* [PATCH] Partial ILP32 support for aarch64
@ 2016-11-19 0:40 Steve Ellcey
2016-11-21 16:51 ` Joseph Myers
0 siblings, 1 reply; 12+ messages in thread
From: Steve Ellcey @ 2016-11-19 0:40 UTC (permalink / raw)
To: libc-alpha
[-- Attachment #1: Type: text/plain, Size: 4037 bytes --]
This is the first of the aarch64 ILP32 patches.  I don't know if this
can be approved and checked in before the rest of the ILP32 patches but
I was hoping it could be so that we can get everything checked in before
the glibc code freeze. Â
This part of the patch does not include any of the structural changes
to support ILP32 builds but it contains the majority of the assembly
language changes and macro changes.  This patch has been tested to make
sure it does not break the aarch64 LP64 build or cause any regressions.
Almost all LP64 objects are identical before and after this patch,
there are a couple of exceptions due to changes in the add_system_dir
macro in sysdeps/unix/sysv/linux/aarch64/dl-cache.h.
OK to checkin?
Steve Ellcey
sellcey@caviumnetworks.com
2016-11-18  Andrew Pinski  <andrew.pinski@caviumnetworks.com>
    Yury Norov  <ynorov@caviumnetworks.com>
    Steve Ellcey  <sellcey@caviumnetworks.com>
* sysdeps/aarch64/crti.S: Add include of sysdep.h.
(call_weak_fn): Use PTR_REG to get correct reg name in ILP32.
* sysdeps/aarch64/dl-irel.h: Add include of sysdep.h.
(elf_irela): Use AARCH64_R macro to get correct relocation in ILP32.
* sysdeps/aarch64/dl-machine.h: Add include of sysdep.h.
(elf_machine_load_address, RTLD_START, RTLD_START_1, RTLD_START,
elf_machine_type_class, ELF_MACHINE_JMP_SLOT, elf_machine_rela,
elf_machine_lazy_rel): Add ifdef's for ILP32 support.
* sysdeps/aarch64/dl-tlsdesc.S (_dl_tlsdesc_return,
_dl_tlsdesc_return_lazy, _dl_tlsdesc_dynamic,
_dl_tlsdesc_resolve_hold): Extend pointers in ILP32, use PTR_REG
to get correct reg name for ILP32.
* sysdeps/aarch64/dl-trampoline.S (ip01): New Macro.
(RELA_SIZE): New Macro.
(_dl_runtime_resolve, _dl_runtime_profile): Use new macros and PTR_REG
to support ILP32.
* sysdeps/aarch64/jmpbuf-unwind.h (_JMPBUF_CFA_UNWINDS_ADJ): Add
cast for ILP32 mode.
* sysdeps/aarch64/memcmp.S (memcmp): Extend arg pointers for ILP32 mode.
* sysdeps/aarch64/memcpy.S (memmove, memcpy): Ditto.
* sysdeps/aarch64/memset.S (__memset): Ditto.
* sysdeps/aarch64/strchr.S (strchr): Ditto.
* sysdeps/aarch64/strchrnul.S (__strchrnul): Ditto.
* sysdeps/aarch64/strcmp.S (strcmp): Ditto.
* sysdeps/aarch64/strcpy.S (strcpy): Ditto.
* sysdeps/aarch64/strlen.S (__strlen): Ditto.
* sysdeps/aarch64/strncmp.S (strncmp): Ditto.
* sysdeps/aarch64/strnlen.S (strnlen): Ditto.
* sysdeps/aarch64/strrchr.S (strrchr): Ditto.
* sysdeps/unix/sysv/linux/aarch64/clone.S: Ditto.
* sysdeps/unix/sysv/linux/aarch64/setcontext.S (__setcontext): Ditto.
* sysdeps/unix/sysv/linux/aarch64/swapcontext.S (__swapcontext): Ditto.
* sysdeps/aarch64/__longjmp.S (__longjmp): Extend pointers in ILP32,
change PTR_MANGLE call to use register numbers instead of names.
* sysdeps/unix/sysv/linux/aarch64/getcontext.S (__getcontext): Ditto.
* sysdeps/aarch64/setjmp.S (__sigsetjmp): Extend arg pointers for
ILP32 mode, change PTR_MANGLE calls to use register numbers.
* sysdeps/aarch64/start.S (_start): Ditto.
* sysdeps/aarch64/nptl/bits/pthreadtypes.h
(__PTHREAD_RWLOCK_INT_FLAGS_SHARED): New define.
* sysdeps/aarch64/nptl/bits/semaphore.h (__SIZEOF_SEM_T): Change define.
* sysdeps/aarch64/sysdep.h (AARCH64_R, PTR_REG, PTR_LOG_SIZE, DELOUSE,
PTR_SIZE): New Macros.
(LDST_PCREL, LDST_GLOBAL) Update to use PTR_REG.
* sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h (O_LARGEFILE):
Set when in ILP32 mode.
(F_GETLK64, F_SETLK64, F_SETLKW64): Only set in LP64 mode.
* sysdeps/unix/sysv/linux/aarch64/dl-cache.h (DL_CACHE_DEFAULT_ID):
Set elf flags for ILP32.
(add_system_dir): Set ILP32 library directories.
* sysdeps/unix/sysv/linux/aarch64/init-first.c
(_libc_vdso_platform_setup): Set minimum kernel version for ILP32.
* sysdeps/unix/sysv/linux/aarch64/ldconfig.h
(SYSDEP_KNOWN_INTERPRETER_NAMES): Add ILP32 names.
* sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h (GET_PC, SET_PC):
New Macros.
* sysdeps/unix/sysv/linux/aarch64/sysdep.h: Handle ILP32 pointers.
[-- Attachment #2: aarch64_ilp32.patch --]
[-- Type: text/x-patch, Size: 36007 bytes --]
diff --git a/sysdeps/aarch64/__longjmp.S b/sysdeps/aarch64/__longjmp.S
index 65116be..4d411fe 100644
--- a/sysdeps/aarch64/__longjmp.S
+++ b/sysdeps/aarch64/__longjmp.S
@@ -46,6 +46,8 @@ ENTRY (__longjmp)
cfi_offset(d14, JB_D14<<3)
cfi_offset(d15, JB_D15<<3)
+ DELOUSE (0)
+
ldp x19, x20, [x0, #JB_X19<<3]
ldp x21, x22, [x0, #JB_X21<<3]
ldp x23, x24, [x0, #JB_X23<<3]
@@ -53,7 +55,7 @@ ENTRY (__longjmp)
ldp x27, x28, [x0, #JB_X27<<3]
#ifdef PTR_DEMANGLE
ldp x29, x4, [x0, #JB_X29<<3]
- PTR_DEMANGLE (x30, x4, x3, x2)
+ PTR_DEMANGLE (30, 4, 3, 2)
#else
ldp x29, x30, [x0, #JB_X29<<3]
#endif
@@ -98,7 +100,7 @@ ENTRY (__longjmp)
cfi_same_value(d15)
#ifdef PTR_DEMANGLE
ldr x4, [x0, #JB_SP<<3]
- PTR_DEMANGLE (x5, x4, x3, x2)
+ PTR_DEMANGLE (5, 4, 3, 2)
#else
ldr x5, [x0, #JB_SP<<3]
#endif
diff --git a/sysdeps/aarch64/crti.S b/sysdeps/aarch64/crti.S
index 53ccb42..5c42fd5 100644
--- a/sysdeps/aarch64/crti.S
+++ b/sysdeps/aarch64/crti.S
@@ -39,6 +39,7 @@
they can be called as functions. The symbols _init and _fini are
magic and cause the linker to emit DT_INIT and DT_FINI. */
+#include <sysdep.h>
#include <libc-symbols.h>
#ifndef PREINIT_FUNCTION
@@ -60,7 +61,7 @@
.type call_weak_fn, %function
call_weak_fn:
adrp x0, :got:PREINIT_FUNCTION
- ldr x0, [x0, #:got_lo12:PREINIT_FUNCTION]
+ ldr PTR_REG (0), [x0, #:got_lo12:PREINIT_FUNCTION]
cbz x0, 1f
b PREINIT_FUNCTION
1:
diff --git a/sysdeps/aarch64/dl-irel.h b/sysdeps/aarch64/dl-irel.h
index 63a8e50..2effca4 100644
--- a/sysdeps/aarch64/dl-irel.h
+++ b/sysdeps/aarch64/dl-irel.h
@@ -23,6 +23,7 @@
#include <stdio.h>
#include <unistd.h>
#include <ldsodefs.h>
+#include <sysdep.h>
#define ELF_MACHINE_IRELA 1
@@ -40,7 +41,7 @@ elf_irela (const ElfW(Rela) *reloc)
ElfW(Addr) *const reloc_addr = (void *) reloc->r_offset;
const unsigned long int r_type = ELFW(R_TYPE) (reloc->r_info);
- if (__glibc_likely (r_type == R_AARCH64_IRELATIVE))
+ if (__glibc_likely (r_type == AARCH64_R (IRELATIVE)))
{
ElfW(Addr) value = elf_ifunc_invoke (reloc->r_addend);
*reloc_addr = value;
diff --git a/sysdeps/aarch64/dl-machine.h b/sysdeps/aarch64/dl-machine.h
index 282805e..b5ea7a8 100644
--- a/sysdeps/aarch64/dl-machine.h
+++ b/sysdeps/aarch64/dl-machine.h
@@ -21,6 +21,7 @@
#define ELF_MACHINE_NAME "aarch64"
+#include <sysdep.h>
#include <tls.h>
#include <dl-tlsdesc.h>
#include <dl-irel.h>
@@ -53,19 +54,33 @@ elf_machine_load_address (void)
by constructing a non GOT reference to the symbol, the dynamic
address of the symbol we compute using adrp/add to compute the
symbol's address relative to the PC.
- This depends on 32bit relocations being resolved at link time
- and that the static address fits in the 32bits. */
+ This depends on 32/16bit relocations being resolved at link time
+ and that the static address fits in the 32/16 bits. */
ElfW(Addr) static_addr;
ElfW(Addr) dynamic_addr;
asm (" \n"
" adrp %1, _dl_start; \n"
+#ifdef __LP64__
" add %1, %1, #:lo12:_dl_start \n"
+#else
+" add %w1, %w1, #:lo12:_dl_start \n"
+#endif
" ldr %w0, 1f \n"
" b 2f \n"
"1: \n"
+#ifdef __LP64__
" .word _dl_start \n"
+#else
+# ifdef __AARCH64EB__
+" .short 0 \n"
+# endif
+" .short _dl_start \n"
+# ifndef __AARCH64EB__
+" .short 0 \n"
+# endif
+#endif
"2: \n"
: "=r" (static_addr), "=r" (dynamic_addr));
return dynamic_addr - static_addr;
@@ -125,80 +140,86 @@ elf_machine_runtime_setup (struct link_map *l, int lazy, int profile)
/* Initial entry point for the dynamic linker. The C function
_dl_start is the real entry point, its return value is the user
program's entry point */
+#ifdef __LP64__
+# define RTLD_START RTLD_START_1 ("x", "3", "sp")
+#else
+# define RTLD_START RTLD_START_1 ("w", "2", "wsp")
+#endif
-#define RTLD_START asm ("\
-.text \n\
-.globl _start \n\
-.type _start, %function \n\
-.globl _dl_start_user \n\
-.type _dl_start_user, %function \n\
-_start: \n\
- mov x0, sp \n\
- bl _dl_start \n\
- // returns user entry point in x0 \n\
- mov x21, x0 \n\
-_dl_start_user: \n\
- // get the original arg count \n\
- ldr x1, [sp] \n\
- // get the argv address \n\
- add x2, sp, #8 \n\
- // get _dl_skip_args to see if we were \n\
- // invoked as an executable \n\
- adrp x4, _dl_skip_args \n\
- ldr w4, [x4, #:lo12:_dl_skip_args] \n\
- // do we need to adjust argc/argv \n\
- cmp w4, 0 \n\
- beq .L_done_stack_adjust \n\
- // subtract _dl_skip_args from original arg count \n\
- sub x1, x1, x4 \n\
- // store adjusted argc back to stack \n\
- str x1, [sp] \n\
- // find the first unskipped argument \n\
- mov x3, x2 \n\
- add x4, x2, x4, lsl #3 \n\
- // shuffle argv down \n\
-1: ldr x5, [x4], #8 \n\
- str x5, [x3], #8 \n\
- cmp x5, #0 \n\
- bne 1b \n\
- // shuffle envp down \n\
-1: ldr x5, [x4], #8 \n\
- str x5, [x3], #8 \n\
- cmp x5, #0 \n\
- bne 1b \n\
- // shuffle auxv down \n\
-1: ldp x0, x5, [x4, #16]! \n\
- stp x0, x5, [x3], #16 \n\
- cmp x0, #0 \n\
- bne 1b \n\
- // Update _dl_argv \n\
- adrp x3, _dl_argv \n\
- str x2, [x3, #:lo12:_dl_argv] \n\
-.L_done_stack_adjust: \n\
- // compute envp \n\
- add x3, x2, x1, lsl #3 \n\
- add x3, x3, #8 \n\
- adrp x16, _rtld_local \n\
- add x16, x16, #:lo12:_rtld_local \n\
- ldr x0, [x16] \n\
- bl _dl_init \n\
- // load the finalizer function \n\
- adrp x0, _dl_fini \n\
- add x0, x0, #:lo12:_dl_fini \n\
- // jump to the user_s entry point \n\
- br x21 \n\
+
+#define RTLD_START_1(PTR, PTR_SIZE_LOG, PTR_SP) asm ("\
+.text \n\
+.globl _start \n\
+.type _start, %function \n\
+.globl _dl_start_user \n\
+.type _dl_start_user, %function \n\
+_start: \n\
+ mov " PTR "0, " PTR_SP " \n\
+ bl _dl_start \n\
+ // returns user entry point in x0 \n\
+ mov x21, x0 \n\
+_dl_start_user: \n\
+ // get the original arg count \n\
+ ldr " PTR "1, [sp] \n\
+ // get the argv address \n\
+ add " PTR "2, " PTR_SP ", #(1<<" PTR_SIZE_LOG ") \n\
+ // get _dl_skip_args to see if we were \n\
+ // invoked as an executable \n\
+ adrp x4, _dl_skip_args \n\
+ ldr w4, [x4, #:lo12:_dl_skip_args] \n\
+ // do we need to adjust argc/argv \n\
+ cmp w4, 0 \n\
+ beq .L_done_stack_adjust \n\
+ // subtract _dl_skip_args from original arg count \n\
+ sub " PTR "1, " PTR "1, " PTR "4 \n\
+ // store adjusted argc back to stack \n\
+ str " PTR "1, [sp] \n\
+ // find the first unskipped argument \n\
+ mov " PTR "3, " PTR "2 \n\
+ add " PTR "4, " PTR "2, " PTR "4, lsl #" PTR_SIZE_LOG " \n\
+ // shuffle argv down \n\
+1: ldr " PTR "5, [x4], #(1<<" PTR_SIZE_LOG ") \n\
+ str " PTR "5, [x3], #(1<<" PTR_SIZE_LOG ") \n\
+ cmp " PTR "5, #0 \n\
+ bne 1b \n\
+ // shuffle envp down \n\
+1: ldr " PTR "5, [x4], #(1<<" PTR_SIZE_LOG ") \n\
+ str " PTR "5, [x3], #(1<<" PTR_SIZE_LOG ") \n\
+ cmp " PTR "5, #0 \n\
+ bne 1b \n\
+ // shuffle auxv down \n\
+1: ldp " PTR "0, " PTR "5, [x4, #(2<<" PTR_SIZE_LOG ")]! \n\
+ stp " PTR "0, " PTR "5, [x3], #(2<<" PTR_SIZE_LOG ") \n\
+ cmp " PTR "0, #0 \n\
+ bne 1b \n\
+ // Update _dl_argv \n\
+ adrp x3, _dl_argv \n\
+ str " PTR "2, [x3, #:lo12:_dl_argv] \n\
+.L_done_stack_adjust: \n\
+ // compute envp \n\
+ add " PTR "3, " PTR "2, " PTR "1, lsl #" PTR_SIZE_LOG " \n\
+ add " PTR "3, " PTR "3, #(1<<" PTR_SIZE_LOG ") \n\
+ adrp x16, _rtld_local \n\
+ add " PTR "16, " PTR "16, #:lo12:_rtld_local \n\
+ ldr " PTR "0, [x16] \n\
+ bl _dl_init \n\
+ // load the finalizer function \n\
+ adrp x0, _dl_fini \n\
+ add " PTR "0, " PTR "0, #:lo12:_dl_fini \n\
+ // jump to the user_s entry point \n\
+ br x21 \n\
");
#define elf_machine_type_class(type) \
- ((((type) == R_AARCH64_JUMP_SLOT || \
- (type) == R_AARCH64_TLS_DTPMOD || \
- (type) == R_AARCH64_TLS_DTPREL || \
- (type) == R_AARCH64_TLS_TPREL || \
- (type) == R_AARCH64_TLSDESC) * ELF_RTYPE_CLASS_PLT) \
- | (((type) == R_AARCH64_COPY) * ELF_RTYPE_CLASS_COPY) \
- | (((type) == R_AARCH64_GLOB_DAT) * ELF_RTYPE_CLASS_EXTERN_PROTECTED_DATA))
+ ((((type) == AARCH64_R (JUMP_SLOT) \
+ || (type) == AARCH64_R (TLS_DTPMOD) \
+ || (type) == AARCH64_R (TLS_DTPREL) \
+ || (type) == AARCH64_R (TLS_TPREL) \
+ || (type) == AARCH64_R (TLSDESC)) * ELF_RTYPE_CLASS_PLT) \
+ | (((type) == AARCH64_R (COPY)) * ELF_RTYPE_CLASS_COPY) \
+ | (((type) == AARCH64_R (GLOB_DAT)) * ELF_RTYPE_CLASS_EXTERN_PROTECTED_DATA))
-#define ELF_MACHINE_JMP_SLOT R_AARCH64_JUMP_SLOT
+#define ELF_MACHINE_JMP_SLOT AARCH64_R (JUMP_SLOT)
/* AArch64 uses RELA not REL */
#define ELF_MACHINE_NO_REL 1
@@ -237,9 +258,9 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
void *const reloc_addr_arg, int skip_ifunc)
{
ElfW(Addr) *const reloc_addr = reloc_addr_arg;
- const unsigned int r_type = ELF64_R_TYPE (reloc->r_info);
+ const unsigned int r_type = ELFW (R_TYPE) (reloc->r_info);
- if (__builtin_expect (r_type == R_AARCH64_RELATIVE, 0))
+ if (__builtin_expect (r_type == AARCH64_R (RELATIVE), 0))
*reloc_addr = map->l_addr + reloc->r_addend;
else if (__builtin_expect (r_type == R_AARCH64_NONE, 0))
return;
@@ -257,7 +278,7 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
switch (r_type)
{
- case R_AARCH64_COPY:
+ case AARCH64_R (COPY):
if (sym == NULL)
break;
@@ -275,15 +296,17 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
MIN (sym->st_size, refsym->st_size));
break;
- case R_AARCH64_RELATIVE:
- case R_AARCH64_GLOB_DAT:
- case R_AARCH64_JUMP_SLOT:
- case R_AARCH64_ABS32:
- case R_AARCH64_ABS64:
+ case AARCH64_R (RELATIVE):
+ case AARCH64_R (GLOB_DAT):
+ case AARCH64_R (JUMP_SLOT):
+ case AARCH64_R (ABS32):
+#ifdef __LP64__
+ case AARCH64_R (ABS64):
+#endif
*reloc_addr = value + reloc->r_addend;
break;
- case R_AARCH64_TLSDESC:
+ case AARCH64_R (TLSDESC):
{
struct tlsdesc volatile *td =
(struct tlsdesc volatile *)reloc_addr;
@@ -318,7 +341,7 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
break;
}
- case R_AARCH64_TLS_DTPMOD:
+ case AARCH64_R (TLS_DTPMOD):
#ifdef RTLD_BOOTSTRAP
*reloc_addr = 1;
#else
@@ -329,12 +352,12 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
#endif
break;
- case R_AARCH64_TLS_DTPREL:
+ case AARCH64_R (TLS_DTPREL):
if (sym)
*reloc_addr = sym->st_value + reloc->r_addend;
break;
- case R_AARCH64_TLS_TPREL:
+ case AARCH64_R (TLS_TPREL):
if (sym)
{
CHECK_STATIC_TLS (map, sym_map);
@@ -343,7 +366,7 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
}
break;
- case R_AARCH64_IRELATIVE:
+ case AARCH64_R (IRELATIVE):
value = map->l_addr + reloc->r_addend;
value = elf_ifunc_invoke (value);
*reloc_addr = value;
@@ -374,16 +397,16 @@ elf_machine_lazy_rel (struct link_map *map,
int skip_ifunc)
{
ElfW(Addr) *const reloc_addr = (void *) (l_addr + reloc->r_offset);
- const unsigned int r_type = ELF64_R_TYPE (reloc->r_info);
+ const unsigned int r_type = ELFW (R_TYPE) (reloc->r_info);
/* Check for unexpected PLT reloc type. */
- if (__builtin_expect (r_type == R_AARCH64_JUMP_SLOT, 1))
+ if (__builtin_expect (r_type == AARCH64_R (JUMP_SLOT), 1))
{
if (__builtin_expect (map->l_mach.plt, 0) == 0)
*reloc_addr += l_addr;
else
*reloc_addr = map->l_mach.plt;
}
- else if (__builtin_expect (r_type == R_AARCH64_TLSDESC, 1))
+ else if (__builtin_expect (r_type == AARCH64_R (TLSDESC), 1))
{
struct tlsdesc volatile *td =
(struct tlsdesc volatile *)reloc_addr;
@@ -392,7 +415,7 @@ elf_machine_lazy_rel (struct link_map *map,
td->entry = (void*)(D_PTR (map, l_info[ADDRIDX (DT_TLSDESC_PLT)])
+ map->l_addr);
}
- else if (__glibc_unlikely (r_type == R_AARCH64_IRELATIVE))
+ else if (__glibc_unlikely (r_type == AARCH64_R (IRELATIVE)))
{
ElfW(Addr) value = map->l_addr + reloc->r_addend;
if (__glibc_likely (!skip_ifunc))
diff --git a/sysdeps/aarch64/dl-tlsdesc.S b/sysdeps/aarch64/dl-tlsdesc.S
index 05be370..42fa943 100644
--- a/sysdeps/aarch64/dl-tlsdesc.S
+++ b/sysdeps/aarch64/dl-tlsdesc.S
@@ -74,7 +74,8 @@
cfi_startproc
.align 2
_dl_tlsdesc_return:
- ldr x0, [x0, #8]
+ DELOUSE (0)
+ ldr PTR_REG (0), [x0, #PTR_SIZE]
RET
cfi_endproc
.size _dl_tlsdesc_return, .-_dl_tlsdesc_return
@@ -95,9 +96,10 @@ _dl_tlsdesc_return_lazy:
so it reads the same value (this function is the final value of
td->entry) and thus it synchronizes with the release store to
td->entry in _dl_tlsdesc_resolve_rela_fixup ensuring that the load
- from [x0,#8] here happens after the initialization of td->arg. */
- ldar xzr, [x0]
- ldr x0, [x0, #8]
+ from [x0,#PTR_SIZE] here happens after the initialization of td->arg. */
+ DELOUSE (0)
+ ldar PTR_REG (zr), [x0]
+ ldr PTR_REG (0), [x0, #PTR_SIZE]
RET
cfi_endproc
.size _dl_tlsdesc_return_lazy, .-_dl_tlsdesc_return_lazy
@@ -125,10 +127,11 @@ _dl_tlsdesc_undefweak:
td->entry) and thus it synchronizes with the release store to
td->entry in _dl_tlsdesc_resolve_rela_fixup ensuring that the load
from [x0,#8] here happens after the initialization of td->arg. */
- ldar xzr, [x0]
- ldr x0, [x0, #8]
+ DELOUSE (0)
+ ldar PTR_REG (zr), [x0]
+ ldr PTR_REG (0), [x0, #PTR_SIZE]
mrs x1, tpidr_el0
- sub x0, x0, x1
+ sub PTR_REG (0), PTR_REG (0), PTR_REG (1)
ldr x1, [sp], #16
cfi_adjust_cfa_offset (-16)
RET
@@ -174,6 +177,7 @@ _dl_tlsdesc_dynamic:
stp x29, x30, [sp,#-(32+16*NSAVEXREGPAIRS)]!
cfi_adjust_cfa_offset (32+16*NSAVEXREGPAIRS)
mov x29, sp
+ DELOUSE (0)
/* Save just enough registers to support fast path, if we fall
into slow path we will save additional registers. */
@@ -187,22 +191,22 @@ _dl_tlsdesc_dynamic:
so it reads the same value (this function is the final value of
td->entry) and thus it synchronizes with the release store to
td->entry in _dl_tlsdesc_resolve_rela_fixup ensuring that the load
- from [x0,#8] here happens after the initialization of td->arg. */
- ldar xzr, [x0]
- ldr x1, [x0,#8]
- ldr x0, [x4]
- ldr x3, [x1,#16]
- ldr x2, [x0]
- cmp x3, x2
+ from [x0,#PTR_SIZE] here happens after the initialization of td->arg. */
+ ldar PTR_REG (zr), [x0]
+ ldr PTR_REG (1), [x0,#PTR_SIZE]
+ ldr PTR_REG (0), [x4]
+ ldr PTR_REG (3), [x1,#(PTR_SIZE * 2)]
+ ldr PTR_REG (2), [x0]
+ cmp PTR_REG (3), PTR_REG (2)
b.hi 2f
- ldr x2, [x1]
- add x0, x0, x2, lsl #4
- ldr x0, [x0]
+ ldr PTR_REG (2), [x1]
+ add PTR_REG (0), PTR_REG (0), PTR_REG (2), lsl #(PTR_LOG_SIZE + 1)
+ ldr PTR_REG (0), [x0]
cmn x0, #0x1
b.eq 2f
- ldr x1, [x1,#8]
- add x0, x0, x1
- sub x0, x0, x4
+ ldr PTR_REG (1), [x1,#(PTR_SIZE * 2)]
+ add PTR_REG (0), PTR_REG (0), PTR_REG (1)
+ sub PTR_REG (0), PTR_REG (0), PTR_REG (4)
1:
ldp x1, x2, [sp, #32+16*0]
ldp x3, x4, [sp, #32+16*1]
@@ -233,7 +237,7 @@ _dl_tlsdesc_dynamic:
bl __tls_get_addr
mrs x1, tpidr_el0
- sub x0, x0, x1
+ sub PTR_REG (0), PTR_REG (0), PTR_REG (1)
RESTORE_Q_REGISTERS
@@ -279,13 +283,15 @@ _dl_tlsdesc_resolve_rela:
SAVE_Q_REGISTERS
- ldr x1, [x3, #8]
+ DELOUSE (3)
+ ldr PTR_REG (1), [x3, #PTR_SIZE]
bl _dl_tlsdesc_resolve_rela_fixup
RESTORE_Q_REGISTERS
ldr x0, [sp, #32+16*8]
- ldr x1, [x0]
+ DELOUSE (0)
+ ldr PTR_REG (1), [x0]
blr x1
ldp x1, x4, [sp, #32+16*0]
@@ -346,7 +352,8 @@ _dl_tlsdesc_resolve_hold:
RESTORE_Q_REGISTERS
ldr x0, [sp, #32+16*9]
- ldr x1, [x0]
+ DELOUSE (0)
+ ldr PTR_REG (1), [x0]
blr x1
ldp x1, x2, [sp, #32+16*0]
diff --git a/sysdeps/aarch64/dl-trampoline.S b/sysdeps/aarch64/dl-trampoline.S
index 947a515..63ef6f7 100644
--- a/sysdeps/aarch64/dl-trampoline.S
+++ b/sysdeps/aarch64/dl-trampoline.S
@@ -22,9 +22,13 @@
#include "dl-link.h"
#define ip0 x16
+#define ip0l PTR_REG (16)
#define ip1 x17
#define lr x30
+/* RELA relocatons are 3 pointers */
+#define RELA_SIZE (PTR_SIZE * 3)
+
.text
.globl _dl_runtime_resolve
.type _dl_runtime_resolve, #function
@@ -79,7 +83,7 @@ _dl_runtime_resolve:
cfi_rel_offset (q1, 80+7*16)
/* Get pointer to linker struct. */
- ldr x0, [ip0, #-8]
+ ldr PTR_REG (0), [ip0, #-PTR_SIZE]
/* Prepare to call _dl_fixup(). */
ldr x1, [sp, 80+8*16] /* Recover &PLTGOT[n] */
@@ -87,7 +91,7 @@ _dl_runtime_resolve:
sub x1, x1, ip0
add x1, x1, x1, lsl #1
lsl x1, x1, #3
- sub x1, x1, #192
+ sub x1, x1, #(RELA_SIZE<<3)
lsr x1, x1, #3
/* Call fixup routine. */
@@ -191,7 +195,7 @@ _dl_runtime_profile:
stp x0, x1, [x29, #OFFSET_RG + DL_OFFSET_RG_SP]
/* Get pointer to linker struct. */
- ldr x0, [ip0, #-8]
+ ldr PTR_REG (0), [ip0, #-PTR_SIZE]
/* Prepare to call _dl_profile_fixup(). */
ldr x1, [x29, OFFSET_PLTGOTN] /* Recover &PLTGOT[n] */
@@ -199,7 +203,7 @@ _dl_runtime_profile:
sub x1, x1, ip0
add x1, x1, x1, lsl #1
lsl x1, x1, #3
- sub x1, x1, #192
+ sub x1, x1, #(RELA_SIZE<<3)
lsr x1, x1, #3
stp x0, x1, [x29, #OFFSET_SAVED_CALL_X0]
@@ -210,8 +214,8 @@ _dl_runtime_profile:
add x4, x29, #OFFSET_FS /* address of framesize */
bl _dl_profile_fixup
- ldr ip0, [x29, #OFFSET_FS] /* framesize == 0 */
- cmp ip0, #0
+ ldr ip0l, [x29, #OFFSET_FS] /* framesize == 0 */
+ cmp ip0l, #0
bge 1f
cfi_remember_state
@@ -243,7 +247,7 @@ _dl_runtime_profile:
1:
/* The new frame size is in ip0. */
- sub x1, x29, ip0
+ sub PTR_REG (1), PTR_REG (29), ip0l
and sp, x1, #0xfffffffffffffff0
str x0, [x29, #OFFSET_T1]
diff --git a/sysdeps/aarch64/jmpbuf-unwind.h b/sysdeps/aarch64/jmpbuf-unwind.h
index 3e0a37d..11ace17 100644
--- a/sysdeps/aarch64/jmpbuf-unwind.h
+++ b/sysdeps/aarch64/jmpbuf-unwind.h
@@ -27,7 +27,7 @@
((void *) (address) < (void *) demangle (jmpbuf[JB_SP]))
#define _JMPBUF_CFA_UNWINDS_ADJ(jmpbuf, context, adj) \
- _JMPBUF_UNWINDS_ADJ (jmpbuf, (void *) _Unwind_GetCFA (context), adj)
+ _JMPBUF_UNWINDS_ADJ (jmpbuf, (void *) (size_t) _Unwind_GetCFA (context), adj)
#define _JMPBUF_UNWINDS_ADJ(_jmpbuf, _address, _adj) \
((uintptr_t) (_address) - (_adj) < _jmpbuf_sp (_jmpbuf) - (_adj))
diff --git a/sysdeps/aarch64/memcmp.S b/sysdeps/aarch64/memcmp.S
index ae2d997..8b87e9b 100644
--- a/sysdeps/aarch64/memcmp.S
+++ b/sysdeps/aarch64/memcmp.S
@@ -47,6 +47,9 @@
#define mask x13
ENTRY_ALIGN (memcmp, 6)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
cbz limit, L(ret0)
eor tmp1, src1, src2
tst tmp1, #7
diff --git a/sysdeps/aarch64/memcpy.S b/sysdeps/aarch64/memcpy.S
index de73f0f..b269316 100644
--- a/sysdeps/aarch64/memcpy.S
+++ b/sysdeps/aarch64/memcpy.S
@@ -61,6 +61,10 @@
ENTRY_ALIGN (memmove, 6)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
+
sub tmp1, dstin, src
cmp count, 96
ccmp tmp1, count, 2, hi
@@ -71,6 +75,10 @@ END (memmove)
libc_hidden_builtin_def (memmove)
ENTRY (memcpy)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
+
prfm PLDL1KEEP, [src]
add srcend, src, count
add dstend, dstin, count
diff --git a/sysdeps/aarch64/memset.S b/sysdeps/aarch64/memset.S
index 4d222c5..7bad29a 100644
--- a/sysdeps/aarch64/memset.S
+++ b/sysdeps/aarch64/memset.S
@@ -39,6 +39,9 @@
ENTRY_ALIGN (__memset, 6)
+ DELOUSE (0)
+ DELOUSE (2)
+
dup v0.16B, valw
add dstend, dstin, count
diff --git a/sysdeps/aarch64/nptl/bits/pthreadtypes.h b/sysdeps/aarch64/nptl/bits/pthreadtypes.h
index c376e64..9dcf8d9 100644
--- a/sysdeps/aarch64/nptl/bits/pthreadtypes.h
+++ b/sysdeps/aarch64/nptl/bits/pthreadtypes.h
@@ -32,6 +32,8 @@
#define __SIZEOF_PTHREAD_BARRIER_T 32
#define __SIZEOF_PTHREAD_BARRIERATTR_T 8
+#define __PTHREAD_RWLOCK_INT_FLAGS_SHARED 1
+
/* Thread identifiers. The structure of the attribute type is not
exposed on purpose. */
diff --git a/sysdeps/aarch64/nptl/bits/semaphore.h b/sysdeps/aarch64/nptl/bits/semaphore.h
index 3cc5b37..1d1389c 100644
--- a/sysdeps/aarch64/nptl/bits/semaphore.h
+++ b/sysdeps/aarch64/nptl/bits/semaphore.h
@@ -21,7 +21,11 @@
#endif
+#ifdef __ILP32__
+#define __SIZEOF_SEM_T 16
+#else
#define __SIZEOF_SEM_T 32
+#endif
/* Value returned if `sem_open' failed. */
diff --git a/sysdeps/aarch64/setjmp.S b/sysdeps/aarch64/setjmp.S
index 22f4368..e03b3b5 100644
--- a/sysdeps/aarch64/setjmp.S
+++ b/sysdeps/aarch64/setjmp.S
@@ -33,6 +33,7 @@ END (_setjmp)
libc_hidden_def (_setjmp)
ENTRY (__sigsetjmp)
+ DELOUSE (0)
1:
stp x19, x20, [x0, #JB_X19<<3]
@@ -42,7 +43,7 @@ ENTRY (__sigsetjmp)
stp x27, x28, [x0, #JB_X27<<3]
#ifdef PTR_MANGLE
- PTR_MANGLE (x4, x30, x3, x2)
+ PTR_MANGLE (4, 30, 3, 2)
stp x29, x4, [x0, #JB_X29<<3]
#else
stp x29, x30, [x0, #JB_X29<<3]
@@ -57,7 +58,7 @@ ENTRY (__sigsetjmp)
stp d14, d15, [x0, #JB_D14<<3]
#ifdef PTR_MANGLE
mov x4, sp
- PTR_MANGLE (x5, x4, x3, x2)
+ PTR_MANGLE (5, 4, 3, 2)
str x5, [x0, #JB_SP<<3]
#else
mov x2, sp
diff --git a/sysdeps/aarch64/start.S b/sysdeps/aarch64/start.S
index efe2474..9198c57 100644
--- a/sysdeps/aarch64/start.S
+++ b/sysdeps/aarch64/start.S
@@ -16,6 +16,8 @@
License along with the GNU C Library. If not, see
<http://www.gnu.org/licenses/>. */
+#include <sysdep.h>
+
/* This is the canonical entry point, usually the first thing in the text
segment.
@@ -25,7 +27,7 @@
At this entry point, most registers' values are unspecified, except:
- x0 Contains a function pointer to be registered with `atexit'.
+ x0/w0 Contains a function pointer to be registered with `atexit'.
This is how the dynamic linker arranges to have DT_FINI
functions called for shared libraries that have been loaded
before this code runs.
@@ -52,26 +54,26 @@ _start:
mov x5, x0
/* Load argc and a pointer to argv */
- ldr x1, [sp, #0]
- add x2, sp, #8
+ ldr PTR_REG (1), [sp, #0]
+ add x2, sp, #PTR_SIZE
/* Setup stack limit in argument register */
mov x6, sp
#ifdef SHARED
adrp x0, :got:main
- ldr x0, [x0, #:got_lo12:main]
+ ldr PTR_REG (0), [x0, #:got_lo12:main]
adrp x3, :got:__libc_csu_init
- ldr x3, [x3, #:got_lo12:__libc_csu_init]
+ ldr PTR_REG (3), [x3, #:got_lo12:__libc_csu_init]
adrp x4, :got:__libc_csu_fini
- ldr x4, [x4, #:got_lo12:__libc_csu_fini]
+ ldr PTR_REG (4), [x4, #:got_lo12:__libc_csu_fini]
#else
/* Set up the other arguments in registers */
- ldr x0, =main
- ldr x3, =__libc_csu_init
- ldr x4, =__libc_csu_fini
+ ldr PTR_REG (0), =main
+ ldr PTR_REG (3), =__libc_csu_init
+ ldr PTR_REG (4), =__libc_csu_fini
#endif
/* __libc_start_main (main, argc, argv, init, fini, rtld_fini,
diff --git a/sysdeps/aarch64/strchr.S b/sysdeps/aarch64/strchr.S
index 5e3aecf..c66fea3 100644
--- a/sysdeps/aarch64/strchr.S
+++ b/sysdeps/aarch64/strchr.S
@@ -62,6 +62,7 @@
/* Locals and temporaries. */
ENTRY (strchr)
+ DELOUSE (0)
mov wtmp2, #0x0401
movk wtmp2, #0x4010, lsl #16
dup vrepchr.16b, chrin
diff --git a/sysdeps/aarch64/strchrnul.S b/sysdeps/aarch64/strchrnul.S
index a624c8d..c2cc47e 100644
--- a/sysdeps/aarch64/strchrnul.S
+++ b/sysdeps/aarch64/strchrnul.S
@@ -60,6 +60,7 @@
identify exactly which byte is causing the termination. */
ENTRY (__strchrnul)
+ DELOUSE (0)
/* Magic constant 0x40100401 to allow us to identify which lane
matches the termination condition. */
mov wtmp2, #0x0401
diff --git a/sysdeps/aarch64/strcmp.S b/sysdeps/aarch64/strcmp.S
index ba0ccb4..49e528b 100644
--- a/sysdeps/aarch64/strcmp.S
+++ b/sysdeps/aarch64/strcmp.S
@@ -49,6 +49,8 @@
/* Start of performance-critical section -- one 64B cache line. */
ENTRY_ALIGN(strcmp, 6)
+ DELOUSE (0)
+ DELOUSE (1)
eor tmp1, src1, src2
mov zeroones, #REP8_01
tst tmp1, #7
diff --git a/sysdeps/aarch64/strcpy.S b/sysdeps/aarch64/strcpy.S
index 0694199..45809e8 100644
--- a/sysdeps/aarch64/strcpy.S
+++ b/sysdeps/aarch64/strcpy.S
@@ -91,6 +91,8 @@
#define MIN_PAGE_SIZE (1 << MIN_PAGE_P2)
ENTRY_ALIGN (STRCPY, 6)
+ DELOUSE (0)
+ DELOUSE (1)
/* For moderately short strings, the fastest way to do the copy is to
calculate the length of the string in the same way as strlen, then
essentially do a memcpy of the result. This avoids the need for
diff --git a/sysdeps/aarch64/strlen.S b/sysdeps/aarch64/strlen.S
index a07834b..5fb653a 100644
--- a/sysdeps/aarch64/strlen.S
+++ b/sysdeps/aarch64/strlen.S
@@ -85,6 +85,8 @@
boundary. */
ENTRY_ALIGN (__strlen, 6)
+ DELOUSE (0)
+ DELOUSE (1)
and tmp1, srcin, MIN_PAGE_SIZE - 1
mov zeroones, REP8_01
cmp tmp1, MIN_PAGE_SIZE - 16
diff --git a/sysdeps/aarch64/strncmp.S b/sysdeps/aarch64/strncmp.S
index f6a17fd..02de93c 100644
--- a/sysdeps/aarch64/strncmp.S
+++ b/sysdeps/aarch64/strncmp.S
@@ -51,6 +51,9 @@
#define endloop x15
ENTRY_ALIGN_AND_PAD (strncmp, 6, 7)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
cbz limit, L(ret0)
eor tmp1, src1, src2
mov zeroones, #REP8_01
diff --git a/sysdeps/aarch64/strnlen.S b/sysdeps/aarch64/strnlen.S
index 4cce45f..af765f1 100644
--- a/sysdeps/aarch64/strnlen.S
+++ b/sysdeps/aarch64/strnlen.S
@@ -50,6 +50,9 @@
#define REP8_80 0x8080808080808080
ENTRY_ALIGN_AND_PAD (__strnlen, 6, 9)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
cbz limit, L(hit_limit)
mov zeroones, #REP8_01
bic src, srcin, #15
diff --git a/sysdeps/aarch64/strrchr.S b/sysdeps/aarch64/strrchr.S
index 44c1917..ea37968 100644
--- a/sysdeps/aarch64/strrchr.S
+++ b/sysdeps/aarch64/strrchr.S
@@ -68,6 +68,7 @@
identify exactly which byte is causing the termination, and why. */
ENTRY(strrchr)
+ DELOUSE (0)
cbz x1, L(null_search)
/* Magic constant 0x40100401 to allow us to identify which lane
matches the requested byte. Magic constant 0x80200802 used
diff --git a/sysdeps/aarch64/sysdep.h b/sysdeps/aarch64/sysdep.h
index e045759..0a7dccb 100644
--- a/sysdeps/aarch64/sysdep.h
+++ b/sysdeps/aarch64/sysdep.h
@@ -16,8 +16,25 @@
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
+#ifndef _AARCH64_SYSDEP_H
+#define _AARCH64_SYSDEP_H
+
#include <sysdeps/generic/sysdep.h>
+#ifdef __LP64__
+# define AARCH64_R(NAME) R_AARCH64_ ## NAME
+# define PTR_REG(n) x##n
+# define PTR_LOG_SIZE 3
+# define DELOUSE(n)
+#else
+# define AARCH64_R(NAME) R_AARCH64_P32_ ## NAME
+# define PTR_REG(n) w##n
+# define PTR_LOG_SIZE 2
+# define DELOUSE(n) mov w##n, w##n
+#endif
+
+#define PTR_SIZE (1<<PTR_LOG_SIZE)
+
#ifdef __ASSEMBLER__
/* Syntactic details of assembler. */
@@ -107,16 +124,18 @@
# define L(name) .L##name
#endif
-/* Load or store to/from a pc-relative EXPR into/from R, using T. */
-#define LDST_PCREL(OP, R, T, EXPR) \
- adrp T, EXPR; \
- OP R, [T, #:lo12:EXPR];\
+/* Load or store to/from a pc-relative EXPR into/from R, using T.
+ Note R and T are register numbers and not register names. */
+#define LDST_PCREL(OP, R, T, EXPR) \
+ adrp x##T, EXPR; \
+ OP PTR_REG (R), [x##T, #:lo12:EXPR]; \
-/* Load or store to/from a got-relative EXPR into/from R, using T. */
-#define LDST_GLOBAL(OP, R, T, EXPR) \
- adrp T, :got:EXPR; \
- ldr T, [T, #:got_lo12:EXPR];\
- OP R, [T];
+/* Load or store to/from a got-relative EXPR into/from R, using T.
+ Note R and T are register numbers and not register names. */
+#define LDST_GLOBAL(OP, R, T, EXPR) \
+ adrp x##T, :got:EXPR; \
+ ldr PTR_REG (T), [x##T, #:got_lo12:EXPR]; \
+ OP PTR_REG (R), [x##T];
/* Since C identifiers are not normally prefixed with an underscore
on this system, the asm identifier `syscall_error' intrudes on the
@@ -125,3 +144,5 @@
#define mcount _mcount
#endif /* __ASSEMBLER__ */
+
+#endif /* _AARCH64_SYSDEP_H */
diff --git a/sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h b/sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h
index 658f696..1717c35 100644
--- a/sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h
+++ b/sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h
@@ -25,11 +25,17 @@
#define __O_NOFOLLOW 0100000
#define __O_DIRECT 0200000
-#define __O_LARGEFILE 0
+#ifdef __ILP32__
+# define __O_LARGEFILE 0400000
+#else
+# define __O_LARGEFILE 0
+#endif
+#ifdef __LP64__
# define F_GETLK64 5
# define F_SETLK64 6
# define F_SETLKW64 7
+#endif
struct flock
{
diff --git a/sysdeps/unix/sysv/linux/aarch64/clone.S b/sysdeps/unix/sysv/linux/aarch64/clone.S
index 76baa7a..220ec11 100644
--- a/sysdeps/unix/sysv/linux/aarch64/clone.S
+++ b/sysdeps/unix/sysv/linux/aarch64/clone.S
@@ -39,6 +39,13 @@
*/
.text
ENTRY(__clone)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
+ DELOUSE (3)
+ DELOUSE (4)
+ DELOUSE (5)
+ DELOUSE (6)
/* Save args for the child. */
mov x10, x0
mov x11, x2
diff --git a/sysdeps/unix/sysv/linux/aarch64/dl-cache.h b/sysdeps/unix/sysv/linux/aarch64/dl-cache.h
index 9c7b271..044fdba 100644
--- a/sysdeps/unix/sysv/linux/aarch64/dl-cache.h
+++ b/sysdeps/unix/sysv/linux/aarch64/dl-cache.h
@@ -18,7 +18,11 @@
#include <ldconfig.h>
+#ifdef __LP64__
#define _DL_CACHE_DEFAULT_ID (FLAG_AARCH64_LIB64 | FLAG_ELF_LIBC6)
+#else
+#define _DL_CACHE_DEFAULT_ID (FLAG_AARCH64_LIB32 | FLAG_ELF_LIBC6)
+#endif
#define _dl_cache_check_flags(flags) \
((flags) == _DL_CACHE_DEFAULT_ID)
@@ -27,18 +31,25 @@
do \
{ \
size_t len = strlen (dir); \
- char path[len + 3]; \
+ char path[len + 6]; \
memcpy (path, dir, len + 1); \
if (len >= 6 && ! memcmp (path + len - 6, "/lib64", 6)) \
{ \
len -= 2; \
path[len] = '\0'; \
} \
+ if (len >= 9 && ! memcmp (path + len - 9, "/libilp32", 9))\
+ { \
+ len -= 5; \
+ path[len] = '\0'; \
+ } \
add_dir (path); \
if (len >= 4 && ! memcmp (path + len - 4, "/lib", 4)) \
{ \
memcpy (path + len, "64", 3); \
add_dir (path); \
+ memcpy (path + len, "ilp32", 6); \
+ add_dir (path); \
} \
} while (0)
diff --git a/sysdeps/unix/sysv/linux/aarch64/getcontext.S b/sysdeps/unix/sysv/linux/aarch64/getcontext.S
index c2dd5b8..f6bf24f 100644
--- a/sysdeps/unix/sysv/linux/aarch64/getcontext.S
+++ b/sysdeps/unix/sysv/linux/aarch64/getcontext.S
@@ -30,6 +30,7 @@
.text
ENTRY(__getcontext)
+ DELOUSE (0)
/* The saved context will return to the getcontext() call point
with a return value of 0 */
str xzr, [x0, oX0 + 0 * SZREG]
@@ -90,7 +91,7 @@ ENTRY(__getcontext)
/* Grab the signal mask */
/* rt_sigprocmask (SIG_BLOCK, NULL, &ucp->uc_sigmask, _NSIG8) */
- add x2, x0, #UCONTEXT_SIGMASK
+ add PTR_REG (2), PTR_REG (0), #UCONTEXT_SIGMASK
mov x0, SIG_BLOCK
mov x1, 0
mov x3, _NSIG8
diff --git a/sysdeps/unix/sysv/linux/aarch64/init-first.c b/sysdeps/unix/sysv/linux/aarch64/init-first.c
index f7224a2..f7bfc4d 100644
--- a/sysdeps/unix/sysv/linux/aarch64/init-first.c
+++ b/sysdeps/unix/sysv/linux/aarch64/init-first.c
@@ -27,17 +27,21 @@ int (*VDSO_SYMBOL(clock_getres)) (clockid_t, struct timespec *);
static inline void
_libc_vdso_platform_setup (void)
{
- PREPARE_VERSION (linux2639, "LINUX_2.6.39", 123718537);
+#ifdef __LP64__
+ PREPARE_VERSION (linux_version, "LINUX_2.6.39", 123718537);
+#else
+ PREPARE_VERSION (linux_version, "LINUX_4.9", 61765625);
+#endif
- void *p = _dl_vdso_vsym ("__kernel_gettimeofday", &linux2639);
+ void *p = _dl_vdso_vsym ("__kernel_gettimeofday", &linux_version);
PTR_MANGLE (p);
VDSO_SYMBOL(gettimeofday) = p;
- p = _dl_vdso_vsym ("__kernel_clock_gettime", &linux2639);
+ p = _dl_vdso_vsym ("__kernel_clock_gettime", &linux_version);
PTR_MANGLE (p);
VDSO_SYMBOL(clock_gettime) = p;
- p = _dl_vdso_vsym ("__kernel_clock_getres", &linux2639);
+ p = _dl_vdso_vsym ("__kernel_clock_getres", &linux_version);
PTR_MANGLE (p);
VDSO_SYMBOL(clock_getres) = p;
}
diff --git a/sysdeps/unix/sysv/linux/aarch64/ldconfig.h b/sysdeps/unix/sysv/linux/aarch64/ldconfig.h
index ee91ef8..ac84194 100644
--- a/sysdeps/unix/sysv/linux/aarch64/ldconfig.h
+++ b/sysdeps/unix/sysv/linux/aarch64/ldconfig.h
@@ -21,6 +21,8 @@
#define SYSDEP_KNOWN_INTERPRETER_NAMES \
{ "/lib/ld-linux-aarch64.so.1", FLAG_ELF_LIBC6 }, \
{ "/lib/ld-linux-aarch64_be.so.1", FLAG_ELF_LIBC6 }, \
+ { "/lib/ld-linux-aarch64_ilp32.so.1", FLAG_ELF_LIBC6 }, \
+ { "/lib/ld-linux-aarch64_be_ilp32.so.1", FLAG_ELF_LIBC6 }, \
{ "/lib/ld-linux.so.3", FLAG_ELF_LIBC6 }, \
{ "/lib/ld-linux-armhf.so.3", FLAG_ELF_LIBC6 },
#define SYSDEP_KNOWN_LIBRARY_NAMES \
diff --git a/sysdeps/unix/sysv/linux/aarch64/setcontext.S b/sysdeps/unix/sysv/linux/aarch64/setcontext.S
index d17f8c8..c2bca26 100644
--- a/sysdeps/unix/sysv/linux/aarch64/setcontext.S
+++ b/sysdeps/unix/sysv/linux/aarch64/setcontext.S
@@ -34,6 +34,7 @@
.text
ENTRY (__setcontext)
+ DELOUSE (0)
/* Save a copy of UCP. */
mov x9, x0
diff --git a/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h b/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h
index a579501..ee54222 100644
--- a/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h
+++ b/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h
@@ -19,7 +19,7 @@
#include <sys/ucontext.h>
#define SIGCONTEXT siginfo_t *_si, struct ucontext *
-#define GET_PC(ctx) ((void *) (ctx)->uc_mcontext.pc)
+#define GET_PC(ctx) ((void *) (size_t) (ctx)->uc_mcontext.pc)
/* There is no reliable way to get the sigcontext unless we use a
three-argument signal handler. */
diff --git a/sysdeps/unix/sysv/linux/aarch64/swapcontext.S b/sysdeps/unix/sysv/linux/aarch64/swapcontext.S
index c1a16f3..8e2cadd 100644
--- a/sysdeps/unix/sysv/linux/aarch64/swapcontext.S
+++ b/sysdeps/unix/sysv/linux/aarch64/swapcontext.S
@@ -27,6 +27,7 @@
.text
ENTRY(__swapcontext)
+ DELOUSE (0)
/* Set the value returned when swapcontext() returns in this context. */
str xzr, [x0, oX0 + 0 * SZREG]
diff --git a/sysdeps/unix/sysv/linux/aarch64/sysdep.h b/sysdeps/unix/sysv/linux/aarch64/sysdep.h
index a397e50..1ffabc2 100644
--- a/sysdeps/unix/sysv/linux/aarch64/sysdep.h
+++ b/sysdeps/unix/sysv/linux/aarch64/sysdep.h
@@ -250,12 +250,14 @@
(!defined SHARED && (IS_IN (libc) \
|| IS_IN (libpthread))))
# ifdef __ASSEMBLER__
+/* Note, dst, src, guard, and tmp are all register numbers rather than
+ register names so they will work with both ILP32 and LP64. */
# define PTR_MANGLE(dst, src, guard, tmp) \
LDST_PCREL (ldr, guard, tmp, C_SYMBOL_NAME(__pointer_chk_guard_local)); \
PTR_MANGLE2 (dst, src, guard)
/* Use PTR_MANGLE2 for efficiency if guard is already loaded. */
# define PTR_MANGLE2(dst, src, guard)\
- eor dst, src, guard
+ eor x##dst, x##src, x##guard
# define PTR_DEMANGLE(dst, src, guard, tmp)\
PTR_MANGLE (dst, src, guard, tmp)
# define PTR_DEMANGLE2(dst, src, guard)\
@@ -268,12 +270,14 @@ extern uintptr_t __pointer_chk_guard_local attribute_relro attribute_hidden;
# endif
#else
# ifdef __ASSEMBLER__
+/* Note, dst, src, guard, and tmp are all register numbers rather than
+ register names so they will work with both ILP32 and LP64. */
# define PTR_MANGLE(dst, src, guard, tmp) \
LDST_GLOBAL (ldr, guard, tmp, C_SYMBOL_NAME(__pointer_chk_guard)); \
PTR_MANGLE2 (dst, src, guard)
/* Use PTR_MANGLE2 for efficiency if guard is already loaded. */
# define PTR_MANGLE2(dst, src, guard)\
- eor dst, src, guard
+ eor x##dst, x##src, x##guard
# define PTR_DEMANGLE(dst, src, guard, tmp)\
PTR_MANGLE (dst, src, guard, tmp)
# define PTR_DEMANGLE2(dst, src, guard)\
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] Partial ILP32 support for aarch64
2016-11-19 0:40 [PATCH] Partial ILP32 support for aarch64 Steve Ellcey
@ 2016-11-21 16:51 ` Joseph Myers
2016-11-21 20:50 ` Steve Ellcey
0 siblings, 1 reply; 12+ messages in thread
From: Joseph Myers @ 2016-11-21 16:51 UTC (permalink / raw)
To: Steve Ellcey; +Cc: libc-alpha
On Fri, 18 Nov 2016, Steve Ellcey wrote:
> diff --git a/sysdeps/aarch64/nptl/bits/semaphore.h b/sysdeps/aarch64/nptl/bits/semaphore.h
> index 3cc5b37..1d1389c 100644
> --- a/sysdeps/aarch64/nptl/bits/semaphore.h
> +++ b/sysdeps/aarch64/nptl/bits/semaphore.h
> @@ -21,7 +21,11 @@
> #endif
>
>
> +#ifdef __ILP32__
> +#define __SIZEOF_SEM_T 16
> +#else
> #define __SIZEOF_SEM_T 32
> +#endif
Missing preprocessor indentation inside #ifdef, should be "# define" in
both branches of the #ifdef.
> diff --git a/sysdeps/unix/sysv/linux/aarch64/dl-cache.h b/sysdeps/unix/sysv/linux/aarch64/dl-cache.h
> index 9c7b271..044fdba 100644
> --- a/sysdeps/unix/sysv/linux/aarch64/dl-cache.h
> +++ b/sysdeps/unix/sysv/linux/aarch64/dl-cache.h
> @@ -18,7 +18,11 @@
>
> #include <ldconfig.h>
>
> +#ifdef __LP64__
> #define _DL_CACHE_DEFAULT_ID (FLAG_AARCH64_LIB64 | FLAG_ELF_LIBC6)
> +#else
> +#define _DL_CACHE_DEFAULT_ID (FLAG_AARCH64_LIB32 | FLAG_ELF_LIBC6)
> +#endif
Likewise.
--
Joseph S. Myers
joseph@codesourcery.com
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] Partial ILP32 support for aarch64
2016-11-21 16:51 ` Joseph Myers
@ 2016-11-21 20:50 ` Steve Ellcey
2016-11-23 9:44 ` Andreas Schwab
0 siblings, 1 reply; 12+ messages in thread
From: Steve Ellcey @ 2016-11-21 20:50 UTC (permalink / raw)
To: Joseph Myers; +Cc: libc-alpha
[-- Attachment #1: Type: text/plain, Size: 3343 bytes --]
Here is an updated version with the indentation fixed.
Steve Ellcey
sellcey@caviumnetworks.com
2016-11-21  Andrew Pinski  <andrew.pinski@caviumnetworks.com>
    Yury Norov  <ynorov@caviumnetworks.com>
    Steve Ellcey  <sellcey@caviumnetworks.com>
* sysdeps/aarch64/crti.S: Add include of sysdep.h.
(call_weak_fn): Use PTR_REG to get correct reg name in ILP32.
* sysdeps/aarch64/dl-irel.h: Add include of sysdep.h.
(elf_irela): Use AARCH64_R macro to get correct relocation in ILP32.
* sysdeps/aarch64/dl-machine.h: Add include of sysdep.h.
(elf_machine_load_address, RTLD_START, RTLD_START_1, RTLD_START,
elf_machine_type_class, ELF_MACHINE_JMP_SLOT, elf_machine_rela,
elf_machine_lazy_rel): Add ifdef's for ILP32 support.
* sysdeps/aarch64/dl-tlsdesc.S (_dl_tlsdesc_return,
_dl_tlsdesc_return_lazy, _dl_tlsdesc_dynamic,
_dl_tlsdesc_resolve_hold): Extend pointers in ILP32, use PTR_REG
to get correct reg name for ILP32.
* sysdeps/aarch64/dl-trampoline.S (ip01): New Macro.
(RELA_SIZE): New Macro.
(_dl_runtime_resolve, _dl_runtime_profile): Use new macros and PTR_REG
to support ILP32.
* sysdeps/aarch64/jmpbuf-unwind.h (_JMPBUF_CFA_UNWINDS_ADJ): Add
cast for ILP32 mode.
* sysdeps/aarch64/memcmp.S (memcmp): Extend arg pointers for ILP32 mode.
* sysdeps/aarch64/memcpy.S (memmove, memcpy): Ditto.
* sysdeps/aarch64/memset.S (__memset): Ditto.
* sysdeps/aarch64/strchr.S (strchr): Ditto.
* sysdeps/aarch64/strchrnul.S (__strchrnul): Ditto.
* sysdeps/aarch64/strcmp.S (strcmp): Ditto.
* sysdeps/aarch64/strcpy.S (strcpy): Ditto.
* sysdeps/aarch64/strlen.S (__strlen): Ditto.
* sysdeps/aarch64/strncmp.S (strncmp): Ditto.
* sysdeps/aarch64/strnlen.S (strnlen): Ditto.
* sysdeps/aarch64/strrchr.S (strrchr): Ditto.
* sysdeps/unix/sysv/linux/aarch64/clone.S: Ditto.
* sysdeps/unix/sysv/linux/aarch64/setcontext.S (__setcontext): Ditto.
* sysdeps/unix/sysv/linux/aarch64/swapcontext.S (__swapcontext): Ditto.
* sysdeps/aarch64/__longjmp.S (__longjmp): Extend pointers in ILP32,
change PTR_MANGLE call to use register numbers instead of names.
* sysdeps/unix/sysv/linux/aarch64/getcontext.S (__getcontext): Ditto.
* sysdeps/aarch64/setjmp.S (__sigsetjmp): Extend arg pointers for
ILP32 mode, change PTR_MANGLE calls to use register numbers.
* sysdeps/aarch64/start.S (_start): Ditto.
* sysdeps/aarch64/nptl/bits/pthreadtypes.h
(__PTHREAD_RWLOCK_INT_FLAGS_SHARED): New define.
* sysdeps/aarch64/nptl/bits/semaphore.h (__SIZEOF_SEM_T): Change define.
* sysdeps/aarch64/sysdep.h (AARCH64_R, PTR_REG, PTR_LOG_SIZE, DELOUSE,
PTR_SIZE): New Macros.
(LDST_PCREL, LDST_GLOBAL) Update to use PTR_REG.
* sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h (O_LARGEFILE):
Set when in ILP32 mode.
(F_GETLK64, F_SETLK64, F_SETLKW64): Only set in LP64 mode.
* sysdeps/unix/sysv/linux/aarch64/dl-cache.h (DL_CACHE_DEFAULT_ID):
Set elf flags for ILP32.
(add_system_dir): Set ILP32 library directories.
* sysdeps/unix/sysv/linux/aarch64/init-first.c
(_libc_vdso_platform_setup): Set minimum kernel version for ILP32.
* sysdeps/unix/sysv/linux/aarch64/ldconfig.h
(SYSDEP_KNOWN_INTERPRETER_NAMES): Add ILP32 names.
* sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h (GET_PC, SET_PC):
New Macros.
* sysdeps/unix/sysv/linux/aarch64/sysdep.h: Handle ILP32 pointers.
[-- Attachment #2: aarch64_ilp32.patch --]
[-- Type: text/x-patch, Size: 36109 bytes --]
diff --git a/sysdeps/aarch64/__longjmp.S b/sysdeps/aarch64/__longjmp.S
index 65116be..4d411fe 100644
--- a/sysdeps/aarch64/__longjmp.S
+++ b/sysdeps/aarch64/__longjmp.S
@@ -46,6 +46,8 @@ ENTRY (__longjmp)
cfi_offset(d14, JB_D14<<3)
cfi_offset(d15, JB_D15<<3)
+ DELOUSE (0)
+
ldp x19, x20, [x0, #JB_X19<<3]
ldp x21, x22, [x0, #JB_X21<<3]
ldp x23, x24, [x0, #JB_X23<<3]
@@ -53,7 +55,7 @@ ENTRY (__longjmp)
ldp x27, x28, [x0, #JB_X27<<3]
#ifdef PTR_DEMANGLE
ldp x29, x4, [x0, #JB_X29<<3]
- PTR_DEMANGLE (x30, x4, x3, x2)
+ PTR_DEMANGLE (30, 4, 3, 2)
#else
ldp x29, x30, [x0, #JB_X29<<3]
#endif
@@ -98,7 +100,7 @@ ENTRY (__longjmp)
cfi_same_value(d15)
#ifdef PTR_DEMANGLE
ldr x4, [x0, #JB_SP<<3]
- PTR_DEMANGLE (x5, x4, x3, x2)
+ PTR_DEMANGLE (5, 4, 3, 2)
#else
ldr x5, [x0, #JB_SP<<3]
#endif
diff --git a/sysdeps/aarch64/crti.S b/sysdeps/aarch64/crti.S
index 53ccb42..5c42fd5 100644
--- a/sysdeps/aarch64/crti.S
+++ b/sysdeps/aarch64/crti.S
@@ -39,6 +39,7 @@
they can be called as functions. The symbols _init and _fini are
magic and cause the linker to emit DT_INIT and DT_FINI. */
+#include <sysdep.h>
#include <libc-symbols.h>
#ifndef PREINIT_FUNCTION
@@ -60,7 +61,7 @@
.type call_weak_fn, %function
call_weak_fn:
adrp x0, :got:PREINIT_FUNCTION
- ldr x0, [x0, #:got_lo12:PREINIT_FUNCTION]
+ ldr PTR_REG (0), [x0, #:got_lo12:PREINIT_FUNCTION]
cbz x0, 1f
b PREINIT_FUNCTION
1:
diff --git a/sysdeps/aarch64/dl-irel.h b/sysdeps/aarch64/dl-irel.h
index 63a8e50..2effca4 100644
--- a/sysdeps/aarch64/dl-irel.h
+++ b/sysdeps/aarch64/dl-irel.h
@@ -23,6 +23,7 @@
#include <stdio.h>
#include <unistd.h>
#include <ldsodefs.h>
+#include <sysdep.h>
#define ELF_MACHINE_IRELA 1
@@ -40,7 +41,7 @@ elf_irela (const ElfW(Rela) *reloc)
ElfW(Addr) *const reloc_addr = (void *) reloc->r_offset;
const unsigned long int r_type = ELFW(R_TYPE) (reloc->r_info);
- if (__glibc_likely (r_type == R_AARCH64_IRELATIVE))
+ if (__glibc_likely (r_type == AARCH64_R (IRELATIVE)))
{
ElfW(Addr) value = elf_ifunc_invoke (reloc->r_addend);
*reloc_addr = value;
diff --git a/sysdeps/aarch64/dl-machine.h b/sysdeps/aarch64/dl-machine.h
index 282805e..b5ea7a8 100644
--- a/sysdeps/aarch64/dl-machine.h
+++ b/sysdeps/aarch64/dl-machine.h
@@ -21,6 +21,7 @@
#define ELF_MACHINE_NAME "aarch64"
+#include <sysdep.h>
#include <tls.h>
#include <dl-tlsdesc.h>
#include <dl-irel.h>
@@ -53,19 +54,33 @@ elf_machine_load_address (void)
by constructing a non GOT reference to the symbol, the dynamic
address of the symbol we compute using adrp/add to compute the
symbol's address relative to the PC.
- This depends on 32bit relocations being resolved at link time
- and that the static address fits in the 32bits. */
+ This depends on 32/16bit relocations being resolved at link time
+ and that the static address fits in the 32/16 bits. */
ElfW(Addr) static_addr;
ElfW(Addr) dynamic_addr;
asm (" \n"
" adrp %1, _dl_start; \n"
+#ifdef __LP64__
" add %1, %1, #:lo12:_dl_start \n"
+#else
+" add %w1, %w1, #:lo12:_dl_start \n"
+#endif
" ldr %w0, 1f \n"
" b 2f \n"
"1: \n"
+#ifdef __LP64__
" .word _dl_start \n"
+#else
+# ifdef __AARCH64EB__
+" .short 0 \n"
+# endif
+" .short _dl_start \n"
+# ifndef __AARCH64EB__
+" .short 0 \n"
+# endif
+#endif
"2: \n"
: "=r" (static_addr), "=r" (dynamic_addr));
return dynamic_addr - static_addr;
@@ -125,80 +140,86 @@ elf_machine_runtime_setup (struct link_map *l, int lazy, int profile)
/* Initial entry point for the dynamic linker. The C function
_dl_start is the real entry point, its return value is the user
program's entry point */
+#ifdef __LP64__
+# define RTLD_START RTLD_START_1 ("x", "3", "sp")
+#else
+# define RTLD_START RTLD_START_1 ("w", "2", "wsp")
+#endif
-#define RTLD_START asm ("\
-.text \n\
-.globl _start \n\
-.type _start, %function \n\
-.globl _dl_start_user \n\
-.type _dl_start_user, %function \n\
-_start: \n\
- mov x0, sp \n\
- bl _dl_start \n\
- // returns user entry point in x0 \n\
- mov x21, x0 \n\
-_dl_start_user: \n\
- // get the original arg count \n\
- ldr x1, [sp] \n\
- // get the argv address \n\
- add x2, sp, #8 \n\
- // get _dl_skip_args to see if we were \n\
- // invoked as an executable \n\
- adrp x4, _dl_skip_args \n\
- ldr w4, [x4, #:lo12:_dl_skip_args] \n\
- // do we need to adjust argc/argv \n\
- cmp w4, 0 \n\
- beq .L_done_stack_adjust \n\
- // subtract _dl_skip_args from original arg count \n\
- sub x1, x1, x4 \n\
- // store adjusted argc back to stack \n\
- str x1, [sp] \n\
- // find the first unskipped argument \n\
- mov x3, x2 \n\
- add x4, x2, x4, lsl #3 \n\
- // shuffle argv down \n\
-1: ldr x5, [x4], #8 \n\
- str x5, [x3], #8 \n\
- cmp x5, #0 \n\
- bne 1b \n\
- // shuffle envp down \n\
-1: ldr x5, [x4], #8 \n\
- str x5, [x3], #8 \n\
- cmp x5, #0 \n\
- bne 1b \n\
- // shuffle auxv down \n\
-1: ldp x0, x5, [x4, #16]! \n\
- stp x0, x5, [x3], #16 \n\
- cmp x0, #0 \n\
- bne 1b \n\
- // Update _dl_argv \n\
- adrp x3, _dl_argv \n\
- str x2, [x3, #:lo12:_dl_argv] \n\
-.L_done_stack_adjust: \n\
- // compute envp \n\
- add x3, x2, x1, lsl #3 \n\
- add x3, x3, #8 \n\
- adrp x16, _rtld_local \n\
- add x16, x16, #:lo12:_rtld_local \n\
- ldr x0, [x16] \n\
- bl _dl_init \n\
- // load the finalizer function \n\
- adrp x0, _dl_fini \n\
- add x0, x0, #:lo12:_dl_fini \n\
- // jump to the user_s entry point \n\
- br x21 \n\
+
+#define RTLD_START_1(PTR, PTR_SIZE_LOG, PTR_SP) asm ("\
+.text \n\
+.globl _start \n\
+.type _start, %function \n\
+.globl _dl_start_user \n\
+.type _dl_start_user, %function \n\
+_start: \n\
+ mov " PTR "0, " PTR_SP " \n\
+ bl _dl_start \n\
+ // returns user entry point in x0 \n\
+ mov x21, x0 \n\
+_dl_start_user: \n\
+ // get the original arg count \n\
+ ldr " PTR "1, [sp] \n\
+ // get the argv address \n\
+ add " PTR "2, " PTR_SP ", #(1<<" PTR_SIZE_LOG ") \n\
+ // get _dl_skip_args to see if we were \n\
+ // invoked as an executable \n\
+ adrp x4, _dl_skip_args \n\
+ ldr w4, [x4, #:lo12:_dl_skip_args] \n\
+ // do we need to adjust argc/argv \n\
+ cmp w4, 0 \n\
+ beq .L_done_stack_adjust \n\
+ // subtract _dl_skip_args from original arg count \n\
+ sub " PTR "1, " PTR "1, " PTR "4 \n\
+ // store adjusted argc back to stack \n\
+ str " PTR "1, [sp] \n\
+ // find the first unskipped argument \n\
+ mov " PTR "3, " PTR "2 \n\
+ add " PTR "4, " PTR "2, " PTR "4, lsl #" PTR_SIZE_LOG " \n\
+ // shuffle argv down \n\
+1: ldr " PTR "5, [x4], #(1<<" PTR_SIZE_LOG ") \n\
+ str " PTR "5, [x3], #(1<<" PTR_SIZE_LOG ") \n\
+ cmp " PTR "5, #0 \n\
+ bne 1b \n\
+ // shuffle envp down \n\
+1: ldr " PTR "5, [x4], #(1<<" PTR_SIZE_LOG ") \n\
+ str " PTR "5, [x3], #(1<<" PTR_SIZE_LOG ") \n\
+ cmp " PTR "5, #0 \n\
+ bne 1b \n\
+ // shuffle auxv down \n\
+1: ldp " PTR "0, " PTR "5, [x4, #(2<<" PTR_SIZE_LOG ")]! \n\
+ stp " PTR "0, " PTR "5, [x3], #(2<<" PTR_SIZE_LOG ") \n\
+ cmp " PTR "0, #0 \n\
+ bne 1b \n\
+ // Update _dl_argv \n\
+ adrp x3, _dl_argv \n\
+ str " PTR "2, [x3, #:lo12:_dl_argv] \n\
+.L_done_stack_adjust: \n\
+ // compute envp \n\
+ add " PTR "3, " PTR "2, " PTR "1, lsl #" PTR_SIZE_LOG " \n\
+ add " PTR "3, " PTR "3, #(1<<" PTR_SIZE_LOG ") \n\
+ adrp x16, _rtld_local \n\
+ add " PTR "16, " PTR "16, #:lo12:_rtld_local \n\
+ ldr " PTR "0, [x16] \n\
+ bl _dl_init \n\
+ // load the finalizer function \n\
+ adrp x0, _dl_fini \n\
+ add " PTR "0, " PTR "0, #:lo12:_dl_fini \n\
+ // jump to the user_s entry point \n\
+ br x21 \n\
");
#define elf_machine_type_class(type) \
- ((((type) == R_AARCH64_JUMP_SLOT || \
- (type) == R_AARCH64_TLS_DTPMOD || \
- (type) == R_AARCH64_TLS_DTPREL || \
- (type) == R_AARCH64_TLS_TPREL || \
- (type) == R_AARCH64_TLSDESC) * ELF_RTYPE_CLASS_PLT) \
- | (((type) == R_AARCH64_COPY) * ELF_RTYPE_CLASS_COPY) \
- | (((type) == R_AARCH64_GLOB_DAT) * ELF_RTYPE_CLASS_EXTERN_PROTECTED_DATA))
+ ((((type) == AARCH64_R (JUMP_SLOT) \
+ || (type) == AARCH64_R (TLS_DTPMOD) \
+ || (type) == AARCH64_R (TLS_DTPREL) \
+ || (type) == AARCH64_R (TLS_TPREL) \
+ || (type) == AARCH64_R (TLSDESC)) * ELF_RTYPE_CLASS_PLT) \
+ | (((type) == AARCH64_R (COPY)) * ELF_RTYPE_CLASS_COPY) \
+ | (((type) == AARCH64_R (GLOB_DAT)) * ELF_RTYPE_CLASS_EXTERN_PROTECTED_DATA))
-#define ELF_MACHINE_JMP_SLOT R_AARCH64_JUMP_SLOT
+#define ELF_MACHINE_JMP_SLOT AARCH64_R (JUMP_SLOT)
/* AArch64 uses RELA not REL */
#define ELF_MACHINE_NO_REL 1
@@ -237,9 +258,9 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
void *const reloc_addr_arg, int skip_ifunc)
{
ElfW(Addr) *const reloc_addr = reloc_addr_arg;
- const unsigned int r_type = ELF64_R_TYPE (reloc->r_info);
+ const unsigned int r_type = ELFW (R_TYPE) (reloc->r_info);
- if (__builtin_expect (r_type == R_AARCH64_RELATIVE, 0))
+ if (__builtin_expect (r_type == AARCH64_R (RELATIVE), 0))
*reloc_addr = map->l_addr + reloc->r_addend;
else if (__builtin_expect (r_type == R_AARCH64_NONE, 0))
return;
@@ -257,7 +278,7 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
switch (r_type)
{
- case R_AARCH64_COPY:
+ case AARCH64_R (COPY):
if (sym == NULL)
break;
@@ -275,15 +296,17 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
MIN (sym->st_size, refsym->st_size));
break;
- case R_AARCH64_RELATIVE:
- case R_AARCH64_GLOB_DAT:
- case R_AARCH64_JUMP_SLOT:
- case R_AARCH64_ABS32:
- case R_AARCH64_ABS64:
+ case AARCH64_R (RELATIVE):
+ case AARCH64_R (GLOB_DAT):
+ case AARCH64_R (JUMP_SLOT):
+ case AARCH64_R (ABS32):
+#ifdef __LP64__
+ case AARCH64_R (ABS64):
+#endif
*reloc_addr = value + reloc->r_addend;
break;
- case R_AARCH64_TLSDESC:
+ case AARCH64_R (TLSDESC):
{
struct tlsdesc volatile *td =
(struct tlsdesc volatile *)reloc_addr;
@@ -318,7 +341,7 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
break;
}
- case R_AARCH64_TLS_DTPMOD:
+ case AARCH64_R (TLS_DTPMOD):
#ifdef RTLD_BOOTSTRAP
*reloc_addr = 1;
#else
@@ -329,12 +352,12 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
#endif
break;
- case R_AARCH64_TLS_DTPREL:
+ case AARCH64_R (TLS_DTPREL):
if (sym)
*reloc_addr = sym->st_value + reloc->r_addend;
break;
- case R_AARCH64_TLS_TPREL:
+ case AARCH64_R (TLS_TPREL):
if (sym)
{
CHECK_STATIC_TLS (map, sym_map);
@@ -343,7 +366,7 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
}
break;
- case R_AARCH64_IRELATIVE:
+ case AARCH64_R (IRELATIVE):
value = map->l_addr + reloc->r_addend;
value = elf_ifunc_invoke (value);
*reloc_addr = value;
@@ -374,16 +397,16 @@ elf_machine_lazy_rel (struct link_map *map,
int skip_ifunc)
{
ElfW(Addr) *const reloc_addr = (void *) (l_addr + reloc->r_offset);
- const unsigned int r_type = ELF64_R_TYPE (reloc->r_info);
+ const unsigned int r_type = ELFW (R_TYPE) (reloc->r_info);
/* Check for unexpected PLT reloc type. */
- if (__builtin_expect (r_type == R_AARCH64_JUMP_SLOT, 1))
+ if (__builtin_expect (r_type == AARCH64_R (JUMP_SLOT), 1))
{
if (__builtin_expect (map->l_mach.plt, 0) == 0)
*reloc_addr += l_addr;
else
*reloc_addr = map->l_mach.plt;
}
- else if (__builtin_expect (r_type == R_AARCH64_TLSDESC, 1))
+ else if (__builtin_expect (r_type == AARCH64_R (TLSDESC), 1))
{
struct tlsdesc volatile *td =
(struct tlsdesc volatile *)reloc_addr;
@@ -392,7 +415,7 @@ elf_machine_lazy_rel (struct link_map *map,
td->entry = (void*)(D_PTR (map, l_info[ADDRIDX (DT_TLSDESC_PLT)])
+ map->l_addr);
}
- else if (__glibc_unlikely (r_type == R_AARCH64_IRELATIVE))
+ else if (__glibc_unlikely (r_type == AARCH64_R (IRELATIVE)))
{
ElfW(Addr) value = map->l_addr + reloc->r_addend;
if (__glibc_likely (!skip_ifunc))
diff --git a/sysdeps/aarch64/dl-tlsdesc.S b/sysdeps/aarch64/dl-tlsdesc.S
index 05be370..42fa943 100644
--- a/sysdeps/aarch64/dl-tlsdesc.S
+++ b/sysdeps/aarch64/dl-tlsdesc.S
@@ -74,7 +74,8 @@
cfi_startproc
.align 2
_dl_tlsdesc_return:
- ldr x0, [x0, #8]
+ DELOUSE (0)
+ ldr PTR_REG (0), [x0, #PTR_SIZE]
RET
cfi_endproc
.size _dl_tlsdesc_return, .-_dl_tlsdesc_return
@@ -95,9 +96,10 @@ _dl_tlsdesc_return_lazy:
so it reads the same value (this function is the final value of
td->entry) and thus it synchronizes with the release store to
td->entry in _dl_tlsdesc_resolve_rela_fixup ensuring that the load
- from [x0,#8] here happens after the initialization of td->arg. */
- ldar xzr, [x0]
- ldr x0, [x0, #8]
+ from [x0,#PTR_SIZE] here happens after the initialization of td->arg. */
+ DELOUSE (0)
+ ldar PTR_REG (zr), [x0]
+ ldr PTR_REG (0), [x0, #PTR_SIZE]
RET
cfi_endproc
.size _dl_tlsdesc_return_lazy, .-_dl_tlsdesc_return_lazy
@@ -125,10 +127,11 @@ _dl_tlsdesc_undefweak:
td->entry) and thus it synchronizes with the release store to
td->entry in _dl_tlsdesc_resolve_rela_fixup ensuring that the load
from [x0,#8] here happens after the initialization of td->arg. */
- ldar xzr, [x0]
- ldr x0, [x0, #8]
+ DELOUSE (0)
+ ldar PTR_REG (zr), [x0]
+ ldr PTR_REG (0), [x0, #PTR_SIZE]
mrs x1, tpidr_el0
- sub x0, x0, x1
+ sub PTR_REG (0), PTR_REG (0), PTR_REG (1)
ldr x1, [sp], #16
cfi_adjust_cfa_offset (-16)
RET
@@ -174,6 +177,7 @@ _dl_tlsdesc_dynamic:
stp x29, x30, [sp,#-(32+16*NSAVEXREGPAIRS)]!
cfi_adjust_cfa_offset (32+16*NSAVEXREGPAIRS)
mov x29, sp
+ DELOUSE (0)
/* Save just enough registers to support fast path, if we fall
into slow path we will save additional registers. */
@@ -187,22 +191,22 @@ _dl_tlsdesc_dynamic:
so it reads the same value (this function is the final value of
td->entry) and thus it synchronizes with the release store to
td->entry in _dl_tlsdesc_resolve_rela_fixup ensuring that the load
- from [x0,#8] here happens after the initialization of td->arg. */
- ldar xzr, [x0]
- ldr x1, [x0,#8]
- ldr x0, [x4]
- ldr x3, [x1,#16]
- ldr x2, [x0]
- cmp x3, x2
+ from [x0,#PTR_SIZE] here happens after the initialization of td->arg. */
+ ldar PTR_REG (zr), [x0]
+ ldr PTR_REG (1), [x0,#PTR_SIZE]
+ ldr PTR_REG (0), [x4]
+ ldr PTR_REG (3), [x1,#(PTR_SIZE * 2)]
+ ldr PTR_REG (2), [x0]
+ cmp PTR_REG (3), PTR_REG (2)
b.hi 2f
- ldr x2, [x1]
- add x0, x0, x2, lsl #4
- ldr x0, [x0]
+ ldr PTR_REG (2), [x1]
+ add PTR_REG (0), PTR_REG (0), PTR_REG (2), lsl #(PTR_LOG_SIZE + 1)
+ ldr PTR_REG (0), [x0]
cmn x0, #0x1
b.eq 2f
- ldr x1, [x1,#8]
- add x0, x0, x1
- sub x0, x0, x4
+ ldr PTR_REG (1), [x1,#(PTR_SIZE * 2)]
+ add PTR_REG (0), PTR_REG (0), PTR_REG (1)
+ sub PTR_REG (0), PTR_REG (0), PTR_REG (4)
1:
ldp x1, x2, [sp, #32+16*0]
ldp x3, x4, [sp, #32+16*1]
@@ -233,7 +237,7 @@ _dl_tlsdesc_dynamic:
bl __tls_get_addr
mrs x1, tpidr_el0
- sub x0, x0, x1
+ sub PTR_REG (0), PTR_REG (0), PTR_REG (1)
RESTORE_Q_REGISTERS
@@ -279,13 +283,15 @@ _dl_tlsdesc_resolve_rela:
SAVE_Q_REGISTERS
- ldr x1, [x3, #8]
+ DELOUSE (3)
+ ldr PTR_REG (1), [x3, #PTR_SIZE]
bl _dl_tlsdesc_resolve_rela_fixup
RESTORE_Q_REGISTERS
ldr x0, [sp, #32+16*8]
- ldr x1, [x0]
+ DELOUSE (0)
+ ldr PTR_REG (1), [x0]
blr x1
ldp x1, x4, [sp, #32+16*0]
@@ -346,7 +352,8 @@ _dl_tlsdesc_resolve_hold:
RESTORE_Q_REGISTERS
ldr x0, [sp, #32+16*9]
- ldr x1, [x0]
+ DELOUSE (0)
+ ldr PTR_REG (1), [x0]
blr x1
ldp x1, x2, [sp, #32+16*0]
diff --git a/sysdeps/aarch64/dl-trampoline.S b/sysdeps/aarch64/dl-trampoline.S
index 947a515..63ef6f7 100644
--- a/sysdeps/aarch64/dl-trampoline.S
+++ b/sysdeps/aarch64/dl-trampoline.S
@@ -22,9 +22,13 @@
#include "dl-link.h"
#define ip0 x16
+#define ip0l PTR_REG (16)
#define ip1 x17
#define lr x30
+/* RELA relocatons are 3 pointers */
+#define RELA_SIZE (PTR_SIZE * 3)
+
.text
.globl _dl_runtime_resolve
.type _dl_runtime_resolve, #function
@@ -79,7 +83,7 @@ _dl_runtime_resolve:
cfi_rel_offset (q1, 80+7*16)
/* Get pointer to linker struct. */
- ldr x0, [ip0, #-8]
+ ldr PTR_REG (0), [ip0, #-PTR_SIZE]
/* Prepare to call _dl_fixup(). */
ldr x1, [sp, 80+8*16] /* Recover &PLTGOT[n] */
@@ -87,7 +91,7 @@ _dl_runtime_resolve:
sub x1, x1, ip0
add x1, x1, x1, lsl #1
lsl x1, x1, #3
- sub x1, x1, #192
+ sub x1, x1, #(RELA_SIZE<<3)
lsr x1, x1, #3
/* Call fixup routine. */
@@ -191,7 +195,7 @@ _dl_runtime_profile:
stp x0, x1, [x29, #OFFSET_RG + DL_OFFSET_RG_SP]
/* Get pointer to linker struct. */
- ldr x0, [ip0, #-8]
+ ldr PTR_REG (0), [ip0, #-PTR_SIZE]
/* Prepare to call _dl_profile_fixup(). */
ldr x1, [x29, OFFSET_PLTGOTN] /* Recover &PLTGOT[n] */
@@ -199,7 +203,7 @@ _dl_runtime_profile:
sub x1, x1, ip0
add x1, x1, x1, lsl #1
lsl x1, x1, #3
- sub x1, x1, #192
+ sub x1, x1, #(RELA_SIZE<<3)
lsr x1, x1, #3
stp x0, x1, [x29, #OFFSET_SAVED_CALL_X0]
@@ -210,8 +214,8 @@ _dl_runtime_profile:
add x4, x29, #OFFSET_FS /* address of framesize */
bl _dl_profile_fixup
- ldr ip0, [x29, #OFFSET_FS] /* framesize == 0 */
- cmp ip0, #0
+ ldr ip0l, [x29, #OFFSET_FS] /* framesize == 0 */
+ cmp ip0l, #0
bge 1f
cfi_remember_state
@@ -243,7 +247,7 @@ _dl_runtime_profile:
1:
/* The new frame size is in ip0. */
- sub x1, x29, ip0
+ sub PTR_REG (1), PTR_REG (29), ip0l
and sp, x1, #0xfffffffffffffff0
str x0, [x29, #OFFSET_T1]
diff --git a/sysdeps/aarch64/jmpbuf-unwind.h b/sysdeps/aarch64/jmpbuf-unwind.h
index 3e0a37d..11ace17 100644
--- a/sysdeps/aarch64/jmpbuf-unwind.h
+++ b/sysdeps/aarch64/jmpbuf-unwind.h
@@ -27,7 +27,7 @@
((void *) (address) < (void *) demangle (jmpbuf[JB_SP]))
#define _JMPBUF_CFA_UNWINDS_ADJ(jmpbuf, context, adj) \
- _JMPBUF_UNWINDS_ADJ (jmpbuf, (void *) _Unwind_GetCFA (context), adj)
+ _JMPBUF_UNWINDS_ADJ (jmpbuf, (void *) (size_t) _Unwind_GetCFA (context), adj)
#define _JMPBUF_UNWINDS_ADJ(_jmpbuf, _address, _adj) \
((uintptr_t) (_address) - (_adj) < _jmpbuf_sp (_jmpbuf) - (_adj))
diff --git a/sysdeps/aarch64/memcmp.S b/sysdeps/aarch64/memcmp.S
index ae2d997..8b87e9b 100644
--- a/sysdeps/aarch64/memcmp.S
+++ b/sysdeps/aarch64/memcmp.S
@@ -47,6 +47,9 @@
#define mask x13
ENTRY_ALIGN (memcmp, 6)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
cbz limit, L(ret0)
eor tmp1, src1, src2
tst tmp1, #7
diff --git a/sysdeps/aarch64/memcpy.S b/sysdeps/aarch64/memcpy.S
index de73f0f..b269316 100644
--- a/sysdeps/aarch64/memcpy.S
+++ b/sysdeps/aarch64/memcpy.S
@@ -61,6 +61,10 @@
ENTRY_ALIGN (memmove, 6)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
+
sub tmp1, dstin, src
cmp count, 96
ccmp tmp1, count, 2, hi
@@ -71,6 +75,10 @@ END (memmove)
libc_hidden_builtin_def (memmove)
ENTRY (memcpy)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
+
prfm PLDL1KEEP, [src]
add srcend, src, count
add dstend, dstin, count
diff --git a/sysdeps/aarch64/memset.S b/sysdeps/aarch64/memset.S
index 4d222c5..7bad29a 100644
--- a/sysdeps/aarch64/memset.S
+++ b/sysdeps/aarch64/memset.S
@@ -39,6 +39,9 @@
ENTRY_ALIGN (__memset, 6)
+ DELOUSE (0)
+ DELOUSE (2)
+
dup v0.16B, valw
add dstend, dstin, count
diff --git a/sysdeps/aarch64/nptl/bits/pthreadtypes.h b/sysdeps/aarch64/nptl/bits/pthreadtypes.h
index c376e64..9dcf8d9 100644
--- a/sysdeps/aarch64/nptl/bits/pthreadtypes.h
+++ b/sysdeps/aarch64/nptl/bits/pthreadtypes.h
@@ -32,6 +32,8 @@
#define __SIZEOF_PTHREAD_BARRIER_T 32
#define __SIZEOF_PTHREAD_BARRIERATTR_T 8
+#define __PTHREAD_RWLOCK_INT_FLAGS_SHARED 1
+
/* Thread identifiers. The structure of the attribute type is not
exposed on purpose. */
diff --git a/sysdeps/aarch64/nptl/bits/semaphore.h b/sysdeps/aarch64/nptl/bits/semaphore.h
index 3cc5b37..05333ea 100644
--- a/sysdeps/aarch64/nptl/bits/semaphore.h
+++ b/sysdeps/aarch64/nptl/bits/semaphore.h
@@ -21,7 +21,11 @@
#endif
-#define __SIZEOF_SEM_T 32
+#ifdef __ILP32__
+# define __SIZEOF_SEM_T 16
+#else
+# define __SIZEOF_SEM_T 32
+#endif
/* Value returned if `sem_open' failed. */
diff --git a/sysdeps/aarch64/setjmp.S b/sysdeps/aarch64/setjmp.S
index 22f4368..e03b3b5 100644
--- a/sysdeps/aarch64/setjmp.S
+++ b/sysdeps/aarch64/setjmp.S
@@ -33,6 +33,7 @@ END (_setjmp)
libc_hidden_def (_setjmp)
ENTRY (__sigsetjmp)
+ DELOUSE (0)
1:
stp x19, x20, [x0, #JB_X19<<3]
@@ -42,7 +43,7 @@ ENTRY (__sigsetjmp)
stp x27, x28, [x0, #JB_X27<<3]
#ifdef PTR_MANGLE
- PTR_MANGLE (x4, x30, x3, x2)
+ PTR_MANGLE (4, 30, 3, 2)
stp x29, x4, [x0, #JB_X29<<3]
#else
stp x29, x30, [x0, #JB_X29<<3]
@@ -57,7 +58,7 @@ ENTRY (__sigsetjmp)
stp d14, d15, [x0, #JB_D14<<3]
#ifdef PTR_MANGLE
mov x4, sp
- PTR_MANGLE (x5, x4, x3, x2)
+ PTR_MANGLE (5, 4, 3, 2)
str x5, [x0, #JB_SP<<3]
#else
mov x2, sp
diff --git a/sysdeps/aarch64/start.S b/sysdeps/aarch64/start.S
index efe2474..9198c57 100644
--- a/sysdeps/aarch64/start.S
+++ b/sysdeps/aarch64/start.S
@@ -16,6 +16,8 @@
License along with the GNU C Library. If not, see
<http://www.gnu.org/licenses/>. */
+#include <sysdep.h>
+
/* This is the canonical entry point, usually the first thing in the text
segment.
@@ -25,7 +27,7 @@
At this entry point, most registers' values are unspecified, except:
- x0 Contains a function pointer to be registered with `atexit'.
+ x0/w0 Contains a function pointer to be registered with `atexit'.
This is how the dynamic linker arranges to have DT_FINI
functions called for shared libraries that have been loaded
before this code runs.
@@ -52,26 +54,26 @@ _start:
mov x5, x0
/* Load argc and a pointer to argv */
- ldr x1, [sp, #0]
- add x2, sp, #8
+ ldr PTR_REG (1), [sp, #0]
+ add x2, sp, #PTR_SIZE
/* Setup stack limit in argument register */
mov x6, sp
#ifdef SHARED
adrp x0, :got:main
- ldr x0, [x0, #:got_lo12:main]
+ ldr PTR_REG (0), [x0, #:got_lo12:main]
adrp x3, :got:__libc_csu_init
- ldr x3, [x3, #:got_lo12:__libc_csu_init]
+ ldr PTR_REG (3), [x3, #:got_lo12:__libc_csu_init]
adrp x4, :got:__libc_csu_fini
- ldr x4, [x4, #:got_lo12:__libc_csu_fini]
+ ldr PTR_REG (4), [x4, #:got_lo12:__libc_csu_fini]
#else
/* Set up the other arguments in registers */
- ldr x0, =main
- ldr x3, =__libc_csu_init
- ldr x4, =__libc_csu_fini
+ ldr PTR_REG (0), =main
+ ldr PTR_REG (3), =__libc_csu_init
+ ldr PTR_REG (4), =__libc_csu_fini
#endif
/* __libc_start_main (main, argc, argv, init, fini, rtld_fini,
diff --git a/sysdeps/aarch64/strchr.S b/sysdeps/aarch64/strchr.S
index 5e3aecf..c66fea3 100644
--- a/sysdeps/aarch64/strchr.S
+++ b/sysdeps/aarch64/strchr.S
@@ -62,6 +62,7 @@
/* Locals and temporaries. */
ENTRY (strchr)
+ DELOUSE (0)
mov wtmp2, #0x0401
movk wtmp2, #0x4010, lsl #16
dup vrepchr.16b, chrin
diff --git a/sysdeps/aarch64/strchrnul.S b/sysdeps/aarch64/strchrnul.S
index a624c8d..c2cc47e 100644
--- a/sysdeps/aarch64/strchrnul.S
+++ b/sysdeps/aarch64/strchrnul.S
@@ -60,6 +60,7 @@
identify exactly which byte is causing the termination. */
ENTRY (__strchrnul)
+ DELOUSE (0)
/* Magic constant 0x40100401 to allow us to identify which lane
matches the termination condition. */
mov wtmp2, #0x0401
diff --git a/sysdeps/aarch64/strcmp.S b/sysdeps/aarch64/strcmp.S
index ba0ccb4..49e528b 100644
--- a/sysdeps/aarch64/strcmp.S
+++ b/sysdeps/aarch64/strcmp.S
@@ -49,6 +49,8 @@
/* Start of performance-critical section -- one 64B cache line. */
ENTRY_ALIGN(strcmp, 6)
+ DELOUSE (0)
+ DELOUSE (1)
eor tmp1, src1, src2
mov zeroones, #REP8_01
tst tmp1, #7
diff --git a/sysdeps/aarch64/strcpy.S b/sysdeps/aarch64/strcpy.S
index 0694199..45809e8 100644
--- a/sysdeps/aarch64/strcpy.S
+++ b/sysdeps/aarch64/strcpy.S
@@ -91,6 +91,8 @@
#define MIN_PAGE_SIZE (1 << MIN_PAGE_P2)
ENTRY_ALIGN (STRCPY, 6)
+ DELOUSE (0)
+ DELOUSE (1)
/* For moderately short strings, the fastest way to do the copy is to
calculate the length of the string in the same way as strlen, then
essentially do a memcpy of the result. This avoids the need for
diff --git a/sysdeps/aarch64/strlen.S b/sysdeps/aarch64/strlen.S
index a07834b..5fb653a 100644
--- a/sysdeps/aarch64/strlen.S
+++ b/sysdeps/aarch64/strlen.S
@@ -85,6 +85,8 @@
boundary. */
ENTRY_ALIGN (__strlen, 6)
+ DELOUSE (0)
+ DELOUSE (1)
and tmp1, srcin, MIN_PAGE_SIZE - 1
mov zeroones, REP8_01
cmp tmp1, MIN_PAGE_SIZE - 16
diff --git a/sysdeps/aarch64/strncmp.S b/sysdeps/aarch64/strncmp.S
index f6a17fd..02de93c 100644
--- a/sysdeps/aarch64/strncmp.S
+++ b/sysdeps/aarch64/strncmp.S
@@ -51,6 +51,9 @@
#define endloop x15
ENTRY_ALIGN_AND_PAD (strncmp, 6, 7)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
cbz limit, L(ret0)
eor tmp1, src1, src2
mov zeroones, #REP8_01
diff --git a/sysdeps/aarch64/strnlen.S b/sysdeps/aarch64/strnlen.S
index 4cce45f..af765f1 100644
--- a/sysdeps/aarch64/strnlen.S
+++ b/sysdeps/aarch64/strnlen.S
@@ -50,6 +50,9 @@
#define REP8_80 0x8080808080808080
ENTRY_ALIGN_AND_PAD (__strnlen, 6, 9)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
cbz limit, L(hit_limit)
mov zeroones, #REP8_01
bic src, srcin, #15
diff --git a/sysdeps/aarch64/strrchr.S b/sysdeps/aarch64/strrchr.S
index 44c1917..ea37968 100644
--- a/sysdeps/aarch64/strrchr.S
+++ b/sysdeps/aarch64/strrchr.S
@@ -68,6 +68,7 @@
identify exactly which byte is causing the termination, and why. */
ENTRY(strrchr)
+ DELOUSE (0)
cbz x1, L(null_search)
/* Magic constant 0x40100401 to allow us to identify which lane
matches the requested byte. Magic constant 0x80200802 used
diff --git a/sysdeps/aarch64/sysdep.h b/sysdeps/aarch64/sysdep.h
index e045759..0a7dccb 100644
--- a/sysdeps/aarch64/sysdep.h
+++ b/sysdeps/aarch64/sysdep.h
@@ -16,8 +16,25 @@
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
+#ifndef _AARCH64_SYSDEP_H
+#define _AARCH64_SYSDEP_H
+
#include <sysdeps/generic/sysdep.h>
+#ifdef __LP64__
+# define AARCH64_R(NAME) R_AARCH64_ ## NAME
+# define PTR_REG(n) x##n
+# define PTR_LOG_SIZE 3
+# define DELOUSE(n)
+#else
+# define AARCH64_R(NAME) R_AARCH64_P32_ ## NAME
+# define PTR_REG(n) w##n
+# define PTR_LOG_SIZE 2
+# define DELOUSE(n) mov w##n, w##n
+#endif
+
+#define PTR_SIZE (1<<PTR_LOG_SIZE)
+
#ifdef __ASSEMBLER__
/* Syntactic details of assembler. */
@@ -107,16 +124,18 @@
# define L(name) .L##name
#endif
-/* Load or store to/from a pc-relative EXPR into/from R, using T. */
-#define LDST_PCREL(OP, R, T, EXPR) \
- adrp T, EXPR; \
- OP R, [T, #:lo12:EXPR];\
+/* Load or store to/from a pc-relative EXPR into/from R, using T.
+ Note R and T are register numbers and not register names. */
+#define LDST_PCREL(OP, R, T, EXPR) \
+ adrp x##T, EXPR; \
+ OP PTR_REG (R), [x##T, #:lo12:EXPR]; \
-/* Load or store to/from a got-relative EXPR into/from R, using T. */
-#define LDST_GLOBAL(OP, R, T, EXPR) \
- adrp T, :got:EXPR; \
- ldr T, [T, #:got_lo12:EXPR];\
- OP R, [T];
+/* Load or store to/from a got-relative EXPR into/from R, using T.
+ Note R and T are register numbers and not register names. */
+#define LDST_GLOBAL(OP, R, T, EXPR) \
+ adrp x##T, :got:EXPR; \
+ ldr PTR_REG (T), [x##T, #:got_lo12:EXPR]; \
+ OP PTR_REG (R), [x##T];
/* Since C identifiers are not normally prefixed with an underscore
on this system, the asm identifier `syscall_error' intrudes on the
@@ -125,3 +144,5 @@
#define mcount _mcount
#endif /* __ASSEMBLER__ */
+
+#endif /* _AARCH64_SYSDEP_H */
diff --git a/sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h b/sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h
index 658f696..1717c35 100644
--- a/sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h
+++ b/sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h
@@ -25,11 +25,17 @@
#define __O_NOFOLLOW 0100000
#define __O_DIRECT 0200000
-#define __O_LARGEFILE 0
+#ifdef __ILP32__
+# define __O_LARGEFILE 0400000
+#else
+# define __O_LARGEFILE 0
+#endif
+#ifdef __LP64__
# define F_GETLK64 5
# define F_SETLK64 6
# define F_SETLKW64 7
+#endif
struct flock
{
diff --git a/sysdeps/unix/sysv/linux/aarch64/clone.S b/sysdeps/unix/sysv/linux/aarch64/clone.S
index 76baa7a..220ec11 100644
--- a/sysdeps/unix/sysv/linux/aarch64/clone.S
+++ b/sysdeps/unix/sysv/linux/aarch64/clone.S
@@ -39,6 +39,13 @@
*/
.text
ENTRY(__clone)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
+ DELOUSE (3)
+ DELOUSE (4)
+ DELOUSE (5)
+ DELOUSE (6)
/* Save args for the child. */
mov x10, x0
mov x11, x2
diff --git a/sysdeps/unix/sysv/linux/aarch64/dl-cache.h b/sysdeps/unix/sysv/linux/aarch64/dl-cache.h
index 9c7b271..86905ed 100644
--- a/sysdeps/unix/sysv/linux/aarch64/dl-cache.h
+++ b/sysdeps/unix/sysv/linux/aarch64/dl-cache.h
@@ -18,7 +18,11 @@
#include <ldconfig.h>
-#define _DL_CACHE_DEFAULT_ID (FLAG_AARCH64_LIB64 | FLAG_ELF_LIBC6)
+#ifdef __LP64__
+# define _DL_CACHE_DEFAULT_ID (FLAG_AARCH64_LIB64 | FLAG_ELF_LIBC6)
+#else
+# define _DL_CACHE_DEFAULT_ID (FLAG_AARCH64_LIB32 | FLAG_ELF_LIBC6)
+#endif
#define _dl_cache_check_flags(flags) \
((flags) == _DL_CACHE_DEFAULT_ID)
@@ -27,18 +31,25 @@
do \
{ \
size_t len = strlen (dir); \
- char path[len + 3]; \
+ char path[len + 6]; \
memcpy (path, dir, len + 1); \
if (len >= 6 && ! memcmp (path + len - 6, "/lib64", 6)) \
{ \
len -= 2; \
path[len] = '\0'; \
} \
+ if (len >= 9 && ! memcmp (path + len - 9, "/libilp32", 9))\
+ { \
+ len -= 5; \
+ path[len] = '\0'; \
+ } \
add_dir (path); \
if (len >= 4 && ! memcmp (path + len - 4, "/lib", 4)) \
{ \
memcpy (path + len, "64", 3); \
add_dir (path); \
+ memcpy (path + len, "ilp32", 6); \
+ add_dir (path); \
} \
} while (0)
diff --git a/sysdeps/unix/sysv/linux/aarch64/getcontext.S b/sysdeps/unix/sysv/linux/aarch64/getcontext.S
index c2dd5b8..f6bf24f 100644
--- a/sysdeps/unix/sysv/linux/aarch64/getcontext.S
+++ b/sysdeps/unix/sysv/linux/aarch64/getcontext.S
@@ -30,6 +30,7 @@
.text
ENTRY(__getcontext)
+ DELOUSE (0)
/* The saved context will return to the getcontext() call point
with a return value of 0 */
str xzr, [x0, oX0 + 0 * SZREG]
@@ -90,7 +91,7 @@ ENTRY(__getcontext)
/* Grab the signal mask */
/* rt_sigprocmask (SIG_BLOCK, NULL, &ucp->uc_sigmask, _NSIG8) */
- add x2, x0, #UCONTEXT_SIGMASK
+ add PTR_REG (2), PTR_REG (0), #UCONTEXT_SIGMASK
mov x0, SIG_BLOCK
mov x1, 0
mov x3, _NSIG8
diff --git a/sysdeps/unix/sysv/linux/aarch64/init-first.c b/sysdeps/unix/sysv/linux/aarch64/init-first.c
index f7224a2..f7bfc4d 100644
--- a/sysdeps/unix/sysv/linux/aarch64/init-first.c
+++ b/sysdeps/unix/sysv/linux/aarch64/init-first.c
@@ -27,17 +27,21 @@ int (*VDSO_SYMBOL(clock_getres)) (clockid_t, struct timespec *);
static inline void
_libc_vdso_platform_setup (void)
{
- PREPARE_VERSION (linux2639, "LINUX_2.6.39", 123718537);
+#ifdef __LP64__
+ PREPARE_VERSION (linux_version, "LINUX_2.6.39", 123718537);
+#else
+ PREPARE_VERSION (linux_version, "LINUX_4.9", 61765625);
+#endif
- void *p = _dl_vdso_vsym ("__kernel_gettimeofday", &linux2639);
+ void *p = _dl_vdso_vsym ("__kernel_gettimeofday", &linux_version);
PTR_MANGLE (p);
VDSO_SYMBOL(gettimeofday) = p;
- p = _dl_vdso_vsym ("__kernel_clock_gettime", &linux2639);
+ p = _dl_vdso_vsym ("__kernel_clock_gettime", &linux_version);
PTR_MANGLE (p);
VDSO_SYMBOL(clock_gettime) = p;
- p = _dl_vdso_vsym ("__kernel_clock_getres", &linux2639);
+ p = _dl_vdso_vsym ("__kernel_clock_getres", &linux_version);
PTR_MANGLE (p);
VDSO_SYMBOL(clock_getres) = p;
}
diff --git a/sysdeps/unix/sysv/linux/aarch64/ldconfig.h b/sysdeps/unix/sysv/linux/aarch64/ldconfig.h
index ee91ef8..ac84194 100644
--- a/sysdeps/unix/sysv/linux/aarch64/ldconfig.h
+++ b/sysdeps/unix/sysv/linux/aarch64/ldconfig.h
@@ -21,6 +21,8 @@
#define SYSDEP_KNOWN_INTERPRETER_NAMES \
{ "/lib/ld-linux-aarch64.so.1", FLAG_ELF_LIBC6 }, \
{ "/lib/ld-linux-aarch64_be.so.1", FLAG_ELF_LIBC6 }, \
+ { "/lib/ld-linux-aarch64_ilp32.so.1", FLAG_ELF_LIBC6 }, \
+ { "/lib/ld-linux-aarch64_be_ilp32.so.1", FLAG_ELF_LIBC6 }, \
{ "/lib/ld-linux.so.3", FLAG_ELF_LIBC6 }, \
{ "/lib/ld-linux-armhf.so.3", FLAG_ELF_LIBC6 },
#define SYSDEP_KNOWN_LIBRARY_NAMES \
diff --git a/sysdeps/unix/sysv/linux/aarch64/setcontext.S b/sysdeps/unix/sysv/linux/aarch64/setcontext.S
index d17f8c8..c2bca26 100644
--- a/sysdeps/unix/sysv/linux/aarch64/setcontext.S
+++ b/sysdeps/unix/sysv/linux/aarch64/setcontext.S
@@ -34,6 +34,7 @@
.text
ENTRY (__setcontext)
+ DELOUSE (0)
/* Save a copy of UCP. */
mov x9, x0
diff --git a/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h b/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h
index a579501..ee54222 100644
--- a/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h
+++ b/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h
@@ -19,7 +19,7 @@
#include <sys/ucontext.h>
#define SIGCONTEXT siginfo_t *_si, struct ucontext *
-#define GET_PC(ctx) ((void *) (ctx)->uc_mcontext.pc)
+#define GET_PC(ctx) ((void *) (size_t) (ctx)->uc_mcontext.pc)
/* There is no reliable way to get the sigcontext unless we use a
three-argument signal handler. */
diff --git a/sysdeps/unix/sysv/linux/aarch64/swapcontext.S b/sysdeps/unix/sysv/linux/aarch64/swapcontext.S
index c1a16f3..8e2cadd 100644
--- a/sysdeps/unix/sysv/linux/aarch64/swapcontext.S
+++ b/sysdeps/unix/sysv/linux/aarch64/swapcontext.S
@@ -27,6 +27,7 @@
.text
ENTRY(__swapcontext)
+ DELOUSE (0)
/* Set the value returned when swapcontext() returns in this context. */
str xzr, [x0, oX0 + 0 * SZREG]
diff --git a/sysdeps/unix/sysv/linux/aarch64/sysdep.h b/sysdeps/unix/sysv/linux/aarch64/sysdep.h
index a397e50..1ffabc2 100644
--- a/sysdeps/unix/sysv/linux/aarch64/sysdep.h
+++ b/sysdeps/unix/sysv/linux/aarch64/sysdep.h
@@ -250,12 +250,14 @@
(!defined SHARED && (IS_IN (libc) \
|| IS_IN (libpthread))))
# ifdef __ASSEMBLER__
+/* Note, dst, src, guard, and tmp are all register numbers rather than
+ register names so they will work with both ILP32 and LP64. */
# define PTR_MANGLE(dst, src, guard, tmp) \
LDST_PCREL (ldr, guard, tmp, C_SYMBOL_NAME(__pointer_chk_guard_local)); \
PTR_MANGLE2 (dst, src, guard)
/* Use PTR_MANGLE2 for efficiency if guard is already loaded. */
# define PTR_MANGLE2(dst, src, guard)\
- eor dst, src, guard
+ eor x##dst, x##src, x##guard
# define PTR_DEMANGLE(dst, src, guard, tmp)\
PTR_MANGLE (dst, src, guard, tmp)
# define PTR_DEMANGLE2(dst, src, guard)\
@@ -268,12 +270,14 @@ extern uintptr_t __pointer_chk_guard_local attribute_relro attribute_hidden;
# endif
#else
# ifdef __ASSEMBLER__
+/* Note, dst, src, guard, and tmp are all register numbers rather than
+ register names so they will work with both ILP32 and LP64. */
# define PTR_MANGLE(dst, src, guard, tmp) \
LDST_GLOBAL (ldr, guard, tmp, C_SYMBOL_NAME(__pointer_chk_guard)); \
PTR_MANGLE2 (dst, src, guard)
/* Use PTR_MANGLE2 for efficiency if guard is already loaded. */
# define PTR_MANGLE2(dst, src, guard)\
- eor dst, src, guard
+ eor x##dst, x##src, x##guard
# define PTR_DEMANGLE(dst, src, guard, tmp)\
PTR_MANGLE (dst, src, guard, tmp)
# define PTR_DEMANGLE2(dst, src, guard)\
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] Partial ILP32 support for aarch64
2016-11-21 20:50 ` Steve Ellcey
@ 2016-11-23 9:44 ` Andreas Schwab
2016-11-23 23:01 ` Steve Ellcey
0 siblings, 1 reply; 12+ messages in thread
From: Andreas Schwab @ 2016-11-23 9:44 UTC (permalink / raw)
To: Steve Ellcey; +Cc: Joseph Myers, libc-alpha
On Nov 21 2016, Steve Ellcey <sellcey@caviumnetworks.com> wrote:
> diff --git a/sysdeps/aarch64/dl-irel.h b/sysdeps/aarch64/dl-irel.h
> index 63a8e50..2effca4 100644
> --- a/sysdeps/aarch64/dl-irel.h
> +++ b/sysdeps/aarch64/dl-irel.h
> @@ -23,6 +23,7 @@
> #include <stdio.h>
> #include <unistd.h>
> #include <ldsodefs.h>
> +#include <sysdep.h>
>
> #define ELF_MACHINE_IRELA 1
>
> @@ -40,7 +41,7 @@ elf_irela (const ElfW(Rela) *reloc)
> ElfW(Addr) *const reloc_addr = (void *) reloc->r_offset;
> const unsigned long int r_type = ELFW(R_TYPE) (reloc->r_info);
>
> - if (__glibc_likely (r_type == R_AARCH64_IRELATIVE))
> + if (__glibc_likely (r_type == AARCH64_R (IRELATIVE)))
While we generally favor a space before parens, for macros that produce
an identifier we don't write one, similar to ElfW.
> #define elf_machine_type_class(type) \
> - ((((type) == R_AARCH64_JUMP_SLOT || \
> - (type) == R_AARCH64_TLS_DTPMOD || \
> - (type) == R_AARCH64_TLS_DTPREL || \
> - (type) == R_AARCH64_TLS_TPREL || \
> - (type) == R_AARCH64_TLSDESC) * ELF_RTYPE_CLASS_PLT) \
> - | (((type) == R_AARCH64_COPY) * ELF_RTYPE_CLASS_COPY) \
> - | (((type) == R_AARCH64_GLOB_DAT) * ELF_RTYPE_CLASS_EXTERN_PROTECTED_DATA))
> + ((((type) == AARCH64_R (JUMP_SLOT) \
> + || (type) == AARCH64_R (TLS_DTPMOD) \
> + || (type) == AARCH64_R (TLS_DTPREL) \
> + || (type) == AARCH64_R (TLS_TPREL) \
> + || (type) == AARCH64_R (TLSDESC)) * ELF_RTYPE_CLASS_PLT) \
Indentation is off.
> diff --git a/sysdeps/aarch64/jmpbuf-unwind.h b/sysdeps/aarch64/jmpbuf-unwind.h
> index 3e0a37d..11ace17 100644
> --- a/sysdeps/aarch64/jmpbuf-unwind.h
> +++ b/sysdeps/aarch64/jmpbuf-unwind.h
> @@ -27,7 +27,7 @@
> ((void *) (address) < (void *) demangle (jmpbuf[JB_SP]))
>
> #define _JMPBUF_CFA_UNWINDS_ADJ(jmpbuf, context, adj) \
> - _JMPBUF_UNWINDS_ADJ (jmpbuf, (void *) _Unwind_GetCFA (context), adj)
> + _JMPBUF_UNWINDS_ADJ (jmpbuf, (void *) (size_t) _Unwind_GetCFA (context), adj)
That should be uintptr_t.
> diff --git a/sysdeps/aarch64/nptl/bits/pthreadtypes.h b/sysdeps/aarch64/nptl/bits/pthreadtypes.h
> index c376e64..9dcf8d9 100644
> --- a/sysdeps/aarch64/nptl/bits/pthreadtypes.h
> +++ b/sysdeps/aarch64/nptl/bits/pthreadtypes.h
> @@ -32,6 +32,8 @@
> #define __SIZEOF_PTHREAD_BARRIER_T 32
> #define __SIZEOF_PTHREAD_BARRIERATTR_T 8
>
> +#define __PTHREAD_RWLOCK_INT_FLAGS_SHARED 1
> +
>
> /* Thread identifiers. The structure of the attribute type is not
> exposed on purpose. */
The pthread type sizes should be adjusted, similar to x32, to avoid
excessive padding:
#ifdef __ILP32__
# define __SIZEOF_PTHREAD_ATTR_T 32
# define __SIZEOF_PTHREAD_MUTEX_T 32
# define __SIZEOF_PTHREAD_MUTEXATTR_T 4
# define __SIZEOF_PTHREAD_COND_T 48
# define __SIZEOF_PTHREAD_CONDATTR_T 4
# define __SIZEOF_PTHREAD_RWLOCK_T 44
# define __SIZEOF_PTHREAD_RWLOCKATTR_T 8
# define __SIZEOF_PTHREAD_BARRIER_T 20
# define __SIZEOF_PTHREAD_BARRIERATTR_T 4
#else
# define __SIZEOF_PTHREAD_ATTR_T 64
# define __SIZEOF_PTHREAD_MUTEX_T 48
# define __SIZEOF_PTHREAD_MUTEXATTR_T 8
# define __SIZEOF_PTHREAD_COND_T 48
# define __SIZEOF_PTHREAD_CONDATTR_T 8
# define __SIZEOF_PTHREAD_RWLOCK_T 56
# define __SIZEOF_PTHREAD_RWLOCKATTR_T 8
# define __SIZEOF_PTHREAD_BARRIER_T 32
# define __SIZEOF_PTHREAD_BARRIERATTR_T 8
#endif
> diff --git a/sysdeps/aarch64/nptl/bits/semaphore.h b/sysdeps/aarch64/nptl/bits/semaphore.h
> index 3cc5b37..05333ea 100644
> --- a/sysdeps/aarch64/nptl/bits/semaphore.h
> +++ b/sysdeps/aarch64/nptl/bits/semaphore.h
> @@ -21,7 +21,11 @@
> #endif
>
>
> -#define __SIZEOF_SEM_T 32
> +#ifdef __ILP32__
> +# define __SIZEOF_SEM_T 16
> +#else
> +# define __SIZEOF_SEM_T 32
> +#endif
sem_t needs to be 64-bit aligned, to match struct new_sem.
typedef union
{
char __size[__SIZEOF_SEM_T];
- long int __align;
+ long long int __align;
} sem_t;
> diff --git a/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h b/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h
> index a579501..ee54222 100644
> --- a/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h
> +++ b/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h
> @@ -19,7 +19,7 @@
> #include <sys/ucontext.h>
>
> #define SIGCONTEXT siginfo_t *_si, struct ucontext *
> -#define GET_PC(ctx) ((void *) (ctx)->uc_mcontext.pc)
> +#define GET_PC(ctx) ((void *) (size_t) (ctx)->uc_mcontext.pc)
uintptr_t
Andreas.
--
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE 1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] Partial ILP32 support for aarch64
2016-11-23 9:44 ` Andreas Schwab
@ 2016-11-23 23:01 ` Steve Ellcey
2016-11-28 9:05 ` Andreas Schwab
0 siblings, 1 reply; 12+ messages in thread
From: Steve Ellcey @ 2016-11-23 23:01 UTC (permalink / raw)
To: Andreas Schwab; +Cc: Joseph Myers, libc-alpha
[-- Attachment #1: Type: text/plain, Size: 3826 bytes --]
Here is a new version of this patch with Andreas's changes incorporated
(as well as Joseph's earlier changes). Â I retested on aarch64 and
everything still looks good.
Steve Ellcey
sellcey@caviumnetworks.com
2016-11-23  Andrew Pinski  <andrew.pinski@caviumnetworks.com>
    Yury Norov  <ynorov@caviumnetworks.com>
    Steve Ellcey  <sellcey@caviumnetworks.com>
* sysdeps/aarch64/crti.S: Add include of sysdep.h.
(call_weak_fn): Use PTR_REG to get correct reg name in ILP32.
* sysdeps/aarch64/dl-irel.h: Add include of sysdep.h.
(elf_irela): Use AARCH64_R macro to get correct relocation in ILP32.
* sysdeps/aarch64/dl-machine.h: Add include of sysdep.h.
(elf_machine_load_address, RTLD_START, RTLD_START_1, RTLD_START,
elf_machine_type_class, ELF_MACHINE_JMP_SLOT, elf_machine_rela,
elf_machine_lazy_rel): Add ifdef's for ILP32 support.
* sysdeps/aarch64/dl-tlsdesc.S (_dl_tlsdesc_return,
_dl_tlsdesc_return_lazy, _dl_tlsdesc_dynamic,
_dl_tlsdesc_resolve_hold): Extend pointers in ILP32, use PTR_REG
to get correct reg name for ILP32.
* sysdeps/aarch64/dl-trampoline.S (ip01): New Macro.
(RELA_SIZE): New Macro.
(_dl_runtime_resolve, _dl_runtime_profile): Use new macros and PTR_REG
to support ILP32.
* sysdeps/aarch64/jmpbuf-unwind.h (_JMPBUF_CFA_UNWINDS_ADJ): Add
cast for ILP32 mode.
* sysdeps/aarch64/memcmp.S (memcmp): Extend arg pointers for ILP32 mode.
* sysdeps/aarch64/memcpy.S (memmove, memcpy): Ditto.
* sysdeps/aarch64/memset.S (__memset): Ditto.
* sysdeps/aarch64/strchr.S (strchr): Ditto.
* sysdeps/aarch64/strchrnul.S (__strchrnul): Ditto.
* sysdeps/aarch64/strcmp.S (strcmp): Ditto.
* sysdeps/aarch64/strcpy.S (strcpy): Ditto.
* sysdeps/aarch64/strlen.S (__strlen): Ditto.
* sysdeps/aarch64/strncmp.S (strncmp): Ditto.
* sysdeps/aarch64/strnlen.S (strnlen): Ditto.
* sysdeps/aarch64/strrchr.S (strrchr): Ditto.
* sysdeps/unix/sysv/linux/aarch64/clone.S: Ditto.
* sysdeps/unix/sysv/linux/aarch64/setcontext.S (__setcontext): Ditto.
* sysdeps/unix/sysv/linux/aarch64/swapcontext.S (__swapcontext): Ditto.
* sysdeps/aarch64/__longjmp.S (__longjmp): Extend pointers in ILP32,
change PTR_MANGLE call to use register numbers instead of names.
* sysdeps/unix/sysv/linux/aarch64/getcontext.S (__getcontext): Ditto.
* sysdeps/aarch64/setjmp.S (__sigsetjmp): Extend arg pointers for
ILP32 mode, change PTR_MANGLE calls to use register numbers.
* sysdeps/aarch64/start.S (_start): Ditto.
* sysdeps/aarch64/nptl/bits/pthreadtypes.h
(__PTHREAD_RWLOCK_INT_FLAGS_SHARED): New define.
(__SIZEOF_PTHREAD_ATTR_T, __SIZEOF_PTHREAD_MUTEX_T,
__SIZEOF_PTHREAD_MUTEXATTR_T, __SIZEOF_PTHREAD_COND_T,
__SIZEOF_PTHREAD_COND_COMPAT_T, __SIZEOF_PTHREAD_CONDATTR_T,
__SIZEOF_PTHREAD_RWLOCK_T, __SIZEOF_PTHREAD_RWLOCKATTR_T,
__SIZEOF_PTHREAD_BARRIER_T, __SIZEOF_PTHREAD_BARRIERATTR_T):
Make defined values dependent on __ILP32__.
* sysdeps/aarch64/nptl/bits/semaphore.h (__SIZEOF_SEM_T): Change define.
(sem_t): Change __align type.
* sysdeps/aarch64/sysdep.h (AARCH64_R, PTR_REG, PTR_LOG_SIZE, DELOUSE,
PTR_SIZE): New Macros.
(LDST_PCREL, LDST_GLOBAL) Update to use PTR_REG.
* sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h (O_LARGEFILE):
Set when in ILP32 mode.
(F_GETLK64, F_SETLK64, F_SETLKW64): Only set in LP64 mode.
* sysdeps/unix/sysv/linux/aarch64/dl-cache.h (DL_CACHE_DEFAULT_ID):
Set elf flags for ILP32.
(add_system_dir): Set ILP32 library directories.
* sysdeps/unix/sysv/linux/aarch64/init-first.c
(_libc_vdso_platform_setup): Set minimum kernel version for ILP32.
* sysdeps/unix/sysv/linux/aarch64/ldconfig.h
(SYSDEP_KNOWN_INTERPRETER_NAMES): Add ILP32 names.
* sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h (GET_PC, SET_PC):
New Macros.
* sysdeps/unix/sysv/linux/aarch64/sysdep.h: Handle ILP32 pointers.
[-- Attachment #2: aarch64_ilp32.patch --]
[-- Type: text/x-patch, Size: 37641 bytes --]
diff --git a/sysdeps/aarch64/__longjmp.S b/sysdeps/aarch64/__longjmp.S
index 65116be..4d411fe 100644
--- a/sysdeps/aarch64/__longjmp.S
+++ b/sysdeps/aarch64/__longjmp.S
@@ -46,6 +46,8 @@ ENTRY (__longjmp)
cfi_offset(d14, JB_D14<<3)
cfi_offset(d15, JB_D15<<3)
+ DELOUSE (0)
+
ldp x19, x20, [x0, #JB_X19<<3]
ldp x21, x22, [x0, #JB_X21<<3]
ldp x23, x24, [x0, #JB_X23<<3]
@@ -53,7 +55,7 @@ ENTRY (__longjmp)
ldp x27, x28, [x0, #JB_X27<<3]
#ifdef PTR_DEMANGLE
ldp x29, x4, [x0, #JB_X29<<3]
- PTR_DEMANGLE (x30, x4, x3, x2)
+ PTR_DEMANGLE (30, 4, 3, 2)
#else
ldp x29, x30, [x0, #JB_X29<<3]
#endif
@@ -98,7 +100,7 @@ ENTRY (__longjmp)
cfi_same_value(d15)
#ifdef PTR_DEMANGLE
ldr x4, [x0, #JB_SP<<3]
- PTR_DEMANGLE (x5, x4, x3, x2)
+ PTR_DEMANGLE (5, 4, 3, 2)
#else
ldr x5, [x0, #JB_SP<<3]
#endif
diff --git a/sysdeps/aarch64/crti.S b/sysdeps/aarch64/crti.S
index 53ccb42..5c42fd5 100644
--- a/sysdeps/aarch64/crti.S
+++ b/sysdeps/aarch64/crti.S
@@ -39,6 +39,7 @@
they can be called as functions. The symbols _init and _fini are
magic and cause the linker to emit DT_INIT and DT_FINI. */
+#include <sysdep.h>
#include <libc-symbols.h>
#ifndef PREINIT_FUNCTION
@@ -60,7 +61,7 @@
.type call_weak_fn, %function
call_weak_fn:
adrp x0, :got:PREINIT_FUNCTION
- ldr x0, [x0, #:got_lo12:PREINIT_FUNCTION]
+ ldr PTR_REG (0), [x0, #:got_lo12:PREINIT_FUNCTION]
cbz x0, 1f
b PREINIT_FUNCTION
1:
diff --git a/sysdeps/aarch64/dl-irel.h b/sysdeps/aarch64/dl-irel.h
index 63a8e50..460454f 100644
--- a/sysdeps/aarch64/dl-irel.h
+++ b/sysdeps/aarch64/dl-irel.h
@@ -23,6 +23,7 @@
#include <stdio.h>
#include <unistd.h>
#include <ldsodefs.h>
+#include <sysdep.h>
#define ELF_MACHINE_IRELA 1
@@ -40,7 +41,7 @@ elf_irela (const ElfW(Rela) *reloc)
ElfW(Addr) *const reloc_addr = (void *) reloc->r_offset;
const unsigned long int r_type = ELFW(R_TYPE) (reloc->r_info);
- if (__glibc_likely (r_type == R_AARCH64_IRELATIVE))
+ if (__glibc_likely (r_type == AARCH64_R(IRELATIVE)))
{
ElfW(Addr) value = elf_ifunc_invoke (reloc->r_addend);
*reloc_addr = value;
diff --git a/sysdeps/aarch64/dl-machine.h b/sysdeps/aarch64/dl-machine.h
index 282805e..24b51a9 100644
--- a/sysdeps/aarch64/dl-machine.h
+++ b/sysdeps/aarch64/dl-machine.h
@@ -21,6 +21,7 @@
#define ELF_MACHINE_NAME "aarch64"
+#include <sysdep.h>
#include <tls.h>
#include <dl-tlsdesc.h>
#include <dl-irel.h>
@@ -53,19 +54,33 @@ elf_machine_load_address (void)
by constructing a non GOT reference to the symbol, the dynamic
address of the symbol we compute using adrp/add to compute the
symbol's address relative to the PC.
- This depends on 32bit relocations being resolved at link time
- and that the static address fits in the 32bits. */
+ This depends on 32/16bit relocations being resolved at link time
+ and that the static address fits in the 32/16 bits. */
ElfW(Addr) static_addr;
ElfW(Addr) dynamic_addr;
asm (" \n"
" adrp %1, _dl_start; \n"
+#ifdef __LP64__
" add %1, %1, #:lo12:_dl_start \n"
+#else
+" add %w1, %w1, #:lo12:_dl_start \n"
+#endif
" ldr %w0, 1f \n"
" b 2f \n"
"1: \n"
+#ifdef __LP64__
" .word _dl_start \n"
+#else
+# ifdef __AARCH64EB__
+" .short 0 \n"
+# endif
+" .short _dl_start \n"
+# ifndef __AARCH64EB__
+" .short 0 \n"
+# endif
+#endif
"2: \n"
: "=r" (static_addr), "=r" (dynamic_addr));
return dynamic_addr - static_addr;
@@ -125,80 +140,86 @@ elf_machine_runtime_setup (struct link_map *l, int lazy, int profile)
/* Initial entry point for the dynamic linker. The C function
_dl_start is the real entry point, its return value is the user
program's entry point */
+#ifdef __LP64__
+# define RTLD_START RTLD_START_1 ("x", "3", "sp")
+#else
+# define RTLD_START RTLD_START_1 ("w", "2", "wsp")
+#endif
-#define RTLD_START asm ("\
-.text \n\
-.globl _start \n\
-.type _start, %function \n\
-.globl _dl_start_user \n\
-.type _dl_start_user, %function \n\
-_start: \n\
- mov x0, sp \n\
- bl _dl_start \n\
- // returns user entry point in x0 \n\
- mov x21, x0 \n\
-_dl_start_user: \n\
- // get the original arg count \n\
- ldr x1, [sp] \n\
- // get the argv address \n\
- add x2, sp, #8 \n\
- // get _dl_skip_args to see if we were \n\
- // invoked as an executable \n\
- adrp x4, _dl_skip_args \n\
- ldr w4, [x4, #:lo12:_dl_skip_args] \n\
- // do we need to adjust argc/argv \n\
- cmp w4, 0 \n\
- beq .L_done_stack_adjust \n\
- // subtract _dl_skip_args from original arg count \n\
- sub x1, x1, x4 \n\
- // store adjusted argc back to stack \n\
- str x1, [sp] \n\
- // find the first unskipped argument \n\
- mov x3, x2 \n\
- add x4, x2, x4, lsl #3 \n\
- // shuffle argv down \n\
-1: ldr x5, [x4], #8 \n\
- str x5, [x3], #8 \n\
- cmp x5, #0 \n\
- bne 1b \n\
- // shuffle envp down \n\
-1: ldr x5, [x4], #8 \n\
- str x5, [x3], #8 \n\
- cmp x5, #0 \n\
- bne 1b \n\
- // shuffle auxv down \n\
-1: ldp x0, x5, [x4, #16]! \n\
- stp x0, x5, [x3], #16 \n\
- cmp x0, #0 \n\
- bne 1b \n\
- // Update _dl_argv \n\
- adrp x3, _dl_argv \n\
- str x2, [x3, #:lo12:_dl_argv] \n\
-.L_done_stack_adjust: \n\
- // compute envp \n\
- add x3, x2, x1, lsl #3 \n\
- add x3, x3, #8 \n\
- adrp x16, _rtld_local \n\
- add x16, x16, #:lo12:_rtld_local \n\
- ldr x0, [x16] \n\
- bl _dl_init \n\
- // load the finalizer function \n\
- adrp x0, _dl_fini \n\
- add x0, x0, #:lo12:_dl_fini \n\
- // jump to the user_s entry point \n\
- br x21 \n\
+
+#define RTLD_START_1(PTR, PTR_SIZE_LOG, PTR_SP) asm ("\
+.text \n\
+.globl _start \n\
+.type _start, %function \n\
+.globl _dl_start_user \n\
+.type _dl_start_user, %function \n\
+_start: \n\
+ mov " PTR "0, " PTR_SP " \n\
+ bl _dl_start \n\
+ // returns user entry point in x0 \n\
+ mov x21, x0 \n\
+_dl_start_user: \n\
+ // get the original arg count \n\
+ ldr " PTR "1, [sp] \n\
+ // get the argv address \n\
+ add " PTR "2, " PTR_SP ", #(1<<" PTR_SIZE_LOG ") \n\
+ // get _dl_skip_args to see if we were \n\
+ // invoked as an executable \n\
+ adrp x4, _dl_skip_args \n\
+ ldr w4, [x4, #:lo12:_dl_skip_args] \n\
+ // do we need to adjust argc/argv \n\
+ cmp w4, 0 \n\
+ beq .L_done_stack_adjust \n\
+ // subtract _dl_skip_args from original arg count \n\
+ sub " PTR "1, " PTR "1, " PTR "4 \n\
+ // store adjusted argc back to stack \n\
+ str " PTR "1, [sp] \n\
+ // find the first unskipped argument \n\
+ mov " PTR "3, " PTR "2 \n\
+ add " PTR "4, " PTR "2, " PTR "4, lsl #" PTR_SIZE_LOG " \n\
+ // shuffle argv down \n\
+1: ldr " PTR "5, [x4], #(1<<" PTR_SIZE_LOG ") \n\
+ str " PTR "5, [x3], #(1<<" PTR_SIZE_LOG ") \n\
+ cmp " PTR "5, #0 \n\
+ bne 1b \n\
+ // shuffle envp down \n\
+1: ldr " PTR "5, [x4], #(1<<" PTR_SIZE_LOG ") \n\
+ str " PTR "5, [x3], #(1<<" PTR_SIZE_LOG ") \n\
+ cmp " PTR "5, #0 \n\
+ bne 1b \n\
+ // shuffle auxv down \n\
+1: ldp " PTR "0, " PTR "5, [x4, #(2<<" PTR_SIZE_LOG ")]! \n\
+ stp " PTR "0, " PTR "5, [x3], #(2<<" PTR_SIZE_LOG ") \n\
+ cmp " PTR "0, #0 \n\
+ bne 1b \n\
+ // Update _dl_argv \n\
+ adrp x3, _dl_argv \n\
+ str " PTR "2, [x3, #:lo12:_dl_argv] \n\
+.L_done_stack_adjust: \n\
+ // compute envp \n\
+ add " PTR "3, " PTR "2, " PTR "1, lsl #" PTR_SIZE_LOG " \n\
+ add " PTR "3, " PTR "3, #(1<<" PTR_SIZE_LOG ") \n\
+ adrp x16, _rtld_local \n\
+ add " PTR "16, " PTR "16, #:lo12:_rtld_local \n\
+ ldr " PTR "0, [x16] \n\
+ bl _dl_init \n\
+ // load the finalizer function \n\
+ adrp x0, _dl_fini \n\
+ add " PTR "0, " PTR "0, #:lo12:_dl_fini \n\
+ // jump to the user_s entry point \n\
+ br x21 \n\
");
#define elf_machine_type_class(type) \
- ((((type) == R_AARCH64_JUMP_SLOT || \
- (type) == R_AARCH64_TLS_DTPMOD || \
- (type) == R_AARCH64_TLS_DTPREL || \
- (type) == R_AARCH64_TLS_TPREL || \
- (type) == R_AARCH64_TLSDESC) * ELF_RTYPE_CLASS_PLT) \
- | (((type) == R_AARCH64_COPY) * ELF_RTYPE_CLASS_COPY) \
- | (((type) == R_AARCH64_GLOB_DAT) * ELF_RTYPE_CLASS_EXTERN_PROTECTED_DATA))
+ ((((type) == AARCH64_R(JUMP_SLOT) \
+ || (type) == AARCH64_R(TLS_DTPMOD) \
+ || (type) == AARCH64_R(TLS_DTPREL) \
+ || (type) == AARCH64_R(TLS_TPREL) \
+ || (type) == AARCH64_R(TLSDESC)) * ELF_RTYPE_CLASS_PLT) \
+ | (((type) == AARCH64_R(COPY)) * ELF_RTYPE_CLASS_COPY) \
+ | (((type) == AARCH64_R(GLOB_DAT)) * ELF_RTYPE_CLASS_EXTERN_PROTECTED_DATA))
-#define ELF_MACHINE_JMP_SLOT R_AARCH64_JUMP_SLOT
+#define ELF_MACHINE_JMP_SLOT AARCH64_R(JUMP_SLOT)
/* AArch64 uses RELA not REL */
#define ELF_MACHINE_NO_REL 1
@@ -237,9 +258,9 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
void *const reloc_addr_arg, int skip_ifunc)
{
ElfW(Addr) *const reloc_addr = reloc_addr_arg;
- const unsigned int r_type = ELF64_R_TYPE (reloc->r_info);
+ const unsigned int r_type = ELFW (R_TYPE) (reloc->r_info);
- if (__builtin_expect (r_type == R_AARCH64_RELATIVE, 0))
+ if (__builtin_expect (r_type == AARCH64_R(RELATIVE), 0))
*reloc_addr = map->l_addr + reloc->r_addend;
else if (__builtin_expect (r_type == R_AARCH64_NONE, 0))
return;
@@ -257,7 +278,7 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
switch (r_type)
{
- case R_AARCH64_COPY:
+ case AARCH64_R(COPY):
if (sym == NULL)
break;
@@ -275,15 +296,17 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
MIN (sym->st_size, refsym->st_size));
break;
- case R_AARCH64_RELATIVE:
- case R_AARCH64_GLOB_DAT:
- case R_AARCH64_JUMP_SLOT:
- case R_AARCH64_ABS32:
- case R_AARCH64_ABS64:
+ case AARCH64_R(RELATIVE):
+ case AARCH64_R(GLOB_DAT):
+ case AARCH64_R(JUMP_SLOT):
+ case AARCH64_R(ABS32):
+#ifdef __LP64__
+ case AARCH64_R(ABS64):
+#endif
*reloc_addr = value + reloc->r_addend;
break;
- case R_AARCH64_TLSDESC:
+ case AARCH64_R(TLSDESC):
{
struct tlsdesc volatile *td =
(struct tlsdesc volatile *)reloc_addr;
@@ -318,7 +341,7 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
break;
}
- case R_AARCH64_TLS_DTPMOD:
+ case AARCH64_R(TLS_DTPMOD):
#ifdef RTLD_BOOTSTRAP
*reloc_addr = 1;
#else
@@ -329,12 +352,12 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
#endif
break;
- case R_AARCH64_TLS_DTPREL:
+ case AARCH64_R(TLS_DTPREL):
if (sym)
*reloc_addr = sym->st_value + reloc->r_addend;
break;
- case R_AARCH64_TLS_TPREL:
+ case AARCH64_R(TLS_TPREL):
if (sym)
{
CHECK_STATIC_TLS (map, sym_map);
@@ -343,7 +366,7 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
}
break;
- case R_AARCH64_IRELATIVE:
+ case AARCH64_R(IRELATIVE):
value = map->l_addr + reloc->r_addend;
value = elf_ifunc_invoke (value);
*reloc_addr = value;
@@ -374,16 +397,16 @@ elf_machine_lazy_rel (struct link_map *map,
int skip_ifunc)
{
ElfW(Addr) *const reloc_addr = (void *) (l_addr + reloc->r_offset);
- const unsigned int r_type = ELF64_R_TYPE (reloc->r_info);
+ const unsigned int r_type = ELFW (R_TYPE) (reloc->r_info);
/* Check for unexpected PLT reloc type. */
- if (__builtin_expect (r_type == R_AARCH64_JUMP_SLOT, 1))
+ if (__builtin_expect (r_type == AARCH64_R(JUMP_SLOT), 1))
{
if (__builtin_expect (map->l_mach.plt, 0) == 0)
*reloc_addr += l_addr;
else
*reloc_addr = map->l_mach.plt;
}
- else if (__builtin_expect (r_type == R_AARCH64_TLSDESC, 1))
+ else if (__builtin_expect (r_type == AARCH64_R(TLSDESC), 1))
{
struct tlsdesc volatile *td =
(struct tlsdesc volatile *)reloc_addr;
@@ -392,7 +415,7 @@ elf_machine_lazy_rel (struct link_map *map,
td->entry = (void*)(D_PTR (map, l_info[ADDRIDX (DT_TLSDESC_PLT)])
+ map->l_addr);
}
- else if (__glibc_unlikely (r_type == R_AARCH64_IRELATIVE))
+ else if (__glibc_unlikely (r_type == AARCH64_R(IRELATIVE)))
{
ElfW(Addr) value = map->l_addr + reloc->r_addend;
if (__glibc_likely (!skip_ifunc))
diff --git a/sysdeps/aarch64/dl-tlsdesc.S b/sysdeps/aarch64/dl-tlsdesc.S
index 05be370..42fa943 100644
--- a/sysdeps/aarch64/dl-tlsdesc.S
+++ b/sysdeps/aarch64/dl-tlsdesc.S
@@ -74,7 +74,8 @@
cfi_startproc
.align 2
_dl_tlsdesc_return:
- ldr x0, [x0, #8]
+ DELOUSE (0)
+ ldr PTR_REG (0), [x0, #PTR_SIZE]
RET
cfi_endproc
.size _dl_tlsdesc_return, .-_dl_tlsdesc_return
@@ -95,9 +96,10 @@ _dl_tlsdesc_return_lazy:
so it reads the same value (this function is the final value of
td->entry) and thus it synchronizes with the release store to
td->entry in _dl_tlsdesc_resolve_rela_fixup ensuring that the load
- from [x0,#8] here happens after the initialization of td->arg. */
- ldar xzr, [x0]
- ldr x0, [x0, #8]
+ from [x0,#PTR_SIZE] here happens after the initialization of td->arg. */
+ DELOUSE (0)
+ ldar PTR_REG (zr), [x0]
+ ldr PTR_REG (0), [x0, #PTR_SIZE]
RET
cfi_endproc
.size _dl_tlsdesc_return_lazy, .-_dl_tlsdesc_return_lazy
@@ -125,10 +127,11 @@ _dl_tlsdesc_undefweak:
td->entry) and thus it synchronizes with the release store to
td->entry in _dl_tlsdesc_resolve_rela_fixup ensuring that the load
from [x0,#8] here happens after the initialization of td->arg. */
- ldar xzr, [x0]
- ldr x0, [x0, #8]
+ DELOUSE (0)
+ ldar PTR_REG (zr), [x0]
+ ldr PTR_REG (0), [x0, #PTR_SIZE]
mrs x1, tpidr_el0
- sub x0, x0, x1
+ sub PTR_REG (0), PTR_REG (0), PTR_REG (1)
ldr x1, [sp], #16
cfi_adjust_cfa_offset (-16)
RET
@@ -174,6 +177,7 @@ _dl_tlsdesc_dynamic:
stp x29, x30, [sp,#-(32+16*NSAVEXREGPAIRS)]!
cfi_adjust_cfa_offset (32+16*NSAVEXREGPAIRS)
mov x29, sp
+ DELOUSE (0)
/* Save just enough registers to support fast path, if we fall
into slow path we will save additional registers. */
@@ -187,22 +191,22 @@ _dl_tlsdesc_dynamic:
so it reads the same value (this function is the final value of
td->entry) and thus it synchronizes with the release store to
td->entry in _dl_tlsdesc_resolve_rela_fixup ensuring that the load
- from [x0,#8] here happens after the initialization of td->arg. */
- ldar xzr, [x0]
- ldr x1, [x0,#8]
- ldr x0, [x4]
- ldr x3, [x1,#16]
- ldr x2, [x0]
- cmp x3, x2
+ from [x0,#PTR_SIZE] here happens after the initialization of td->arg. */
+ ldar PTR_REG (zr), [x0]
+ ldr PTR_REG (1), [x0,#PTR_SIZE]
+ ldr PTR_REG (0), [x4]
+ ldr PTR_REG (3), [x1,#(PTR_SIZE * 2)]
+ ldr PTR_REG (2), [x0]
+ cmp PTR_REG (3), PTR_REG (2)
b.hi 2f
- ldr x2, [x1]
- add x0, x0, x2, lsl #4
- ldr x0, [x0]
+ ldr PTR_REG (2), [x1]
+ add PTR_REG (0), PTR_REG (0), PTR_REG (2), lsl #(PTR_LOG_SIZE + 1)
+ ldr PTR_REG (0), [x0]
cmn x0, #0x1
b.eq 2f
- ldr x1, [x1,#8]
- add x0, x0, x1
- sub x0, x0, x4
+ ldr PTR_REG (1), [x1,#(PTR_SIZE * 2)]
+ add PTR_REG (0), PTR_REG (0), PTR_REG (1)
+ sub PTR_REG (0), PTR_REG (0), PTR_REG (4)
1:
ldp x1, x2, [sp, #32+16*0]
ldp x3, x4, [sp, #32+16*1]
@@ -233,7 +237,7 @@ _dl_tlsdesc_dynamic:
bl __tls_get_addr
mrs x1, tpidr_el0
- sub x0, x0, x1
+ sub PTR_REG (0), PTR_REG (0), PTR_REG (1)
RESTORE_Q_REGISTERS
@@ -279,13 +283,15 @@ _dl_tlsdesc_resolve_rela:
SAVE_Q_REGISTERS
- ldr x1, [x3, #8]
+ DELOUSE (3)
+ ldr PTR_REG (1), [x3, #PTR_SIZE]
bl _dl_tlsdesc_resolve_rela_fixup
RESTORE_Q_REGISTERS
ldr x0, [sp, #32+16*8]
- ldr x1, [x0]
+ DELOUSE (0)
+ ldr PTR_REG (1), [x0]
blr x1
ldp x1, x4, [sp, #32+16*0]
@@ -346,7 +352,8 @@ _dl_tlsdesc_resolve_hold:
RESTORE_Q_REGISTERS
ldr x0, [sp, #32+16*9]
- ldr x1, [x0]
+ DELOUSE (0)
+ ldr PTR_REG (1), [x0]
blr x1
ldp x1, x2, [sp, #32+16*0]
diff --git a/sysdeps/aarch64/dl-trampoline.S b/sysdeps/aarch64/dl-trampoline.S
index 947a515..63ef6f7 100644
--- a/sysdeps/aarch64/dl-trampoline.S
+++ b/sysdeps/aarch64/dl-trampoline.S
@@ -22,9 +22,13 @@
#include "dl-link.h"
#define ip0 x16
+#define ip0l PTR_REG (16)
#define ip1 x17
#define lr x30
+/* RELA relocatons are 3 pointers */
+#define RELA_SIZE (PTR_SIZE * 3)
+
.text
.globl _dl_runtime_resolve
.type _dl_runtime_resolve, #function
@@ -79,7 +83,7 @@ _dl_runtime_resolve:
cfi_rel_offset (q1, 80+7*16)
/* Get pointer to linker struct. */
- ldr x0, [ip0, #-8]
+ ldr PTR_REG (0), [ip0, #-PTR_SIZE]
/* Prepare to call _dl_fixup(). */
ldr x1, [sp, 80+8*16] /* Recover &PLTGOT[n] */
@@ -87,7 +91,7 @@ _dl_runtime_resolve:
sub x1, x1, ip0
add x1, x1, x1, lsl #1
lsl x1, x1, #3
- sub x1, x1, #192
+ sub x1, x1, #(RELA_SIZE<<3)
lsr x1, x1, #3
/* Call fixup routine. */
@@ -191,7 +195,7 @@ _dl_runtime_profile:
stp x0, x1, [x29, #OFFSET_RG + DL_OFFSET_RG_SP]
/* Get pointer to linker struct. */
- ldr x0, [ip0, #-8]
+ ldr PTR_REG (0), [ip0, #-PTR_SIZE]
/* Prepare to call _dl_profile_fixup(). */
ldr x1, [x29, OFFSET_PLTGOTN] /* Recover &PLTGOT[n] */
@@ -199,7 +203,7 @@ _dl_runtime_profile:
sub x1, x1, ip0
add x1, x1, x1, lsl #1
lsl x1, x1, #3
- sub x1, x1, #192
+ sub x1, x1, #(RELA_SIZE<<3)
lsr x1, x1, #3
stp x0, x1, [x29, #OFFSET_SAVED_CALL_X0]
@@ -210,8 +214,8 @@ _dl_runtime_profile:
add x4, x29, #OFFSET_FS /* address of framesize */
bl _dl_profile_fixup
- ldr ip0, [x29, #OFFSET_FS] /* framesize == 0 */
- cmp ip0, #0
+ ldr ip0l, [x29, #OFFSET_FS] /* framesize == 0 */
+ cmp ip0l, #0
bge 1f
cfi_remember_state
@@ -243,7 +247,7 @@ _dl_runtime_profile:
1:
/* The new frame size is in ip0. */
- sub x1, x29, ip0
+ sub PTR_REG (1), PTR_REG (29), ip0l
and sp, x1, #0xfffffffffffffff0
str x0, [x29, #OFFSET_T1]
diff --git a/sysdeps/aarch64/jmpbuf-unwind.h b/sysdeps/aarch64/jmpbuf-unwind.h
index 3e0a37d..6a03fe6 100644
--- a/sysdeps/aarch64/jmpbuf-unwind.h
+++ b/sysdeps/aarch64/jmpbuf-unwind.h
@@ -27,7 +27,8 @@
((void *) (address) < (void *) demangle (jmpbuf[JB_SP]))
#define _JMPBUF_CFA_UNWINDS_ADJ(jmpbuf, context, adj) \
- _JMPBUF_UNWINDS_ADJ (jmpbuf, (void *) _Unwind_GetCFA (context), adj)
+ _JMPBUF_UNWINDS_ADJ (jmpbuf, (void *) (uintptr_t) _Unwind_GetCFA (context), \
+ adj)
#define _JMPBUF_UNWINDS_ADJ(_jmpbuf, _address, _adj) \
((uintptr_t) (_address) - (_adj) < _jmpbuf_sp (_jmpbuf) - (_adj))
diff --git a/sysdeps/aarch64/memcmp.S b/sysdeps/aarch64/memcmp.S
index ae2d997..8b87e9b 100644
--- a/sysdeps/aarch64/memcmp.S
+++ b/sysdeps/aarch64/memcmp.S
@@ -47,6 +47,9 @@
#define mask x13
ENTRY_ALIGN (memcmp, 6)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
cbz limit, L(ret0)
eor tmp1, src1, src2
tst tmp1, #7
diff --git a/sysdeps/aarch64/memcpy.S b/sysdeps/aarch64/memcpy.S
index de73f0f..b269316 100644
--- a/sysdeps/aarch64/memcpy.S
+++ b/sysdeps/aarch64/memcpy.S
@@ -61,6 +61,10 @@
ENTRY_ALIGN (memmove, 6)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
+
sub tmp1, dstin, src
cmp count, 96
ccmp tmp1, count, 2, hi
@@ -71,6 +75,10 @@ END (memmove)
libc_hidden_builtin_def (memmove)
ENTRY (memcpy)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
+
prfm PLDL1KEEP, [src]
add srcend, src, count
add dstend, dstin, count
diff --git a/sysdeps/aarch64/memset.S b/sysdeps/aarch64/memset.S
index 4d222c5..7bad29a 100644
--- a/sysdeps/aarch64/memset.S
+++ b/sysdeps/aarch64/memset.S
@@ -39,6 +39,9 @@
ENTRY_ALIGN (__memset, 6)
+ DELOUSE (0)
+ DELOUSE (2)
+
dup v0.16B, valw
add dstend, dstin, count
diff --git a/sysdeps/aarch64/nptl/bits/pthreadtypes.h b/sysdeps/aarch64/nptl/bits/pthreadtypes.h
index c376e64..5bf2eb2 100644
--- a/sysdeps/aarch64/nptl/bits/pthreadtypes.h
+++ b/sysdeps/aarch64/nptl/bits/pthreadtypes.h
@@ -21,17 +21,31 @@
#include <endian.h>
-#define __SIZEOF_PTHREAD_ATTR_T 64
-#define __SIZEOF_PTHREAD_MUTEX_T 48
-#define __SIZEOF_PTHREAD_MUTEXATTR_T 8
-#define __SIZEOF_PTHREAD_COND_T 48
-#define __SIZEOF_PTHREAD_COND_COMPAT_T 48
-#define __SIZEOF_PTHREAD_CONDATTR_T 8
-#define __SIZEOF_PTHREAD_RWLOCK_T 56
-#define __SIZEOF_PTHREAD_RWLOCKATTR_T 8
-#define __SIZEOF_PTHREAD_BARRIER_T 32
-#define __SIZEOF_PTHREAD_BARRIERATTR_T 8
+#ifdef __ILP32__
+# define __SIZEOF_PTHREAD_ATTR_T 32
+# define __SIZEOF_PTHREAD_MUTEX_T 32
+# define __SIZEOF_PTHREAD_MUTEXATTR_T 4
+# define __SIZEOF_PTHREAD_COND_T 48
+# define __SIZEOF_PTHREAD_COND_COMPAT_T 48
+# define __SIZEOF_PTHREAD_CONDATTR_T 4
+# define __SIZEOF_PTHREAD_RWLOCK_T 48
+# define __SIZEOF_PTHREAD_RWLOCKATTR_T 8
+# define __SIZEOF_PTHREAD_BARRIER_T 20
+# define __SIZEOF_PTHREAD_BARRIERATTR_T 4
+#else
+# define __SIZEOF_PTHREAD_ATTR_T 64
+# define __SIZEOF_PTHREAD_MUTEX_T 48
+# define __SIZEOF_PTHREAD_MUTEXATTR_T 8
+# define __SIZEOF_PTHREAD_COND_T 48
+# define __SIZEOF_PTHREAD_COND_COMPAT_T 48
+# define __SIZEOF_PTHREAD_CONDATTR_T 8
+# define __SIZEOF_PTHREAD_RWLOCK_T 56
+# define __SIZEOF_PTHREAD_RWLOCKATTR_T 8
+# define __SIZEOF_PTHREAD_BARRIER_T 32
+# define __SIZEOF_PTHREAD_BARRIERATTR_T 8
+#endif
+#define __PTHREAD_RWLOCK_INT_FLAGS_SHARED 1
/* Thread identifiers. The structure of the attribute type is not
exposed on purpose. */
diff --git a/sysdeps/aarch64/nptl/bits/semaphore.h b/sysdeps/aarch64/nptl/bits/semaphore.h
index 3cc5b37..6ff8d47 100644
--- a/sysdeps/aarch64/nptl/bits/semaphore.h
+++ b/sysdeps/aarch64/nptl/bits/semaphore.h
@@ -21,7 +21,11 @@
#endif
-#define __SIZEOF_SEM_T 32
+#ifdef __ILP32__
+# define __SIZEOF_SEM_T 16
+#else
+# define __SIZEOF_SEM_T 32
+#endif
/* Value returned if `sem_open' failed. */
@@ -31,5 +35,5 @@
typedef union
{
char __size[__SIZEOF_SEM_T];
- long int __align;
+ long long int __align;
} sem_t;
diff --git a/sysdeps/aarch64/setjmp.S b/sysdeps/aarch64/setjmp.S
index 22f4368..e03b3b5 100644
--- a/sysdeps/aarch64/setjmp.S
+++ b/sysdeps/aarch64/setjmp.S
@@ -33,6 +33,7 @@ END (_setjmp)
libc_hidden_def (_setjmp)
ENTRY (__sigsetjmp)
+ DELOUSE (0)
1:
stp x19, x20, [x0, #JB_X19<<3]
@@ -42,7 +43,7 @@ ENTRY (__sigsetjmp)
stp x27, x28, [x0, #JB_X27<<3]
#ifdef PTR_MANGLE
- PTR_MANGLE (x4, x30, x3, x2)
+ PTR_MANGLE (4, 30, 3, 2)
stp x29, x4, [x0, #JB_X29<<3]
#else
stp x29, x30, [x0, #JB_X29<<3]
@@ -57,7 +58,7 @@ ENTRY (__sigsetjmp)
stp d14, d15, [x0, #JB_D14<<3]
#ifdef PTR_MANGLE
mov x4, sp
- PTR_MANGLE (x5, x4, x3, x2)
+ PTR_MANGLE (5, 4, 3, 2)
str x5, [x0, #JB_SP<<3]
#else
mov x2, sp
diff --git a/sysdeps/aarch64/start.S b/sysdeps/aarch64/start.S
index efe2474..9198c57 100644
--- a/sysdeps/aarch64/start.S
+++ b/sysdeps/aarch64/start.S
@@ -16,6 +16,8 @@
License along with the GNU C Library. If not, see
<http://www.gnu.org/licenses/>. */
+#include <sysdep.h>
+
/* This is the canonical entry point, usually the first thing in the text
segment.
@@ -25,7 +27,7 @@
At this entry point, most registers' values are unspecified, except:
- x0 Contains a function pointer to be registered with `atexit'.
+ x0/w0 Contains a function pointer to be registered with `atexit'.
This is how the dynamic linker arranges to have DT_FINI
functions called for shared libraries that have been loaded
before this code runs.
@@ -52,26 +54,26 @@ _start:
mov x5, x0
/* Load argc and a pointer to argv */
- ldr x1, [sp, #0]
- add x2, sp, #8
+ ldr PTR_REG (1), [sp, #0]
+ add x2, sp, #PTR_SIZE
/* Setup stack limit in argument register */
mov x6, sp
#ifdef SHARED
adrp x0, :got:main
- ldr x0, [x0, #:got_lo12:main]
+ ldr PTR_REG (0), [x0, #:got_lo12:main]
adrp x3, :got:__libc_csu_init
- ldr x3, [x3, #:got_lo12:__libc_csu_init]
+ ldr PTR_REG (3), [x3, #:got_lo12:__libc_csu_init]
adrp x4, :got:__libc_csu_fini
- ldr x4, [x4, #:got_lo12:__libc_csu_fini]
+ ldr PTR_REG (4), [x4, #:got_lo12:__libc_csu_fini]
#else
/* Set up the other arguments in registers */
- ldr x0, =main
- ldr x3, =__libc_csu_init
- ldr x4, =__libc_csu_fini
+ ldr PTR_REG (0), =main
+ ldr PTR_REG (3), =__libc_csu_init
+ ldr PTR_REG (4), =__libc_csu_fini
#endif
/* __libc_start_main (main, argc, argv, init, fini, rtld_fini,
diff --git a/sysdeps/aarch64/strchr.S b/sysdeps/aarch64/strchr.S
index 5e3aecf..c66fea3 100644
--- a/sysdeps/aarch64/strchr.S
+++ b/sysdeps/aarch64/strchr.S
@@ -62,6 +62,7 @@
/* Locals and temporaries. */
ENTRY (strchr)
+ DELOUSE (0)
mov wtmp2, #0x0401
movk wtmp2, #0x4010, lsl #16
dup vrepchr.16b, chrin
diff --git a/sysdeps/aarch64/strchrnul.S b/sysdeps/aarch64/strchrnul.S
index a624c8d..c2cc47e 100644
--- a/sysdeps/aarch64/strchrnul.S
+++ b/sysdeps/aarch64/strchrnul.S
@@ -60,6 +60,7 @@
identify exactly which byte is causing the termination. */
ENTRY (__strchrnul)
+ DELOUSE (0)
/* Magic constant 0x40100401 to allow us to identify which lane
matches the termination condition. */
mov wtmp2, #0x0401
diff --git a/sysdeps/aarch64/strcmp.S b/sysdeps/aarch64/strcmp.S
index ba0ccb4..49e528b 100644
--- a/sysdeps/aarch64/strcmp.S
+++ b/sysdeps/aarch64/strcmp.S
@@ -49,6 +49,8 @@
/* Start of performance-critical section -- one 64B cache line. */
ENTRY_ALIGN(strcmp, 6)
+ DELOUSE (0)
+ DELOUSE (1)
eor tmp1, src1, src2
mov zeroones, #REP8_01
tst tmp1, #7
diff --git a/sysdeps/aarch64/strcpy.S b/sysdeps/aarch64/strcpy.S
index 0694199..45809e8 100644
--- a/sysdeps/aarch64/strcpy.S
+++ b/sysdeps/aarch64/strcpy.S
@@ -91,6 +91,8 @@
#define MIN_PAGE_SIZE (1 << MIN_PAGE_P2)
ENTRY_ALIGN (STRCPY, 6)
+ DELOUSE (0)
+ DELOUSE (1)
/* For moderately short strings, the fastest way to do the copy is to
calculate the length of the string in the same way as strlen, then
essentially do a memcpy of the result. This avoids the need for
diff --git a/sysdeps/aarch64/strlen.S b/sysdeps/aarch64/strlen.S
index a07834b..5fb653a 100644
--- a/sysdeps/aarch64/strlen.S
+++ b/sysdeps/aarch64/strlen.S
@@ -85,6 +85,8 @@
boundary. */
ENTRY_ALIGN (__strlen, 6)
+ DELOUSE (0)
+ DELOUSE (1)
and tmp1, srcin, MIN_PAGE_SIZE - 1
mov zeroones, REP8_01
cmp tmp1, MIN_PAGE_SIZE - 16
diff --git a/sysdeps/aarch64/strncmp.S b/sysdeps/aarch64/strncmp.S
index f6a17fd..02de93c 100644
--- a/sysdeps/aarch64/strncmp.S
+++ b/sysdeps/aarch64/strncmp.S
@@ -51,6 +51,9 @@
#define endloop x15
ENTRY_ALIGN_AND_PAD (strncmp, 6, 7)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
cbz limit, L(ret0)
eor tmp1, src1, src2
mov zeroones, #REP8_01
diff --git a/sysdeps/aarch64/strnlen.S b/sysdeps/aarch64/strnlen.S
index 4cce45f..af765f1 100644
--- a/sysdeps/aarch64/strnlen.S
+++ b/sysdeps/aarch64/strnlen.S
@@ -50,6 +50,9 @@
#define REP8_80 0x8080808080808080
ENTRY_ALIGN_AND_PAD (__strnlen, 6, 9)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
cbz limit, L(hit_limit)
mov zeroones, #REP8_01
bic src, srcin, #15
diff --git a/sysdeps/aarch64/strrchr.S b/sysdeps/aarch64/strrchr.S
index 44c1917..ea37968 100644
--- a/sysdeps/aarch64/strrchr.S
+++ b/sysdeps/aarch64/strrchr.S
@@ -68,6 +68,7 @@
identify exactly which byte is causing the termination, and why. */
ENTRY(strrchr)
+ DELOUSE (0)
cbz x1, L(null_search)
/* Magic constant 0x40100401 to allow us to identify which lane
matches the requested byte. Magic constant 0x80200802 used
diff --git a/sysdeps/aarch64/sysdep.h b/sysdeps/aarch64/sysdep.h
index e045759..0a7dccb 100644
--- a/sysdeps/aarch64/sysdep.h
+++ b/sysdeps/aarch64/sysdep.h
@@ -16,8 +16,25 @@
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
+#ifndef _AARCH64_SYSDEP_H
+#define _AARCH64_SYSDEP_H
+
#include <sysdeps/generic/sysdep.h>
+#ifdef __LP64__
+# define AARCH64_R(NAME) R_AARCH64_ ## NAME
+# define PTR_REG(n) x##n
+# define PTR_LOG_SIZE 3
+# define DELOUSE(n)
+#else
+# define AARCH64_R(NAME) R_AARCH64_P32_ ## NAME
+# define PTR_REG(n) w##n
+# define PTR_LOG_SIZE 2
+# define DELOUSE(n) mov w##n, w##n
+#endif
+
+#define PTR_SIZE (1<<PTR_LOG_SIZE)
+
#ifdef __ASSEMBLER__
/* Syntactic details of assembler. */
@@ -107,16 +124,18 @@
# define L(name) .L##name
#endif
-/* Load or store to/from a pc-relative EXPR into/from R, using T. */
-#define LDST_PCREL(OP, R, T, EXPR) \
- adrp T, EXPR; \
- OP R, [T, #:lo12:EXPR];\
+/* Load or store to/from a pc-relative EXPR into/from R, using T.
+ Note R and T are register numbers and not register names. */
+#define LDST_PCREL(OP, R, T, EXPR) \
+ adrp x##T, EXPR; \
+ OP PTR_REG (R), [x##T, #:lo12:EXPR]; \
-/* Load or store to/from a got-relative EXPR into/from R, using T. */
-#define LDST_GLOBAL(OP, R, T, EXPR) \
- adrp T, :got:EXPR; \
- ldr T, [T, #:got_lo12:EXPR];\
- OP R, [T];
+/* Load or store to/from a got-relative EXPR into/from R, using T.
+ Note R and T are register numbers and not register names. */
+#define LDST_GLOBAL(OP, R, T, EXPR) \
+ adrp x##T, :got:EXPR; \
+ ldr PTR_REG (T), [x##T, #:got_lo12:EXPR]; \
+ OP PTR_REG (R), [x##T];
/* Since C identifiers are not normally prefixed with an underscore
on this system, the asm identifier `syscall_error' intrudes on the
@@ -125,3 +144,5 @@
#define mcount _mcount
#endif /* __ASSEMBLER__ */
+
+#endif /* _AARCH64_SYSDEP_H */
diff --git a/sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h b/sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h
index 658f696..1717c35 100644
--- a/sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h
+++ b/sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h
@@ -25,11 +25,17 @@
#define __O_NOFOLLOW 0100000
#define __O_DIRECT 0200000
-#define __O_LARGEFILE 0
+#ifdef __ILP32__
+# define __O_LARGEFILE 0400000
+#else
+# define __O_LARGEFILE 0
+#endif
+#ifdef __LP64__
# define F_GETLK64 5
# define F_SETLK64 6
# define F_SETLKW64 7
+#endif
struct flock
{
diff --git a/sysdeps/unix/sysv/linux/aarch64/clone.S b/sysdeps/unix/sysv/linux/aarch64/clone.S
index 76baa7a..220ec11 100644
--- a/sysdeps/unix/sysv/linux/aarch64/clone.S
+++ b/sysdeps/unix/sysv/linux/aarch64/clone.S
@@ -39,6 +39,13 @@
*/
.text
ENTRY(__clone)
+ DELOUSE (0)
+ DELOUSE (1)
+ DELOUSE (2)
+ DELOUSE (3)
+ DELOUSE (4)
+ DELOUSE (5)
+ DELOUSE (6)
/* Save args for the child. */
mov x10, x0
mov x11, x2
diff --git a/sysdeps/unix/sysv/linux/aarch64/dl-cache.h b/sysdeps/unix/sysv/linux/aarch64/dl-cache.h
index 9c7b271..86905ed 100644
--- a/sysdeps/unix/sysv/linux/aarch64/dl-cache.h
+++ b/sysdeps/unix/sysv/linux/aarch64/dl-cache.h
@@ -18,7 +18,11 @@
#include <ldconfig.h>
-#define _DL_CACHE_DEFAULT_ID (FLAG_AARCH64_LIB64 | FLAG_ELF_LIBC6)
+#ifdef __LP64__
+# define _DL_CACHE_DEFAULT_ID (FLAG_AARCH64_LIB64 | FLAG_ELF_LIBC6)
+#else
+# define _DL_CACHE_DEFAULT_ID (FLAG_AARCH64_LIB32 | FLAG_ELF_LIBC6)
+#endif
#define _dl_cache_check_flags(flags) \
((flags) == _DL_CACHE_DEFAULT_ID)
@@ -27,18 +31,25 @@
do \
{ \
size_t len = strlen (dir); \
- char path[len + 3]; \
+ char path[len + 6]; \
memcpy (path, dir, len + 1); \
if (len >= 6 && ! memcmp (path + len - 6, "/lib64", 6)) \
{ \
len -= 2; \
path[len] = '\0'; \
} \
+ if (len >= 9 && ! memcmp (path + len - 9, "/libilp32", 9))\
+ { \
+ len -= 5; \
+ path[len] = '\0'; \
+ } \
add_dir (path); \
if (len >= 4 && ! memcmp (path + len - 4, "/lib", 4)) \
{ \
memcpy (path + len, "64", 3); \
add_dir (path); \
+ memcpy (path + len, "ilp32", 6); \
+ add_dir (path); \
} \
} while (0)
diff --git a/sysdeps/unix/sysv/linux/aarch64/getcontext.S b/sysdeps/unix/sysv/linux/aarch64/getcontext.S
index c2dd5b8..f6bf24f 100644
--- a/sysdeps/unix/sysv/linux/aarch64/getcontext.S
+++ b/sysdeps/unix/sysv/linux/aarch64/getcontext.S
@@ -30,6 +30,7 @@
.text
ENTRY(__getcontext)
+ DELOUSE (0)
/* The saved context will return to the getcontext() call point
with a return value of 0 */
str xzr, [x0, oX0 + 0 * SZREG]
@@ -90,7 +91,7 @@ ENTRY(__getcontext)
/* Grab the signal mask */
/* rt_sigprocmask (SIG_BLOCK, NULL, &ucp->uc_sigmask, _NSIG8) */
- add x2, x0, #UCONTEXT_SIGMASK
+ add PTR_REG (2), PTR_REG (0), #UCONTEXT_SIGMASK
mov x0, SIG_BLOCK
mov x1, 0
mov x3, _NSIG8
diff --git a/sysdeps/unix/sysv/linux/aarch64/init-first.c b/sysdeps/unix/sysv/linux/aarch64/init-first.c
index f7224a2..f7bfc4d 100644
--- a/sysdeps/unix/sysv/linux/aarch64/init-first.c
+++ b/sysdeps/unix/sysv/linux/aarch64/init-first.c
@@ -27,17 +27,21 @@ int (*VDSO_SYMBOL(clock_getres)) (clockid_t, struct timespec *);
static inline void
_libc_vdso_platform_setup (void)
{
- PREPARE_VERSION (linux2639, "LINUX_2.6.39", 123718537);
+#ifdef __LP64__
+ PREPARE_VERSION (linux_version, "LINUX_2.6.39", 123718537);
+#else
+ PREPARE_VERSION (linux_version, "LINUX_4.9", 61765625);
+#endif
- void *p = _dl_vdso_vsym ("__kernel_gettimeofday", &linux2639);
+ void *p = _dl_vdso_vsym ("__kernel_gettimeofday", &linux_version);
PTR_MANGLE (p);
VDSO_SYMBOL(gettimeofday) = p;
- p = _dl_vdso_vsym ("__kernel_clock_gettime", &linux2639);
+ p = _dl_vdso_vsym ("__kernel_clock_gettime", &linux_version);
PTR_MANGLE (p);
VDSO_SYMBOL(clock_gettime) = p;
- p = _dl_vdso_vsym ("__kernel_clock_getres", &linux2639);
+ p = _dl_vdso_vsym ("__kernel_clock_getres", &linux_version);
PTR_MANGLE (p);
VDSO_SYMBOL(clock_getres) = p;
}
diff --git a/sysdeps/unix/sysv/linux/aarch64/ldconfig.h b/sysdeps/unix/sysv/linux/aarch64/ldconfig.h
index ee91ef8..ac84194 100644
--- a/sysdeps/unix/sysv/linux/aarch64/ldconfig.h
+++ b/sysdeps/unix/sysv/linux/aarch64/ldconfig.h
@@ -21,6 +21,8 @@
#define SYSDEP_KNOWN_INTERPRETER_NAMES \
{ "/lib/ld-linux-aarch64.so.1", FLAG_ELF_LIBC6 }, \
{ "/lib/ld-linux-aarch64_be.so.1", FLAG_ELF_LIBC6 }, \
+ { "/lib/ld-linux-aarch64_ilp32.so.1", FLAG_ELF_LIBC6 }, \
+ { "/lib/ld-linux-aarch64_be_ilp32.so.1", FLAG_ELF_LIBC6 }, \
{ "/lib/ld-linux.so.3", FLAG_ELF_LIBC6 }, \
{ "/lib/ld-linux-armhf.so.3", FLAG_ELF_LIBC6 },
#define SYSDEP_KNOWN_LIBRARY_NAMES \
diff --git a/sysdeps/unix/sysv/linux/aarch64/setcontext.S b/sysdeps/unix/sysv/linux/aarch64/setcontext.S
index d17f8c8..c2bca26 100644
--- a/sysdeps/unix/sysv/linux/aarch64/setcontext.S
+++ b/sysdeps/unix/sysv/linux/aarch64/setcontext.S
@@ -34,6 +34,7 @@
.text
ENTRY (__setcontext)
+ DELOUSE (0)
/* Save a copy of UCP. */
mov x9, x0
diff --git a/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h b/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h
index a579501..d1185fa 100644
--- a/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h
+++ b/sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h
@@ -16,10 +16,11 @@
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
+#include <stdint.h>
#include <sys/ucontext.h>
#define SIGCONTEXT siginfo_t *_si, struct ucontext *
-#define GET_PC(ctx) ((void *) (ctx)->uc_mcontext.pc)
+#define GET_PC(ctx) ((void *) (uintptr_t) (ctx)->uc_mcontext.pc)
/* There is no reliable way to get the sigcontext unless we use a
three-argument signal handler. */
diff --git a/sysdeps/unix/sysv/linux/aarch64/swapcontext.S b/sysdeps/unix/sysv/linux/aarch64/swapcontext.S
index c1a16f3..8e2cadd 100644
--- a/sysdeps/unix/sysv/linux/aarch64/swapcontext.S
+++ b/sysdeps/unix/sysv/linux/aarch64/swapcontext.S
@@ -27,6 +27,7 @@
.text
ENTRY(__swapcontext)
+ DELOUSE (0)
/* Set the value returned when swapcontext() returns in this context. */
str xzr, [x0, oX0 + 0 * SZREG]
diff --git a/sysdeps/unix/sysv/linux/aarch64/sysdep.h b/sysdeps/unix/sysv/linux/aarch64/sysdep.h
index a397e50..1ffabc2 100644
--- a/sysdeps/unix/sysv/linux/aarch64/sysdep.h
+++ b/sysdeps/unix/sysv/linux/aarch64/sysdep.h
@@ -250,12 +250,14 @@
(!defined SHARED && (IS_IN (libc) \
|| IS_IN (libpthread))))
# ifdef __ASSEMBLER__
+/* Note, dst, src, guard, and tmp are all register numbers rather than
+ register names so they will work with both ILP32 and LP64. */
# define PTR_MANGLE(dst, src, guard, tmp) \
LDST_PCREL (ldr, guard, tmp, C_SYMBOL_NAME(__pointer_chk_guard_local)); \
PTR_MANGLE2 (dst, src, guard)
/* Use PTR_MANGLE2 for efficiency if guard is already loaded. */
# define PTR_MANGLE2(dst, src, guard)\
- eor dst, src, guard
+ eor x##dst, x##src, x##guard
# define PTR_DEMANGLE(dst, src, guard, tmp)\
PTR_MANGLE (dst, src, guard, tmp)
# define PTR_DEMANGLE2(dst, src, guard)\
@@ -268,12 +270,14 @@ extern uintptr_t __pointer_chk_guard_local attribute_relro attribute_hidden;
# endif
#else
# ifdef __ASSEMBLER__
+/* Note, dst, src, guard, and tmp are all register numbers rather than
+ register names so they will work with both ILP32 and LP64. */
# define PTR_MANGLE(dst, src, guard, tmp) \
LDST_GLOBAL (ldr, guard, tmp, C_SYMBOL_NAME(__pointer_chk_guard)); \
PTR_MANGLE2 (dst, src, guard)
/* Use PTR_MANGLE2 for efficiency if guard is already loaded. */
# define PTR_MANGLE2(dst, src, guard)\
- eor dst, src, guard
+ eor x##dst, x##src, x##guard
# define PTR_DEMANGLE(dst, src, guard, tmp)\
PTR_MANGLE (dst, src, guard, tmp)
# define PTR_DEMANGLE2(dst, src, guard)\
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] Partial ILP32 support for aarch64
2016-11-23 23:01 ` Steve Ellcey
@ 2016-11-28 9:05 ` Andreas Schwab
0 siblings, 0 replies; 12+ messages in thread
From: Andreas Schwab @ 2016-11-28 9:05 UTC (permalink / raw)
To: Steve Ellcey; +Cc: Joseph Myers, libc-alpha
On Nov 24 2016, Steve Ellcey <sellcey@caviumnetworks.com> wrote:
> 2016-11-23  Andrew Pinski  <andrew.pinski@caviumnetworks.com>
>     Yury Norov  <ynorov@caviumnetworks.com>
>     Steve Ellcey  <sellcey@caviumnetworks.com>
>
> * sysdeps/aarch64/crti.S: Add include of sysdep.h.
> (call_weak_fn): Use PTR_REG to get correct reg name in ILP32.
> * sysdeps/aarch64/dl-irel.h: Add include of sysdep.h.
> (elf_irela): Use AARCH64_R macro to get correct relocation in ILP32.
> * sysdeps/aarch64/dl-machine.h: Add include of sysdep.h.
> (elf_machine_load_address, RTLD_START, RTLD_START_1, RTLD_START,
> elf_machine_type_class, ELF_MACHINE_JMP_SLOT, elf_machine_rela,
> elf_machine_lazy_rel): Add ifdef's for ILP32 support.
> * sysdeps/aarch64/dl-tlsdesc.S (_dl_tlsdesc_return,
> _dl_tlsdesc_return_lazy, _dl_tlsdesc_dynamic,
> _dl_tlsdesc_resolve_hold): Extend pointers in ILP32, use PTR_REG
> to get correct reg name for ILP32.
> * sysdeps/aarch64/dl-trampoline.S (ip01): New Macro.
> (RELA_SIZE): New Macro.
> (_dl_runtime_resolve, _dl_runtime_profile): Use new macros and PTR_REG
> to support ILP32.
> * sysdeps/aarch64/jmpbuf-unwind.h (_JMPBUF_CFA_UNWINDS_ADJ): Add
> cast for ILP32 mode.
> * sysdeps/aarch64/memcmp.S (memcmp): Extend arg pointers for ILP32 mode.
> * sysdeps/aarch64/memcpy.S (memmove, memcpy): Ditto.
> * sysdeps/aarch64/memset.S (__memset): Ditto.
> * sysdeps/aarch64/strchr.S (strchr): Ditto.
> * sysdeps/aarch64/strchrnul.S (__strchrnul): Ditto.
> * sysdeps/aarch64/strcmp.S (strcmp): Ditto.
> * sysdeps/aarch64/strcpy.S (strcpy): Ditto.
> * sysdeps/aarch64/strlen.S (__strlen): Ditto.
> * sysdeps/aarch64/strncmp.S (strncmp): Ditto.
> * sysdeps/aarch64/strnlen.S (strnlen): Ditto.
> * sysdeps/aarch64/strrchr.S (strrchr): Ditto.
> * sysdeps/unix/sysv/linux/aarch64/clone.S: Ditto.
> * sysdeps/unix/sysv/linux/aarch64/setcontext.S (__setcontext): Ditto.
> * sysdeps/unix/sysv/linux/aarch64/swapcontext.S (__swapcontext): Ditto.
> * sysdeps/aarch64/__longjmp.S (__longjmp): Extend pointers in ILP32,
> change PTR_MANGLE call to use register numbers instead of names.
> * sysdeps/unix/sysv/linux/aarch64/getcontext.S (__getcontext): Ditto.
> * sysdeps/aarch64/setjmp.S (__sigsetjmp): Extend arg pointers for
> ILP32 mode, change PTR_MANGLE calls to use register numbers.
> * sysdeps/aarch64/start.S (_start): Ditto.
> * sysdeps/aarch64/nptl/bits/pthreadtypes.h
> (__PTHREAD_RWLOCK_INT_FLAGS_SHARED): New define.
> (__SIZEOF_PTHREAD_ATTR_T, __SIZEOF_PTHREAD_MUTEX_T,
> __SIZEOF_PTHREAD_MUTEXATTR_T, __SIZEOF_PTHREAD_COND_T,
> __SIZEOF_PTHREAD_COND_COMPAT_T, __SIZEOF_PTHREAD_CONDATTR_T,
> __SIZEOF_PTHREAD_RWLOCK_T, __SIZEOF_PTHREAD_RWLOCKATTR_T,
> __SIZEOF_PTHREAD_BARRIER_T, __SIZEOF_PTHREAD_BARRIERATTR_T):
> Make defined values dependent on __ILP32__.
> * sysdeps/aarch64/nptl/bits/semaphore.h (__SIZEOF_SEM_T): Change define.
> (sem_t): Change __align type.
> * sysdeps/aarch64/sysdep.h (AARCH64_R, PTR_REG, PTR_LOG_SIZE, DELOUSE,
> PTR_SIZE): New Macros.
> (LDST_PCREL, LDST_GLOBAL) Update to use PTR_REG.
> * sysdeps/unix/sysv/linux/aarch64/bits/fcntl.h (O_LARGEFILE):
> Set when in ILP32 mode.
> (F_GETLK64, F_SETLK64, F_SETLKW64): Only set in LP64 mode.
> * sysdeps/unix/sysv/linux/aarch64/dl-cache.h (DL_CACHE_DEFAULT_ID):
> Set elf flags for ILP32.
> (add_system_dir): Set ILP32 library directories.
> * sysdeps/unix/sysv/linux/aarch64/init-first.c
> (_libc_vdso_platform_setup): Set minimum kernel version for ILP32.
> * sysdeps/unix/sysv/linux/aarch64/ldconfig.h
> (SYSDEP_KNOWN_INTERPRETER_NAMES): Add ILP32 names.
> * sysdeps/unix/sysv/linux/aarch64/sigcontextinfo.h (GET_PC, SET_PC):
> New Macros.
> * sysdeps/unix/sysv/linux/aarch64/sysdep.h: Handle ILP32 pointers.
Ok.
Andreas.
--
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE 1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] Partial ILP32 support for aarch64
2016-12-06 14:08 Wilco Dijkstra
2016-12-06 14:47 ` Florian Weimer
@ 2016-12-06 16:10 ` Szabolcs Nagy
1 sibling, 0 replies; 12+ messages in thread
From: Szabolcs Nagy @ 2016-12-06 16:10 UTC (permalink / raw)
To: Wilco Dijkstra, sellcey; +Cc: nd, libc-alpha, Florian Weimer
On 06/12/16 14:08, Wilco Dijkstra wrote:
> This commit breaks any use of LD_LIBRARY_PATH:
>
> LD_LIBRARY_PATH=<newglibc>/lib/ /bin/true
> /bin/true: 8�����: ��: Error 18446744072501075648
>
> It seems there is some incompatibility introduced in the dynamic loader that I presume wasn't intended?
i think this was caused by Florian's patch
"Implement _dl_catch_error, _dl_signal_error in libc.so [BZ #16628]"
https://sourceware.org/ml/libc-alpha/2016-11/msg00465.html
an old dynamic linker fails to load a new libc.so because
the internal abi changed, so hacks that use a newly built
glibc, but old dynamic linker no longer work.
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] Partial ILP32 support for aarch64
2016-12-06 14:57 ` Wilco Dijkstra
2016-12-06 15:12 ` Yury Norov
@ 2016-12-06 15:15 ` Florian Weimer
1 sibling, 0 replies; 12+ messages in thread
From: Florian Weimer @ 2016-12-06 15:15 UTC (permalink / raw)
To: Wilco Dijkstra, sellcey; +Cc: libc-alpha, nd, Szabolcs Nagy
On 12/06/2016 03:57 PM, Wilco Dijkstra wrote:
> Florian Weimer wrote:
>> So you are running the new libc against the system dynamic linker?
>> That's not supported.
>
> So LD_LIBRARY_PATH is not guaranteed to do the correct thing?
No, not at all.
>> You should rather use the testrun.sh wrapper.
>
> For running benchmarks, eg. SPEC?
Yes, it's quite realistic.
>>> It seems there is some incompatibility introduced in the dynamic loader that I presume wasn't intended?
>>
>> The dynamic linker and libc have to be updated in lock-step, and such
>> discrepancies are expected.
>
> So how do I run a binary against a non-system GLIBC?
Either with the script above, or you can build the binaries you a
interested in with special paths, just like
--enable-hardcoded-paths-in-tests does.
For more realistic testing (which is still reasonably easy to set up), I
create a chroot with mock and run “make install DESTDIR=/path/to/chroot”
from outside of the chroot.
Florian
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] Partial ILP32 support for aarch64
2016-12-06 14:57 ` Wilco Dijkstra
@ 2016-12-06 15:12 ` Yury Norov
2016-12-06 15:15 ` Florian Weimer
1 sibling, 0 replies; 12+ messages in thread
From: Yury Norov @ 2016-12-06 15:12 UTC (permalink / raw)
To: Wilco Dijkstra; +Cc: Florian Weimer, sellcey, libc-alpha, nd, Szabolcs Nagy
On Tue, Dec 06, 2016 at 02:57:17PM +0000, Wilco Dijkstra wrote:
> Florian Weimer wrote:
> > So you are running the new libc against the system dynamic linker?
> > That's not supported.
>
> So LD_LIBRARY_PATH is not guaranteed to do the correct thing?
>
> > You should rather use the testrun.sh wrapper.
>
> For running benchmarks, eg. SPEC?
>
> >> It seems there is some incompatibility introduced in the dynamic loader that I presume wasn't intended?
> >
> > The dynamic linker and libc have to be updated in lock-step, and such
> > discrepancies are expected.
>
> So how do I run a binary against a non-system GLIBC?
>
> Wilco
>
>
I provide the path to linker and sys-root at configure step.
Check this, there's no SPEC targets, but it will probably be
helpful.
https://github.com/norov/build-glibc
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] Partial ILP32 support for aarch64
2016-12-06 14:47 ` Florian Weimer
@ 2016-12-06 14:57 ` Wilco Dijkstra
2016-12-06 15:12 ` Yury Norov
2016-12-06 15:15 ` Florian Weimer
0 siblings, 2 replies; 12+ messages in thread
From: Wilco Dijkstra @ 2016-12-06 14:57 UTC (permalink / raw)
To: Florian Weimer, sellcey; +Cc: libc-alpha, nd, Szabolcs Nagy
Florian Weimer wrote:
> So you are running the new libc against the system dynamic linker?
> That's not supported.
So LD_LIBRARY_PATH is not guaranteed to do the correct thing?
> You should rather use the testrun.sh wrapper.
For running benchmarks, eg. SPEC?
>> It seems there is some incompatibility introduced in the dynamic loader that I presume wasn't intended?
>
> The dynamic linker and libc have to be updated in lock-step, and such
> discrepancies are expected.
So how do I run a binary against a non-system GLIBC?
Wilco
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] Partial ILP32 support for aarch64
2016-12-06 14:08 Wilco Dijkstra
@ 2016-12-06 14:47 ` Florian Weimer
2016-12-06 14:57 ` Wilco Dijkstra
2016-12-06 16:10 ` Szabolcs Nagy
1 sibling, 1 reply; 12+ messages in thread
From: Florian Weimer @ 2016-12-06 14:47 UTC (permalink / raw)
To: Wilco Dijkstra, sellcey; +Cc: libc-alpha, nd, Szabolcs Nagy
On 12/06/2016 03:08 PM, Wilco Dijkstra wrote:
> Hi,
>
> This commit breaks any use of LD_LIBRARY_PATH:
>
> LD_LIBRARY_PATH=<newglibc>/lib/ /bin/true
> /bin/true: 8�����: ��: Error 18446744072501075648
So you are running the new libc against the system dynamic linker?
That's not supported.
You should rather use the testrun.sh wrapper.
> It seems there is some incompatibility introduced in the dynamic loader that I presume wasn't intended?
The dynamic linker and libc have to be updated in lock-step, and such
discrepancies are expected.
Florian
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] Partial ILP32 support for aarch64
@ 2016-12-06 14:08 Wilco Dijkstra
2016-12-06 14:47 ` Florian Weimer
2016-12-06 16:10 ` Szabolcs Nagy
0 siblings, 2 replies; 12+ messages in thread
From: Wilco Dijkstra @ 2016-12-06 14:08 UTC (permalink / raw)
To: sellcey; +Cc: libc-alpha, nd, Szabolcs Nagy
Hi,
This commit breaks any use of LD_LIBRARY_PATH:
LD_LIBRARY_PATH=<newglibc>/lib/ /bin/true
/bin/true: 8�����: ��: Error 18446744072501075648
It seems there is some incompatibility introduced in the dynamic loader that I presume wasn't intended?
Wilco
^ permalink raw reply [flat|nested] 12+ messages in thread
end of thread, other threads:[~2016-12-06 16:10 UTC | newest]
Thread overview: 12+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-11-19 0:40 [PATCH] Partial ILP32 support for aarch64 Steve Ellcey
2016-11-21 16:51 ` Joseph Myers
2016-11-21 20:50 ` Steve Ellcey
2016-11-23 9:44 ` Andreas Schwab
2016-11-23 23:01 ` Steve Ellcey
2016-11-28 9:05 ` Andreas Schwab
2016-12-06 14:08 Wilco Dijkstra
2016-12-06 14:47 ` Florian Weimer
2016-12-06 14:57 ` Wilco Dijkstra
2016-12-06 15:12 ` Yury Norov
2016-12-06 15:15 ` Florian Weimer
2016-12-06 16:10 ` Szabolcs Nagy
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).