Skip to contentSkip to conversation

tilegx: Can load elf64 tilegx binary successfully for linux-user

PATCH In review Chen Gang S 5 participants 24 messages Feb 22, 2015 – Feb 28, 2015

lore
Share & tools
lore
#0 lore raw

v4 cover tilegx: Can load elf64 tilegx binary successfully for linux-user

After load elf64 tilegx binary for linux-user, the working flow reaches 1st correct instruction "__start". Next, we shall load all instructions for qemu using.

This patch is based on Linux kernel tile architecture tilegx 64-bit implementation, and also based on tilegx architecture ABI reference.

The related test:

[root@localhost qemu]# ./configure --target-list=tilegx-linux-user && make [root@localhost qemu]# ./tilegx-linux-user/qemu-tilegx -d all ./test.tgx CPU Reset (CPU 0)
CPU Reset (CPU 0)
host mmap_min_addr=0x10000
Reserved 0xe0000 bytes of guest address space
Relocating guest address space from 0x0000000000010000 to 0x10000 guest_base 0x0
start end size prot
0000000000010000-00000000000e0000 00000000000d0000 r-x 00000000000e0000-00000000000f0000 0000000000010000 rw- 0000004000000000-0000004000010000 0000000000010000 --- 0000004000010000-0000004000810000 0000000000800000 rw-

  start_brk   0x0000000000000000
  end_code    0x00000000000d86f7
  start_code  0x0000000000010000
  start_data  0x00000000000e86f8
  end_data    0x00000000000ea208
  start_stack 0x000000400080f250
  brk         0x00000000000ec2b0
  entry       0x0000000000010f60
  PROLOGUE: [size=40]
  0x7fcc44c716f0:  push   %rbp 
  0x7fcc44c716f1:  push   %rbx 
  0x7fcc44c716f2:  push   %r12 
  0x7fcc44c716f4:  push   %r13 
  0x7fcc44c716f6:  push   %r14 
  0x7fcc44c716f8:  push   %r15 
  0x7fcc44c716fa:  mov    %rdi,%r14
  0x7fcc44c716fd:  add    $0xfffffffffffffb78,%rsp
  0x7fcc44c71704:  jmpq   *%rsi
  0x7fcc44c71706:  add    $0x488,%rsp
  0x7fcc44c7170d:  pop    %r15 
  0x7fcc44c7170f:  pop    %r14 
  0x7fcc44c71711:  pop    %r13 
  0x7fcc44c71713:  pop    %r12 
  0x7fcc44c71715:  pop    %rbx 
  0x7fcc44c71716:  pop    %rbp 
  0x7fcc44c71717:  retq 

Load elf64 tilegx successfully
reach code start position: [0000000000010f60] _start

[root@localhost qemu]# echo $?
0
[root@localhost qemu]#

Chen Gang (6):
target-tilegx: Firstly add to qemu with minimized features
linux-user: tilegx: Firstly add architecture related features
linux-user: tilegx: Add target features support within qemu
linux-user: Support tilegx architecture in syscall
linux-user: Support tilegx architecture in linux-user linux-user/syscall.c: Switch all macros which are not defined in
tilegx

 configure                             |   7 +
 default-configs/tilegx-linux-user.mak |   1 +
 include/elf.h                         |   2 +
 linux-user/elfload.c                  |  23 +++
 linux-user/main.c                     |  74 +++++++++
 linux-user/syscall.c                  |  50 +++++-
 linux-user/syscall_defs.h             |  38 ++++-
 linux-user/tilegx/syscall.h           |  80 ++++++++++
 linux-user/tilegx/syscall_nr.h        | 278 +++++++++++++++++++++++++++++++++
 linux-user/tilegx/target_cpu.h        |  35 +++++
 linux-user/tilegx/target_signal.h     |  28 ++++
 linux-user/tilegx/target_structs.h    |  48 ++++++
 linux-user/tilegx/termbits.h          | 285 ++++++++++++++++++++++++++++++++++
 target-tilegx/Makefile.objs           |   1 +
 target-tilegx/cpu-qom.h               |  72 +++++++++
 target-tilegx/cpu.c                   | 162 +++++++++++++++++++
 target-tilegx/cpu.h                   |  85 ++++++++++
 target-tilegx/helper.h                |   0
 target-tilegx/translate.c             |  54 +++++++
 19 files changed, 1318 insertions(+), 5 deletions(-)
 create mode 100644 default-configs/tilegx-linux-user.mak
 create mode 100644 linux-user/tilegx/syscall.h
 create mode 100644 linux-user/tilegx/syscall_nr.h
 create mode 100644 linux-user/tilegx/target_cpu.h
 create mode 100644 linux-user/tilegx/target_signal.h
 create mode 100644 linux-user/tilegx/target_structs.h
 create mode 100644 linux-user/tilegx/termbits.h
 create mode 100644 target-tilegx/Makefile.objs
 create mode 100644 target-tilegx/cpu-qom.h
 create mode 100644 target-tilegx/cpu.c
 create mode 100644 target-tilegx/cpu.h
 create mode 100644 target-tilegx/helper.h
 create mode 100644 target-tilegx/translate.c
Signature · 1 line
1.9.3
#1 lore raw

PATCH v4 1/6 target-tilegx: Firstly add to qemu with minimized features

It almost likes a template for adding an architecture target.

---
 configure                             |   7 ++
 default-configs/tilegx-linux-user.mak |   1 +
 target-tilegx/Makefile.objs           |   1 +
 target-tilegx/cpu-qom.h               |  72 +++++++++++++++
 target-tilegx/cpu.c                   | 162 ++++++++++++++++++++++++++++++++++
 target-tilegx/cpu.h                   |  85 ++++++++++++++++++
 target-tilegx/helper.h                |   0
 target-tilegx/translate.c             |  54 ++++++++++++
 8 files changed, 382 insertions(+)
 create mode 100644 default-configs/tilegx-linux-user.mak
 create mode 100644 target-tilegx/Makefile.objs
 create mode 100644 target-tilegx/cpu-qom.h
 create mode 100644 target-tilegx/cpu.c
 create mode 100644 target-tilegx/cpu.h
 create mode 100644 target-tilegx/helper.h
 create mode 100644 target-tilegx/translate.c
configure +7 -0
index 7ba4bcb..23aa8f6 100755
@@ -5191,6 +5191,9 @@ case "$target_name" in
   s390x)
     gdb_xml_files="s390x-core64.xml s390-acr.xml s390-fpr.xml"
   ;;
+  tilegx)
+    TARGET_ARCH=tilegx
+  ;;
   unicore32)
   ;;
   xtensa|xtensaeb)
@@ -5363,6 +5366,10 @@ for i in $ARCH $TARGET_BASE_ARCH ; do
     echo "CONFIG_SPARC_DIS=y"  >> $config_target_mak
     echo "CONFIG_SPARC_DIS=y"  >> config-all-disas.mak
   ;;
+  tilegx*)
+    echo "CONFIG_TILEGX_DIS=y"  >> $config_target_mak
+    echo "CONFIG_TILEGX_DIS=y"  >> config-all-disas.mak
+  ;;
   xtensa*)
     echo "CONFIG_XTENSA_DIS=y"  >> $config_target_mak
     echo "CONFIG_XTENSA_DIS=y"  >> config-all-disas.mak
default-configs/tilegx-linux-user.mak new file +1 -0
new file mode 100644
index 0000000..3e47493
@@ -0,0 +1 @@
+# Default configuration for tilegx-linux-user
target-tilegx/Makefile.objs new file +1 -0
new file mode 100644
index 0000000..dcf2fe4
@@ -0,0 +1 @@
+obj-y += cpu.o translate.o
target-tilegx/cpu-qom.h new file +72 -0

72 lines not on this page · raw message

target-tilegx/cpu.c new file +162 -0

162 lines not on this page · raw message

target-tilegx/cpu.h new file +85 -0

85 lines not on this page · raw message

target-tilegx/helper.h new file +0 -0
new file mode 100644
index 0000000..e69de29
target-tilegx/translate.c new file +54 -0

54 lines not on this page · raw message

Signature · 1 line
1.9.3
#2 lore raw

Re: [Qemu-devel] [PATCH 1/6 v4] target-tilegx: Firstly add to qemu with minimized features

Hi,

"target-tilegx: Initial stub" or "...support"? No need to mention QEMU (spelling!) in a QEMU commit.

Am 22.02.2015 um 14:33 schrieb Chen Gang S:

Quoted 1 line

It almost likes a template for adding an architecture target.

That's a comment that's not really descriptive of what the patch does. Instead, maybe mention that this is the configure and build system support etc. and what name to use for enabling it from configure, that it's linux-user only for now, ...?

Quoted 35 lines
---
 configure                             |   7 ++
 default-configs/tilegx-linux-user.mak |   1 +
 target-tilegx/Makefile.objs           |   1 +
 target-tilegx/cpu-qom.h               |  72 +++++++++++++++
 target-tilegx/cpu.c                   | 162 ++++++++++++++++++++++++++++++++++
 target-tilegx/cpu.h                   |  85 ++++++++++++++++++
 target-tilegx/helper.h                |   0
 target-tilegx/translate.c             |  54 ++++++++++++
 8 files changed, 382 insertions(+)
 create mode 100644 default-configs/tilegx-linux-user.mak
 create mode 100644 target-tilegx/Makefile.objs
 create mode 100644 target-tilegx/cpu-qom.h
 create mode 100644 target-tilegx/cpu.c
 create mode 100644 target-tilegx/cpu.h
 create mode 100644 target-tilegx/helper.h
 create mode 100644 target-tilegx/translate.c
configure +3 -0
index 7ba4bcb..23aa8f6 100755
@@ -5191,6 +5191,9 @@ case "$target_name" in
   s390x)
     gdb_xml_files="s390x-core64.xml s390-acr.xml s390-fpr.xml"
   ;;
+  tilegx)
+    TARGET_ARCH=tilegx
+  ;;
   unicore32)
   ;;
   xtensa|xtensaeb)

configure for i in $ARCH $TARGET_BASE_ARCH ; do · line 5372

Quoted 7 lines
configure +4 -0
@@ -5363,6 +5366,10 @@ for i in $ARCH $TARGET_BASE_ARCH ; do
     echo "CONFIG_SPARC_DIS=y"  >> $config_target_mak
     echo "CONFIG_SPARC_DIS=y"  >> config-all-disas.mak
   ;;
+  tilegx*)
+    echo "CONFIG_TILEGX_DIS=y"  >> $config_target_mak
+    echo "CONFIG_TILEGX_DIS=y"  >> config-all-disas.mak
+  ;;

Hadn't you been asked to drop these lines, as you are not yet adding any disassembler code that uses it?

Quoted 3 lines
configure +0 -0
   xtensa*)
     echo "CONFIG_XTENSA_DIS=y"  >> $config_target_mak
     echo "CONFIG_XTENSA_DIS=y"  >> config-all-disas.mak

[...]

target-tilegx/cpu-qom.h line 2

Quoted 8 lines
target-tilegx/cpu-qom.h new file +2 -0
new file mode 100644
index 0000000..e15a8b8
@@ -0,0 +1,72 @@
+/*
+ * QEMU Tilegx CPU

"TILE-Gx" according to
http://www.tilera.com/products/?ezchip=585&spage=614 - please fix wherever used in textual form.

Quoted 29 lines
target-tilegx/cpu-qom.h new file +29 -0
+ *
+ * Copyright (c) 2015 Chen Gang
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see
+ * <http://www.gnu.org/licenses/lgpl-2.1.html>
+ */
+#ifndef QEMU_TILEGX_CPU_QOM_H
+#define QEMU_TILEGX_CPU_QOM_H
+
+#include "qom/cpu.h"
+
+#define TYPE_TILEGX_CPU "tilegx-cpu"
+
+#define TILEGX_CPU_CLASS(klass) \
+    OBJECT_CLASS_CHECK(TilegxCPUClass, (klass), TYPE_TILEGX_CPU)
+#define TILEGX_CPU(obj) \
+    OBJECT_CHECK(TilegxCPU, (obj), TYPE_TILEGX_CPU)
+#define TILEGX_CPU_GET_CLASS(obj) \

target-tilegx/cpu-qom.h line 41

Quoted 10 lines
target-tilegx/cpu-qom.h new file +10 -0
+    OBJECT_GET_CLASS(TilegxCPUClass, (obj), TYPE_TILEGX_CPU)
+
+/**
+ * TilegxCPUClass:
+ * @parent_realize: The parent class' realize handler.
+ * @parent_reset: The parent class' reset handler.
+ *
+ * A Tilegx CPU model.
+ */
+typedef struct TilegxCPUClass {

For the benefit of readers, please call this TileGXCPUClass ...

Quoted 5 lines
target-tilegx/cpu-qom.h new file +5 -0
+    /*< private >*/
+    CPUClass parent_class;
+    /*< public >*/
+
+    DeviceRealize parent_realize;

target-tilegx/cpu-qom.h line 56

Quoted 10 lines
target-tilegx/cpu-qom.h new file +10 -0
+    void (*parent_reset)(CPUState *cpu);
+} TilegxCPUClass;
+
+/**
+ * TilegxCPU:
+ * @env: #CPUTLState
+ *
+ * A Tilegx CPU.
+ */
+typedef struct TilegxCPU {

... and TileGXCPU. (or TileGx...)

target-tilegx/cpu-qom.h line 59

Quoted 3 lines
target-tilegx/cpu-qom.h new file +3 -0
+    /*< private >*/
+    CPUState parent_obj;
+    uint64_t base_vectors;

This should not be in here, the private section serves to hide the parent field from documentation.

base_vectors should also probably be after env, for performance reasons. rth?

target-tilegx/cpu-qom.h line 62

Quoted 3 lines
target-tilegx/cpu-qom.h new file +3 -0
+    /*< public >*/
+
+    CPUTLState env;

Can this be more telling than TL please?

Quoted 58 lines
target-tilegx/cpu-qom.h new file +10 -0
+} TilegxCPU;
+
+static inline TilegxCPU *tilegx_env_get_cpu(CPUTLState *env)
+{
+    return container_of(env, TilegxCPU, env);
+}
+
+#define ENV_GET_CPU(e) CPU(tilegx_env_get_cpu(e))
+
+#endif
target-tilegx/cpu.c new file +42 -0
new file mode 100644
index 0000000..3dd66b5
@@ -0,0 +1,162 @@
+/*
+ * QEMU Tilegx CPU
+ *
+ *  Copyright (c) 2015 Chen Gang
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see
+ * <http://www.gnu.org/licenses/lgpl-2.1.html>
+ */
+
+#include "cpu.h"
+#include "qemu-common.h"
+#include "hw/qdev-properties.h"
+#include "migration/vmstate.h"
+
+TilegxCPU *cpu_tilegx_init(const char *cpu_model)
+{
+    TilegxCPU *cpu;
+
+    cpu = TILEGX_CPU(object_new(TYPE_TILEGX_CPU));
+
+    object_property_set_bool(OBJECT(cpu), true, "realized", NULL);
+
+    return cpu;
+}
+
+static void tilegx_cpu_set_pc(CPUState *cs, vaddr value)
+{
+    TilegxCPU *cpu = TILEGX_CPU(cs);
+
+    cpu->env.pc = value;
+}

target-tilegx/cpu.c line 52

Quoted 10 lines
target-tilegx/cpu.c new file +10 -0
+
+static bool tilegx_cpu_has_work(CPUState *cs)
+{
+    return true;
+}
+
+static void tilegx_cpu_reset(CPUState *s)
+{
+    TilegxCPU *cpu = TILEGX_CPU(s);
+    TilegxCPUClass *mcc = TILEGX_CPU_GET_CLASS(cpu);

Why mcc? Probably copied from a CPU with M.

target-tilegx/cpu.c line 64

Quoted 12 lines
target-tilegx/cpu.c new file +12 -0
+    CPUTLState *env = &cpu->env;
+
+    mcc->parent_reset(s);
+
+    memset(env, 0, sizeof(CPUTLState));
+    tlb_flush(s, 1);
+}
+
+static void tilegx_cpu_realizefn(DeviceState *dev, Error **errp)
+{
+    CPUState *cs = CPU(dev);
+    TilegxCPUClass *mcc = TILEGX_CPU_GET_CLASS(dev);

Ditto.

Quoted 23 lines
target-tilegx/cpu.c new file +23 -0
+
+    cpu_reset(cs);
+    qemu_init_vcpu(cs);
+
+    mcc->parent_realize(dev, errp);
+}
+
+static void tilegx_tcg_init(void)
+{
+}
+
+static void tilegx_cpu_initfn(Object *obj)
+{
+    CPUState *cs = CPU(obj);
+    TilegxCPU *cpu = TILEGX_CPU(obj);
+    CPUTLState *env = &cpu->env;
+    static bool tcg_initialized;
+
+    cs->env_ptr = env;
+    cpu_exec_init(env);
+
+    if (tcg_enabled() && !tcg_initialized) {
+        tcg_initialized = true;

target-tilegx/cpu.c line 97

Quoted 10 lines
target-tilegx/cpu.c new file +10 -0
+        tilegx_tcg_init();
+    }
+}
+
+static const VMStateDescription vmstate_tilegx_cpu = {
+    .name = "cpu",
+    .unmigratable = 1,
+};
+
+static Property tilegx_properties[] = {

"tilegx_cpu_properties" for consistency?

target-tilegx/cpu.c line 98

Quoted 1 line
target-tilegx/cpu.c new file +1 -0
+    DEFINE_PROP_UINT64("tilegx.base-vectors", TilegxCPU, base_vectors, 0),

Why "tilegx." prefix? That would likely interfere with our QMP tools.

What is this actually used for? To someone not knowing the architecture it may sound a bit strange to have a variable/property ...vectors that is actually just one uint64 value. Someone recently added an API to add a description for a property, and this seems to be calling for one. Moving it to a patch that actually uses it would be another idea.

target-tilegx/cpu.c line 108

Quoted 10 lines
target-tilegx/cpu.c new file +10 -0
+    DEFINE_PROP_END_OF_LIST(),
+};
+
+static void tilegx_cpu_do_interrupt(CPUState *cs)
+{
+    cs->exception_index = -1;
+}
+
+static int tilegx_cpu_handle_mmu_fault(CPUState *cs, vaddr address, int rw,
+                            int mmu_idx)

Indentation.

Quoted 23 lines
target-tilegx/cpu.c new file +23 -0
+{
+    cpu_dump_state(cs, stderr, fprintf, 0);
+    return 1;
+}
+
+static bool tilegx_cpu_exec_interrupt(CPUState *cs, int interrupt_request)
+{
+    if (interrupt_request & CPU_INTERRUPT_HARD) {
+        tilegx_cpu_do_interrupt(cs);
+        return true;
+    }
+    return false;
+}
+
+static void tilegx_cpu_class_init(ObjectClass *oc, void *data)
+{
+    DeviceClass *dc = DEVICE_CLASS(oc);
+    CPUClass *cc = CPU_CLASS(oc);
+    TilegxCPUClass *mcc = TILEGX_CPU_CLASS(oc);
+
+    mcc->parent_realize = dc->realize;
+    dc->realize = tilegx_cpu_realizefn;
+

target-tilegx/cpu.c line 141

Quoted 10 lines
target-tilegx/cpu.c new file +10 -0
+    mcc->parent_reset = cc->reset;
+    cc->reset = tilegx_cpu_reset;
+
+    cc->has_work = tilegx_cpu_has_work;
+    cc->do_interrupt = tilegx_cpu_do_interrupt;
+    cc->cpu_exec_interrupt = tilegx_cpu_exec_interrupt;
+    cc->dump_state = NULL;
+    cc->set_pc = tilegx_cpu_set_pc;
+    cc->gdb_read_register = NULL;
+    cc->gdb_write_register = NULL;

Is this really safe to do? If so, all fields are zero-initialized at this point already, so no need to assign NULL or 0.

target-tilegx/cpu.c line 152

Quoted 11 lines
target-tilegx/cpu.c new file +11 -0
+    cc->handle_mmu_fault = tilegx_cpu_handle_mmu_fault;
+    dc->vmsd = &vmstate_tilegx_cpu;
+    dc->props = tilegx_properties;
+    cc->gdb_num_core_regs = 0;
+}
+
+static const TypeInfo tilegx_cpu_type_info = {
+    .name = TYPE_TILEGX_CPU,
+    .parent = TYPE_CPU,
+    .instance_size = sizeof(TilegxCPU),
+    .instance_init = tilegx_cpu_initfn,

What about CPU models? Do the Gx72, Gx36, etc. differ only in core count or also in instruction set features?

Quoted 33 lines
target-tilegx/cpu.c new file +10 -0
+    .class_size = sizeof(TilegxCPUClass),
+    .class_init = tilegx_cpu_class_init,
+};
+
+static void tilegx_cpu_register_types(void)
+{
+    type_register_static(&tilegx_cpu_type_info);
+}
+
+type_init(tilegx_cpu_register_types)
target-tilegx/cpu.h new file +17 -0
new file mode 100644
index 0000000..09a2b26
@@ -0,0 +1,85 @@
+/*
+ *  Tilegx virtual CPU header
+ *
+ *  Copyright (c) 2015 Chen Gang
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see <http://www.gnu.org/licenses/>.

target-tilegx/cpu.h line 27

Quoted 10 lines
target-tilegx/cpu.h new file +10 -0
+ */
+#ifndef CPU_TILEGX_H
+#define CPU_TILEGX_H
+
+#include "config.h"
+#include "qemu-common.h"
+
+#define TARGET_LONG_BITS 64
+
+#define CPUArchState struct CPUTLState

Drop the struct here? You have a typedef below.

Quoted 5 lines
target-tilegx/cpu.h new file +5 -0
+
+#include "exec/cpu-defs.h"
+#include "fpu/softfloat.h"
+
+/* Tilegx register alias */

target-tilegx/cpu.h line 42

Quoted 10 lines
target-tilegx/cpu.h new file +10 -0
+#define TILEGX_R_RE  0   /*  0 register, for function/syscall return value */
+#define TILEGX_R_NR  10  /* 10 register, for syscall number */
+#define TILEGX_R_BP  52  /* 52 register, optional frame pointer */
+#define TILEGX_R_TP  53  /* TP register, thread local storage data */
+#define TILEGX_R_SP  54  /* SP register, stack pointer */
+#define TILEGX_R_LR  55  /* LR register, may save pc, but it is not pc */
+
+typedef struct CPUTLState {
+    uint64_t regs[56];
+    uint64_t pc;

You have marked the CPU unmigratable. Might it make sense to still prepare the corresponding VMState fields so that it does not get forgotten?

target-tilegx/helper.h

Quoted 46 lines
target-tilegx/cpu.h new file +43 -0
+    CPU_COMMON
+} CPUTLState;
+
+#include "cpu-qom.h"
+
+/* Tilegx memory attributes */
+#define TARGET_PAGE_BITS 16  /* Tilegx uses 64KB page size */
+#define MMAP_SHIFT TARGET_PAGE_BITS
+#define TARGET_PHYS_ADDR_SPACE_BITS 42 /* Tilegx is 42 bit physical address */
+#define TARGET_VIRT_ADDR_SPACE_BITS 64 /* Tilegx has 64 bit virtual address */
+#define MMU_USER_IDX    0  /* independent from both qemu and architecture */
+
+#include "exec/cpu-all.h"
+
+int cpu_tilegx_exec(CPUTLState *s);
+int cpu_tilegx_signal_handler(int host_signum, void *pinfo, void *puc);
+
+#define cpu_exec cpu_tilegx_exec
+#define cpu_gen_code cpu_tilegx_gen_code
+#define cpu_signal_handler cpu_tilegx_signal_handler
+
+TilegxCPU *cpu_tilegx_init(const char *cpu_model);
+
+static inline CPUTLState *cpu_init(const char *cpu_model)
+{
+    TilegxCPU *cpu = cpu_tilegx_init(cpu_model);
+    if (cpu == NULL) {
+        return NULL;
+    }
+    return &cpu->env;
+}
+
+static inline void cpu_get_tb_cpu_state(CPUTLState *env, target_ulong *pc,
+                                        target_ulong *cs_base, int *flags)
+{
+    *pc = env->pc;
+    *cs_base = 0;
+    *flags = 0;
+}
+
+#include "exec/exec-all.h"
+
+#endif
target-tilegx/helper.h new file +0 -0
new file mode 100644
index 0000000..e69de29

Is this empty file actually included from somewhere in this patch?

Quoted 32 lines
target-tilegx/translate.c new file +26 -0
new file mode 100644
index 0000000..5131fa7
@@ -0,0 +1,54 @@
+/*
+ * QEMU Tilegx CPU
+ *
+ *  Copyright (c) 2015 Chen Gang
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see
+ * <http://www.gnu.org/licenses/lgpl-2.1.html>
+ */
+
+#include "cpu.h"
+#include "disas/disas.h"
+#include "tcg-op.h"
+#include "exec/helper-proto.h"
+#include "exec/cpu_ldst.h"
+#include "exec/helper-gen.h"

target-tilegx/translate.c line 36

Quoted 10 lines
target-tilegx/translate.c new file +10 -0
+
+static inline void gen_intermediate_code_internal(TilegxCPU *cpu,
+                                                  TranslationBlock *tb,
+                                                  bool search_pc)
+{
+    /*
+     * FIXME: after load elf64 tilegx binary successfully, it will quit, at
+     * present, and will implement the related features next.
+     */
+    fprintf(stderr, "\nLoad elf64 tilegx successfully\n");

"Loaded"

target-tilegx/translate.c line 37

Quoted 1 line
target-tilegx/translate.c new file +1 -0
+    fprintf(stderr, "reach code start position: [" TARGET_FMT_lx "] %s\n\n",

"reached"

target-tilegx/translate.c line 38

Quoted 1 line
target-tilegx/translate.c new file +1 -0
+            tb->pc, lookup_symbol(tb->pc));

This output is hopefully only temporary. But as a general reminder, there is error_report() for error messages, and for debug messages please define your own macros or use the qemu_log infrastructure instead of fprintf(stderr, ...).

Quoted 6 lines
target-tilegx/translate.c new file +6 -0
+    exit(0);
+}
+
+void gen_intermediate_code(CPUTLState *env, struct TranslationBlock *tb)
+{
+    gen_intermediate_code_internal(tilegx_env_get_cpu(env), tb, false);

target-tilegx/translate.c line 54

Quoted 10 lines
target-tilegx/translate.c new file +10 -0
+}
+
+void gen_intermediate_code_pc(CPUTLState *env, struct TranslationBlock *tb)
+{
+    gen_intermediate_code_internal(tilegx_env_get_cpu(env), tb, true);
+}
+
+void restore_state_to_opc(CPUTLState *env, TranslationBlock *tb, int pc_pos)
+{
+}

Regards,
Andreas

Signature · 3 lines
SUSE Linux GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer, Jane Smithard, Jennifer Guild, Dilip Upmanyu,
Graham Norton; HRB 21284 (AG Nürnberg)
#3 lore raw

Re: [Qemu-devel] [PATCH 1/6 v4] target-tilegx: Firstly add to qemu with minimized features

Quoted 3 lines · Andreas Färber

"TILE-Gx" according to
http://www.tilera.com/products/?ezchip=585&spage=614 - please fix wherever used in textual form.

Yes, really only "tilegx" or "TILE-Gx" should be used. In type names I agree that TileGX is probably clearer than any alternatives.

Quoted 2 lines

What about CPU models? Do the Gx72, Gx36, etc. differ only in core count or also in instruction set features?

From a userspace perspective, it's mostly the core count. They also have a different set of on-chip hardware accelerators, etc., but that's out of scope for this project.

Signature · 2 lines
Chris Metcalf, EZChip Semiconductor
http://www.ezchip.com
#4 lore raw

Re: [Qemu-devel] [PATCH 1/6 v4] target-tilegx: Firstly add to qemu with minimized features

Quoted 7 lines · Chris Metcalf
Quoted 3 lines · Andreas Färber

"TILE-Gx" according to
http://www.tilera.com/products/?ezchip=585&spage=614 - please fix wherever used in textual form.

Yes, really only "tilegx" or "TILE-Gx" should be used. In type names I agree that TileGX is probably clearer than any alternatives.

OK, thanks. I shall use TILE-Gx for it.

And next, for the type names, I shall use TileGX.

Quoted 5 lines
Quoted 2 lines

What about CPU models? Do the Gx72, Gx36, etc. differ only in core count or also in instruction set features?

From a userspace perspective, it's mostly the core count. They also have a different set of on-chip hardware accelerators, etc., but that's out of scope for this project.

OK, thanks. I guess your meaning is "For qemu, we needn't consider about Gx72, Gx36 ..., we can treat all of them as TILE-Gx".

By the way, does Gx8000 mean 8000 core count?

Thanks.

Signature · 3 lines
Chen Gang

Open, share, and attitude like air, water, and life which God blessed
#5 lore raw

Re: [Qemu-devel] [PATCH 1/6 v4] target-tilegx: Firstly add to qemu with minimized features

Quoted 1 line · Chen Gang S

By the way, does Gx8000 mean 8000 core count?

It's marketing, so it doesn't have to mean anything. The TILE-Gx 8036 is the 36-core part, the 8072 is the 72-core part, 8009 is the 9-core part, etc. There is no "8000" chip, it's just a way of describing the TILE-Gx chip series.

Signature · 2 lines
Chris Metcalf, EZChip Semiconductor
http://www.ezchip.com
#6 lore raw

Re: [Qemu-devel] [PATCH 1/6 v4] target-tilegx: Firstly add to qemu with minimized features

Quoted 5 lines · Chris Metcalf
Quoted 1 line · Chen Gang S

By the way, does Gx8000 mean 8000 core count?

It's marketing, so it doesn't have to mean anything. The TILE-Gx 8036 is the 36-core part, the 8072 is the 72-core part, 8009 is the 9-core part, etc. There is no "8000" chip, it's just a way of describing the TILE-Gx chip series.

OK, thanks.

Signature · 3 lines
Chen Gang

Open, share, and attitude like air, water, and life which God blessed
#7 lore raw

Re: [Qemu-devel] [PATCH 1/6 v4] target-tilegx: Firstly add to qemu with minimized features

In reply to Andreas Färber

Firstly, thank you very much for your patient work!

Quoted 5 lines · Andreas Färber

Hi,

"target-tilegx: Initial stub" or "...support"? No need to mention QEMU (spelling!) in a QEMU commit.

OK, thanks.

Quoted 8 lines

Am 22.02.2015 um 14:33 schrieb Chen Gang S:

Quoted 1 line

It almost likes a template for adding an architecture target.

That's a comment that's not really descriptive of what the patch does. Instead, maybe mention that this is the configure and build system support etc. and what name to use for enabling it from configure, that it's linux-user only for now, ...?

OK, thanks.

Quoted 37 lines
Quoted 37 lines
---
 configure                             |   7 ++
 default-configs/tilegx-linux-user.mak |   1 +
 target-tilegx/Makefile.objs           |   1 +
 target-tilegx/cpu-qom.h               |  72 +++++++++++++++
 target-tilegx/cpu.c                   | 162 ++++++++++++++++++++++++++++++++++
 target-tilegx/cpu.h                   |  85 ++++++++++++++++++
 target-tilegx/helper.h                |   0
 target-tilegx/translate.c             |  54 ++++++++++++
 8 files changed, 382 insertions(+)
 create mode 100644 default-configs/tilegx-linux-user.mak
 create mode 100644 target-tilegx/Makefile.objs
 create mode 100644 target-tilegx/cpu-qom.h
 create mode 100644 target-tilegx/cpu.c
 create mode 100644 target-tilegx/cpu.h
 create mode 100644 target-tilegx/helper.h
 create mode 100644 target-tilegx/translate.c
configure +3 -0
index 7ba4bcb..23aa8f6 100755
@@ -5191,6 +5191,9 @@ case "$target_name" in
   s390x)
     gdb_xml_files="s390x-core64.xml s390-acr.xml s390-fpr.xml"
   ;;
+  tilegx)
+    TARGET_ARCH=tilegx
+  ;;
   unicore32)
   ;;
   xtensa|xtensaeb)
@@ -5363,6 +5366,10 @@ for i in $ARCH $TARGET_BASE_ARCH ; do
     echo "CONFIG_SPARC_DIS=y"  >> $config_target_mak
     echo "CONFIG_SPARC_DIS=y"  >> config-all-disas.mak
Quoted 9 lines

configure for i in $ARCH $TARGET_BASE_ARCH ; do · line 5372

Quoted 5 lines
configure +4 -0
   ;;
+  tilegx*)
+    echo "CONFIG_TILEGX_DIS=y"  >> $config_target_mak
+    echo "CONFIG_TILEGX_DIS=y"  >> config-all-disas.mak
+  ;;

Hadn't you been asked to drop these lines, as you are not yet adding any disassembler code that uses it?

NO. And next, I shall drop them.

Quoted 4 lines
Quoted 3 lines
configure +0 -0
   xtensa*)
     echo "CONFIG_XTENSA_DIS=y"  >> $config_target_mak
     echo "CONFIG_XTENSA_DIS=y"  >> config-all-disas.mak

[...]

Quoted 13 lines

target-tilegx/cpu-qom.h line 2

Quoted 8 lines
target-tilegx/cpu-qom.h new file +2 -0
new file mode 100644
index 0000000..e15a8b8
@@ -0,0 +1,72 @@
+/*
+ * QEMU Tilegx CPU

"TILE-Gx" according to
http://www.tilera.com/products/?ezchip=585&spage=614 - please fix wherever used in textual form.

OK, thanks.

Quoted 33 lines
Quoted 33 lines
target-tilegx/cpu-qom.h new file +33 -0
+ *
+ * Copyright (c) 2015 Chen Gang
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see
+ * <http://www.gnu.org/licenses/lgpl-2.1.html>
+ */
+#ifndef QEMU_TILEGX_CPU_QOM_H
+#define QEMU_TILEGX_CPU_QOM_H
+
+#include "qom/cpu.h"
+
+#define TYPE_TILEGX_CPU "tilegx-cpu"
+
+#define TILEGX_CPU_CLASS(klass) \
+    OBJECT_CLASS_CHECK(TilegxCPUClass, (klass), TYPE_TILEGX_CPU)
+#define TILEGX_CPU(obj) \
+    OBJECT_CHECK(TilegxCPU, (obj), TYPE_TILEGX_CPU)
+#define TILEGX_CPU_GET_CLASS(obj) \
+    OBJECT_GET_CLASS(TilegxCPUClass, (obj), TYPE_TILEGX_CPU)
+
+/**
+ * TilegxCPUClass:
Quoted 9 lines

target-tilegx/cpu-qom.h line 41

Quoted 6 lines
target-tilegx/cpu-qom.h new file +6 -0
+ * @parent_realize: The parent class' realize handler.
+ * @parent_reset: The parent class' reset handler.
+ *
+ * A Tilegx CPU model.
+ */
+typedef struct TilegxCPUClass {

For the benefit of readers, please call this TileGXCPUClass ...

OK, thanks.

Quoted 9 lines
Quoted 9 lines
target-tilegx/cpu-qom.h new file +9 -0
+    /*< private >*/
+    CPUClass parent_class;
+    /*< public >*/
+
+    DeviceRealize parent_realize;
+    void (*parent_reset)(CPUState *cpu);
+} TilegxCPUClass;
+
+/**
Quoted 9 lines

target-tilegx/cpu-qom.h line 56

Quoted 6 lines
target-tilegx/cpu-qom.h new file +6 -0
+ * TilegxCPU:
+ * @env: #CPUTLState
+ *
+ * A Tilegx CPU.
+ */
+typedef struct TilegxCPU {

... and TileGXCPU. (or TileGx...)

OK, thanks. I shall call it TileGXCPU.

Quoted 10 lines

target-tilegx/cpu-qom.h line 59

Quoted 3 lines
target-tilegx/cpu-qom.h new file +3 -0
+    /*< private >*/
+    CPUState parent_obj;
+    uint64_t base_vectors;

This should not be in here, the private section serves to hide the parent field from documentation.

base_vectors should also probably be after env, for performance reasons. rth?

OK, thanks. At present, we do not use base_vectors, so I guess, we can just remove it, and the related implementation will be:

typedef struct TileGXCPU {
    /*< private >*/
    CPUState parent_obj;
    /*< public >*/

    CPUS390XState env;
} TileGXCPU;
Quoted 6 lines

target-tilegx/cpu-qom.h line 62

Quoted 3 lines
target-tilegx/cpu-qom.h new file +3 -0
+    /*< public >*/
+
+    CPUTLState env;

Can this be more telling than TL please?

How about CPUTLGState?

Quoted 62 lines
Quoted 62 lines
target-tilegx/cpu-qom.h new file +10 -0
+} TilegxCPU;
+
+static inline TilegxCPU *tilegx_env_get_cpu(CPUTLState *env)
+{
+    return container_of(env, TilegxCPU, env);
+}
+
+#define ENV_GET_CPU(e) CPU(tilegx_env_get_cpu(e))
+
+#endif
target-tilegx/cpu.c new file +46 -0
new file mode 100644
index 0000000..3dd66b5
@@ -0,0 +1,162 @@
+/*
+ * QEMU Tilegx CPU
+ *
+ *  Copyright (c) 2015 Chen Gang
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see
+ * <http://www.gnu.org/licenses/lgpl-2.1.html>
+ */
+
+#include "cpu.h"
+#include "qemu-common.h"
+#include "hw/qdev-properties.h"
+#include "migration/vmstate.h"
+
+TilegxCPU *cpu_tilegx_init(const char *cpu_model)
+{
+    TilegxCPU *cpu;
+
+    cpu = TILEGX_CPU(object_new(TYPE_TILEGX_CPU));
+
+    object_property_set_bool(OBJECT(cpu), true, "realized", NULL);
+
+    return cpu;
+}
+
+static void tilegx_cpu_set_pc(CPUState *cs, vaddr value)
+{
+    TilegxCPU *cpu = TILEGX_CPU(cs);
+
+    cpu->env.pc = value;
+}
+
+static bool tilegx_cpu_has_work(CPUState *cs)
+{
+    return true;
Quoted 9 lines

target-tilegx/cpu.c line 52

Quoted 6 lines
target-tilegx/cpu.c new file +6 -0
+}
+
+static void tilegx_cpu_reset(CPUState *s)
+{
+    TilegxCPU *cpu = TILEGX_CPU(s);
+    TilegxCPUClass *mcc = TILEGX_CPU_GET_CLASS(cpu);

Why mcc? Probably copied from a CPU with M.

Yes, we need another name for it. How about tcc?

Quoted 6 lines
Quoted 6 lines
target-tilegx/cpu.c new file +6 -0
+    CPUTLState *env = &cpu->env;
+
+    mcc->parent_reset(s);
+
+    memset(env, 0, sizeof(CPUTLState));
+    tlb_flush(s, 1);
Quoted 9 lines

target-tilegx/cpu.c line 64

Quoted 6 lines
target-tilegx/cpu.c new file +6 -0
+}
+
+static void tilegx_cpu_realizefn(DeviceState *dev, Error **errp)
+{
+    CPUState *cs = CPU(dev);
+    TilegxCPUClass *mcc = TILEGX_CPU_GET_CLASS(dev);

Ditto.

OK.

Quoted 27 lines
Quoted 27 lines
target-tilegx/cpu.c new file +27 -0
+
+    cpu_reset(cs);
+    qemu_init_vcpu(cs);
+
+    mcc->parent_realize(dev, errp);
+}
+
+static void tilegx_tcg_init(void)
+{
+}
+
+static void tilegx_cpu_initfn(Object *obj)
+{
+    CPUState *cs = CPU(obj);
+    TilegxCPU *cpu = TILEGX_CPU(obj);
+    CPUTLState *env = &cpu->env;
+    static bool tcg_initialized;
+
+    cs->env_ptr = env;
+    cpu_exec_init(env);
+
+    if (tcg_enabled() && !tcg_initialized) {
+        tcg_initialized = true;
+        tilegx_tcg_init();
+    }
+}
+
Quoted 9 lines

target-tilegx/cpu.c line 97

Quoted 6 lines
target-tilegx/cpu.c new file +6 -0
+static const VMStateDescription vmstate_tilegx_cpu = {
+    .name = "cpu",
+    .unmigratable = 1,
+};
+
+static Property tilegx_properties[] = {

"tilegx_cpu_properties" for consistency?

Originally, I copied from Microblaze, but at present, I guess, we can drop it (then also drop base_vectors).

Quoted 4 lines

target-tilegx/cpu.c line 98

Quoted 1 line
target-tilegx/cpu.c new file +1 -0
+    DEFINE_PROP_UINT64("tilegx.base-vectors", TilegxCPU, base_vectors, 0),

Why "tilegx." prefix? That would likely interfere with our QMP tools.

May it be "tilera.base-vectors"?

But after all, I guess, we can just drop it, at present.

Quoted 6 lines

What is this actually used for? To someone not knowing the architecture it may sound a bit strange to have a variable/property ...vectors that is actually just one uint64 value. Someone recently added an API to add a description for a property, and this seems to be calling for one. Moving it to a patch that actually uses it would be another idea.

Quoted 13 lines

target-tilegx/cpu.c line 108

Quoted 10 lines
target-tilegx/cpu.c new file +10 -0
+    DEFINE_PROP_END_OF_LIST(),
+};
+
+static void tilegx_cpu_do_interrupt(CPUState *cs)
+{
+    cs->exception_index = -1;
+}
+
+static int tilegx_cpu_handle_mmu_fault(CPUState *cs, vaddr address, int rw,
+                            int mmu_idx)

Indentation.

OK, thanks.

Quoted 37 lines
Quoted 23 lines
target-tilegx/cpu.c new file +23 -0
+{
+    cpu_dump_state(cs, stderr, fprintf, 0);
+    return 1;
+}
+
+static bool tilegx_cpu_exec_interrupt(CPUState *cs, int interrupt_request)
+{
+    if (interrupt_request & CPU_INTERRUPT_HARD) {
+        tilegx_cpu_do_interrupt(cs);
+        return true;
+    }
+    return false;
+}
+
+static void tilegx_cpu_class_init(ObjectClass *oc, void *data)
+{
+    DeviceClass *dc = DEVICE_CLASS(oc);
+    CPUClass *cc = CPU_CLASS(oc);
+    TilegxCPUClass *mcc = TILEGX_CPU_CLASS(oc);
+
+    mcc->parent_realize = dc->realize;
+    dc->realize = tilegx_cpu_realizefn;
+

target-tilegx/cpu.c line 141

Quoted 10 lines
target-tilegx/cpu.c new file +10 -0
+    mcc->parent_reset = cc->reset;
+    cc->reset = tilegx_cpu_reset;
+
+    cc->has_work = tilegx_cpu_has_work;
+    cc->do_interrupt = tilegx_cpu_do_interrupt;
+    cc->cpu_exec_interrupt = tilegx_cpu_exec_interrupt;
+    cc->dump_state = NULL;
+    cc->set_pc = tilegx_cpu_set_pc;
+    cc->gdb_read_register = NULL;
+    cc->gdb_write_register = NULL;

Is this really safe to do? If so, all fields are zero-initialized at this point already, so no need to assign NULL or 0.

Quoted 15 lines

target-tilegx/cpu.c line 152

Quoted 11 lines
target-tilegx/cpu.c new file +11 -0
+    cc->handle_mmu_fault = tilegx_cpu_handle_mmu_fault;
+    dc->vmsd = &vmstate_tilegx_cpu;
+    dc->props = tilegx_properties;
+    cc->gdb_num_core_regs = 0;
+}
+
+static const TypeInfo tilegx_cpu_type_info = {
+    .name = TYPE_TILEGX_CPU,
+    .parent = TYPE_CPU,
+    .instance_size = sizeof(TilegxCPU),
+    .instance_init = tilegx_cpu_initfn,

What about CPU models? Do the Gx72, Gx36, etc. differ only in core count or also in instruction set features?

OK, thanks. It is already replied in another mail thread.

For me, we need still let TYPE_TILEGX_CPU be "tilegx-cpu" (according to microblaze implementation).

Quoted 37 lines
Quoted 37 lines
target-tilegx/cpu.c new file +10 -0
+    .class_size = sizeof(TilegxCPUClass),
+    .class_init = tilegx_cpu_class_init,
+};
+
+static void tilegx_cpu_register_types(void)
+{
+    type_register_static(&tilegx_cpu_type_info);
+}
+
+type_init(tilegx_cpu_register_types)
target-tilegx/cpu.h new file +21 -0
new file mode 100644
index 0000000..09a2b26
@@ -0,0 +1,85 @@
+/*
+ *  Tilegx virtual CPU header
+ *
+ *  Copyright (c) 2015 Chen Gang
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef CPU_TILEGX_H
+#define CPU_TILEGX_H
+
Quoted 9 lines

target-tilegx/cpu.h line 27

Quoted 6 lines
target-tilegx/cpu.h new file +6 -0
+#include "config.h"
+#include "qemu-common.h"
+
+#define TARGET_LONG_BITS 64
+
+#define CPUArchState struct CPUTLState

Drop the struct here? You have a typedef below.

Hmm... originally, I really noticed about it, but after check the other target implementation (e.g microblaze, alpha, s390), I remained it.

Quoted 10 lines
Quoted 10 lines
target-tilegx/cpu.h new file +10 -0
+
+#include "exec/cpu-defs.h"
+#include "fpu/softfloat.h"
+
+/* Tilegx register alias */
+#define TILEGX_R_RE  0   /*  0 register, for function/syscall return value */
+#define TILEGX_R_NR  10  /* 10 register, for syscall number */
+#define TILEGX_R_BP  52  /* 52 register, optional frame pointer */
+#define TILEGX_R_TP  53  /* TP register, thread local storage data */
+#define TILEGX_R_SP  54  /* SP register, stack pointer */
Quoted 9 lines

target-tilegx/cpu.h line 42

Quoted 5 lines
target-tilegx/cpu.h new file +5 -0
+#define TILEGX_R_LR  55  /* LR register, may save pc, but it is not pc */
+
+typedef struct CPUTLState {
+    uint64_t regs[56];
+    uint64_t pc;

You have marked the CPU unmigratable. Might it make sense to still prepare the corresponding VMState fields so that it does not get forgotten?

Excuse me, I am not quite familiar with it, could you provide more details?

Quoted 40 lines
Quoted 40 lines
target-tilegx/cpu.h new file +40 -0
+    CPU_COMMON
+} CPUTLState;
+
+#include "cpu-qom.h"
+
+/* Tilegx memory attributes */
+#define TARGET_PAGE_BITS 16  /* Tilegx uses 64KB page size */
+#define MMAP_SHIFT TARGET_PAGE_BITS
+#define TARGET_PHYS_ADDR_SPACE_BITS 42 /* Tilegx is 42 bit physical address */
+#define TARGET_VIRT_ADDR_SPACE_BITS 64 /* Tilegx has 64 bit virtual address */
+#define MMU_USER_IDX    0  /* independent from both qemu and architecture */
+
+#include "exec/cpu-all.h"
+
+int cpu_tilegx_exec(CPUTLState *s);
+int cpu_tilegx_signal_handler(int host_signum, void *pinfo, void *puc);
+
+#define cpu_exec cpu_tilegx_exec
+#define cpu_gen_code cpu_tilegx_gen_code
+#define cpu_signal_handler cpu_tilegx_signal_handler
+
+TilegxCPU *cpu_tilegx_init(const char *cpu_model);
+
+static inline CPUTLState *cpu_init(const char *cpu_model)
+{
+    TilegxCPU *cpu = cpu_tilegx_init(cpu_model);
+    if (cpu == NULL) {
+        return NULL;
+    }
+    return &cpu->env;
+}
+
+static inline void cpu_get_tb_cpu_state(CPUTLState *env, target_ulong *pc,
+                                        target_ulong *cs_base, int *flags)
+{
+    *pc = env->pc;
+    *cs_base = 0;
+    *flags = 0;
+}
+
Quoted 9 lines

target-tilegx/helper.h

Quoted 6 lines
target-tilegx/cpu.h new file +3 -0
+#include "exec/exec-all.h"
+
+#endif
target-tilegx/helper.h new file +0 -0
new file mode 100644
index 0000000..e69de29

Is this empty file actually included from somewhere in this patch?

Yes, it is for future use, I guess, at present, we can just drop it:

- "translate.c" includes "exec/helper-gen.h".

- "exec/helper-gen.h" includes "helper.h".

Quoted 36 lines
Quoted 36 lines
target-tilegx/translate.c new file +30 -0
new file mode 100644
index 0000000..5131fa7
@@ -0,0 +1,54 @@
+/*
+ * QEMU Tilegx CPU
+ *
+ *  Copyright (c) 2015 Chen Gang
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see
+ * <http://www.gnu.org/licenses/lgpl-2.1.html>
+ */
+
+#include "cpu.h"
+#include "disas/disas.h"
+#include "tcg-op.h"
+#include "exec/helper-proto.h"
+#include "exec/cpu_ldst.h"
+#include "exec/helper-gen.h"
+
+static inline void gen_intermediate_code_internal(TilegxCPU *cpu,
+                                                  TranslationBlock *tb,
+                                                  bool search_pc)
Quoted 9 lines

target-tilegx/translate.c line 36

Quoted 6 lines
target-tilegx/translate.c new file +6 -0
+{
+    /*
+     * FIXME: after load elf64 tilegx binary successfully, it will quit, at
+     * present, and will implement the related features next.
+     */
+    fprintf(stderr, "\nLoad elf64 tilegx successfully\n");

"Loaded"

OK, thanks.

Quoted 4 lines

target-tilegx/translate.c line 37

Quoted 1 line
target-tilegx/translate.c new file +1 -0
+    fprintf(stderr, "reach code start position: [" TARGET_FMT_lx "] %s\n\n",

"reached"

OK, thanks.

Quoted 7 lines

target-tilegx/translate.c line 38

Quoted 1 line
target-tilegx/translate.c new file +1 -0
+            tb->pc, lookup_symbol(tb->pc));

This output is hopefully only temporary. But as a general reminder, there is error_report() for error messages, and for debug messages please define your own macros or use the qemu_log infrastructure instead of fprintf(stderr, ...).

OK, thanks, I shall use qemu_log for it.

Thanks

Signature · 3 lines
Chen Gang

Open, share, and attitude like air, water, and life which God blessed
#8 lore raw

Re: [Qemu-devel] [PATCH 1/6 v4] target-tilegx: Firstly add to qemu with minimized features

Quoted 4 lines · Chen Gang S
Quoted 4 lines
Quoted 4 lines
target-tilegx/helper.h +0 -0
Quoted 2 lines
Quoted 2 lines

new file mode 100644
index 0000000..e69de29

Quoted 9 lines
Quoted 2 lines
Quoted 2 lines

Is this empty file actually included from somewhere in this patch?

Yes, it is for future use, I guess, at present, we can just drop it:

- "translate.c" includes "exec/helper-gen.h".

- "exec/helper-gen.h" includes "helper.h".

Oh, sorry, I forgot, "tcg/tcg-op.h" -> "helper-proto.h" -> "helper.h"

CC tilegx-linux-user/tcg/tcg.o
In file included from /upstream/qemu/tcg/tcg-op.h:26:0,

                   from /upstream/qemu/tcg/tcg.c:49:
  /upstream/qemu/include/exec/helper-proto.h:29:20: fatal error: helper.h: No such file or directory
   #include "helper.h"
                      ^

So for me, we have to still use an empty "helper.h" file for it.

Thanks.

Signature · 3 lines
Chen Gang

Open, share, and attitude like air, water, and life which God blessed
#9 lore raw

Re: [Qemu-devel] [PATCH 1/6 v4] target-tilegx: Firstly add to qemu with minimized features

In reply to Andreas Färber
Quoted 22 lines · Andreas Färber
Quoted 11 lines

+static void tilegx_cpu_class_init(ObjectClass *oc, void *data)

Quoted 10 lines
target-tilegx/cpu.c new file +10 -0
+{
+    DeviceClass *dc = DEVICE_CLASS(oc);
+    CPUClass *cc = CPU_CLASS(oc);
+    TilegxCPUClass *mcc = TILEGX_CPU_CLASS(oc);
+
+    mcc->parent_realize = dc->realize;
+    dc->realize = tilegx_cpu_realizefn;
+
+    mcc->parent_reset = cc->reset;
+    cc->reset = tilegx_cpu_reset;
Quoted 8 lines

target-tilegx/cpu.c line 141

Quoted 8 lines
target-tilegx/cpu.c new file +8 -0
+
+    cc->has_work = tilegx_cpu_has_work;
+    cc->do_interrupt = tilegx_cpu_do_interrupt;
+    cc->cpu_exec_interrupt = tilegx_cpu_exec_interrupt;
+    cc->dump_state = NULL;
+    cc->set_pc = tilegx_cpu_set_pc;
+    cc->gdb_read_register = NULL;
+    cc->gdb_write_register = NULL;

Is this really safe to do? If so, all fields are zero-initialized at this point already, so no need to assign NULL or 0.

Quoted 6 lines
Quoted 6 lines

target-tilegx/cpu.c line 147

Quoted 6 lines
target-tilegx/cpu.c new file +6 -0
+    cc->handle_mmu_fault = tilegx_cpu_handle_mmu_fault;
+    dc->vmsd = &vmstate_tilegx_cpu;
+    dc->props = tilegx_properties;
+    cc->gdb_num_core_regs = 0;
+}
+

When we set gdb_num_core_regs be 0, gdb_read/write_register should not be called, and they assume the callers always pass valid parameters.

- For alpha, s390 have considered about the invalid parameter.

- arm, i386 and microblaze do not consider about the invalid parameter.

For dump_state, the caller will check whether it is NULL, before use it.

Thanks.

Signature · 3 lines
Chen Gang

Open, share, and attitude like air, water, and life which God blessed
#10 lore raw

PATCH v4 3/6 linux-user: tilegx: Add target features support within qemu

In reply to Chen Gang S

They are for target features within qemu which independent from outside.

---
 linux-user/tilegx/target_cpu.h     | 35 +++++++++++++++++++++++++++
 linux-user/tilegx/target_signal.h  | 28 ++++++++++++++++++++++
 linux-user/tilegx/target_structs.h | 48 ++++++++++++++++++++++++++++++++++++++
 3 files changed, 111 insertions(+)
 create mode 100644 linux-user/tilegx/target_cpu.h
 create mode 100644 linux-user/tilegx/target_signal.h
 create mode 100644 linux-user/tilegx/target_structs.h
linux-user/tilegx/target_cpu.h new file +35 -0
new file mode 100644
index 0000000..93329fc
@@ -0,0 +1,35 @@
+/*
+ * Tilegx specific CPU ABI and functions for linux-user
+ *
+ * Copyright (c) 2015 Chen Gang
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef TARGET_CPU_H
+#define TARGET_CPU_H
+
+static inline void cpu_clone_regs(CPUTLState *env, target_ulong newsp)
+{
+    if (newsp) {
+        env->regs[TILEGX_R_SP] = newsp;
+    }
+    env->regs[TILEGX_R_RE] = 0;
+}
+
+static inline void cpu_set_tls(CPUTLState *env, target_ulong newtls)
+{
+    env->regs[TILEGX_R_TP] = newtls;
+}
+
+#endif
linux-user/tilegx/target_signal.h new file +28 -0
new file mode 100644
index 0000000..8c7b02a
@@ -0,0 +1,28 @@
+#ifndef TARGET_SIGNAL_H
+#define TARGET_SIGNAL_H
+
+#include "cpu.h"
+
+/* this struct defines a stack used during syscall handling */
+
+typedef struct target_sigaltstack {
+    abi_ulong ss_sp;
+    abi_ulong ss_size;
+    abi_long ss_flags;
+} target_stack_t;
+
+/*
+ * sigaltstack controls
+ */
+#define TARGET_SS_ONSTACK     1
+#define TARGET_SS_DISABLE     2
+
+#define TARGET_MINSIGSTKSZ    2048
+#define TARGET_SIGSTKSZ       8192
+
+static inline abi_ulong get_sp_from_cpustate(CPUTLState *state)
+{
+    return state->regs[TILEGX_R_SP];
+}
+
+#endif /* TARGET_SIGNAL_H */
linux-user/tilegx/target_structs.h new file +48 -0

48 lines not on this page · raw message

Signature · 1 line
1.9.3
#11 lore raw

PATCH v4 4/6 linux-user: Support tilegx architecture in syscall

In reply to Chen Gang S

Add tilegx architecture in "syscall_defs.h", all related features (ioctrl, and stat) are based on Linux kernel tilegx 64-bit implementation.

---
 linux-user/syscall_defs.h | 38 ++++++++++++++++++++++++++++++++++----
 1 file changed, 34 insertions(+), 4 deletions(-)
linux-user/syscall_defs.h +34 -4
index 9ed6de8..a0d9d77 100644
@@ -64,8 +64,9 @@
 #endif
 
 #if defined(TARGET_I386) || defined(TARGET_ARM) || defined(TARGET_SH4) \
-    || defined(TARGET_M68K) || defined(TARGET_CRIS) || defined(TARGET_UNICORE32) \
-    || defined(TARGET_S390X) || defined(TARGET_OPENRISC)
+    || defined(TARGET_M68K) || defined(TARGET_CRIS) \
+    || defined(TARGET_UNICORE32) || defined(TARGET_S390X) \
+    || defined(TARGET_OPENRISC) || defined(TARGET_TILEGX)
 
 #define TARGET_IOC_SIZEBITS	14
 #define TARGET_IOC_DIRBITS	2
@@ -365,7 +366,8 @@ int do_sigaction(int sig, const struct target_sigaction *act,
     || defined(TARGET_PPC) || defined(TARGET_MIPS) || defined(TARGET_SH4) \
     || defined(TARGET_M68K) || defined(TARGET_ALPHA) || defined(TARGET_CRIS) \
     || defined(TARGET_MICROBLAZE) || defined(TARGET_UNICORE32) \
-    || defined(TARGET_S390X) || defined(TARGET_OPENRISC)
+    || defined(TARGET_S390X) || defined(TARGET_OPENRISC) \
+    || defined(TARGET_TILEGX)
 
 #if defined(TARGET_SPARC)
 #define TARGET_SA_NOCLDSTOP    8u
@@ -1970,6 +1972,32 @@ struct target_stat64 {
     unsigned int __unused5;
 };
 
+#elif defined(TARGET_TILEGX)
+
+/* Copy from Linux kernel "uapi/asm-generic/stat.h" */
+struct target_stat {
+        abi_ulong st_dev;               /* Device.  */
+        abi_ulong st_ino;               /* File serial number.  */
+        unsigned int st_mode;           /* File mode.  */
+        unsigned int st_nlink;          /* Link count.  */
+        unsigned int st_uid;            /* User ID of the file's owner.  */
+        unsigned int st_gid;            /* Group ID of the file's group. */
+        abi_ulong st_rdev;              /* Device number, if device.  */
+        abi_ulong __pad1;
+        abi_long  st_size;              /* Size of file, in bytes.  */
+        int st_blksize;                 /* Optimal block size for I/O.  */
+        int __pad2;
+        abi_long st_blocks;             /* Number 512-byte blocks allocated. */
+        abi_long target_st_atime;       /* Time of last access.  */
+        abi_ulong target_st_atime_nsec;
+        abi_long target_st_mtime;       /* Time of last modification.  */
+        abi_ulong target_st_mtime_nsec;
+        abi_long target_st_ctime;       /* Time of last status change.  */
+        abi_ulong target_st_ctime_nsec;
+        unsigned int __unused4;
+        unsigned int __unused5;
+};
+
 #else
 #error unsupported CPU
 #endif
@@ -2312,7 +2340,9 @@ struct target_flock {
 struct target_flock64 {
 	short  l_type;
 	short  l_whence;
-#if defined(TARGET_PPC) || defined(TARGET_X86_64) || defined(TARGET_MIPS) || defined(TARGET_SPARC) || defined(TARGET_HPPA) || defined (TARGET_MICROBLAZE)
+#if defined(TARGET_PPC) || defined(TARGET_X86_64) || defined(TARGET_MIPS) || \
+    defined(TARGET_SPARC) || defined(TARGET_HPPA) \
+    || defined(TARGET_MICROBLAZE) || defined(TARGET_TILEGX)
         int __pad;
 #endif
 	unsigned long long l_start;
Signature · 1 line
1.9.3
#12 lore raw

Re: [Qemu-devel] [PATCH 4/6 v4] linux-user: Support tilegx architecture in syscall

Am 22.02.2015 um 14:36 schrieb Chen Gang S:

Quoted 12 lines

Add tilegx architecture in "syscall_defs.h", all related features (ioctrl, and stat) are based on Linux kernel tilegx 64-bit implementation.

---
 linux-user/syscall_defs.h | 38 ++++++++++++++++++++++++++++++++++----
 1 file changed, 34 insertions(+), 4 deletions(-)
linux-user/syscall_defs.h +0 -0
index 9ed6de8..a0d9d77 100644

[...]

linux-user/syscall_defs.h struct target_flock { · line 2345

Quoted 8 lines
linux-user/syscall_defs.h +3 -1
@@ -2312,7 +2340,9 @@ struct target_flock {
 struct target_flock64 {
 	short  l_type;
 	short  l_whence;
-#if defined(TARGET_PPC) || defined(TARGET_X86_64) || defined(TARGET_MIPS) || defined(TARGET_SPARC) || defined(TARGET_HPPA) || defined (TARGET_MICROBLAZE)
+#if defined(TARGET_PPC) || defined(TARGET_X86_64) || defined(TARGET_MIPS) || \
+    defined(TARGET_SPARC) || defined(TARGET_HPPA) \
+    || defined(TARGET_MICROBLAZE) || defined(TARGET_TILEGX)

Nit: You're inconsistent as to where you're placing ||.

Regards,
Andreas

Quoted 4 lines
linux-user/syscall_defs.h +0 -0
         int __pad;
 #endif
 	unsigned long long l_start;
Signature · 3 lines
SUSE Linux GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer, Jane Smithard, Jennifer Guild, Dilip Upmanyu,
Graham Norton; HRB 21284 (AG Nürnberg)
#13 lore raw

Re: [Qemu-devel] [PATCH 4/6 v4] linux-user: Support tilegx architecture in syscall

Quoted 14 lines · Andreas Färber

Am 22.02.2015 um 14:36 schrieb Chen Gang S:

Quoted 12 lines

Add tilegx architecture in "syscall_defs.h", all related features (ioctrl, and stat) are based on Linux kernel tilegx 64-bit implementation.

---
 linux-user/syscall_defs.h | 38 ++++++++++++++++++++++++++++++++++----
 1 file changed, 34 insertions(+), 4 deletions(-)
linux-user/syscall_defs.h +0 -0
index 9ed6de8..a0d9d77 100644

[...]

Quoted 11 lines

linux-user/syscall_defs.h struct target_flock { · line 2345

Quoted 8 lines
linux-user/syscall_defs.h +3 -1
@@ -2312,7 +2340,9 @@ struct target_flock {
 struct target_flock64 {
 	short  l_type;
 	short  l_whence;
-#if defined(TARGET_PPC) || defined(TARGET_X86_64) || defined(TARGET_MIPS) || defined(TARGET_SPARC) || defined(TARGET_HPPA) || defined (TARGET_MICROBLAZE)
+#if defined(TARGET_PPC) || defined(TARGET_X86_64) || defined(TARGET_MIPS) || \
+    defined(TARGET_SPARC) || defined(TARGET_HPPA) \
+    || defined(TARGET_MICROBLAZE) || defined(TARGET_TILEGX)

Nit: You're inconsistent as to where you're placing ||.

OK.

Signature · 1 line
Open, share, and attitude like air, water, and life which God blessed.
#14 lore raw

PATCH v4 5/6 linux-user: Support tilegx architecture in linux-user

In reply to Chen Gang S

Add main working flow feature and loading elf64 tilegx binary feature, based on Linux kernel tilegx 64-bit implementation.

After this patch, qemu can successfully load elf64 tilegx binary for linux-user, and the working flow reaches the first correct instruction position "__start".

---
 include/elf.h        |  2 ++
 linux-user/elfload.c | 23 ++++++++++++++++
 linux-user/main.c    | 74 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 99 insertions(+)
include/elf.h +2 -0
index a516584..e70512c 100644
@@ -133,6 +133,8 @@ typedef int64_t  Elf64_Sxword;
 
 #define EM_AARCH64  183
 
+#define EM_TILEGX   191 /* Tilegx */
+
 /* This is the info that is needed to parse the dynamic section of the file */
 #define DT_NULL		0
 #define DT_NEEDED	1
linux-user/elfload.c +23 -0
index 399c021..12e3ab8 100644
@@ -1189,6 +1189,29 @@ static inline void init_thread(struct target_pt_regs *regs, struct image_info *i
 
 #endif /* TARGET_S390X */
 
+#ifdef TARGET_TILEGX
+
+/* 42 bits real used address, a half for user mode */
+#define ELF_START_MMAP (0x00000020000000000ULL)
+
+#define elf_check_arch(x) ((x) == EM_TILEGX)
+
+#define ELF_CLASS   ELFCLASS64
+#define ELF_DATA    ELFDATA2LSB
+#define ELF_ARCH    EM_TILEGX
+
+static inline void init_thread(struct target_pt_regs *regs,
+                               struct image_info *infop)
+{
+    regs->lr = infop->entry;
+    regs->sp = infop->start_stack;
+
+}
+
+#define ELF_EXEC_PAGESIZE        65536 /* Tilegx page size is 64KB */
+
+#endif /* TARGET_TILEGX */
+
 #ifndef ELF_PLATFORM
 #define ELF_PLATFORM (NULL)
 #endif
linux-user/main.c +74 -0

86 lines not on this page · raw message

Signature · 1 line
1.9.3
#15 lore raw

PATCH v4 6/6 linux-user/syscall.c: Switch all macros which are not defined in tilegx

In reply to Chen Gang S

For tilegx, several syscall macros are not supported, so switch them to avoid building break.

---
 linux-user/syscall.c | 50 +++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 49 insertions(+), 1 deletion(-)
linux-user/syscall.c +49 -1
index 5720195..a3c3792 100644
@@ -213,7 +213,7 @@ static int gettid(void) {
     return -ENOSYS;
 }
 #endif
-#ifdef __NR_getdents
+#if defined(TARGET_NR_getdents) && defined(__NR_getdents)
 _syscall3(int, sys_getdents, uint, fd, struct linux_dirent *, dirp, uint, count);
 #endif
 #if !defined(__NR_getdents) || \
@@ -5580,6 +5580,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         ret = get_errno(write(arg1, p, arg3));
         unlock_user(p, arg2, 0);
         break;
+#ifdef TARGET_NR_open
     case TARGET_NR_open:
         if (!(p = lock_user_string(arg1)))
             goto efault;
@@ -5588,6 +5589,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
                                   arg3));
         unlock_user(p, arg1, 0);
         break;
+#endif
     case TARGET_NR_openat:
         if (!(p = lock_user_string(arg2)))
             goto efault;
@@ -5602,9 +5604,11 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
     case TARGET_NR_brk:
         ret = do_brk(arg1);
         break;
+#ifdef TARGET_NR_fork
     case TARGET_NR_fork:
         ret = get_errno(do_fork(cpu_env, SIGCHLD, 0, 0, 0, 0));
         break;
+#endif
 #ifdef TARGET_NR_waitpid
     case TARGET_NR_waitpid:
         {
@@ -5639,6 +5643,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         unlock_user(p, arg1, 0);
         break;
 #endif
+#ifdef TARGET_NR_link
     case TARGET_NR_link:
         {
             void * p2;
@@ -5652,6 +5657,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
             unlock_user(p, arg1, 0);
         }
         break;
+#endif
 #if defined(TARGET_NR_linkat)
     case TARGET_NR_linkat:
         {
@@ -5669,12 +5675,14 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         }
         break;
 #endif
+#ifdef TARGET_NR_unlink
     case TARGET_NR_unlink:
         if (!(p = lock_user_string(arg1)))
             goto efault;
         ret = get_errno(unlink(p));
         unlock_user(p, arg1, 0);
         break;
+#endif
 #if defined(TARGET_NR_unlinkat)
     case TARGET_NR_unlinkat:
         if (!(p = lock_user_string(arg2)))
@@ -5791,12 +5799,14 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         }
         break;
 #endif
+#ifdef TARGET_NR_mknod
     case TARGET_NR_mknod:
         if (!(p = lock_user_string(arg1)))
             goto efault;
         ret = get_errno(mknod(p, arg2, arg3));
         unlock_user(p, arg1, 0);
         break;
+#endif
 #if defined(TARGET_NR_mknodat)
     case TARGET_NR_mknodat:
         if (!(p = lock_user_string(arg2)))
@@ -5805,12 +5815,14 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         unlock_user(p, arg2, 0);
         break;
 #endif
+#ifdef TARGET_NR_chmod
     case TARGET_NR_chmod:
         if (!(p = lock_user_string(arg1)))
             goto efault;
         ret = get_errno(chmod(p, arg2));
         unlock_user(p, arg1, 0);
         break;
+#endif
 #ifdef TARGET_NR_break
     case TARGET_NR_break:
         goto unimplemented;
@@ -5945,6 +5957,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         }
         break;
 #endif
+#ifdef TARGET_NR_utimes
     case TARGET_NR_utimes:
         {
             struct timeval *tvp, tv[2];
@@ -5963,6 +5976,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
             unlock_user(p, arg1, 0);
         }
         break;
+#endif
 #if defined(TARGET_NR_futimesat)
     case TARGET_NR_futimesat:
         {
@@ -5991,12 +6005,14 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
     case TARGET_NR_gtty:
         goto unimplemented;
 #endif
+#ifdef TARGET_NR_access
     case TARGET_NR_access:
         if (!(p = lock_user_string(arg1)))
             goto efault;
         ret = get_errno(access(path(p), arg2));
         unlock_user(p, arg1, 0);
         break;
+#endif
 #if defined(TARGET_NR_faccessat) && defined(__NR_faccessat)
     case TARGET_NR_faccessat:
         if (!(p = lock_user_string(arg2)))
@@ -6021,6 +6037,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
     case TARGET_NR_kill:
         ret = get_errno(kill(arg1, target_to_host_signal(arg2)));
         break;
+#ifdef TARGET_NR_rename
     case TARGET_NR_rename:
         {
             void *p2;
@@ -6034,6 +6051,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
             unlock_user(p, arg1, 0);
         }
         break;
+#endif
 #if defined(TARGET_NR_renameat)
     case TARGET_NR_renameat:
         {
@@ -6049,12 +6067,14 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         }
         break;
 #endif
+#ifdef TARGET_NR_mkdir
     case TARGET_NR_mkdir:
         if (!(p = lock_user_string(arg1)))
             goto efault;
         ret = get_errno(mkdir(p, arg2));
         unlock_user(p, arg1, 0);
         break;
+#endif
 #if defined(TARGET_NR_mkdirat)
     case TARGET_NR_mkdirat:
         if (!(p = lock_user_string(arg2)))
@@ -6063,18 +6083,22 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         unlock_user(p, arg2, 0);
         break;
 #endif
+#ifdef TARGET_NR_rmdir
     case TARGET_NR_rmdir:
         if (!(p = lock_user_string(arg1)))
             goto efault;
         ret = get_errno(rmdir(p));
         unlock_user(p, arg1, 0);
         break;
+#endif
     case TARGET_NR_dup:
         ret = get_errno(dup(arg1));
         break;
+#ifdef TARGET_NR_pipe
     case TARGET_NR_pipe:
         ret = do_pipe(cpu_env, arg1, 0, 0);
         break;
+#endif
 #ifdef TARGET_NR_pipe2
     case TARGET_NR_pipe2:
         ret = do_pipe(cpu_env, arg1,
@@ -6159,11 +6183,15 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         ret = get_errno(chroot(p));
         unlock_user(p, arg1, 0);
         break;
+#ifdef TARGET_NR_ustat
     case TARGET_NR_ustat:
         goto unimplemented;
+#endif
+#ifdef TARGET_NR_dup2
     case TARGET_NR_dup2:
         ret = get_errno(dup2(arg1, arg2));
         break;
+#endif
 #if defined(CONFIG_DUP3) && defined(TARGET_NR_dup3)
     case TARGET_NR_dup3:
         ret = get_errno(dup3(arg1, arg2, arg3));
@@ -6174,12 +6202,14 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         ret = get_errno(getppid());
         break;
 #endif
+#ifdef TARGET_NR_getpgrp
     case TARGET_NR_getpgrp:
         ret = get_errno(getpgrp());
         break;
     case TARGET_NR_setsid:
         ret = get_errno(setsid());
         break;
+#endif
 #ifdef TARGET_NR_sigaction
     case TARGET_NR_sigaction:
         {
@@ -6752,6 +6782,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         }
         break;
 #endif
+#ifdef TARGET_NR_symlink
     case TARGET_NR_symlink:
         {
             void *p2;
@@ -6765,6 +6796,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
             unlock_user(p, arg1, 0);
         }
         break;
+#endif
 #if defined(TARGET_NR_symlinkat)
     case TARGET_NR_symlinkat:
         {
@@ -6784,6 +6816,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
     case TARGET_NR_oldlstat:
         goto unimplemented;
 #endif
+#ifdef TARGET_NR_readlink
     case TARGET_NR_readlink:
         {
             void *p2;
@@ -6814,6 +6847,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
             unlock_user(p, arg1, 0);
         }
         break;
+#endif
 #if defined(TARGET_NR_readlinkat)
     case TARGET_NR_readlinkat:
         {
@@ -7213,22 +7247,28 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
             }
         }
         break;
+#ifdef TARGET_NR_stat
     case TARGET_NR_stat:
         if (!(p = lock_user_string(arg1)))
             goto efault;
         ret = get_errno(stat(path(p), &st));
         unlock_user(p, arg1, 0);
         goto do_stat;
+#endif
+#ifdef TARGET_NR_lstat
     case TARGET_NR_lstat:
         if (!(p = lock_user_string(arg1)))
             goto efault;
         ret = get_errno(lstat(path(p), &st));
         unlock_user(p, arg1, 0);
         goto do_stat;
+#endif
     case TARGET_NR_fstat:
         {
             ret = get_errno(fstat(arg1, &st));
+#if defined(TARGET_NR_stat) || defined(TARGET_NR_lstat)
         do_stat:
+#endif
             if (!is_error(ret)) {
                 struct target_stat *target_st;
 
@@ -7516,6 +7556,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         }
         break;
 #endif
+#ifdef TARGET_NR_getdents
     case TARGET_NR_getdents:
 #ifdef __NR_getdents
 #if TARGET_ABI_BITS == 32 && HOST_LONG_BITS == 64
@@ -7646,6 +7687,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         }
 #endif
         break;
+#endif /* TARGET_NR_getdents */
 #if defined(TARGET_NR_getdents64) && defined(__NR_getdents64)
     case TARGET_NR_getdents64:
         {
@@ -7785,11 +7827,13 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         ret = get_errno(fdatasync(arg1));
         break;
 #endif
+#ifdef TARGET_NR__sysctl
     case TARGET_NR__sysctl:
         /* We don't implement this, but ENOTDIR is always a safe
            return value. */
         ret = -TARGET_ENOTDIR;
         break;
+#endif
     case TARGET_NR_sched_getaffinity:
         {
             unsigned int mask_size;
@@ -8236,12 +8280,14 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
             ret = host_to_target_stat64(cpu_env, arg3, &st);
         break;
 #endif
+#ifdef TARGET_NR_lchown
     case TARGET_NR_lchown:
         if (!(p = lock_user_string(arg1)))
             goto efault;
         ret = get_errno(lchown(p, low2highuid(arg2), low2highgid(arg3)));
         unlock_user(p, arg1, 0);
         break;
+#endif
 #ifdef TARGET_NR_getuid
     case TARGET_NR_getuid:
         ret = get_errno(high2lowuid(getuid()));
@@ -8364,12 +8410,14 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
         }
         break;
 #endif
+#ifdef TARGET_NR_chown
     case TARGET_NR_chown:
         if (!(p = lock_user_string(arg1)))
             goto efault;
         ret = get_errno(chown(p, low2highuid(arg2), low2highgid(arg3)));
         unlock_user(p, arg1, 0);
         break;
+#endif
     case TARGET_NR_setuid:
         ret = get_errno(setuid(low2highuid(arg1)));
         break;
Signature · 1 line
1.9.3
Reviewed-by on v4 6/6
#16 lore raw

Re: [Qemu-devel] [PATCH 6/6 v4] linux-user/syscall.c: Switch all macros which are not defined in tilegx

Am 22.02.2015 um 14:37 schrieb Chen Gang S:

Quoted 7 lines

For tilegx, several syscall macros are not supported, so switch them to avoid building break.

---
 linux-user/syscall.c | 50 +++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 49 insertions(+), 1 deletion(-)

LGTM,

However I suggest "conditionalize syscalls" or "disable syscalls" over "switch ... macros/them".

Regards,
Andreas

Signature · 3 lines
SUSE Linux GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer, Jane Smithard, Jennifer Guild, Dilip Upmanyu,
Graham Norton; HRB 21284 (AG Nürnberg)
#17 lore raw

Re: [Qemu-devel] [PATCH 6/6 v4] linux-user/syscall.c: Switch all macros which are not defined in tilegx

Quoted 9 lines · Andreas Färber

Am 22.02.2015 um 14:37 schrieb Chen Gang S:

Quoted 7 lines

For tilegx, several syscall macros are not supported, so switch them to avoid building break.

---
 linux-user/syscall.c | 50 +++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 49 insertions(+), 1 deletion(-)
Quoted 7 lines

LGTM,

However I suggest "conditionalize syscalls" or "disable syscalls" over "switch ... macros/them".

OK, thanks. What you said above sounds reasonable to me.

Thanks.

Signature · 1 line
Open, share, and attitude like air, water, and life which God blessed.
Reviewed-by on the v4 series
#18 lore raw
In reply to Chen Gang S
Quoted 8 lines · Chen Gang S

Chen Gang (6):
target-tilegx: Firstly add to qemu with minimized features
linux-user: tilegx: Firstly add architecture related features
linux-user: tilegx: Add target features support within qemu
linux-user: Support tilegx architecture in syscall
linux-user: Support tilegx architecture in linux-user linux-user/syscall.c: Switch all macros which are not defined in
tilegx

r~

#19 lore raw
Quoted 10 lines · Richard Henderson
Quoted 8 lines · Chen Gang S

Chen Gang (6):
target-tilegx: Firstly add to qemu with minimized features
linux-user: tilegx: Firstly add architecture related features
linux-user: tilegx: Add target features support within qemu
linux-user: Support tilegx architecture in syscall
linux-user: Support tilegx architecture in linux-user linux-user/syscall.c: Switch all macros which are not defined in
tilegx

Quoted 2 lines

Thank all of you, next, I shall try to send patch for executing one TB block within 3 days (2015-02-25)

Thanks.

Signature · 3 lines
Chen Gang

Open, share, and attitude like air, water, and life which God blessed
#20 lore raw
In reply to Chen Gang S
Quoted 29 lines · Chen Gang S

Chen Gang (6):
target-tilegx: Firstly add to qemu with minimized features
linux-user: tilegx: Firstly add architecture related features
linux-user: tilegx: Add target features support within qemu
linux-user: Support tilegx architecture in syscall
linux-user: Support tilegx architecture in linux-user linux-user/syscall.c: Switch all macros which are not defined in
tilegx

  configure                             |   7 +
  default-configs/tilegx-linux-user.mak |   1 +
  include/elf.h                         |   2 +
  linux-user/elfload.c                  |  23 +++
  linux-user/main.c                     |  74 +++++++++
  linux-user/syscall.c                  |  50 +++++-
  linux-user/syscall_defs.h             |  38 ++++-
  linux-user/tilegx/syscall.h           |  80 ++++++++++
  linux-user/tilegx/syscall_nr.h        | 278 +++++++++++++++++++++++++++++++++
  linux-user/tilegx/target_cpu.h        |  35 +++++
  linux-user/tilegx/target_signal.h     |  28 ++++
  linux-user/tilegx/target_structs.h    |  48 ++++++
  linux-user/tilegx/termbits.h          | 285 ++++++++++++++++++++++++++++++++++
  target-tilegx/Makefile.objs           |   1 +
  target-tilegx/cpu-qom.h               |  72 +++++++++
  target-tilegx/cpu.c                   | 162 +++++++++++++++++++
  target-tilegx/cpu.h                   |  85 ++++++++++
  target-tilegx/helper.h                |   0
  target-tilegx/translate.c             |  54 +++++++
  19 files changed, 1318 insertions(+), 5 deletions(-)
Quoted 14 lines

create mode 100644 default-configs/tilegx-linux-user.mak create mode 100644 linux-user/tilegx/syscall.h create mode 100644 linux-user/tilegx/syscall_nr.h create mode 100644 linux-user/tilegx/target_cpu.h create mode 100644 linux-user/tilegx/target_signal.h create mode 100644 linux-user/tilegx/target_structs.h create mode 100644 linux-user/tilegx/termbits.h create mode 100644 target-tilegx/Makefile.objs create mode 100644 target-tilegx/cpu-qom.h create mode 100644 target-tilegx/cpu.c
create mode 100644 target-tilegx/cpu.h
create mode 100644 target-tilegx/helper.h create mode 100644 target-tilegx/translate.c

Hi Chen,

while looking at the diffstat, I see no change to the maintainer file. Can you add someone (you?), who will take care of this code you're trying to add?

Btw, cool looking cpu architecture.

Cheers,
Bastian

#21 lore raw
Quoted 45 lines · Bastian Koppelmann
Quoted 29 lines · Chen Gang S

Chen Gang (6):
target-tilegx: Firstly add to qemu with minimized features
linux-user: tilegx: Firstly add architecture related features
linux-user: tilegx: Add target features support within qemu
linux-user: Support tilegx architecture in syscall
linux-user: Support tilegx architecture in linux-user linux-user/syscall.c: Switch all macros which are not defined in
tilegx

  configure                             |   7 +
  default-configs/tilegx-linux-user.mak |   1 +
  include/elf.h                         |   2 +
  linux-user/elfload.c                  |  23 +++
  linux-user/main.c                     |  74 +++++++++
  linux-user/syscall.c                  |  50 +++++-
  linux-user/syscall_defs.h             |  38 ++++-
  linux-user/tilegx/syscall.h           |  80 ++++++++++
  linux-user/tilegx/syscall_nr.h        | 278 +++++++++++++++++++++++++++++++++
  linux-user/tilegx/target_cpu.h        |  35 +++++
  linux-user/tilegx/target_signal.h     |  28 ++++
  linux-user/tilegx/target_structs.h    |  48 ++++++
  linux-user/tilegx/termbits.h          | 285 ++++++++++++++++++++++++++++++++++
  target-tilegx/Makefile.objs           |   1 +
  target-tilegx/cpu-qom.h               |  72 +++++++++
  target-tilegx/cpu.c                   | 162 +++++++++++++++++++
  target-tilegx/cpu.h                   |  85 ++++++++++
  target-tilegx/helper.h                |   0
  target-tilegx/translate.c             |  54 +++++++
  19 files changed, 1318 insertions(+), 5 deletions(-)
Quoted 14 lines

create mode 100644 default-configs/tilegx-linux-user.mak create mode 100644 linux-user/tilegx/syscall.h create mode 100644 linux-user/tilegx/syscall_nr.h create mode 100644 linux-user/tilegx/target_cpu.h create mode 100644 linux-user/tilegx/target_signal.h create mode 100644 linux-user/tilegx/target_structs.h create mode 100644 linux-user/tilegx/termbits.h create mode 100644 target-tilegx/Makefile.objs create mode 100644 target-tilegx/cpu-qom.h create mode 100644 target-tilegx/cpu.c
create mode 100644 target-tilegx/cpu.h
create mode 100644 target-tilegx/helper.h create mode 100644 target-tilegx/translate.c

Quoted 4 lines

Hi Chen,

while looking at the diffstat, I see no change to the maintainer file. Can you add someone (you?), who will take care of this code you're trying to add?

Excuse me, I guess, I am not the suitable maintainer for it.

- At present, in fact, I am not quite familiar with tile, not quite familiar with qemu.

- I am not quite sure I can still have enough time resources on qemu:
In holidays, I can (e.g. Chinese Spring Festival, but in work days, I guess, I can not.

- I suggest to let other tile related members as maintainer (although I am not quite sure whether they have enough time resource or not).

If really nobody have enough time resource as the related maintainer, I should be. Since I start it, I have duty to let it grow up (although I really have no enough time resource on it).

Really welcome any tile related members to be as the related maintainer.

Quoted 2 lines

Btw, cool looking cpu architecture.

For me, tilegx is really cool! :-)

Thanks.

Signature · 3 lines
Chen Gang

Open, share, and attitude like air, water, and life which God blessed
#22 lore raw
In reply to Bastian Koppelmann
Quoted 45 lines · Bastian Koppelmann
Quoted 29 lines · Chen Gang S

Chen Gang (6):
target-tilegx: Firstly add to qemu with minimized features
linux-user: tilegx: Firstly add architecture related features
linux-user: tilegx: Add target features support within qemu
linux-user: Support tilegx architecture in syscall
linux-user: Support tilegx architecture in linux-user linux-user/syscall.c: Switch all macros which are not defined in
tilegx

  configure                             |   7 +
  default-configs/tilegx-linux-user.mak |   1 +
  include/elf.h                         |   2 +
  linux-user/elfload.c                  |  23 +++
  linux-user/main.c                     |  74 +++++++++
  linux-user/syscall.c                  |  50 +++++-
  linux-user/syscall_defs.h             |  38 ++++-
  linux-user/tilegx/syscall.h           |  80 ++++++++++
  linux-user/tilegx/syscall_nr.h        | 278 +++++++++++++++++++++++++++++++++
  linux-user/tilegx/target_cpu.h        |  35 +++++
  linux-user/tilegx/target_signal.h     |  28 ++++
  linux-user/tilegx/target_structs.h    |  48 ++++++
  linux-user/tilegx/termbits.h          | 285 ++++++++++++++++++++++++++++++++++
  target-tilegx/Makefile.objs           |   1 +
  target-tilegx/cpu-qom.h               |  72 +++++++++
  target-tilegx/cpu.c                   | 162 +++++++++++++++++++
  target-tilegx/cpu.h                   |  85 ++++++++++
  target-tilegx/helper.h                |   0
  target-tilegx/translate.c             |  54 +++++++
  19 files changed, 1318 insertions(+), 5 deletions(-)
Quoted 14 lines

create mode 100644 default-configs/tilegx-linux-user.mak create mode 100644 linux-user/tilegx/syscall.h create mode 100644 linux-user/tilegx/syscall_nr.h create mode 100644 linux-user/tilegx/target_cpu.h create mode 100644 linux-user/tilegx/target_signal.h create mode 100644 linux-user/tilegx/target_structs.h create mode 100644 linux-user/tilegx/termbits.h create mode 100644 target-tilegx/Makefile.objs create mode 100644 target-tilegx/cpu-qom.h create mode 100644 target-tilegx/cpu.c
create mode 100644 target-tilegx/cpu.h
create mode 100644 target-tilegx/helper.h create mode 100644 target-tilegx/translate.c

Quoted 4 lines

Hi Chen,

while looking at the diffstat, I see no change to the maintainer file. Can you add someone (you?), who will take care of this code you're trying to add?

At present, please add me for it, firstly. At least, it is my duty.

Excuse me, I am not quite familiar with the related working flow (adding maintainer), if it is necessary to let me do something, please let me know, and I shall follow with.

Thanks.

Signature · 1 line
Open, share, and attitude like air, water, and life which God blessed.
#23 lore raw
Quoted 6 lines · Chen Gang S
Quoted 4 lines · Bastian Koppelmann

Hi Chen,

while looking at the diffstat, I see no change to the maintainer file. Can you add someone (you?), who will take care of this code you're trying to add?

Hello Chris, Richard, Peter:

If any one of you can accept to be as tilegx maintainer, please reply. I welcome any of you to be the tilegx maintainer.

Thanks.

Quoted 8 lines

At present, please add me for it, firstly. At least, it is my duty.

Excuse me, I am not quite familiar with the related working flow (adding maintainer), if it is necessary to let me do something, please let me know, and I shall follow with.

Thanks.

Signature · 1 line
Open, share, and attitude like air, water, and life which God blessed.