAVE_MPX_SUPPORT
# define PRESERVE_BND_REGS_PREFIX bnd
# else
# define PRESERVE_BND_REGS_PREFIX .byte 0xf2
# endif
#endif
+#define REGISTER_SAVE_RAX 0
#define REGISTER_SAVE_RCX (REGISTER_SAVE_RAX + 8)
#define REGISTER_SAVE_RDX (REGISTER_SAVE_RCX + 8)
#define REGISTER_SAVE_RSI (REGISTER_SAVE_RDX + 8)
@@ -80,68 +63,56 @@
#define VEC_SIZE 64
#define VMOVA vmovdqa64
-#if DL_RUNTIME_RESOLVE_REALIGN_STACK || VEC_SIZE <= DL_STACK_ALIGNMENT
-# define VMOV vmovdqa64
-#else
-# define VMOV vmovdqu64
-#endif
#define VEC(i) zmm##i
-#define _dl_runtime_resolve _dl_runtime_resolve_avx512
#define _dl_runtime_profile _dl_runtime_profile_avx512
#include "dl-trampoline.h"
-#undef _dl_runtime_resolve
#undef _dl_runtime_profile
#undef VEC
-#undef VMOV
#undef VMOVA
#undef VEC_SIZE
#define VEC_SIZE 32
#define VMOVA vmovdqa
-#if DL_RUNTIME_RESOLVE_REALIGN_STACK || VEC_SIZE <= DL_STACK_ALIGNMENT
-# define VMOV vmovdqa
-#else
-# define VMOV vmovdqu
-#endif
#define VEC(i) ymm##i
-#define _dl_runtime_resolve _dl_runtime_resolve_avx
-#define _dl_runtime_resolve_opt _dl_runtime_resolve_avx_opt
#define _dl_runtime_profile _dl_runtime_profile_avx
#include "dl-trampoline.h"
-#undef _dl_runtime_resolve
-#undef _dl_runtime_resolve_opt
#undef _dl_runtime_profile
#undef VEC
-#undef VMOV
#undef VMOVA
#undef VEC_SIZE
/* movaps/movups is 1-byte shorter. */
#define VEC_SIZE 16
#define VMOVA movaps
-#if DL_RUNTIME_RESOLVE_REALIGN_STACK || VEC_SIZE <= DL_STACK_ALIGNMENT
-# define VMOV movaps
-#else
-# define VMOV movups
-#endif
#define VEC(i) xmm##i
-#define _dl_runtime_resolve _dl_runtime_resolve_sse
#define _dl_runtime_profile _dl_runtime_profile_sse
#undef RESTORE_AVX
#include "dl-trampoline.h"
-#undef _dl_runtime_resolve
#undef _dl_runtime_profile
-#undef VMOV
+#undef VEC
#undef VMOVA
+#undef VEC_SIZE
-/* Used by _dl_runtime_resolve_avx_opt/_dl_runtime_resolve_avx512_opt
- to preserve the full vector registers with zero upper bits. */
-#define VMOVA vmovdqa
-#if DL_RUNTIME_RESOLVE_REALIGN_STACK || VEC_SIZE <= DL_STACK_ALIGNMENT
-# define VMOV vmovdqa
-#else
-# define VMOV vmovdqu
-#endif
-#define _dl_runtime_resolve _dl_runtime_resolve_sse_vex
-#define _dl_runtime_resolve_opt _dl_runtime_resolve_avx512_opt
+#define USE_FXSAVE
+#define STATE_SAVE_ALIGNMENT 16
+#define _dl_runtime_resolve _dl_runtime_resolve_fxsave
+#include "dl-trampoline.h"
+#undef _dl_runtime_resolve
+#undef USE_FXSAVE
+#undef STATE_SAVE_ALIGNMENT
+
+#define USE_XSAVE
+#define STATE_SAVE_ALIGNMENT 64
+#define _dl_runtime_resolve _dl_runtime_resolve_xsave
+#include "dl-trampoline.h"
+#undef _dl_runtime_resolve
+#undef USE_XSAVE
+#undef STATE_SAVE_ALIGNMENT
+
+#define USE_XSAVEC
+#define STATE_SAVE_ALIGNMENT 64
+#define _dl_runtime_resolve _dl_runtime_resolve_xsavec
#include "dl-trampoline.h"
+#undef _dl_runtime_resolve
+#undef USE_XSAVEC
+#undef STATE_SAVE_ALIGNMENT
Index: glibc-2.26/sysdeps/x86_64/dl-trampoline.h
===================================================================
--- glibc-2.26.orig/sysdeps/x86_64/dl-trampoline.h
+++ glibc-2.26/sysdeps/x86_64/dl-trampoline.h
@@ -16,140 +16,47 @@
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
-#undef REGISTER_SAVE_AREA_RAW
-#ifdef __ILP32__
-/* X32 saves RCX, RDX, RSI, RDI, R8 and R9 plus RAX as well as VEC0 to
- VEC7. */
-# define REGISTER_SAVE_AREA_RAW (8 * 7 + VEC_SIZE * 8)
-#else
-/* X86-64 saves RCX, RDX, RSI, RDI, R8 and R9 plus RAX as well as
- BND0, BND1, BND2, BND3 and VEC0 to VEC7. */
-# define REGISTER_SAVE_AREA_RAW (8 * 7 + 16 * 4 + VEC_SIZE * 8)
-#endif
+ .text
+#ifdef _dl_runtime_resolve
-#undef REGISTER_SAVE_AREA
-#undef LOCAL_STORAGE_AREA
-#undef BASE
-#if DL_RUNTIME_RESOLVE_REALIGN_STACK
-# define REGISTER_SAVE_AREA (REGISTER_SAVE_AREA_RAW + 8)
-/* Local stack area before jumping to function address: RBX. */
-# define LOCAL_STORAGE_AREA 8
-# define BASE rbx
-# if (REGISTER_SAVE_AREA % VEC_SIZE) != 0
-# error REGISTER_SAVE_AREA must be multples of VEC_SIZE
-# endif
-#else
-# define REGISTER_SAVE_AREA REGISTER_SAVE_AREA_RAW
-/* Local stack area before jumping to function address: All saved
- registers. */
-# define LOCAL_STORAGE_AREA REGISTER_SAVE_AREA
-# define BASE rsp
-# if (REGISTER_SAVE_AREA % 16) != 8
-# error REGISTER_SAVE_AREA must be odd multples of 8
-# endif
-#endif
+# undef REGISTER_SAVE_AREA
+# undef LOCAL_STORAGE_AREA
+# undef BASE
- .text
-#ifdef _dl_runtime_resolve_opt
-/* Use the smallest vector registers to preserve the full YMM/ZMM
- registers to avoid SSE transition penalty. */
+# if (STATE_SAVE_ALIGNMENT % 16) != 0
+# error STATE_SAVE_ALIGNMENT must be multples of 16
+# endif
-# if VEC_SIZE == 32
-/* Check if the upper 128 bits in %ymm0 - %ymm7 registers are non-zero
- and preserve %xmm0 - %xmm7 registers with the zero upper bits. Since
- there is no SSE transition penalty on AVX512 processors which don't
- support XGETBV with ECX == 1, _dl_runtime_resolve_avx512_slow isn't
- provided. */
- .globl _dl_runtime_resolve_avx_slow
- .hidden _dl_runtime_resolve_avx_slow
- .type _dl_runtime_resolve_avx_slow, @function
- .align 16
-_dl_runtime_resolve_avx_slow:
- cfi_startproc
- cfi_adjust_cfa_offset(16) # Incorporate PLT
- vorpd %ymm0, %ymm1, %ymm8
- vorpd %ymm2, %ymm3, %ymm9
- vorpd %ymm4, %ymm5, %ymm10
- vorpd %ymm6, %ymm7, %ymm11
- vorpd %ymm8, %ymm9, %ymm9
- vorpd %ymm10, %ymm11, %ymm10
- vpcmpeqd %xmm8, %xmm8, %xmm8
- vorpd %ymm9, %ymm10, %ymm10
- vptest %ymm10, %ymm8
- # Preserve %ymm0 - %ymm7 registers if the upper 128 bits of any
- # %ymm0 - %ymm7 registers aren't zero.
- PRESERVE_BND_REGS_PREFIX
- jnc _dl_runtime_resolve_avx
- # Use vzeroupper to avoid SSE transition penalty.
- vzeroupper
- # Preserve %xmm0 - %xmm7 registers with the zero upper 128 bits
- # when the upper 128 bits of %ymm0 - %ymm7 registers are zero.
- PRESERVE_BND_REGS_PREFIX
- jmp _dl_runtime_resolve_sse_vex
- cfi_adjust_cfa_offset(-16) # Restore PLT adjustment
- cfi_endproc
- .size _dl_runtime_resolve_avx_slow, .-_dl_runtime_resolve_avx_slow
+# if (STATE_SAVE_OFFSET % STATE_SAVE_ALIGNMENT) != 0
+# error STATE_SAVE_OFFSET must be multples of STATE_SAVE_ALIGNMENT
# endif
-/* Use XGETBV with ECX == 1 to check which bits in vector registers are
- non-zero and only preserve the non-zero lower bits with zero upper
- bits. */
- .globl _dl_runtime_resolve_opt
- .hidden _dl_runtime_resolve_opt
- .type _dl_runtime_resolve_opt, @function
- .align 16
-_dl_runtime_resolve_opt:
- cfi_startproc
- cfi_adjust_cfa_offset(16) # Incorporate PLT
- pushq %rax
- cfi_adjust_cfa_offset(8)
- cfi_rel_offset(%rax, 0)
- pushq %rcx
- cfi_adjust_cfa_offset(8)
- cfi_rel_offset(%rcx, 0)
- pushq %rdx
- cfi_adjust_cfa_offset(8)
- cfi_rel_offset(%rdx, 0)
- movl $1, %ecx
- xgetbv
- movl %eax, %r11d
- popq %rdx
- cfi_adjust_cfa_offset(-8)
- cfi_restore (%rdx)
- popq %rcx
- cfi_adjust_cfa_offset(-8)
- cfi_restore (%rcx)
- popq %rax
- cfi_adjust_cfa_offset(-8)
- cfi_restore (%rax)
-# if VEC_SIZE == 32
- # For YMM registers, check if YMM state is in use.
- andl $bit_YMM_state, %r11d
- # Preserve %xmm0 - %xmm7 registers with the zero upper 128 bits if
- # YMM state isn't in use.
- PRESERVE_BND_REGS_PREFIX
- jz _dl_runtime_resolve_sse_vex
-# elif VEC_SIZE == 16
- # For ZMM registers, check if YMM state and ZMM state are in
- # use.
- andl $(bit_YMM_state | bit_ZMM0_15_state), %r11d
- cmpl $bit_YMM_state, %r11d
- # Preserve %zmm0 - %zmm7 registers if ZMM state is in use.
- PRESERVE_BND_REGS_PREFIX
- jg _dl_runtime_resolve_avx512
- # Preserve %ymm0 - %ymm7 registers with the zero upper 256 bits if
- # ZMM state isn't in use.
- PRESERVE_BND_REGS_PREFIX
- je _dl_runtime_resolve_avx
- # Preserve %xmm0 - %xmm7 registers with the zero upper 384 bits if
- # neither YMM state nor ZMM state are in use.
+# if DL_RUNTIME_RESOLVE_REALIGN_STACK
+/* Local stack area before jumping to function address: RBX. */
+# define LOCAL_STORAGE_AREA 8
+# define BASE rbx
+# ifdef USE_FXSAVE
+/* Use fxsave to save XMM registers. */
+# define REGISTER_SAVE_AREA (512 + STATE_SAVE_OFFSET)
+# if (REGISTER_SAVE_AREA % 16) != 0
+# error REGISTER_SAVE_AREA must be multples of 16
+# endif
+# endif
# else
-# error Unsupported VEC_SIZE!
+# ifndef USE_FXSAVE
+# error USE_FXSAVE must be defined
+# endif
+/* Use fxsave to save XMM registers. */
+# define REGISTER_SAVE_AREA (512 + STATE_SAVE_OFFSET + 8)
+/* Local stack area before jumping to function address: All saved
+ registers. */
+# define LOCAL_STORAGE_AREA REGISTER_SAVE_AREA
+# define BASE rsp
+# if (REGISTER_SAVE_AREA % 16) != 8
+# error REGISTER_SAVE_AREA must be odd multples of 8
+# endif
# endif
- cfi_adjust_cfa_offset(-16) # Restore PLT adjustment
- cfi_endproc
- .size _dl_runtime_resolve_opt, .-_dl_runtime_resolve_opt
-#endif
+
.globl _dl_runtime_resolve
.hidden _dl_runtime_resolve
.type _dl_runtime_resolve, @function
@@ -157,21 +64,30 @@ _dl_runtime_resolve_opt:
cfi_startproc
_dl_runtime_resolve:
cfi_adjust_cfa_offset(16) # Incorporate PLT
-#if DL_RUNTIME_RESOLVE_REALIGN_STACK
-# if LOCAL_STORAGE_AREA != 8
-# error LOCAL_STORAGE_AREA must be 8
-# endif
+# if DL_RUNTIME_RESOLVE_REALIGN_STACK
+# if LOCAL_STORAGE_AREA != 8
+# error LOCAL_STORAGE_AREA must be 8
+# endif
pushq %rbx # push subtracts stack by 8.
cfi_adjust_cfa_offset(8)
cfi_rel_offset(%rbx, 0)
mov %RSP_LP, %RBX_LP
cfi_def_cfa_register(%rbx)
- and $-VEC_SIZE, %RSP_LP
-#endif
+ and $-STATE_SAVE_ALIGNMENT, %RSP_LP
+# endif
+# ifdef REGISTER_SAVE_AREA
sub $REGISTER_SAVE_AREA, %RSP_LP
-#if !DL_RUNTIME_RESOLVE_REALIGN_STACK
+# if !DL_RUNTIME_RESOLVE_REALIGN_STACK
cfi_adjust_cfa_offset(REGISTER_SAVE_AREA)
-#endif
+# endif
+# else
+ # Allocate stack space of the required size to save the state.
+# if IS_IN (rtld)
+ sub _rtld_local_ro+RTLD_GLOBAL_RO_DL_X86_CPU_FEATURES_OFFSET+XSAVE_STATE_SIZE_OFFSET(%rip), %RSP_LP
+# else
+ sub _dl_x86_cpu_features+XSAVE_STATE_SIZE_OFFSET(%rip), %RSP_LP
+# endif
+# endif
# Preserve registers otherwise clobbered.
movq %rax, REGISTER_SAVE_RAX(%rsp)
movq %rcx, REGISTER_SAVE_RCX(%rsp)
@@ -180,59 +96,42 @@ _dl_runtime_resolve:
movq %rdi, REGISTER_SAVE_RDI(%rsp)
movq %r8, REGISTER_SAVE_R8(%rsp)
movq %r9, REGISTER_SAVE_R9(%rsp)
- VMOV %VEC(0), (REGISTER_SAVE_VEC_OFF)(%rsp)
- VMOV %VEC(1), (REGISTER_SAVE_VEC_OFF + VEC_SIZE)(%rsp)
- VMOV %VEC(2), (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 2)(%rsp)
- VMOV %VEC(3), (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 3)(%rsp)
- VMOV %VEC(4), (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 4)(%rsp)
- VMOV %VEC(5), (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 5)(%rsp)
- VMOV %VEC(6), (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 6)(%rsp)
- VMOV %VEC(7), (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 7)(%rsp)
-#ifndef __ILP32__
- # We also have to preserve bound registers. These are nops if
- # Intel MPX isn't available or disabled.
-# ifdef HAVE_MPX_SUPPORT
- bndmov %bnd0, REGISTER_SAVE_BND0(%rsp)
- bndmov %bnd1, REGISTER_SAVE_BND1(%rsp)
- bndmov %bnd2, REGISTER_SAVE_BND2(%rsp)
- bndmov %bnd3, REGISTER_SAVE_BND3(%rsp)
+# ifdef USE_FXSAVE
+ fxsave STATE_SAVE_OFFSET(%rsp)
# else
-# if REGISTER_SAVE_BND0 == 0
- .byte 0x66,0x0f,0x1b,0x04,0x24
+ movl $STATE_SAVE_MASK, %eax
+ xorl %edx, %edx
+ # Clear the XSAVE Header.
+# ifdef USE_XSAVE
+ movq %rdx, (STATE_SAVE_OFFSET + 512)(%rsp)
+ movq %rdx, (STATE_SAVE_OFFSET + 512 + 8)(%rsp)
+# endif
+ movq %rdx, (STATE_SAVE_OFFSET + 512 + 8 * 2)(%rsp)
+ movq %rdx, (STATE_SAVE_OFFSET + 512 + 8 * 3)(%rsp)
+ movq %rdx, (STATE_SAVE_OFFSET + 512 + 8 * 4)(%rsp)
+ movq %rdx, (STATE_SAVE_OFFSET + 512 + 8 * 5)(%rsp)
+ movq %rdx, (STATE_SAVE_OFFSET + 512 + 8 * 6)(%rsp)
+ movq %rdx, (STATE_SAVE_OFFSET + 512 + 8 * 7)(%rsp)
+# ifdef USE_XSAVE
+ xsave STATE_SAVE_OFFSET(%rsp)
# else
- .byte 0x66,0x0f,0x1b,0x44,0x24,REGISTER_SAVE_BND0
+ xsavec STATE_SAVE_OFFSET(%rsp)
# endif
- .byte 0x66,0x0f,0x1b,0x4c,0x24,REGISTER_SAVE_BND1
- .byte 0x66,0x0f,0x1b,0x54,0x24,REGISTER_SAVE_BND2
- .byte 0x66,0x0f,0x1b,0x5c,0x24,REGISTER_SAVE_BND3
# endif
-#endif
# Copy args pushed by PLT in register.
# %rdi: link_map, %rsi: reloc_index
mov (LOCAL_STORAGE_AREA + 8)(%BASE), %RSI_LP
mov LOCAL_STORAGE_AREA(%BASE), %RDI_LP
call _dl_fixup # Call resolver.
mov %RAX_LP, %R11_LP # Save return value
-#ifndef __ILP32__
- # Restore bound registers. These are nops if Intel MPX isn't
- # avaiable or disabled.
-# ifdef HAVE_MPX_SUPPORT
- bndmov REGISTER_SAVE_BND3(%rsp), %bnd3
- bndmov REGISTER_SAVE_BND2(%rsp), %bnd2
- bndmov REGISTER_SAVE_BND1(%rsp), %bnd1
- bndmov REGISTER_SAVE_BND0(%rsp), %bnd0
+ # Get register content back.
+# ifdef USE_FXSAVE
+ fxrstor STATE_SAVE_OFFSET(%rsp)
# else
- .byte 0x66,0x0f,0x1a,0x5c,0x24,REGISTER_SAVE_BND3
- .byte 0x66,0x0f,0x1a,0x54,0x24,REGISTER_SAVE_BND2
- .byte 0x66,0x0f,0x1a,0x4c,0x24,REGISTER_SAVE_BND1
-# if REGISTER_SAVE_BND0 == 0
- .byte 0x66,0x0f,0x1a,0x04,0x24
-# else
- .byte 0x66,0x0f,0x1a,0x44,0x24,REGISTER_SAVE_BND0
-# endif
+ movl $STATE_SAVE_MASK, %eax
+ xorl %edx, %edx
+ xrstor STATE_SAVE_OFFSET(%rsp)
# endif
-#endif
- # Get register content back.
movq REGISTER_SAVE_R9(%rsp), %r9
movq REGISTER_SAVE_R8(%rsp), %r8
movq REGISTER_SAVE_RDI(%rsp), %rdi
@@ -240,20 +139,12 @@ _dl_runtime_resolve:
movq REGISTER_SAVE_RDX(%rsp), %rdx
movq REGISTER_SAVE_RCX(%rsp), %rcx
movq REGISTER_SAVE_RAX(%rsp), %rax
- VMOV (REGISTER_SAVE_VEC_OFF)(%rsp), %VEC(0)
- VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE)(%rsp), %VEC(1)
- VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 2)(%rsp), %VEC(2)
- VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 3)(%rsp), %VEC(3)
- VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 4)(%rsp), %VEC(4)
- VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 5)(%rsp), %VEC(5)
- VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 6)(%rsp), %VEC(6)
- VMOV (REGISTER_SAVE_VEC_OFF + VEC_SIZE * 7)(%rsp), %VEC(7)
-#if DL_RUNTIME_RESOLVE_REALIGN_STACK
+# if DL_RUNTIME_RESOLVE_REALIGN_STACK
mov %RBX_LP, %RSP_LP
cfi_def_cfa_register(%rsp)
movq (%rsp), %rbx
cfi_restore(%rbx)
-#endif
+# endif
# Adjust stack(PLT did 2 pushes)
add $(LOCAL_STORAGE_AREA + 16), %RSP_LP
cfi_adjust_cfa_offset(-(LOCAL_STORAGE_AREA + 16))
@@ -262,11 +153,9 @@ _dl_runtime_resolve:
jmp *%r11 # Jump to function address.
cfi_endproc
.size _dl_runtime_resolve, .-_dl_runtime_resolve
+#endif
-/* To preserve %xmm0 - %xmm7 registers, dl-trampoline.h is included
- twice, for _dl_runtime_resolve_sse and _dl_runtime_resolve_sse_vex.
- But we don't need another _dl_runtime_profile for XMM registers. */
#if !defined PROF && defined _dl_runtime_profile
# if (LR_VECTOR_OFFSET % VEC_SIZE) != 0
# error LR_VECTOR_OFFSET must be multples of VEC_SIZE

573
falkor-memcpy-memmove.patch Normal file
View File

@ -0,0 +1,573 @@
2017-10-10 Siddhesh Poyarekar <siddhesh@sourceware.org>
* sysdeps/aarch64/multiarch/Makefile (sysdep_routines): Add
memmove_falkor.
* sysdeps/aarch64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Likewise.
* sysdeps/aarch64/multiarch/memmove.c: Likewise.
* sysdeps/aarch64/multiarch/memmove_falkor.S: New file.
* benchtests/bench-memmove-walk.c: New file.
* benchtests/Makefile (string-benchset): Add it.
* benchtests/bench-memset-walk.c: New file.
* benchtests/Makefile (string-benchset): Add it.
* benchtests/bench-memcpy-walk.c: New file.
* benchtests/Makefile (string-benchset): Add it.
2017-10-10 Siddhesh Poyarekar <siddhesh@sourceware.org>
* manual/tunables.texi (Tunable glibc.tune.cpu): Add falkor.
* sysdeps/aarch64/multiarch/Makefile (sysdep_routines): Add
memcpy_falkor.
* sysdeps/aarch64/multiarch/ifunc-impl-list.c (MAX_IFUNC):
Bump.
(__libc_ifunc_impl_list): Add __memcpy_falkor.
* sysdeps/aarch64/multiarch/memcpy.c: Likewise.
* sysdeps/aarch64/multiarch/memcpy_falkor.S: New file.
* sysdeps/unix/sysv/linux/aarch64/cpu-features.c (cpu_list):
Add falkor.
* sysdeps/unix/sysv/linux/aarch64/cpu-features.h (IS_FALKOR):
New macro.
Index: glibc-2.26/manual/tunables.texi
===================================================================
--- glibc-2.26.orig/manual/tunables.texi
+++ glibc-2.26/manual/tunables.texi
@@ -267,7 +267,7 @@ This tunable is specific to i386 and x86
@deftp Tunable glibc.tune.cpu
The @code{glibc.tune.cpu=xxx} tunable allows the user to tell @theglibc{} to
assume that the CPU is @code{xxx} where xxx may have one of these values:
-@code{generic}, @code{thunderxt88}.
+@code{generic}, @code{falkor}, @code{thunderxt88}.
This tunable is specific to aarch64.
@end deftp
Index: glibc-2.26/sysdeps/aarch64/multiarch/Makefile
===================================================================
--- glibc-2.26.orig/sysdeps/aarch64/multiarch/Makefile
+++ glibc-2.26/sysdeps/aarch64/multiarch/Makefile
@@ -1,3 +1,4 @@
ifeq ($(subdir),string)
-sysdep_routines += memcpy_generic memcpy_thunderx
+sysdep_routines += memcpy_generic memcpy_thunderx memcpy_falkor \
+ memmove_falkor
endif
Index: glibc-2.26/sysdeps/aarch64/multiarch/ifunc-impl-list.c
===================================================================
--- glibc-2.26.orig/sysdeps/aarch64/multiarch/ifunc-impl-list.c
+++ glibc-2.26/sysdeps/aarch64/multiarch/ifunc-impl-list.c
@@ -25,7 +25,7 @@
#include <stdio.h>
/* Maximum number of IFUNC implementations. */
-#define MAX_IFUNC 2
+#define MAX_IFUNC 3
size_t
__libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array,
@@ -40,9 +40,11 @@ __libc_ifunc_impl_list (const char *name
/* Support sysdeps/aarch64/multiarch/memcpy.c and memmove.c. */
IFUNC_IMPL (i, name, memcpy,
IFUNC_IMPL_ADD (array, i, memcpy, 1, __memcpy_thunderx)
+ IFUNC_IMPL_ADD (array, i, memcpy, 1, __memcpy_falkor)
IFUNC_IMPL_ADD (array, i, memcpy, 1, __memcpy_generic))
IFUNC_IMPL (i, name, memmove,
IFUNC_IMPL_ADD (array, i, memmove, 1, __memmove_thunderx)
+ IFUNC_IMPL_ADD (array, i, memmove, 1, __memmove_falkor)
IFUNC_IMPL_ADD (array, i, memmove, 1, __memmove_generic))
return i;
Index: glibc-2.26/sysdeps/aarch64/multiarch/memcpy.c
===================================================================
--- glibc-2.26.orig/sysdeps/aarch64/multiarch/memcpy.c
+++ glibc-2.26/sysdeps/aarch64/multiarch/memcpy.c
@@ -30,9 +30,14 @@ extern __typeof (__redirect_memcpy) __li
extern __typeof (__redirect_memcpy) __memcpy_generic attribute_hidden;
extern __typeof (__redirect_memcpy) __memcpy_thunderx attribute_hidden;
+extern __typeof (__redirect_memcpy) __memcpy_falkor attribute_hidden;
libc_ifunc (__libc_memcpy,
- IS_THUNDERX (midr) ? __memcpy_thunderx : __memcpy_generic);
+ (IS_THUNDERX (midr)
+ ? __memcpy_thunderx
+ : (IS_FALKOR (midr)
+ ? __memcpy_falkor
+ : __memcpy_generic)));
# undef memcpy
strong_alias (__libc_memcpy, memcpy);
Index: glibc-2.26/sysdeps/aarch64/multiarch/memcpy_falkor.S
===================================================================
--- /dev/null
+++ glibc-2.26/sysdeps/aarch64/multiarch/memcpy_falkor.S
@@ -0,0 +1,184 @@
+/* Optimized memcpy for Qualcomm Falkor processor.
+ Copyright (C) 2017 Free Software Foundation, Inc.
+
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library. If not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <sysdep.h>
+
+/* Assumptions:
+
+ ARMv8-a, AArch64, falkor, unaligned accesses. */
+
+#define dstin x0
+#define src x1
+#define count x2
+#define dst x3
+#define srcend x4
+#define dstend x5
+#define A_l x6
+#define A_lw w6
+#define A_h x7
+#define A_hw w7
+#define tmp1 x14
+
+/* Copies are split into 3 main cases:
+
+ 1. Small copies of up to 32 bytes
+ 2. Medium copies of 33..128 bytes which are fully unrolled
+ 3. Large copies of more than 128 bytes.
+
+ Large copies align the sourceto a quad word and use an unrolled loop
+ processing 64 bytes per iteration.
+
+ FALKOR-SPECIFIC DESIGN:
+
+ The smallest copies (32 bytes or less) focus on optimal pipeline usage,
+ which is why the redundant copies of 0-3 bytes have been replaced with
+ conditionals, since the former would unnecessarily break across multiple
+ issue groups. The medium copy group has been enlarged to 128 bytes since
+ bumping up the small copies up to 32 bytes allows us to do that without
+ cost and also allows us to reduce the size of the prep code before loop64.
+
+ All copies are done only via two registers r6 and r7. This is to ensure
+ that all loads hit a single hardware prefetcher which can get correctly
+ trained to prefetch a single stream.
+
+ The non-temporal stores help optimize cache utilization. */
+
+#if IS_IN (libc)
+ENTRY_ALIGN (__memcpy_falkor, 6)
+
+ cmp count, 32
+ add srcend, src, count
+ add dstend, dstin, count
+ b.ls L(copy32)
+ ldp A_l, A_h, [src]
+ cmp count, 128
+ stp A_l, A_h, [dstin]
+ b.hi L(copy_long)
+
+ /* Medium copies: 33..128 bytes. */
+ sub tmp1, count, 1
+ ldp A_l, A_h, [src, 16]
+ stp A_l, A_h, [dstin, 16]
+ tbz tmp1, 6, 1f
+ ldp A_l, A_h, [src, 32]
+ stp A_l, A_h, [dstin, 32]
+ ldp A_l, A_h, [src, 48]
+ stp A_l, A_h, [dstin, 48]
+ ldp A_l, A_h, [srcend, -64]
+ stp A_l, A_h, [dstend, -64]
+ ldp A_l, A_h, [srcend, -48]
+ stp A_l, A_h, [dstend, -48]
+1:
+ ldp A_l, A_h, [srcend, -32]
+ stp A_l, A_h, [dstend, -32]
+ ldp A_l, A_h, [srcend, -16]
+ stp A_l, A_h, [dstend, -16]
+ ret
+
+ .p2align 4
+ /* Small copies: 0..32 bytes. */
+L(copy32):
+ /* 16-32 */
+ cmp count, 16
+ b.lo 1f
+ ldp A_l, A_h, [src]
+ stp A_l, A_h, [dstin]
+ ldp A_l, A_h, [srcend, -16]
+ stp A_l, A_h, [dstend, -16]
+ ret
+ .p2align 4
+1:
+ /* 8-15 */
+ tbz count, 3, 1f
+ ldr A_l, [src]
+ str A_l, [dstin]
+ ldr A_l, [srcend, -8]
+ str A_l, [dstend, -8]
+ ret
+ .p2align 4
+1:
+ /* 4-7 */
+ tbz count, 2, 1f
+ ldr A_lw, [src]
+ str A_lw, [dstin]
+ ldr A_lw, [srcend, -4]
+ str A_lw, [dstend, -4]
+ ret
+ .p2align 4
+1:
+ /* 2-3 */
+ tbz count, 1, 1f
+ ldrh A_lw, [src]
+ strh A_lw, [dstin]
+ ldrh A_lw, [srcend, -2]
+ strh A_lw, [dstend, -2]
+ ret
+ .p2align 4
+1:
+ /* 0-1 */
+ tbz count, 0, 1f
+ ldrb A_lw, [src]
+ strb A_lw, [dstin]
+1:
+ ret
+
+ /* Align SRC to 16 bytes and copy; that way at least one of the
+ accesses is aligned throughout the copy sequence.
+
+ The count is off by 0 to 15 bytes, but this is OK because we trim
+ off the last 64 bytes to copy off from the end. Due to this the
+ loop never runs out of bounds. */
+ .p2align 6
+L(copy_long):
+ sub count, count, 64 + 16
+ and tmp1, src, 15
+ bic src, src, 15
+ sub dst, dstin, tmp1
+ add count, count, tmp1
+
+L(loop64):
+ ldp A_l, A_h, [src, 16]!
+ stnp A_l, A_h, [dst, 16]
+ ldp A_l, A_h, [src, 16]!
+ subs count, count, 64
+ stnp A_l, A_h, [dst, 32]
+ ldp A_l, A_h, [src, 16]!
+ stnp A_l, A_h, [dst, 48]
+ ldp A_l, A_h, [src, 16]!
+ stnp A_l, A_h, [dst, 64]
+ add dst, dst, 64
+ b.hi L(loop64)
+
+ /* Write the last full set of 64 bytes. The remainder is at most 64
+ bytes, so it is safe to always copy 64 bytes from the end even if
+ there is just 1 byte left. */
+L(last64):
+ ldp A_l, A_h, [srcend, -64]
+ stnp A_l, A_h, [dstend, -64]
+ ldp A_l, A_h, [srcend, -48]
+ stnp A_l, A_h, [dstend, -48]
+ ldp A_l, A_h, [srcend, -32]
+ stnp A_l, A_h, [dstend, -32]
+ ldp A_l, A_h, [srcend, -16]
+ stnp A_l, A_h, [dstend, -16]
+ ret
+
+END (__memcpy_falkor)
+libc_hidden_builtin_def (__memcpy_falkor)
+#endif
Index: glibc-2.26/sysdeps/aarch64/multiarch/memmove.c
===================================================================
--- glibc-2.26.orig/sysdeps/aarch64/multiarch/memmove.c
+++ glibc-2.26/sysdeps/aarch64/multiarch/memmove.c
@@ -30,9 +30,14 @@ extern __typeof (__redirect_memmove) __l
extern __typeof (__redirect_memmove) __memmove_generic attribute_hidden;
extern __typeof (__redirect_memmove) __memmove_thunderx attribute_hidden;
+extern __typeof (__redirect_memmove) __memmove_falkor attribute_hidden;
libc_ifunc (__libc_memmove,
- IS_THUNDERX (midr) ? __memmove_thunderx : __memmove_generic);
+ (IS_THUNDERX (midr)
+ ? __memmove_thunderx
+ : (IS_FALKOR (midr)
+ ? __memmove_falkor
+ : __memmove_generic)));
# undef memmove
strong_alias (__libc_memmove, memmove);
Index: glibc-2.26/sysdeps/aarch64/multiarch/memmove_falkor.S
===================================================================
--- /dev/null
+++ glibc-2.26/sysdeps/aarch64/multiarch/memmove_falkor.S
@@ -0,0 +1,232 @@
+/* Copyright (C) 2017 Free Software Foundation, Inc.
+
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library. If not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <sysdep.h>
+
+/* Assumptions: ARMv8-a, AArch64, falkor, unaligned accesses. */
+
+#define dstin x0
+#define src x1
+#define count x2
+#define dstlen x3
+#define dst x3
+#define srcend x4
+#define dstend x5
+#define A_l x6
+#define A_lw w6
+#define A_h x7
+#define A_hw w7
+#define B_l x8
+#define B_lw w8
+#define B_h x9
+#define C_l x10
+#define C_h x11
+#define D_l x12
+#define D_h x13
+#define E_l src
+#define E_h count
+#define F_l srcend
+#define F_h dst
+#define tmp1 x14
+
+/* Alias with A_l and A_h to train the prefetcher. */
+#define Q_l x22
+#define Q_h x23
+
+/* RATIONALE:
+
+ The copy has 4 distinct parts:
+ * Small copies of 16 bytes and under
+ * Medium sized copies of 17-96 bytes
+ * Large copies where the source address is higher than the destination
+ (forward copies)
+ * Large copies where the destination address is higher than the source
+ (copy backward, or move).
+
+ We use only two registerpairs x6,x7 and x22,x23 for the copies and copy 32
+ bytes at a time to correctly train the hardware prefetcher for better
+ throughput. */
+ENTRY_ALIGN (__memmove_falkor, 6)
+
+ sub tmp1, dstin, src
+ add srcend, src, count
+ add dstend, dstin, count
+ cmp count, 96
+ ccmp tmp1, count, 2, hi
+ b.lo L(move_long)
+
+ cmp count, 16
+ b.ls L(copy16)
+ cmp count, 96
+ b.hi L(copy_long)
+
+ /* Medium copies: 17..96 bytes. */
+ sub tmp1, count, 1
+ ldp A_l, A_h, [src]
+ tbnz tmp1, 6, L(copy96)
+ ldp D_l, D_h, [srcend, -16]
+ tbz tmp1, 5, 1f
+ ldp B_l, B_h, [src, 16]
+ ldp C_l, C_h, [srcend, -32]
+ stp B_l, B_h, [dstin, 16]
+ stp C_l, C_h, [dstend, -32]
+1:
+ stp A_l, A_h, [dstin]
+ stp D_l, D_h, [dstend, -16]
+ ret
+
+ .p2align 4
+ /* Small copies: 0..16 bytes. */
+L(copy16):
+ cmp count, 8
+ b.lo 1f
+ ldr A_l, [src]
+ ldr A_h, [srcend, -8]
+ str A_l, [dstin]
+ str A_h, [dstend, -8]
+ ret
+ .p2align 4
+1:
+ /* 4-7 */
+ tbz count, 2, 1f
+ ldr A_lw, [src]
+ ldr A_hw, [srcend, -4]
+ str A_lw, [dstin]
+ str A_hw, [dstend, -4]
+ ret
+ .p2align 4
+1:
+ /* 2-3 */
+ tbz count, 1, 1f
+ ldrh A_lw, [src]
+ ldrh A_hw, [srcend, -2]
+ strh A_lw, [dstin]
+ strh A_hw, [dstend, -2]
+ ret
+ .p2align 4
+1:
+ /* 0-1 */
+ tbz count, 0, 1f
+ ldrb A_lw, [src]
+ strb A_lw, [dstin]
+1: ret
+
+ .p2align 4
+ /* Copy 64..96 bytes. Copy 64 bytes from the start and
+ 32 bytes from the end. */
+L(copy96):
+ ldp B_l, B_h, [src, 16]
+ ldp C_l, C_h, [src, 32]
+ ldp D_l, D_h, [src, 48]
+ ldp E_l, E_h, [srcend, -32]
+ ldp F_l, F_h, [srcend, -16]
+ stp A_l, A_h, [dstin]
+ stp B_l, B_h, [dstin, 16]
+ stp C_l, C_h, [dstin, 32]
+ stp D_l, D_h, [dstin, 48]
+ stp E_l, E_h, [dstend, -32]
+ stp F_l, F_h, [dstend, -16]
+ ret
+
+ /* Align SRC to 16 byte alignment so that we don't cross cache line
+ boundaries on both loads and stores. There are at least 96 bytes
+ to copy, so copy 16 bytes unaligned and then align. The loop
+ copies 32 bytes per iteration and prefetches one iteration ahead. */
+
+ .p2align 4
+L(copy_long):
+ sub count, count, 64 + 16 /* Test and readjust count. */
+ mov B_l, Q_l
+ mov B_h, Q_h
+ ldp A_l, A_h, [src]
+ and tmp1, src, 15
+ bic src, src, 15
+ sub dst, dstin, tmp1
+ add count, count, tmp1 /* Count is now 16 too large. */
+ ldp Q_l, Q_h, [src, 16]!
+ stp A_l, A_h, [dstin]
+ ldp A_l, A_h, [src, 16]!
+
+L(loop64):
+ subs count, count, 32
+ stp Q_l, Q_h, [dst, 16]
+ ldp Q_l, Q_h, [src, 16]!
+ stp A_l, A_h, [dst, 32]!
+ ldp A_l, A_h, [src, 16]!
+ b.hi L(loop64)
+
+ /* Write the last full set of 32 bytes. The remainder is at most 32
+ bytes, so it is safe to always copy 32 bytes from the end even if
+ there is just 1 byte left. */
+L(last64):
+ ldp C_l, C_h, [srcend, -32]
+ stp Q_l, Q_h, [dst, 16]
+ ldp Q_l, Q_h, [srcend, -16]
+ stp A_l, A_h, [dst, 32]
+ stp C_l, C_h, [dstend, -32]
+ stp Q_l, Q_h, [dstend, -16]
+ mov Q_l, B_l
+ mov Q_h, B_h
+ ret
+
+ .p2align 4
+L(move_long):
+ cbz tmp1, 3f
+
+ mov B_l, Q_l
+ mov B_h, Q_h
+
+ /* Align SRCEND to 16 byte alignment so that we don't cross cache line
+ boundaries on both loads and stores. There are at least 96 bytes
+ to copy, so copy 16 bytes unaligned and then align. The loop
+ copies 32 bytes per iteration and prefetches one iteration ahead. */
+
+ ldp A_l, A_h, [srcend, -16]
+ and tmp1, srcend, 15
+ sub srcend, srcend, tmp1
+ ldp Q_l, Q_h, [srcend, -16]!
+ stp A_l, A_h, [dstend, -16]
+ sub count, count, tmp1
+ ldp A_l, A_h, [srcend, -16]!
+ sub dstend, dstend, tmp1
+ sub count, count, 64
+
+1:
+ subs count, count, 32
+ stp Q_l, Q_h, [dstend, -16]
+ ldp Q_l, Q_h, [srcend, -16]!
+ stp A_l, A_h, [dstend, -32]!
+ ldp A_l, A_h, [srcend, -16]!
+ b.hi 1b
+
+ /* Write the last full set of 32 bytes. The remainder is at most 32
+ bytes, so it is safe to always copy 32 bytes from the start even if
+ there is just 1 byte left. */
+2:
+ ldp C_l, C_h, [src, 16]
+ stp Q_l, Q_h, [dstend, -16]
+ ldp Q_l, Q_h, [src]
+ stp A_l, A_h, [dstend, -32]
+ stp C_l, C_h, [dstin, 16]
+ stp Q_l, Q_h, [dstin]
+ mov Q_l, B_l
+ mov Q_h, B_h
+3: ret
+
+END (__memmove_falkor)
+libc_hidden_builtin_def (__memmove_falkor)
Index: glibc-2.26/sysdeps/unix/sysv/linux/aarch64/cpu-features.c
===================================================================
--- glibc-2.26.orig/sysdeps/unix/sysv/linux/aarch64/cpu-features.c
+++ glibc-2.26/sysdeps/unix/sysv/linux/aarch64/cpu-features.c
@@ -28,6 +28,7 @@ struct cpu_list
};
static struct cpu_list cpu_list[] = {
+ {"falkor", 0x510FC000},
{"thunderxt88", 0x430F0A10},
{"generic", 0x0}
};
Index: glibc-2.26/sysdeps/unix/sysv/linux/aarch64/cpu-features.h
===================================================================
--- glibc-2.26.orig/sysdeps/unix/sysv/linux/aarch64/cpu-features.h
+++ glibc-2.26/sysdeps/unix/sysv/linux/aarch64/cpu-features.h
@@ -41,6 +41,9 @@
#define IS_THUNDERX(midr) (MIDR_IMPLEMENTOR(midr) == 'C' \
&& MIDR_PARTNUM(midr) == 0x0a1)
+#define IS_FALKOR(midr) (MIDR_IMPLEMENTOR(midr) == 'Q' \
+ && MIDR_PARTNUM(midr) == 0xc00)
+
struct cpu_features
{
uint64_t midr_el1;

46
fnmatch-collating-elements.patch
View File

@ -10,22 +10,22 @@ Fix fnmatch handling of collating elements (BZ #17396, BZ #16976)
* posix/tst-fnmatch4.c: New file.
* posix/tst-fnmatch5.c: New file.
Index: glibc-2.22/posix/Makefile
Index: glibc-2.26/posix/Makefile
===================================================================
--- glibc-2.22.orig/posix/Makefile
+++ glibc-2.22/posix/Makefile
@@ -87,6 +87,7 @@ tests := tstgetopt testfnm runtests run
--- glibc-2.26.orig/posix/Makefile
+++ glibc-2.26/posix/Makefile
@@ -91,6 +91,7 @@ tests := test-errno tstgetopt testfnm r
bug-getopt5 tst-getopt_long1 bug-regex34 bug-regex35 \
tst-pathconf tst-getaddrinfo4 tst-rxspencer-no-utf8 \
tst-fnmatch3 bug-regex36 tst-getaddrinfo5 \
+ tst-fnmatch4 tst-fnmatch5 \
tst-posix_spawn-fd tst-posix_spawn-setsid \
tst-posix_fadvise tst-posix_fadvise64 \
tst-sysconf-empty-chroot
Index: glibc-2.22/posix/fnmatch.c
tst-sysconf-empty-chroot tst-glob-tilde
Index: glibc-2.26/posix/fnmatch.c
===================================================================
--- glibc-2.22.orig/posix/fnmatch.c
+++ glibc-2.22/posix/fnmatch.c
--- glibc-2.26.orig/posix/fnmatch.c
+++ glibc-2.26/posix/fnmatch.c
@@ -53,7 +53,6 @@
we support a correct implementation only in glibc. */
#ifdef _LIBC
@ -34,11 +34,11 @@ Index: glibc-2.22/posix/fnmatch.c
# include "../locale/coll-lookup.h"
# include <shlib-compat.h>
Index: glibc-2.22/posix/fnmatch_loop.c
Index: glibc-2.26/posix/fnmatch_loop.c
===================================================================
--- glibc-2.22.orig/posix/fnmatch_loop.c
+++ glibc-2.22/posix/fnmatch_loop.c
@@ -503,26 +503,12 @@ FCT (pattern, string, string_end, no_lea
--- glibc-2.26.orig/posix/fnmatch_loop.c
+++ glibc-2.26/posix/fnmatch_loop.c
@@ -497,26 +497,12 @@ FCT (const CHAR *pattern, const CHAR *st
{
int32_t table_size;
const int32_t *symb_table;
@ -67,7 +67,7 @@ Index: glibc-2.22/posix/fnmatch_loop.c
table_size =
_NL_CURRENT_WORD (LC_COLLATE,
@@ -534,71 +520,55 @@ FCT (pattern, string, string_end, no_lea
@@ -528,71 +514,55 @@ FCT (const CHAR *pattern, const CHAR *st
_NL_CURRENT (LC_COLLATE,
_NL_COLLATE_SYMB_EXTRAMB);
@ -180,7 +180,7 @@ Index: glibc-2.22/posix/fnmatch_loop.c
}
/* Get the collation sequence value. */
@@ -606,9 +576,9 @@ FCT (pattern, string, string_end, no_lea
@@ -600,9 +570,9 @@ FCT (const CHAR *pattern, const CHAR *st
# if WIDE_CHAR_VERSION
cold = wextra[1 + wextra[idx]];
# else
@ -192,7 +192,7 @@ Index: glibc-2.22/posix/fnmatch_loop.c
cold = *((int32_t *) &extra[idx]);
# endif
@@ -618,10 +588,10 @@ FCT (pattern, string, string_end, no_lea
@@ -612,10 +582,10 @@ FCT (const CHAR *pattern, const CHAR *st
{
/* No valid character. Match it as a
single byte. */
@ -205,7 +205,7 @@ Index: glibc-2.22/posix/fnmatch_loop.c
c = *p++;
}
else
@@ -629,7 +599,6 @@ FCT (pattern, string, string_end, no_lea
@@ -623,7 +593,6 @@ FCT (const CHAR *pattern, const CHAR *st
}
}
else
@ -213,7 +213,7 @@ Index: glibc-2.22/posix/fnmatch_loop.c
#endif
{
c = FOLD (c);
@@ -721,25 +690,11 @@ FCT (pattern, string, string_end, no_lea
@@ -715,25 +684,11 @@ FCT (const CHAR *pattern, const CHAR *st
{
int32_t table_size;
const int32_t *symb_table;
@ -240,7 +240,7 @@ Index: glibc-2.22/posix/fnmatch_loop.c
# endif
table_size =
@@ -752,51 +707,44 @@ FCT (pattern, string, string_end, no_lea
@@ -746,51 +701,44 @@ FCT (const CHAR *pattern, const CHAR *st
_NL_CURRENT (LC_COLLATE,
_NL_COLLATE_SYMB_EXTRAMB);
@ -328,7 +328,7 @@ Index: glibc-2.22/posix/fnmatch_loop.c
/* Get the collation sequence value. */
is_seqval = 1;
# if WIDE_CHAR_VERSION
@@ -804,19 +752,18 @@ FCT (pattern, string, string_end, no_lea
@@ -798,19 +746,18 @@ FCT (const CHAR *pattern, const CHAR *st
# else
/* Adjust for the alignment. */
idx += 1 + extra[idx];
@ -351,10 +351,10 @@ Index: glibc-2.22/posix/fnmatch_loop.c
}
else
{
Index: glibc-2.22/posix/tst-fnmatch4.c
Index: glibc-2.26/posix/tst-fnmatch4.c
===================================================================
--- /dev/null
+++ glibc-2.22/posix/tst-fnmatch4.c
+++ glibc-2.26/posix/tst-fnmatch4.c
@@ -0,0 +1,51 @@
+/* Test for fnmatch handling of collating elements
+ Copyright (C) 2015 Free Software Foundation, Inc.
@ -407,10 +407,10 @@ Index: glibc-2.22/posix/tst-fnmatch4.c
+
+#define TEST_FUNCTION do_test ()
+#include "../test-skeleton.c"
Index: glibc-2.22/posix/tst-fnmatch5.c
Index: glibc-2.26/posix/tst-fnmatch5.c
===================================================================
--- /dev/null
+++ glibc-2.22/posix/tst-fnmatch5.c
+++ glibc-2.26/posix/tst-fnmatch5.c
@@ -0,0 +1,53 @@
+/* Test for fnmatch handling of collating elements
+ Copyright (C) 2015 Free Software Foundation, Inc.

27
glibc-testsuite.changes
View File

@ -1,3 +1,28 @@
-------------------------------------------------------------------
Thu Oct 26 08:40:52 UTC 2017 - schwab@suse.de
- spawni-assert.patch: Fix improper assert in Linux posix_spawn (BZ
#22273)
-------------------------------------------------------------------
Mon Oct 23 09:35:18 UTC 2017 - schwab@suse.de
- math-c++-compat.patch: Add more C++ compatibility (BZ #22296)
- malloc-tcache-leak.patch: Fix tcache leak after thread destruction (BZ
#22111)
- falkor-memcpy-memmove.patch: Optimized implementation of memcpy/memmove
for Qualcomm Falkor
- aarch64-cpu-features.patch: Fix glibc.tune.cpu tunable handling
- nss-files-large-buffers.patch: Avoid large buffers with many host
addresses (BZ #22078)
- sysconf-uio-maxiov.patch: Fix missing definition of UIO_MAXIOV (BZ
#22321)
- glob-tilde-overflow.patch: Fix buffer overflows with GLOB_TILDE
(CVE-2017-15670, CVE-2017-15671, CVE-2017-15804,
bsc#1064569. bsc#1064580, bsc#1064583, BZ #22320, BZ #22325, BZ #22332)
- dl-runtime-resolve-xsave.patch: Use fxsave/xsave/xsavec in
_dl_runtime_resolve (BZ #21265)
-------------------------------------------------------------------
Tue Oct 10 15:47:05 UTC 2017 - schwab@suse.de
@ -22,7 +47,7 @@ Thu Sep 28 07:57:52 UTC 2017 - schwab@suse.de
- assert-pedantic.patch: Suppress pedantic warning caused by statement
expression (BZ #21242, BZ #21972)
- math-c++-compat.patch: Add more C++ compatibility
- math-c++-compat.patch: Add more C++ compatibility (BZ #22235)
- getaddrinfo-errno.patch: Fix errno and h_errno handling in getaddrinfo
(BZ #21915, BZ #21922)
- resolv-conf-oom.patch: Fix memory handling in OOM situation during

26
glibc-testsuite.spec
View File

@ -273,7 +273,7 @@ Patch1000: resolv-context-leak.patch
Patch1001: dl-runtime-resolve-opt-avx512f.patch
# PATCH-FIX-UPSTREAM Don't use IFUNC resolver for longjmp or system in libpthread (BZ #21041)
Patch1002: libpthread-compat-wrappers.patch
# PATCH-FIX-UPSTREAM Do not use __builtin_types_compatible_p in C++ mode (BZ #21930)
# PATCH-FIX-UPSTREAM Do not use __builtin_types_compatible_p in C++ mode (BZ #21930, BZ #22146, BZ #22235, BZ #22296)
Patch1003: math-c++-compat.patch
# PATCH-FIX-UPSTREAM Remove nis and compat from default NSS configs
Patch1004: remove-nss-nis-compat.patch
@ -295,6 +295,22 @@ Patch1011: nearbyint-inexact.patch
Patch1012: nss-compat.patch
# PATCH-FIX-UPSTREAM Remove reference to libnsl from nscd
Patch1013: nscd-libnsl.patch
# PATCH-FIX-UPSTREAM malloc: Fix tcache leak after thread destruction (BZ #22111)
Patch1014: malloc-tcache-leak.patch
# PATCH-FIX-UPSTREAM aarch64: Optimized implementation of memcpy/memmove for Qualcomm Falkor
Patch1015: falkor-memcpy-memmove.patch
# PATCH-FIX-UPSTREAM aarch64: Fix glibc.tune.cpu tunable handling
Patch1016: aarch64-cpu-features.patch
# PATCH-FIX-UPSTREAM nss_files: Avoid large buffers with many host addresses (BZ #22078)
Patch1017: nss-files-large-buffers.patch
# PATCH-FIX-UPSTREAM sysconf: Fix missing definition of UIO_MAXIOV on Linux (BZ #22321)
Patch1018: sysconf-uio-maxiov.patch
# PATCH-FIX-UPSTREAM glob: Fix buffer overflows (CVE-2017-15670, CVE-2017-15671, CVE-2017-15804, BZ #22320, BZ #22325, BZ #22332)
Patch1019: glob-tilde-overflow.patch
# PATCH-FIX-UPSTREAM x86-64: Use fxsave/xsave/xsavec in _dl_runtime_resolve (BZ #21265)
Patch1020: dl-runtime-resolve-xsave.patch
# PATCH-FIX-UPSTREAM posix: Fix improper assert in Linux posix_spawn (BZ #22273)
Patch1021: spawni-assert.patch
###
# Patches awaiting upstream approval
@ -519,6 +535,14 @@ rm nscd/s-stamp
%patch1011 -p1
%patch1012 -p1
%patch1013 -p1
%patch1014 -p1
%patch1015 -p1
%patch1016 -p1
%patch1017 -p1
%patch1018 -p1
%patch1019 -p1
%patch1020 -p1
%patch1021 -p1
%patch2000 -p1
%patch2001 -p1

27
glibc-utils.changes
View File

@ -1,3 +1,28 @@
-------------------------------------------------------------------
Thu Oct 26 08:40:52 UTC 2017 - schwab@suse.de
- spawni-assert.patch: Fix improper assert in Linux posix_spawn (BZ
#22273)
-------------------------------------------------------------------
Mon Oct 23 09:35:18 UTC 2017 - schwab@suse.de
- math-c++-compat.patch: Add more C++ compatibility (BZ #22296)
- malloc-tcache-leak.patch: Fix tcache leak after thread destruction (BZ
#22111)
- falkor-memcpy-memmove.patch: Optimized implementation of memcpy/memmove
for Qualcomm Falkor
- aarch64-cpu-features.patch: Fix glibc.tune.cpu tunable handling
- nss-files-large-buffers.patch: Avoid large buffers with many host
addresses (BZ #22078)
- sysconf-uio-maxiov.patch: Fix missing definition of UIO_MAXIOV (BZ
#22321)
- glob-tilde-overflow.patch: Fix buffer overflows with GLOB_TILDE
(CVE-2017-15670, CVE-2017-15671, CVE-2017-15804,
bsc#1064569. bsc#1064580, bsc#1064583, BZ #22320, BZ #22325, BZ #22332)
- dl-runtime-resolve-xsave.patch: Use fxsave/xsave/xsavec in
_dl_runtime_resolve (BZ #21265)
-------------------------------------------------------------------
Tue Oct 10 15:47:05 UTC 2017 - schwab@suse.de
@ -22,7 +47,7 @@ Thu Sep 28 07:57:52 UTC 2017 - schwab@suse.de
- assert-pedantic.patch: Suppress pedantic warning caused by statement
expression (BZ #21242, BZ #21972)
- math-c++-compat.patch: Add more C++ compatibility
- math-c++-compat.patch: Add more C++ compatibility (BZ #22235)
- getaddrinfo-errno.patch: Fix errno and h_errno handling in getaddrinfo
(BZ #21915, BZ #21922)
- resolv-conf-oom.patch: Fix memory handling in OOM situation during

26
glibc-utils.spec
View File

@ -273,7 +273,7 @@ Patch1000: resolv-context-leak.patch
Patch1001: dl-runtime-resolve-opt-avx512f.patch
# PATCH-FIX-UPSTREAM Don't use IFUNC resolver for longjmp or system in libpthread (BZ #21041)
Patch1002: libpthread-compat-wrappers.patch
# PATCH-FIX-UPSTREAM Do not use __builtin_types_compatible_p in C++ mode (BZ #21930)
# PATCH-FIX-UPSTREAM Do not use __builtin_types_compatible_p in C++ mode (BZ #21930, BZ #22146, BZ #22235, BZ #22296)
Patch1003: math-c++-compat.patch
# PATCH-FIX-UPSTREAM Remove nis and compat from default NSS configs
Patch1004: remove-nss-nis-compat.patch
@ -295,6 +295,22 @@ Patch1011: nearbyint-inexact.patch
Patch1012: nss-compat.patch
# PATCH-FIX-UPSTREAM Remove reference to libnsl from nscd
Patch1013: nscd-libnsl.patch
# PATCH-FIX-UPSTREAM malloc: Fix tcache leak after thread destruction (BZ #22111)
Patch1014: malloc-tcache-leak.patch
# PATCH-FIX-UPSTREAM aarch64: Optimized implementation of memcpy/memmove for Qualcomm Falkor
Patch1015: falkor-memcpy-memmove.patch
# PATCH-FIX-UPSTREAM aarch64: Fix glibc.tune.cpu tunable handling
Patch1016: aarch64-cpu-features.patch
# PATCH-FIX-UPSTREAM nss_files: Avoid large buffers with many host addresses (BZ #22078)
Patch1017: nss-files-large-buffers.patch
# PATCH-FIX-UPSTREAM sysconf: Fix missing definition of UIO_MAXIOV on Linux (BZ #22321)
Patch1018: sysconf-uio-maxiov.patch
# PATCH-FIX-UPSTREAM glob: Fix buffer overflows (CVE-2017-15670, CVE-2017-15671, CVE-2017-15804, BZ #22320, BZ #22325, BZ #22332)
Patch1019: glob-tilde-overflow.patch
# PATCH-FIX-UPSTREAM x86-64: Use fxsave/xsave/xsavec in _dl_runtime_resolve (BZ #21265)
Patch1020: dl-runtime-resolve-xsave.patch
# PATCH-FIX-UPSTREAM posix: Fix improper assert in Linux posix_spawn (BZ #22273)
Patch1021: spawni-assert.patch
###
# Patches awaiting upstream approval
@ -519,6 +535,14 @@ rm nscd/s-stamp
%patch1011 -p1
%patch1012 -p1
%patch1013 -p1
%patch1014 -p1
%patch1015 -p1
%patch1016 -p1
%patch1017 -p1
%patch1018 -p1
%patch1019 -p1
%patch1020 -p1
%patch1021 -p1
%patch2000 -p1
%patch2001 -p1

27
glibc.changes
View File

@ -1,3 +1,28 @@
-------------------------------------------------------------------
Thu Oct 26 08:40:52 UTC 2017 - schwab@suse.de
- spawni-assert.patch: Fix improper assert in Linux posix_spawn (BZ
#22273)
-------------------------------------------------------------------
Mon Oct 23 09:35:18 UTC 2017 - schwab@suse.de
- math-c++-compat.patch: Add more C++ compatibility (BZ #22296)
- malloc-tcache-leak.patch: Fix tcache leak after thread destruction (BZ
#22111)
- falkor-memcpy-memmove.patch: Optimized implementation of memcpy/memmove
for Qualcomm Falkor
- aarch64-cpu-features.patch: Fix glibc.tune.cpu tunable handling
- nss-files-large-buffers.patch: Avoid large buffers with many host
addresses (BZ #22078)
- sysconf-uio-maxiov.patch: Fix missing definition of UIO_MAXIOV (BZ
#22321)
- glob-tilde-overflow.patch: Fix buffer overflows with GLOB_TILDE
(CVE-2017-15670, CVE-2017-15671, CVE-2017-15804,
bsc#1064569. bsc#1064580, bsc#1064583, BZ #22320, BZ #22325, BZ #22332)
- dl-runtime-resolve-xsave.patch: Use fxsave/xsave/xsavec in
_dl_runtime_resolve (BZ #21265)
-------------------------------------------------------------------
Tue Oct 10 15:47:05 UTC 2017 - schwab@suse.de
@ -22,7 +47,7 @@ Thu Sep 28 07:57:52 UTC 2017 - schwab@suse.de
- assert-pedantic.patch: Suppress pedantic warning caused by statement
expression (BZ #21242, BZ #21972)
- math-c++-compat.patch: Add more C++ compatibility
- math-c++-compat.patch: Add more C++ compatibility (BZ #22235)
- getaddrinfo-errno.patch: Fix errno and h_errno handling in getaddrinfo
(BZ #21915, BZ #21922)
- resolv-conf-oom.patch: Fix memory handling in OOM situation during

26
glibc.spec
View File

@ -279,7 +279,7 @@ Patch1000: resolv-context-leak.patch
Patch1001: dl-runtime-resolve-opt-avx512f.patch
# PATCH-FIX-UPSTREAM Don't use IFUNC resolver for longjmp or system in libpthread (BZ #21041)
Patch1002: libpthread-compat-wrappers.patch
# PATCH-FIX-UPSTREAM Do not use __builtin_types_compatible_p in C++ mode (BZ #21930)
# PATCH-FIX-UPSTREAM Do not use __builtin_types_compatible_p in C++ mode (BZ #21930, BZ #22146, BZ #22235, BZ #22296)
Patch1003: math-c++-compat.patch
# PATCH-FIX-UPSTREAM Remove nis and compat from default NSS configs
Patch1004: remove-nss-nis-compat.patch
@ -301,6 +301,22 @@ Patch1011: nearbyint-inexact.patch
Patch1012: nss-compat.patch
# PATCH-FIX-UPSTREAM Remove reference to libnsl from nscd
Patch1013: nscd-libnsl.patch
# PATCH-FIX-UPSTREAM malloc: Fix tcache leak after thread destruction (BZ #22111)
Patch1014: malloc-tcache-leak.patch
# PATCH-FIX-UPSTREAM aarch64: Optimized implementation of memcpy/memmove for Qualcomm Falkor
Patch1015: falkor-memcpy-memmove.patch
# PATCH-FIX-UPSTREAM aarch64: Fix glibc.tune.cpu tunable handling
Patch1016: aarch64-cpu-features.patch
# PATCH-FIX-UPSTREAM nss_files: Avoid large buffers with many host addresses (BZ #22078)
Patch1017: nss-files-large-buffers.patch
# PATCH-FIX-UPSTREAM sysconf: Fix missing definition of UIO_MAXIOV on Linux (BZ #22321)
Patch1018: sysconf-uio-maxiov.patch
# PATCH-FIX-UPSTREAM glob: Fix buffer overflows (CVE-2017-15670, CVE-2017-15671, CVE-2017-15804, BZ #22320, BZ #22325, BZ #22332)
Patch1019: glob-tilde-overflow.patch
# PATCH-FIX-UPSTREAM x86-64: Use fxsave/xsave/xsavec in _dl_runtime_resolve (BZ #21265)
Patch1020: dl-runtime-resolve-xsave.patch
# PATCH-FIX-UPSTREAM posix: Fix improper assert in Linux posix_spawn (BZ #22273)
Patch1021: spawni-assert.patch
###
# Patches awaiting upstream approval
@ -525,6 +541,14 @@ rm nscd/s-stamp
%patch1011 -p1
%patch1012 -p1
%patch1013 -p1
%patch1014 -p1
%patch1015 -p1
%patch1016 -p1
%patch1017 -p1
%patch1018 -p1
%patch1019 -p1
%patch1020 -p1
%patch1021 -p1
%patch2000 -p1
%patch2001 -p1

2244
glob-tilde-overflow.patch Normal file
View File

File diff suppressed because it is too large Load Diff

179
malloc-tcache-leak.patch Normal file
View File

@ -0,0 +1,179 @@
2017-10-06 Carlos O'Donell <carlos@redhat.com>
[BZ #22111]
* malloc/malloc.c (tcache_shutting_down): Use bool type.
(tcache_thread_freeres): Set tcache_shutting_down before
freeing the tcache.
* malloc/Makefile (tests): Add tst-malloc-tcache-leak.
* malloc/tst-malloc-tcache-leak.c: New file.
Index: glibc-2.26/malloc/Makefile
===================================================================
--- glibc-2.26.orig/malloc/Makefile
+++ glibc-2.26/malloc/Makefile
@@ -34,6 +34,7 @@ tests := mallocbug tst-malloc tst-valloc
tst-interpose-nothread \
tst-interpose-thread \
tst-alloc_buffer \
+ tst-malloc-tcache-leak \
tests-static := \
tst-interpose-static-nothread \
@@ -242,3 +243,5 @@ tst-dynarray-fail-ENV = MALLOC_TRACE=$(o
$(objpfx)tst-dynarray-fail-mem.out: $(objpfx)tst-dynarray-fail.out
$(common-objpfx)malloc/mtrace $(objpfx)tst-dynarray-fail.mtrace > $@; \
$(evaluate-test)
+
+$(objpfx)tst-malloc-tcache-leak: $(shared-thread-library)
Index: glibc-2.26/malloc/malloc.c
===================================================================
--- glibc-2.26.orig/malloc/malloc.c
+++ glibc-2.26/malloc/malloc.c
@@ -2940,7 +2940,7 @@ typedef struct tcache_perthread_struct
tcache_entry *entries[TCACHE_MAX_BINS];
} tcache_perthread_struct;
-static __thread char tcache_shutting_down = 0;
+static __thread bool tcache_shutting_down = false;
static __thread tcache_perthread_struct *tcache = NULL;
/* Caller must ensure that we know tc_idx is valid and there's room
@@ -2977,8 +2977,12 @@ tcache_thread_freeres (void)
if (!tcache)
return;
+ /* Disable the tcache and prevent it from being reinitialized. */
tcache = NULL;
+ tcache_shutting_down = true;
+ /* Free all of the entries and the tcache itself back to the arena
+ heap for coalescing. */
for (i = 0; i < TCACHE_MAX_BINS; ++i)
{
while (tcache_tmp->entries[i])
@@ -2990,8 +2994,6 @@ tcache_thread_freeres (void)
}
__libc_free (tcache_tmp);
-
- tcache_shutting_down = 1;
}
text_set_element (__libc_thread_subfreeres, tcache_thread_freeres);
Index: glibc-2.26/malloc/tst-malloc-tcache-leak.c
===================================================================
--- /dev/null
+++ glibc-2.26/malloc/tst-malloc-tcache-leak.c
@@ -0,0 +1,112 @@
+/* Bug 22111: Test that threads do not leak their per thread cache.
+ Copyright (C) 2015-2017 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+/* The point of this test is to start and exit a large number of
+ threads, while at the same time looking to see if the used
+ memory grows with each round of threads run. If the memory
+ grows above some linear bound we declare the test failed and
+ that the malloc implementation is leaking memory with each
+ thread. This is a good indicator that the thread local cache
+ is leaking chunks. */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <malloc.h>
+#include <pthread.h>
+#include <assert.h>
+
+#include <support/check.h>
+#include <support/support.h>
+#include <support/xthread.h>
+
+void *
+worker (void *data)
+{
+ void *ret;
+ /* Allocate an arbitrary amount of memory that is known to fit into
+ the thread local cache (tcache). If we have at least 64 bins
+ (default e.g. TCACHE_MAX_BINS) we should be able to allocate 32
+ bytes and force malloc to fill the tcache. We are assuming tcahce
+ init happens at the first small alloc, but it might in the future
+ be deferred to some other point. Therefore to future proof this
+ test we include a full alloc/free/alloc cycle for the thread. We
+ need a compiler barrier to avoid the removal of the useless
+ alloc/free. We send some memory back to main to have the memory
+ freed after the thread dies, as just another check that the chunks
+ that were previously in the tcache are still OK to free after
+ thread death. */
+ ret = xmalloc (32);
+ __asm__ volatile ("" ::: "memory");
+ free (ret);
+ return (void *) xmalloc (32);
+}
+
+static int
+do_test (void)
+{
+ pthread_t *thread;
+ struct mallinfo info_before, info_after;
+ void *retval;
+
+ /* This is an arbitrary choice. We choose a total of THREADS
+ threads created and joined. This gives us enough iterations to
+ show a leak. */
+ int threads = 100000;
+
+ /* Avoid there being 0 malloc'd data at this point by allocating the
+ pthread_t required to run the test. */
+ thread = (pthread_t *) xcalloc (1, sizeof (pthread_t));
+
+ info_before = mallinfo ();
+
+ assert (info_before.uordblks != 0);
+
+ printf ("INFO: %d (bytes) are in use before starting threads.\n",
+ info_before.uordblks);
+
+ for (int loop = 0; loop < threads; loop++)
+ {
+ *thread = xpthread_create (NULL, worker, NULL);
+ retval = xpthread_join (*thread);
+ free (retval);
+ }
+
+ info_after = mallinfo ();
+ printf ("INFO: %d (bytes) are in use after all threads joined.\n",
+ info_after.uordblks);
+
+ /* We need to compare the memory in use before and the memory in use
+ after starting and joining THREADS threads. We almost always grow
+ memory slightly, but not much. Consider that if even 1-byte leaked
+ per thread we'd have THREADS bytes of additional memory, and in
+ general the in-use at the start of main is quite low. We will
+ always leak a full malloc chunk, and never just 1-byte, therefore
+ anything above "+ threads" from the start (constant offset) is a
+ leak. Obviously this assumes no thread-related malloc'd internal
+ libc data structures persist beyond the thread death, and any that
+ did would limit the number of times you could call pthread_create,
+ which is a QoI we'd want to detect and fix. */
+ if (info_after.uordblks > (info_before.uordblks + threads))
+ FAIL_EXIT1 ("Memory usage after threads is too high.\n");
+
+ /* Did not detect excessive memory usage. */
+ free (thread);
+ exit (0);
+}
+
+#include <support/test-driver.c>

93
math-c++-compat.patch
View File

@ -1,3 +1,18 @@
2017-10-17 Romain Naour <romain.naour@gmail.com> (tiny change)
[BZ #22296]
* math/math.h: Let signbit use the builtin in C++ mode with gcc
< 6.x
2017-10-03 Gabriel F. T. Gomes <gabriel@inconstante.eti.br>
[BZ #22235]
* sysdeps/ieee754/ldbl-96/bits/iscanonical.h (iscanonical):
Provide a C++ implementation based on function overloading,
rather than using __MATH_TG, which uses C-only builtins.
* sysdeps/ieee754/ldbl-128ibm/bits/iscanonical.h (iscanonical):
Likewise.
2017-09-22 Gabriel F. T. Gomes <gabriel@inconstante.eti.br>
[BZ #22146]
@ -56,7 +71,23 @@ Index: glibc-2.26/math/math.h
# define fpclassify(x) __builtin_fpclassify (FP_NAN, FP_INFINITE, \
FP_NORMAL, FP_SUBNORMAL, FP_ZERO, x)
# else
@@ -442,8 +448,12 @@ enum
@@ -412,6 +418,15 @@ enum
/* Return nonzero value if sign of X is negative. */
# if __GNUC_PREREQ (6,0)
# define signbit(x) __builtin_signbit (x)
+# elif defined __cplusplus
+ /* In C++ mode, __MATH_TG cannot be used, because it relies on
+ __builtin_types_compatible_p, which is a C-only builtin.
+ The check for __cplusplus allows the use of the builtin instead of
+ __MATH_TG. This is provided for libstdc++, only to let its configure
+ test work. No further use of this definition of signbit is expected
+ in C++ mode, since libstdc++ provides its own version of signbit
+ in cmath (which undefines signbit). */
+# define signbit(x) __builtin_signbitl (x)
# elif __GNUC_PREREQ (4,0)
# define signbit(x) __MATH_TG ((x), __builtin_signbit, (x))
# else
@@ -442,8 +457,12 @@ enum
/* Return nonzero value if X is positive or negative infinity. */
# if __HAVE_DISTINCT_FLOAT128 && !__GNUC_PREREQ (7,0) \
@ -71,7 +102,7 @@ Index: glibc-2.26/math/math.h
# define isinf(x) \
(__builtin_types_compatible_p (__typeof (x), _Float128) \
? __isinff128 (x) : __builtin_isinf_sign (x))
@@ -470,7 +480,32 @@ enum
@@ -470,7 +489,32 @@ enum
# include <bits/iscanonical.h>
/* Return nonzero value if X is a signaling NaN. */
@ -105,7 +136,7 @@ Index: glibc-2.26/math/math.h
/* Return nonzero value if X is subnormal. */
# define issubnormal(x) (fpclassify (x) == FP_SUBNORMAL)
@@ -484,15 +519,40 @@ enum
@@ -484,15 +528,40 @@ enum
# endif
# else /* __cplusplus */
extern "C++" {
@ -182,3 +213,59 @@ Index: glibc-2.26/misc/sys/cdefs.h
# define __HAVE_GENERIC_SELECTION 1
#else
# define __HAVE_GENERIC_SELECTION 0
Index: glibc-2.26/sysdeps/ieee754/ldbl-128ibm/bits/iscanonical.h
===================================================================
--- glibc-2.26.orig/sysdeps/ieee754/ldbl-128ibm/bits/iscanonical.h
+++ glibc-2.26/sysdeps/ieee754/ldbl-128ibm/bits/iscanonical.h
@@ -37,5 +37,22 @@ extern int __iscanonicall (long double _
conversion, before being discarded; in IBM long double, there are
encodings that are not consistently handled as corresponding to any
particular value of the type, and we return 0 for those. */
-# define iscanonical(x) __MATH_TG ((x), __iscanonical, (x))
-#endif
+# ifndef __cplusplus
+# define iscanonical(x) __MATH_TG ((x), __iscanonical, (x))
+# else
+/* In C++ mode, __MATH_TG cannot be used, because it relies on
+ __builtin_types_compatible_p, which is a C-only builtin. On the
+ other hand, overloading provides the means to distinguish between
+ the floating-point types. The overloading resolution will match
+ the correct parameter (regardless of type qualifiers (i.e.: const
+ and volatile)). */
+extern "C++" {
+inline int iscanonical (float __val) { return __iscanonicalf (__val); }
+inline int iscanonical (double __val) { return __iscanonical (__val); }
+inline int iscanonical (long double __val) { return __iscanonicall (__val); }
+# if __HAVE_DISTINCT_FLOAT128
+inline int iscanonical (_Float128 __val) { return __iscanonicalf128 (__val); }
+# endif
+}
+# endif /* __cplusplus */
+#endif /* __NO_LONG_DOUBLE_MATH */
Index: glibc-2.26/sysdeps/ieee754/ldbl-96/bits/iscanonical.h
===================================================================
--- glibc-2.26.orig/sysdeps/ieee754/ldbl-96/bits/iscanonical.h
+++ glibc-2.26/sysdeps/ieee754/ldbl-96/bits/iscanonical.h
@@ -34,4 +34,21 @@ extern int __iscanonicall (long double _
conversion, before being discarded; in extended precision, there
are encodings that are not consistently handled as corresponding to
any particular value of the type, and we return 0 for those. */
-#define iscanonical(x) __MATH_TG ((x), __iscanonical, (x))
+#ifndef __cplusplus
+# define iscanonical(x) __MATH_TG ((x), __iscanonical, (x))
+#else
+/* In C++ mode, __MATH_TG cannot be used, because it relies on
+ __builtin_types_compatible_p, which is a C-only builtin. On the
+ other hand, overloading provides the means to distinguish between
+ the floating-point types. The overloading resolution will match
+ the correct parameter (regardless of type qualifiers (i.e.: const
+ and volatile)). */
+extern "C++" {
+inline int iscanonical (float __val) { return __iscanonicalf (__val); }
+inline int iscanonical (double __val) { return __iscanonical (__val); }
+inline int iscanonical (long double __val) { return __iscanonicall (__val); }
+# if __HAVE_DISTINCT_FLOAT128
+inline int iscanonical (_Float128 __val) { return __iscanonicalf128 (__val); }
+# endif
+}
+#endif /* __cplusplus */

831
nss-files-large-buffers.patch Normal file
View File

@ -0,0 +1,831 @@
2017-10-11 Florian Weimer <fweimer@redhat.com>
[BZ #22078]
Avoid large NSS buffers with many addresses, aliases.
* nss/nss_files/files-hosts.c (gethostbyname3_multi): Rewrite
using dynarrays and struct alloc_buffer.
* nss/Makefile (tests): Add tst-nss-files-hosts-multi.
(tst-nss-files-hosts-multi): Link with -ldl.
* nss/tst-nss-files-hosts-multi.c: New file.
2017-10-11 Florian Weimer <fweimer@redhat.com>
[BZ #18023]
* nss/nss_files/files-hosts.c (gethostbyname3_multi): Use struct
scratch_buffer. Eliminate gotos.
2017-10-10 Florian Weimer <fweimer@redhat.com>
* nss/nss_files/files-hosts.c (gethostbyname3_multi): New
function.
(_nss_files_gethostbyname3_r): Call it.
Index: glibc-2.26/nss/Makefile
===================================================================
--- glibc-2.26.orig/nss/Makefile
+++ glibc-2.26/nss/Makefile
@@ -61,6 +61,7 @@ xtests = bug-erange
# Tests which need libdl
ifeq (yes,$(build-shared))
tests += tst-nss-files-hosts-erange
+tests += tst-nss-files-hosts-multi
endif
# If we have a thread library then we can test cancellation against
@@ -165,3 +166,4 @@ $(objpfx)tst-cancel-getpwuid_r: $(shared
endif
$(objpfx)tst-nss-files-hosts-erange: $(libdl)
+$(objpfx)tst-nss-files-hosts-multi: $(libdl)
Index: glibc-2.26/nss/nss_files/files-hosts.c
===================================================================
--- glibc-2.26.orig/nss/nss_files/files-hosts.c
+++ glibc-2.26/nss/nss_files/files-hosts.c
@@ -22,6 +22,8 @@
#include <arpa/nameser.h>
#include <netdb.h>
#include <resolv/resolv-internal.h>
+#include <scratch_buffer.h>
+#include <alloc_buffer.h>
/* Get implementation for some internal functions. */
@@ -110,228 +112,250 @@ DB_LOOKUP (hostbyaddr, ,,,
}, const void *addr, socklen_t len, int af)
#undef EXTRA_ARGS_VALUE
-enum nss_status
-_nss_files_gethostbyname3_r (const char *name, int af, struct hostent *result,
- char *buffer, size_t buflen, int *errnop,
- int *herrnop, int32_t *ttlp, char **canonp)
+/* Type of the address and alias arrays. */
+#define DYNARRAY_STRUCT array
+#define DYNARRAY_ELEMENT char *
+#define DYNARRAY_PREFIX array_
+#include <malloc/dynarray-skeleton.c>
+
+static enum nss_status
+gethostbyname3_multi (FILE * stream, const char *name, int af,
+ struct hostent *result, char *buffer, size_t buflen,
+ int *errnop, int *herrnop, int flags)
{
- FILE *stream = NULL;
- uintptr_t pad = -(uintptr_t) buffer % __alignof__ (struct hostent_data);
- buffer += pad;
- buflen = buflen > pad ? buflen - pad : 0;
+ assert (af == AF_INET || af == AF_INET6);
- /* Open file. */
- enum nss_status status = internal_setent (&stream);
+ /* We have to get all host entries from the file. */
+ struct scratch_buffer tmp_buffer;
+ scratch_buffer_init (&tmp_buffer);
+ struct hostent tmp_result_buf;
+ struct array addresses;
+ array_init (&addresses);
+ struct array aliases;
+ array_init (&aliases);
+ enum nss_status status;
+
+ /* Preserve the addresses and aliases encountered so far. */
+ for (size_t i = 0; result->h_addr_list[i] != NULL; ++i)
+ array_add (&addresses, result->h_addr_list[i]);
+ for (size_t i = 0; result->h_aliases[i] != NULL; ++i)
+ array_add (&aliases, result->h_aliases[i]);
+
+ /* The output buffer re-uses now-unused space at the end of the
+ buffer, starting with the aliases array. It comes last in the
+ data produced by internal_getent. (The alias names themselves
+ are still located in the line read in internal_getent, which is
+ stored at the beginning of the buffer.) */
+ struct alloc_buffer outbuf;
+ {
+ char *bufferend = (char *) result->h_aliases;
+ outbuf = alloc_buffer_create (bufferend, buffer + buflen - bufferend);
+ }
- if (status == NSS_STATUS_SUCCESS)
+ while (true)
{
- /* XXX Is using _res to determine whether we want to convert IPv4
- addresses to IPv6 addresses really the right thing to do? */
- int flags = (res_use_inet6 () ? AI_V4MAPPED : 0);
-
- while ((status = internal_getent (stream, result, buffer, buflen, errnop,
- herrnop, af, flags))
- == NSS_STATUS_SUCCESS)
+ status = internal_getent (stream, &tmp_result_buf, tmp_buffer.data,
+ tmp_buffer.length, errnop, herrnop, af,
+ flags);
+ /* Enlarge the buffer if necessary. */
+ if (status == NSS_STATUS_TRYAGAIN && *herrnop == NETDB_INTERNAL
+ && *errnop == ERANGE)
{
- LOOKUP_NAME_CASE (h_name, h_aliases)
+ if (!scratch_buffer_grow (&tmp_buffer))
+ {
+ *errnop = ENOMEM;
+ /* *herrnop and status already have the right value. */
+ break;
+ }
+ /* Loop around and retry with a larger buffer. */
}
-
- if (status == NSS_STATUS_SUCCESS
- && _res_hconf.flags & HCONF_FLAG_MULTI)
+ else if (status == NSS_STATUS_SUCCESS)
{
- /* We have to get all host entries from the file. */
- size_t tmp_buflen = MIN (buflen, 4096);
- char tmp_buffer_stack[tmp_buflen]
- __attribute__ ((__aligned__ (__alignof__ (struct hostent_data))));
- char *tmp_buffer = tmp_buffer_stack;
- struct hostent tmp_result_buf;
- int naddrs = 1;
- int naliases = 0;
- char *bufferend;
- bool tmp_buffer_malloced = false;
-
- while (result->h_aliases[naliases] != NULL)
- ++naliases;
-
- bufferend = (char *) &result->h_aliases[naliases + 1];
-
- again:
- while ((status = internal_getent (stream, &tmp_result_buf, tmp_buffer,
- tmp_buflen, errnop, herrnop, af,
- flags))
- == NSS_STATUS_SUCCESS)
+ /* A line was read. Check that it matches the search
+ criteria. */
+
+ int matches = 1;
+ struct hostent *old_result = result;
+ result = &tmp_result_buf;
+ /* The following piece is a bit clumsy but we want to use
+ the `LOOKUP_NAME_CASE' value. The optimizer should do
+ its job. */
+ do
{
- int matches = 1;
- struct hostent *old_result = result;
- result = &tmp_result_buf;
- /* The following piece is a bit clumsy but we want to use the
- `LOOKUP_NAME_CASE' value. The optimizer should do its
- job. */
- do
- {
- LOOKUP_NAME_CASE (h_name, h_aliases)
- result = old_result;
- }
- while ((matches = 0));
+ LOOKUP_NAME_CASE (h_name, h_aliases)
+ result = old_result;
+ }
+ while ((matches = 0));
- if (matches)
+ /* If the line matches, we need to copy the addresses and
+ aliases, so that we can reuse tmp_buffer for the next
+ line. */
+ if (matches)
+ {
+ /* Record the addresses. */
+ for (size_t i = 0; tmp_result_buf.h_addr_list[i] != NULL; ++i)
{
- /* We could be very clever and try to recycle a few bytes
- in the buffer instead of generating new arrays. But
- we are not doing this here since it's more work than
- it's worth. Simply let the user provide a bit bigger
- buffer. */
- char **new_h_addr_list;
- char **new_h_aliases;
- int newaliases = 0;
- size_t newstrlen = 0;
- int cnt;
-
- /* Count the new aliases and the length of the strings. */
- while (tmp_result_buf.h_aliases[newaliases] != NULL)
+ /* Allocate the target space in the output buffer,
+ depending on the address family. */
+ void *target;
+ if (af == AF_INET)
{
- char *cp = tmp_result_buf.h_aliases[newaliases];
- ++newaliases;
- newstrlen += strlen (cp) + 1;
+ assert (tmp_result_buf.h_length == 4);
+ target = alloc_buffer_alloc (&outbuf, struct in_addr);
}
- /* If the real name is different add it also to the
- aliases. This means that there is a duplication
- in the alias list but this is really the user's
- problem. */
- if (strcmp (old_result->h_name,
- tmp_result_buf.h_name) != 0)
+ else if (af == AF_INET6)
{
- ++newaliases;
- newstrlen += strlen (tmp_result_buf.h_name) + 1;
+ assert (tmp_result_buf.h_length == 16);
+ target = alloc_buffer_alloc (&outbuf, struct in6_addr);
}
+ else
+ __builtin_unreachable ();
- /* Make sure bufferend is aligned. */
- assert ((bufferend - (char *) 0) % sizeof (char *) == 0);
-
- /* Now we can check whether the buffer is large enough.
- 16 is the maximal size of the IP address. */
- if (bufferend + 16 + (naddrs + 2) * sizeof (char *)
- + roundup (newstrlen, sizeof (char *))
- + (naliases + newaliases + 1) * sizeof (char *)
- >= buffer + buflen)
+ if (target == NULL)
{
+ /* Request a larger output buffer. */
*errnop = ERANGE;
*herrnop = NETDB_INTERNAL;
status = NSS_STATUS_TRYAGAIN;
- goto out;
- }
-
- new_h_addr_list =
- (char **) (bufferend
- + roundup (newstrlen, sizeof (char *))
- + 16);
- new_h_aliases =
- (char **) ((char *) new_h_addr_list
- + (naddrs + 2) * sizeof (char *));
-
- /* Copy the old data in the new arrays. */
- for (cnt = 0; cnt < naddrs; ++cnt)
- new_h_addr_list[cnt] = old_result->h_addr_list[cnt];
-
- for (cnt = 0; cnt < naliases; ++cnt)
- new_h_aliases[cnt] = old_result->h_aliases[cnt];
-
- /* Store the new strings. */
- cnt = 0;
- while (tmp_result_buf.h_aliases[cnt] != NULL)
- {
- new_h_aliases[naliases++] = bufferend;
- bufferend = (__stpcpy (bufferend,
- tmp_result_buf.h_aliases[cnt])
- + 1);
- ++cnt;
+ break;
}
-
- if (cnt < newaliases)
- {
- new_h_aliases[naliases++] = bufferend;
- bufferend = __stpcpy (bufferend,
- tmp_result_buf.h_name) + 1;
- }
-
- /* Final NULL pointer. */
- new_h_aliases[naliases] = NULL;
-
- /* Round up the buffer end address. */
- bufferend += (sizeof (char *)
- - ((bufferend - (char *) 0)
- % sizeof (char *))) % sizeof (char *);
-
- /* Now the new address. */
- new_h_addr_list[naddrs++] =
- memcpy (bufferend, tmp_result_buf.h_addr,
- tmp_result_buf.h_length);
-
- /* Also here a final NULL pointer. */
- new_h_addr_list[naddrs] = NULL;
-
- /* Store the new array pointers. */
- old_result->h_aliases = new_h_aliases;
- old_result->h_addr_list = new_h_addr_list;
-
- /* Compute the new buffer end. */
- bufferend = (char *) &new_h_aliases[naliases + 1];
- assert (bufferend <= buffer + buflen);
-
- result = old_result;
+ memcpy (target, tmp_result_buf.h_addr_list[i],
+ tmp_result_buf.h_length);
+ array_add (&addresses, target);
}
- }
- if (status == NSS_STATUS_TRYAGAIN)
- {
- size_t newsize = 2 * tmp_buflen;
- if (tmp_buffer_malloced)
+ /* Record the aliases. */
+ for (size_t i = 0; tmp_result_buf.h_aliases[i] != NULL; ++i)
{
- char *newp = realloc (tmp_buffer, newsize);
- if (newp != NULL)
- {
- assert ((((uintptr_t) newp)
- & (__alignof__ (struct hostent_data) - 1))
- == 0);
- tmp_buffer = newp;
- tmp_buflen = newsize;
- goto again;
- }
+ char *alias = tmp_result_buf.h_aliases[i];
+ array_add (&aliases,
+ alloc_buffer_copy_string (&outbuf, alias));
}
- else if (!__libc_use_alloca (buflen + newsize))
+
+ /* If the real name is different add, it also to the
+ aliases. This means that there is a duplication in
+ the alias list but this is really the user's
+ problem. */
+ {
+ char *new_name = tmp_result_buf.h_name;
+ if (strcmp (old_result->h_name, new_name) != 0)
+ array_add (&aliases,
+ alloc_buffer_copy_string (&outbuf, new_name));
+ }
+
+ /* Report memory allocation failures during the
+ expansion of the temporary arrays. */
+ if (array_has_failed (&addresses) || array_has_failed (&aliases))
{
- tmp_buffer = malloc (newsize);
- if (tmp_buffer != NULL)
- {
- assert ((((uintptr_t) tmp_buffer)
- & (__alignof__ (struct hostent_data) - 1))
- == 0);
- tmp_buffer_malloced = true;
- tmp_buflen = newsize;
- goto again;
- }
+ *errnop = ENOMEM;
+ *herrnop = NETDB_INTERNAL;
+ status = NSS_STATUS_UNAVAIL;
+ break;
}
- else
+
+ /* Request a larger output buffer if we ran out of room. */
+ if (alloc_buffer_has_failed (&outbuf))
{
- tmp_buffer
- = extend_alloca (tmp_buffer, tmp_buflen,
- newsize
- + __alignof__ (struct hostent_data));
- tmp_buffer = (char *) (((uintptr_t) tmp_buffer
- + __alignof__ (struct hostent_data)
- - 1)
- & ~(__alignof__ (struct hostent_data)
- - 1));
- goto again;
+ *errnop = ERANGE;
+ *herrnop = NETDB_INTERNAL;
+ status = NSS_STATUS_TRYAGAIN;
+ break;
}
- }
- else
- status = NSS_STATUS_SUCCESS;
- out:
- if (tmp_buffer_malloced)
- free (tmp_buffer);
+
+ result = old_result;
+ } /* If match was found. */
+
+ /* If no match is found, loop around and fetch another
+ line. */
+
+ } /* status == NSS_STATUS_SUCCESS. */
+ else
+ /* internal_getent returned an error. */
+ break;
+ } /* while (true) */
+
+ /* Propagate the NSS_STATUS_TRYAGAIN error to the caller. It means
+ that we may not have loaded the complete result.
+ NSS_STATUS_NOTFOUND, however, means that we reached the end of
+ the file successfully. */
+ if (status != NSS_STATUS_TRYAGAIN)
+ status = NSS_STATUS_SUCCESS;
+
+ if (status == NSS_STATUS_SUCCESS)
+ {
+ /* Copy the address and alias arrays into the output buffer and
+ add NULL terminators. The pointed-to elements were directly
+ written into the output buffer above and do not need to be
+ copied again. */
+ size_t addresses_count = array_size (&addresses);
+ size_t aliases_count = array_size (&aliases);
+ char **out_addresses = alloc_buffer_alloc_array
+ (&outbuf, char *, addresses_count + 1);
+ char **out_aliases = alloc_buffer_alloc_array
+ (&outbuf, char *, aliases_count + 1);
+ if (out_addresses == NULL || out_aliases == NULL)
+ {
+ /* The output buffer is not large enough. */
+ *errnop = ERANGE;
+ *herrnop = NETDB_INTERNAL;
+ status = NSS_STATUS_TRYAGAIN;
+ /* Fall through to function exit. */
+ }
+ else
+ {
+ /* Everything is allocated in place. Make the copies and
+ adjust the array pointers. */
+ memcpy (out_addresses, array_begin (&addresses),
+ addresses_count * sizeof (char *));
+ out_addresses[addresses_count] = NULL;
+ memcpy (out_aliases, array_begin (&aliases),
+ aliases_count * sizeof (char *));
+ out_aliases[aliases_count] = NULL;
+
+ result->h_addr_list = out_addresses;
+ result->h_aliases = out_aliases;
+
+ status = NSS_STATUS_SUCCESS;
+ }
+ }
+
+ scratch_buffer_free (&tmp_buffer);
+ array_free (&addresses);
+ array_free (&aliases);
+ return status;
+}
+
+enum nss_status
+_nss_files_gethostbyname3_r (const char *name, int af, struct hostent *result,
+ char *buffer, size_t buflen, int *errnop,
+ int *herrnop, int32_t *ttlp, char **canonp)
+{
+ FILE *stream = NULL;
+ uintptr_t pad = -(uintptr_t) buffer % __alignof__ (struct hostent_data);
+ buffer += pad;
+ buflen = buflen > pad ? buflen - pad : 0;
+
+ /* Open file. */
+ enum nss_status status = internal_setent (&stream);
+
+ if (status == NSS_STATUS_SUCCESS)
+ {
+ /* XXX Is using _res to determine whether we want to convert IPv4
+ addresses to IPv6 addresses really the right thing to do? */
+ int flags = (res_use_inet6 () ? AI_V4MAPPED : 0);
+
+ while ((status = internal_getent (stream, result, buffer, buflen, errnop,
+ herrnop, af, flags))
+ == NSS_STATUS_SUCCESS)
+ {
+ LOOKUP_NAME_CASE (h_name, h_aliases)
}
+ if (status == NSS_STATUS_SUCCESS
+ && _res_hconf.flags & HCONF_FLAG_MULTI)
+ status = gethostbyname3_multi
+ (stream, name, af, result, buffer, buflen, errnop, herrnop, flags);
+
internal_endent (&stream);
}
Index: glibc-2.26/nss/tst-nss-files-hosts-multi.c
===================================================================
--- /dev/null
+++ glibc-2.26/nss/tst-nss-files-hosts-multi.c
@@ -0,0 +1,331 @@
+/* Parse /etc/hosts in multi mode with many addresses/aliases.
+ Copyright (C) 2017 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <dlfcn.h>
+#include <errno.h>
+#include <gnu/lib-names.h>
+#include <netdb.h>
+#include <nss.h>
+#include <stdbool.h>
+#include <stdlib.h>
+#include <string.h>
+#include <support/check.h>
+#include <support/check_nss.h>
+#include <support/namespace.h>
+#include <support/support.h>
+#include <support/test-driver.h>
+#include <support/test-driver.h>
+#include <support/xmemstream.h>
+#include <support/xstdio.h>
+#include <support/xunistd.h>
+#include <sys/resource.h>
+
+struct support_chroot *chroot_env;
+
+static void
+prepare (int argc, char **argv)
+{
+ chroot_env = support_chroot_create
+ ((struct support_chroot_configuration)
+ {
+ .resolv_conf = "",
+ .hosts = "", /* See write_hosts below. */
+ .host_conf = "multi on\n",
+ });
+}
+
+/* Create the /etc/hosts file from outside the chroot. */
+static void
+write_hosts (int count)
+{
+ TEST_VERIFY (count > 0 && count <= 65535);
+ FILE *fp = xfopen (chroot_env->path_hosts, "w");
+ fputs ("127.0.0.1 localhost localhost.localdomain\n"
+ "::1 localhost localhost.localdomain\n",
+ fp);
+ for (int i = 0; i < count; ++i)
+ {
+ fprintf (fp, "10.4.%d.%d www4.example.com\n",
+ (i / 256) & 0xff, i & 0xff);
+ fprintf (fp, "10.46.%d.%d www.example.com\n",
+ (i / 256) & 0xff, i & 0xff);
+ fprintf (fp, "192.0.2.1 alias.example.com v4-%d.example.com\n", i);
+ fprintf (fp, "2001:db8::6:%x www6.example.com\n", i);
+ fprintf (fp, "2001:db8::46:%x www.example.com\n", i);
+ fprintf (fp, "2001:db8::1 alias.example.com v6-%d.example.com\n", i);
+ }
+ xfclose (fp);
+}
+
+/* Parameters of a single test. */
+struct test_params
+{
+ const char *name; /* Name to query. */
+ const char *marker; /* Address marker for the name. */
+ int count; /* Number of addresses/aliases. */
+ int family; /* AF_INET, AF_INET_6 or AF_UNSPEC. */
+ bool canonname; /* True if AI_CANONNAME should be enabled. */
+};
+
+/* Expected result of gethostbyname/gethostbyname2. */
+static char *
+expected_ghbn (const struct test_params *params)
+{
+ TEST_VERIFY (params->family == AF_INET || params->family == AF_INET6);
+
+ struct xmemstream expected;
+ xopen_memstream (&expected);
+ if (strcmp (params->name, "alias.example.com") == 0)
+ {
+ fprintf (expected.out, "name: %s\n", params->name);
+ char af;
+ if (params->family == AF_INET)
+ af = '4';
+ else
+ af = '6';
+ for (int i = 0; i < params->count; ++i)
+ fprintf (expected.out, "alias: v%c-%d.example.com\n", af, i);
+
+ for (int i = 0; i < params->count; ++i)
+ if (params->family == AF_INET)
+ fputs ("address: 192.0.2.1\n", expected.out);
+ else
+ fputs ("address: 2001:db8::1\n", expected.out);
+ }
+ else /* www/www4/www6 name. */
+ {
+ bool do_ipv4 = params->family == AF_INET
+ && strncmp (params->name, "www6", 4) != 0;
+ bool do_ipv6 = params->family == AF_INET6
+ && strncmp (params->name, "www4", 4) != 0;
+ if (do_ipv4 || do_ipv6)
+ {
+ fprintf (expected.out, "name: %s\n", params->name);
+ if (do_ipv4)
+ for (int i = 0; i < params->count; ++i)
+ fprintf (expected.out, "address: 10.%s.%d.%d\n",
+ params->marker, i / 256, i % 256);
+ if (do_ipv6)
+ for (int i = 0; i < params->count; ++i)
+ fprintf (expected.out, "address: 2001:db8::%s:%x\n",
+ params->marker, i);
+ }
+ else
+ fputs ("error: HOST_NOT_FOUND\n", expected.out);
+ }
+ xfclose_memstream (&expected);
+ return expected.buffer;
+}
+
+/* Expected result of getaddrinfo. */
+static char *
+expected_gai (const struct test_params *params)
+{
+ bool do_ipv4 = false;
+ bool do_ipv6 = false;
+ if (params->family == AF_UNSPEC)
+ do_ipv4 = do_ipv6 = true;
+ else if (params->family == AF_INET)
+ do_ipv4 = true;
+ else if (params->family == AF_INET6)
+ do_ipv6 = true;
+
+ struct xmemstream expected;
+ xopen_memstream (&expected);
+ if (strcmp (params->name, "alias.example.com") == 0)
+ {
+ if (params->canonname)
+ fprintf (expected.out,
+ "flags: AI_CANONNAME\n"
+ "canonname: %s\n",
+ params->name);
+
+ if (do_ipv4)
+ for (int i = 0; i < params->count; ++i)
+ fputs ("address: STREAM/TCP 192.0.2.1 80\n", expected.out);
+ if (do_ipv6)
+ for (int i = 0; i < params->count; ++i)
+ fputs ("address: STREAM/TCP 2001:db8::1 80\n", expected.out);
+ }
+ else /* www/www4/www6 name. */
+ {
+ if (strncmp (params->name, "www4", 4) == 0)
+ do_ipv6 = false;
+ else if (strncmp (params->name, "www6", 4) == 0)
+ do_ipv4 = false;
+ /* Otherwise, we have www as the name, so we do both. */
+
+ if (do_ipv4 || do_ipv6)
+ {
+ if (params->canonname)
+ fprintf (expected.out,
+ "flags: AI_CANONNAME\n"
+ "canonname: %s\n",
+ params->name);
+
+ if (do_ipv4)
+ for (int i = 0; i < params->count; ++i)
+ fprintf (expected.out, "address: STREAM/TCP 10.%s.%d.%d 80\n",
+ params->marker, i / 256, i % 256);
+ if (do_ipv6)
+ for (int i = 0; i < params->count; ++i)
+ fprintf (expected.out,
+ "address: STREAM/TCP 2001:db8::%s:%x 80\n",
+ params->marker, i);
+ }
+ else
+ fputs ("error: Name or service not known\n", expected.out);
+ }
+ xfclose_memstream (&expected);
+ return expected.buffer;
+}
+
+static void
+run_gbhn_gai (struct test_params *params)
+{
+ char *ctx = xasprintf ("name=%s marker=%s count=%d family=%d",
+ params->name, params->marker, params->count,
+ params->family);
+ if (test_verbose > 0)
+ printf ("info: %s\n", ctx);
+
+ /* Check gethostbyname, gethostbyname2. */
+ if (params->family == AF_INET)
+ {
+ char *expected = expected_ghbn (params);
+ check_hostent (ctx, gethostbyname (params->name), expected);
+ free (expected);
+ }
+ if (params->family != AF_UNSPEC)
+ {
+ char *expected = expected_ghbn (params);
+ check_hostent (ctx, gethostbyname2 (params->name, params->family),
+ expected);
+ free (expected);
+ }
+
+ /* Check getaddrinfo. */
+ for (int do_canonical = 0; do_canonical < 2; ++do_canonical)
+ {
+ params->canonname = do_canonical;
+ char *expected = expected_gai (params);
+ struct addrinfo hints =
+ {
+ .ai_family = params->family,
+ .ai_socktype = SOCK_STREAM,
+ .ai_protocol = IPPROTO_TCP,
+ };
+ if (do_canonical)
+ hints.ai_flags |= AI_CANONNAME;
+ struct addrinfo *ai;
+ int ret = getaddrinfo (params->name, "80", &hints, &ai);
+ check_addrinfo (ctx, ai, ret, expected);
+ if (ret == 0)
+ freeaddrinfo (ai);
+ free (expected);
+ }
+
+ free (ctx);
+}
+
+/* Callback for the subprocess which runs the test in a chroot. */
+static void
+subprocess (void *closure)
+{
+ struct test_params *params = closure;
+
+ xchroot (chroot_env->path_chroot);
+
+ static const int families[] = { AF_INET, AF_INET6, AF_UNSPEC, -1 };
+ static const char *const names[] =
+ {
+ "www.example.com", "www4.example.com", "www6.example.com",
+ "alias.example.com",
+ NULL
+ };
+ static const char *const names_marker[] = { "46", "4", "6", "" };
+
+ for (int family_idx = 0; families[family_idx] >= 0; ++family_idx)
+ {
+ params->family = families[family_idx];
+ for (int names_idx = 0; names[names_idx] != NULL; ++names_idx)
+ {
+ params->name = names[names_idx];
+ params->marker = names_marker[names_idx];
+ run_gbhn_gai (params);
+ }
+ }
+}
+
+/* Run the test for a specific number of addresses/aliases. */
+static void
+run_test (int count)
+{
+ write_hosts (count);
+
+ struct test_params params =
+ {
+ .count = count,
+ };
+
+ support_isolate_in_subprocess (subprocess, &params);
+}
+
+static int
+do_test (void)
+{
+ support_become_root ();
+ if (!support_can_chroot ())
+ return EXIT_UNSUPPORTED;
+
+ /* This test should not use gigabytes of memory. */
+ {
+ struct rlimit limit;
+ if (getrlimit (RLIMIT_AS, &limit) != 0)
+ {
+ printf ("getrlimit (RLIMIT_AS) failed: %m\n");
+ return 1;
+ }
+ long target = 200 * 1024 * 1024;
+ if (limit.rlim_cur == RLIM_INFINITY || limit.rlim_cur > target)
+ {
+ limit.rlim_cur = target;
+ if (setrlimit (RLIMIT_AS, &limit) != 0)
+ {
+ printf ("setrlimit (RLIMIT_AS) failed: %m\n");
+ return 1;
+ }
+ }
+ }
+
+ __nss_configure_lookup ("hosts", "files");
+ if (dlopen (LIBNSS_FILES_SO, RTLD_LAZY) == NULL)
+ FAIL_EXIT1 ("could not load " LIBNSS_DNS_SO ": %s", dlerror ());
+
+ /* Run the tests with a few different address/alias counts. */
+ for (int count = 1; count <= 111; ++count)
+ run_test (count);
+ run_test (1111);
+ run_test (22222);
+
+ support_chroot_free (chroot_env);
+ return 0;
+}
+
+#define PREPARE prepare
+#include <support/test-driver.c>

71
spawni-assert.patch Normal file
View File

@ -0,0 +1,71 @@
2017-10-23 Adhemerval Zanella <adhemerval.zanella@linaro.org>
* sysdeps/unix/sysv/linux/spawni.c (__spawnix): Use 0 instead of
WNOHANG in waitpid call.
2017-10-20 Adhemerval Zanella <adhemerval.zanella@linaro.org>
[BZ #22273]
* sysdeps/unix/sysv/linux/spawni.c (__spawnix): Handle the case where
the auxiliary process is terminated by a signal before calling _exit
or execve.
Index: glibc-2.26/sysdeps/unix/sysv/linux/spawni.c
===================================================================
--- glibc-2.26.orig/sysdeps/unix/sysv/linux/spawni.c
+++ glibc-2.26/sysdeps/unix/sysv/linux/spawni.c
@@ -17,7 +17,6 @@
<http://www.gnu.org/licenses/>. */
#include <spawn.h>
-#include <assert.h>
#include <fcntl.h>
#include <paths.h>
#include <string.h>
@@ -268,7 +267,6 @@ __spawni_child (void *arguments)
__sigprocmask (SIG_SETMASK, (attr->__flags & POSIX_SPAWN_SETSIGMASK)
? &attr->__ss : &args->oldmask, 0);
- args->err = 0;
args->exec (args->file, args->argv, args->envp);
/* This is compatibility function required to enable posix_spawn run
@@ -339,7 +337,7 @@ __spawnix (pid_t * pid, const char *file
/* Child must set args.err to something non-negative - we rely on
the parent and child sharing VM. */
- args.err = -1;
+ args.err = 0;
args.file = file;
args.exec = exec;
args.fa = file_actions;
@@ -362,12 +360,26 @@ __spawnix (pid_t * pid, const char *file
new_pid = CLONE (__spawni_child, STACK (stack, stack_size), stack_size,
CLONE_VM | CLONE_VFORK | SIGCHLD, &args);
+ /* It needs to collect the case where the auxiliary process was created
+ but failed to execute the file (due either any preparation step or
+ for execve itself). */
if (new_pid > 0)
{
+ /* Also, it handles the unlikely case where the auxiliary process was
+ terminated before calling execve as if it was successfully. The
+ args.err is set to 0 as default and changed to a positive value
+ only in case of failure, so in case of premature termination
+ due a signal args.err will remain zeroed and it will be up to
+ caller to actually collect it. */
ec = args.err;
- assert (ec >= 0);
- if (ec != 0)
- __waitpid (new_pid, NULL, 0);
+ if (ec > 0)
+ /* There still an unlikely case where the child is cancelled after
+ setting args.err, due to a positive error value. Also there is
+ possible pid reuse race (where the kernel allocated the same pid
+ to an unrelated process). Unfortunately due synchronization
+ issues where the kernel might not have the process collected
+ the waitpid below can not use WNOHANG. */
+ __waitpid (new_pid, NULL, 0);
}
else
ec = -new_pid;

125
sysconf-uio-maxiov.patch Normal file
View File

@ -0,0 +1,125 @@
2017-10-20 Florian Weimer <fweimer@redhat.com>
[BZ #22321]
sysconf: Fix missing definition of UIO_MAXIOV on Linux.
* sysdeps/posix/sysconf.c: Include <sys/uio.h>.
* sysdeps/unix/sysv/linux/Makefile (tests): Add tst-sysconf-iov_max.
(tst-sysconf-iov_max): Link with tst-sysconf-iov_max-uapi.o.
* sysdeps/unix/sysv/linux/tst-sysconf-iov_max.c: New file.
* sysdeps/unix/sysv/linux/tst-sysconf-iov_max-uapi.c: Likewise.
Index: glibc-2.26/sysdeps/posix/sysconf.c
===================================================================
--- glibc-2.26.orig/sysdeps/posix/sysconf.c
+++ glibc-2.26/sysdeps/posix/sysconf.c
@@ -29,6 +29,7 @@
#include <sys/stat.h>
#include <sys/sysinfo.h>
#include <sys/types.h>
+#include <sys/uio.h>
#include <regex.h>
#define NEED_SPEC_ARRAY 0
Index: glibc-2.26/sysdeps/unix/sysv/linux/Makefile
===================================================================
--- glibc-2.26.orig/sysdeps/unix/sysv/linux/Makefile
+++ glibc-2.26/sysdeps/unix/sysv/linux/Makefile
@@ -50,7 +50,7 @@ sysdep_headers += sys/mount.h sys/acct.h
bits/siginfo-arch.h bits/siginfo-consts-arch.h
tests += tst-clone tst-clone2 tst-clone3 tst-fanotify tst-personality \
- tst-quota tst-sync_file_range test-errno-linux
+ tst-quota tst-sync_file_range test-errno-linux tst-sysconf-iov_max
# Generate the list of SYS_* macros for the system calls (__NR_* macros).
@@ -120,7 +120,11 @@ ifndef no_deps
-include $(objpfx)bits/syscall.d
endif
generated += bits/syscall.h bits/syscall.d
-endif
+
+# Separate object file for access to the constant from the UAPI header.
+$(objpfx)tst-sysconf-iov_max: $(objpfx)tst-sysconf-iov_max-uapi.o
+
+endif # $(subdir) == misc
ifeq ($(subdir),time)
sysdep_headers += sys/timex.h bits/timex.h
Index: glibc-2.26/sysdeps/unix/sysv/linux/tst-sysconf-iov_max-uapi.c
===================================================================
--- /dev/null
+++ glibc-2.26/sysdeps/unix/sysv/linux/tst-sysconf-iov_max-uapi.c
@@ -0,0 +1,27 @@
+/* Check IOV_MAX definition: Helper function to capture UAPI header value.
+ Copyright (C) 2017 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+/* Use a separate function to avoid header compatibility issues. */
+
+#include <linux/uio.h>
+
+long
+uio_maxiov_value (void)
+{
+ return UIO_MAXIOV;
+}
Index: glibc-2.26/sysdeps/unix/sysv/linux/tst-sysconf-iov_max.c
===================================================================
--- /dev/null
+++ glibc-2.26/sysdeps/unix/sysv/linux/tst-sysconf-iov_max.c
@@ -0,0 +1,40 @@
+/* Check IOV_MAX definition for consistency (bug 22321).
+ Copyright (C) 2017 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+/* Defined in tst-sysconf-iov_max-uapi.c. */
+long uio_maxiov_value (void);
+
+
+#include <limits.h>
+#include <support/check.h>
+#include <sys/uio.h>
+#include <unistd.h>
+
+static int
+do_test (void)
+{
+ TEST_VERIFY (_XOPEN_IOV_MAX == 16); /* Value required by POSIX. */
+ TEST_VERIFY (uio_maxiov_value () >= _XOPEN_IOV_MAX);
+ TEST_VERIFY (IOV_MAX == uio_maxiov_value ());
+ TEST_VERIFY (UIO_MAXIOV == uio_maxiov_value ());
+ TEST_VERIFY (sysconf (_SC_UIO_MAXIOV) == uio_maxiov_value ());
+ TEST_VERIFY (sysconf (_SC_IOV_MAX) == uio_maxiov_value ());
+ return 0;
+}
+
+#include <support/test-driver.c>