mirror_qemu/gdbstub/user.c

877 lines
23 KiB
C
Raw Normal View History

/*
* gdbstub user-mode helper routines.
*
* We know for user-mode we are using TCG so we can call stuff directly.
*
* Copyright (c) 2003-2005 Fabrice Bellard
* Copyright (c) 2022 Linaro Ltd
*
* SPDX-License-Identifier: LGPL-2.0+
*/
#include "qemu/osdep.h"
#include "qemu/bitops.h"
#include "qemu/cutils.h"
#include "qemu/sockets.h"
#include "exec/hwaddr.h"
#include "exec/tb-flush.h"
#include "exec/gdbstub.h"
#include "gdbstub/syscalls.h"
#include "gdbstub/user.h"
#include "hw/core/cpu.h"
#include "trace.h"
#include "internals.h"
#define GDB_NR_SYSCALLS 1024
typedef unsigned long GDBSyscallsMask[BITS_TO_LONGS(GDB_NR_SYSCALLS)];
gdbstub: Implement follow-fork-mode child Currently it's not possible to use gdbstub for debugging linux-user code that runs in a forked child, which is normally done using the `set follow-fork-mode child` GDB command. Purely on the protocol level, the missing piece is the fork-events feature. However, a deeper problem is supporting $Hg switching between different processes - right now it can do only threads. Implementing this for the general case would be quite complicated, but, fortunately, for the follow-fork-mode case there are a few factors that greatly simplify things: fork() happens in the exclusive section, there are only two processes involved, and before one of them is resumed, the second one is detached. This makes it possible to implement a simplified scheme: the parent and the child share the gdbserver socket, it's used only by one of them at any given time, which is coordinated through a separate socketpair. The processes can read from the gdbserver socket only one byte at a time, which is not great for performance, but, fortunately, the follow-fork-mode handling involves only a few messages. Advertise the fork-events support, and remember whether GDB has it as well. Implement the state machine that is initialized on fork(), decides the current owner of the gdbserver socket, and is terminated when one of the two processes is detached. The logic for the parent and the child is the same, only the initial state is different. Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> Message-Id: <20240219141628.246823-12-iii@linux.ibm.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Message-Id: <20240305121005.3528075-13-alex.bennee@linaro.org>
2024-03-05 15:09:48 +03:00
/*
* Forked child talks to its parent in order to let GDB enforce the
* follow-fork-mode. This happens inside a start_exclusive() section, so that
* the other threads, which may be forking too, do not interfere. The
* implementation relies on GDB not sending $vCont until it has detached
* either from the parent (follow-fork-mode child) or from the child
* (follow-fork-mode parent).
*
* The parent and the child share the GDB socket; at any given time only one
* of them is allowed to use it, as is reflected in the respective fork_state.
* This is negotiated via the fork_sockets pair as a reaction to $Hg.
*
* Below is a short summary of the possible state transitions:
*
* ENABLED : Terminal state.
* DISABLED : Terminal state.
* ACTIVE : Parent initial state.
* INACTIVE : Child initial state.
* ACTIVE -> DEACTIVATING: On $Hg.
* ACTIVE -> ENABLING : On $D.
* ACTIVE -> DISABLING : On $D.
* ACTIVE -> DISABLED : On communication error.
* DEACTIVATING -> INACTIVE : On gdb_read_byte() return.
* DEACTIVATING -> DISABLED : On communication error.
* INACTIVE -> ACTIVE : On $Hg in the peer.
* INACTIVE -> ENABLE : On $D in the peer.
* INACTIVE -> DISABLE : On $D in the peer.
* INACTIVE -> DISABLED : On communication error.
* ENABLING -> ENABLED : On gdb_read_byte() return.
* ENABLING -> DISABLED : On communication error.
* DISABLING -> DISABLED : On gdb_read_byte() return.
*/
enum GDBForkState {
/* Fully owning the GDB socket. */
GDB_FORK_ENABLED,
/* Working with the GDB socket; the peer is inactive. */
GDB_FORK_ACTIVE,
/* Handing off the GDB socket to the peer. */
GDB_FORK_DEACTIVATING,
/* The peer is working with the GDB socket. */
GDB_FORK_INACTIVE,
/* Asking the peer to close its GDB socket fd. */
GDB_FORK_ENABLING,
/* Asking the peer to take over, closing our GDB socket fd. */
GDB_FORK_DISABLING,
/* The peer has taken over, our GDB socket fd is closed. */
GDB_FORK_DISABLED,
};
enum GDBForkMessage {
GDB_FORK_ACTIVATE = 'a',
GDB_FORK_ENABLE = 'e',
GDB_FORK_DISABLE = 'd',
};
/* User-mode specific state */
typedef struct {
int fd;
char *socket_path;
int running_state;
/*
* Store syscalls mask without memory allocation in order to avoid
* implementing synchronization.
*/
bool catch_all_syscalls;
GDBSyscallsMask catch_syscalls_mask;
gdbstub: Implement follow-fork-mode child Currently it's not possible to use gdbstub for debugging linux-user code that runs in a forked child, which is normally done using the `set follow-fork-mode child` GDB command. Purely on the protocol level, the missing piece is the fork-events feature. However, a deeper problem is supporting $Hg switching between different processes - right now it can do only threads. Implementing this for the general case would be quite complicated, but, fortunately, for the follow-fork-mode case there are a few factors that greatly simplify things: fork() happens in the exclusive section, there are only two processes involved, and before one of them is resumed, the second one is detached. This makes it possible to implement a simplified scheme: the parent and the child share the gdbserver socket, it's used only by one of them at any given time, which is coordinated through a separate socketpair. The processes can read from the gdbserver socket only one byte at a time, which is not great for performance, but, fortunately, the follow-fork-mode handling involves only a few messages. Advertise the fork-events support, and remember whether GDB has it as well. Implement the state machine that is initialized on fork(), decides the current owner of the gdbserver socket, and is terminated when one of the two processes is detached. The logic for the parent and the child is the same, only the initial state is different. Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> Message-Id: <20240219141628.246823-12-iii@linux.ibm.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Message-Id: <20240305121005.3528075-13-alex.bennee@linaro.org>
2024-03-05 15:09:48 +03:00
bool fork_events;
enum GDBForkState fork_state;
int fork_sockets[2];
pid_t fork_peer_pid, fork_peer_tid;
uint8_t siginfo[MAX_SIGINFO_LENGTH];
unsigned long siginfo_len;
} GDBUserState;
static GDBUserState gdbserver_user_state;
int gdb_get_char(void)
{
uint8_t ch;
int ret;
for (;;) {
ret = recv(gdbserver_user_state.fd, &ch, 1, 0);
if (ret < 0) {
if (errno == ECONNRESET) {
gdbserver_user_state.fd = -1;
}
if (errno != EINTR) {
return -1;
}
} else if (ret == 0) {
close(gdbserver_user_state.fd);
gdbserver_user_state.fd = -1;
return -1;
} else {
break;
}
}
return ch;
}
bool gdb_got_immediate_ack(void)
{
int i;
i = gdb_get_char();
if (i < 0) {
/* no response, continue anyway */
return true;
}
if (i == '+') {
/* received correctly, continue */
return true;
}
/* anything else, including '-' then try again */
return false;
}
void gdb_put_buffer(const uint8_t *buf, int len)
{
int ret;
while (len > 0) {
ret = send(gdbserver_user_state.fd, buf, len, 0);
if (ret < 0) {
if (errno != EINTR) {
return;
}
} else {
buf += ret;
len -= ret;
}
}
}
/* Tell the remote gdb that the process has exited. */
void gdb_exit(int code)
{
char buf[4];
if (!gdbserver_state.init) {
return;
}
if (gdbserver_user_state.socket_path) {
unlink(gdbserver_user_state.socket_path);
}
if (gdbserver_user_state.fd < 0) {
return;
}
trace_gdbstub_op_exiting((uint8_t)code);
if (gdbserver_state.allow_stop_reply) {
snprintf(buf, sizeof(buf), "W%02x", (uint8_t)code);
gdb_put_packet(buf);
gdbserver_state.allow_stop_reply = false;
}
}
void gdb_qemu_exit(int code)
{
exit(code);
}
int gdb_handlesig(CPUState *cpu, int sig, const char *reason, void *siginfo,
int siginfo_len)
{
char buf[256];
int n;
if (!gdbserver_state.init || gdbserver_user_state.fd < 0) {
return sig;
}
if (siginfo) {
/*
* Save target-specific siginfo.
*
* siginfo size, i.e. siginfo_len, is asserted at compile-time to fit in
* gdbserver_user_state.siginfo, usually in the source file calling
* gdb_handlesig. See, for instance, {linux,bsd}-user/signal.c.
*/
memcpy(gdbserver_user_state.siginfo, siginfo, siginfo_len);
gdbserver_user_state.siginfo_len = siginfo_len;
}
/* disable single step if it was enabled */
cpu_single_step(cpu, 0);
tb_flush(cpu);
if (sig != 0) {
gdb_set_stop_cpu(cpu);
if (gdbserver_state.allow_stop_reply) {
g_string_printf(gdbserver_state.str_buf,
"T%02xthread:", gdb_target_signal_to_gdb(sig));
gdb_append_thread_id(cpu, gdbserver_state.str_buf);
g_string_append_c(gdbserver_state.str_buf, ';');
if (reason) {
g_string_append(gdbserver_state.str_buf, reason);
}
gdb_put_strbuf();
gdbserver_state.allow_stop_reply = false;
}
}
/*
* gdb_put_packet() might have detected that the peer terminated the
* connection.
*/
if (gdbserver_user_state.fd < 0) {
return sig;
}
sig = 0;
gdbserver_state.state = RS_IDLE;
gdbserver_user_state.running_state = 0;
while (gdbserver_user_state.running_state == 0) {
n = read(gdbserver_user_state.fd, buf, 256);
if (n > 0) {
int i;
for (i = 0; i < n; i++) {
gdb_read_byte(buf[i]);
}
} else {
/*
* XXX: Connection closed. Should probably wait for another
* connection before continuing.
*/
if (n == 0) {
close(gdbserver_user_state.fd);
}
gdbserver_user_state.fd = -1;
return sig;
}
}
sig = gdbserver_state.signal;
gdbserver_state.signal = 0;
return sig;
}
/* Tell the remote gdb that the process has exited due to SIG. */
void gdb_signalled(CPUArchState *env, int sig)
{
char buf[4];
if (!gdbserver_state.init || gdbserver_user_state.fd < 0 ||
!gdbserver_state.allow_stop_reply) {
return;
}
snprintf(buf, sizeof(buf), "X%02x", gdb_target_signal_to_gdb(sig));
gdb_put_packet(buf);
gdbserver_state.allow_stop_reply = false;
}
static void gdb_accept_init(int fd)
{
gdb_init_gdbserver_state();
gdb_create_default_process(&gdbserver_state);
gdbserver_state.processes[0].attached = true;
gdbserver_state.c_cpu = gdb_first_attached_cpu();
gdbserver_state.g_cpu = gdbserver_state.c_cpu;
gdbserver_user_state.fd = fd;
}
static bool gdb_accept_socket(int gdb_fd)
{
int fd;
for (;;) {
fd = accept(gdb_fd, NULL, NULL);
if (fd < 0 && errno != EINTR) {
perror("accept socket");
return false;
} else if (fd >= 0) {
qemu_set_cloexec(fd);
break;
}
}
gdb_accept_init(fd);
return true;
}
static int gdbserver_open_socket(const char *path)
{
struct sockaddr_un sockaddr = {};
int fd, ret;
fd = socket(AF_UNIX, SOCK_STREAM, 0);
if (fd < 0) {
perror("create socket");
return -1;
}
sockaddr.sun_family = AF_UNIX;
pstrcpy(sockaddr.sun_path, sizeof(sockaddr.sun_path) - 1, path);
ret = bind(fd, (struct sockaddr *)&sockaddr, sizeof(sockaddr));
if (ret < 0) {
perror("bind socket");
close(fd);
return -1;
}
ret = listen(fd, 1);
if (ret < 0) {
perror("listen socket");
close(fd);
return -1;
}
return fd;
}
static bool gdb_accept_tcp(int gdb_fd)
{
struct sockaddr_in sockaddr = {};
socklen_t len;
int fd;
for (;;) {
len = sizeof(sockaddr);
fd = accept(gdb_fd, (struct sockaddr *)&sockaddr, &len);
if (fd < 0 && errno != EINTR) {
perror("accept");
return false;
} else if (fd >= 0) {
qemu_set_cloexec(fd);
break;
}
}
/* set short latency */
if (socket_set_nodelay(fd)) {
perror("setsockopt");
close(fd);
return false;
}
gdb_accept_init(fd);
return true;
}
static int gdbserver_open_port(int port)
{
struct sockaddr_in sockaddr;
int fd, ret;
fd = socket(PF_INET, SOCK_STREAM, 0);
if (fd < 0) {
perror("socket");
return -1;
}
qemu_set_cloexec(fd);
socket_set_fast_reuse(fd);
sockaddr.sin_family = AF_INET;
sockaddr.sin_port = htons(port);
sockaddr.sin_addr.s_addr = 0;
ret = bind(fd, (struct sockaddr *)&sockaddr, sizeof(sockaddr));
if (ret < 0) {
perror("bind");
close(fd);
return -1;
}
ret = listen(fd, 1);
if (ret < 0) {
perror("listen");
close(fd);
return -1;
}
return fd;
}
int gdbserver_start(const char *port_or_path)
{
int port = g_ascii_strtoull(port_or_path, NULL, 10);
int gdb_fd;
if (port > 0) {
gdb_fd = gdbserver_open_port(port);
} else {
gdb_fd = gdbserver_open_socket(port_or_path);
}
if (gdb_fd < 0) {
return -1;
}
if (port > 0 && gdb_accept_tcp(gdb_fd)) {
return 0;
} else if (gdb_accept_socket(gdb_fd)) {
gdbserver_user_state.socket_path = g_strdup(port_or_path);
return 0;
}
/* gone wrong */
close(gdb_fd);
return -1;
}
void gdbserver_fork_start(void)
{
gdbstub: Implement follow-fork-mode child Currently it's not possible to use gdbstub for debugging linux-user code that runs in a forked child, which is normally done using the `set follow-fork-mode child` GDB command. Purely on the protocol level, the missing piece is the fork-events feature. However, a deeper problem is supporting $Hg switching between different processes - right now it can do only threads. Implementing this for the general case would be quite complicated, but, fortunately, for the follow-fork-mode case there are a few factors that greatly simplify things: fork() happens in the exclusive section, there are only two processes involved, and before one of them is resumed, the second one is detached. This makes it possible to implement a simplified scheme: the parent and the child share the gdbserver socket, it's used only by one of them at any given time, which is coordinated through a separate socketpair. The processes can read from the gdbserver socket only one byte at a time, which is not great for performance, but, fortunately, the follow-fork-mode handling involves only a few messages. Advertise the fork-events support, and remember whether GDB has it as well. Implement the state machine that is initialized on fork(), decides the current owner of the gdbserver socket, and is terminated when one of the two processes is detached. The logic for the parent and the child is the same, only the initial state is different. Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> Message-Id: <20240219141628.246823-12-iii@linux.ibm.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Message-Id: <20240305121005.3528075-13-alex.bennee@linaro.org>
2024-03-05 15:09:48 +03:00
if (!gdbserver_state.init || gdbserver_user_state.fd < 0) {
return;
}
if (!gdbserver_user_state.fork_events ||
qemu_socketpair(AF_UNIX, SOCK_STREAM, 0,
gdbserver_user_state.fork_sockets) < 0) {
gdbserver_user_state.fork_state = GDB_FORK_DISABLED;
return;
}
gdbserver_user_state.fork_state = GDB_FORK_INACTIVE;
gdbserver_user_state.fork_peer_pid = getpid();
gdbserver_user_state.fork_peer_tid = qemu_get_thread_id();
}
static void disable_gdbstub(CPUState *thread_cpu)
{
CPUState *cpu;
close(gdbserver_user_state.fd);
gdbserver_user_state.fd = -1;
CPU_FOREACH(cpu) {
cpu_breakpoint_remove_all(cpu, BP_GDB);
/* no cpu_watchpoint_remove_all for user-mode */
cpu_single_step(cpu, 0);
}
tb_flush(thread_cpu);
}
void gdbserver_fork_end(CPUState *cpu, pid_t pid)
{
gdbstub: Implement follow-fork-mode child Currently it's not possible to use gdbstub for debugging linux-user code that runs in a forked child, which is normally done using the `set follow-fork-mode child` GDB command. Purely on the protocol level, the missing piece is the fork-events feature. However, a deeper problem is supporting $Hg switching between different processes - right now it can do only threads. Implementing this for the general case would be quite complicated, but, fortunately, for the follow-fork-mode case there are a few factors that greatly simplify things: fork() happens in the exclusive section, there are only two processes involved, and before one of them is resumed, the second one is detached. This makes it possible to implement a simplified scheme: the parent and the child share the gdbserver socket, it's used only by one of them at any given time, which is coordinated through a separate socketpair. The processes can read from the gdbserver socket only one byte at a time, which is not great for performance, but, fortunately, the follow-fork-mode handling involves only a few messages. Advertise the fork-events support, and remember whether GDB has it as well. Implement the state machine that is initialized on fork(), decides the current owner of the gdbserver socket, and is terminated when one of the two processes is detached. The logic for the parent and the child is the same, only the initial state is different. Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> Message-Id: <20240219141628.246823-12-iii@linux.ibm.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Message-Id: <20240305121005.3528075-13-alex.bennee@linaro.org>
2024-03-05 15:09:48 +03:00
char b;
int fd;
if (!gdbserver_state.init || gdbserver_user_state.fd < 0) {
return;
}
gdbstub: Implement follow-fork-mode child Currently it's not possible to use gdbstub for debugging linux-user code that runs in a forked child, which is normally done using the `set follow-fork-mode child` GDB command. Purely on the protocol level, the missing piece is the fork-events feature. However, a deeper problem is supporting $Hg switching between different processes - right now it can do only threads. Implementing this for the general case would be quite complicated, but, fortunately, for the follow-fork-mode case there are a few factors that greatly simplify things: fork() happens in the exclusive section, there are only two processes involved, and before one of them is resumed, the second one is detached. This makes it possible to implement a simplified scheme: the parent and the child share the gdbserver socket, it's used only by one of them at any given time, which is coordinated through a separate socketpair. The processes can read from the gdbserver socket only one byte at a time, which is not great for performance, but, fortunately, the follow-fork-mode handling involves only a few messages. Advertise the fork-events support, and remember whether GDB has it as well. Implement the state machine that is initialized on fork(), decides the current owner of the gdbserver socket, and is terminated when one of the two processes is detached. The logic for the parent and the child is the same, only the initial state is different. Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> Message-Id: <20240219141628.246823-12-iii@linux.ibm.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Message-Id: <20240305121005.3528075-13-alex.bennee@linaro.org>
2024-03-05 15:09:48 +03:00
if (pid == -1) {
if (gdbserver_user_state.fork_state != GDB_FORK_DISABLED) {
g_assert(gdbserver_user_state.fork_state == GDB_FORK_INACTIVE);
close(gdbserver_user_state.fork_sockets[0]);
close(gdbserver_user_state.fork_sockets[1]);
}
return;
}
if (gdbserver_user_state.fork_state == GDB_FORK_DISABLED) {
if (pid == 0) {
disable_gdbstub(cpu);
}
return;
}
if (pid == 0) {
close(gdbserver_user_state.fork_sockets[0]);
fd = gdbserver_user_state.fork_sockets[1];
g_assert(gdbserver_state.process_num == 1);
g_assert(gdbserver_state.processes[0].pid ==
gdbserver_user_state.fork_peer_pid);
g_assert(gdbserver_state.processes[0].attached);
gdbserver_state.processes[0].pid = getpid();
} else {
close(gdbserver_user_state.fork_sockets[1]);
fd = gdbserver_user_state.fork_sockets[0];
gdbserver_user_state.fork_state = GDB_FORK_ACTIVE;
gdbserver_user_state.fork_peer_pid = pid;
gdbserver_user_state.fork_peer_tid = pid;
if (!gdbserver_state.allow_stop_reply) {
goto fail;
}
g_string_printf(gdbserver_state.str_buf,
"T%02xfork:p%02x.%02x;thread:p%02x.%02x;",
gdb_target_signal_to_gdb(gdb_target_sigtrap()),
pid, pid, (int)getpid(), qemu_get_thread_id());
gdb_put_strbuf();
}
gdbserver_state.state = RS_IDLE;
gdbserver_state.allow_stop_reply = false;
gdbserver_user_state.running_state = 0;
for (;;) {
switch (gdbserver_user_state.fork_state) {
case GDB_FORK_ENABLED:
if (gdbserver_user_state.running_state) {
close(fd);
gdbstub: Implement follow-fork-mode child Currently it's not possible to use gdbstub for debugging linux-user code that runs in a forked child, which is normally done using the `set follow-fork-mode child` GDB command. Purely on the protocol level, the missing piece is the fork-events feature. However, a deeper problem is supporting $Hg switching between different processes - right now it can do only threads. Implementing this for the general case would be quite complicated, but, fortunately, for the follow-fork-mode case there are a few factors that greatly simplify things: fork() happens in the exclusive section, there are only two processes involved, and before one of them is resumed, the second one is detached. This makes it possible to implement a simplified scheme: the parent and the child share the gdbserver socket, it's used only by one of them at any given time, which is coordinated through a separate socketpair. The processes can read from the gdbserver socket only one byte at a time, which is not great for performance, but, fortunately, the follow-fork-mode handling involves only a few messages. Advertise the fork-events support, and remember whether GDB has it as well. Implement the state machine that is initialized on fork(), decides the current owner of the gdbserver socket, and is terminated when one of the two processes is detached. The logic for the parent and the child is the same, only the initial state is different. Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> Message-Id: <20240219141628.246823-12-iii@linux.ibm.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Message-Id: <20240305121005.3528075-13-alex.bennee@linaro.org>
2024-03-05 15:09:48 +03:00
return;
}
QEMU_FALLTHROUGH;
case GDB_FORK_ACTIVE:
if (read(gdbserver_user_state.fd, &b, 1) != 1) {
goto fail;
}
gdb_read_byte(b);
break;
case GDB_FORK_DEACTIVATING:
b = GDB_FORK_ACTIVATE;
if (write(fd, &b, 1) != 1) {
goto fail;
}
gdbserver_user_state.fork_state = GDB_FORK_INACTIVE;
break;
case GDB_FORK_INACTIVE:
if (read(fd, &b, 1) != 1) {
goto fail;
}
switch (b) {
case GDB_FORK_ACTIVATE:
gdbserver_user_state.fork_state = GDB_FORK_ACTIVE;
break;
case GDB_FORK_ENABLE:
gdbserver_user_state.fork_state = GDB_FORK_ENABLED;
break;
case GDB_FORK_DISABLE:
gdbserver_user_state.fork_state = GDB_FORK_DISABLED;
break;
default:
g_assert_not_reached();
}
break;
case GDB_FORK_ENABLING:
b = GDB_FORK_DISABLE;
if (write(fd, &b, 1) != 1) {
goto fail;
}
gdbserver_user_state.fork_state = GDB_FORK_ENABLED;
break;
case GDB_FORK_DISABLING:
b = GDB_FORK_ENABLE;
if (write(fd, &b, 1) != 1) {
goto fail;
}
gdbserver_user_state.fork_state = GDB_FORK_DISABLED;
break;
case GDB_FORK_DISABLED:
close(fd);
disable_gdbstub(cpu);
return;
default:
g_assert_not_reached();
}
}
fail:
close(fd);
if (pid == 0) {
disable_gdbstub(cpu);
}
}
void gdb_handle_query_supported_user(const char *gdb_supported)
{
gdbstub: Implement follow-fork-mode child Currently it's not possible to use gdbstub for debugging linux-user code that runs in a forked child, which is normally done using the `set follow-fork-mode child` GDB command. Purely on the protocol level, the missing piece is the fork-events feature. However, a deeper problem is supporting $Hg switching between different processes - right now it can do only threads. Implementing this for the general case would be quite complicated, but, fortunately, for the follow-fork-mode case there are a few factors that greatly simplify things: fork() happens in the exclusive section, there are only two processes involved, and before one of them is resumed, the second one is detached. This makes it possible to implement a simplified scheme: the parent and the child share the gdbserver socket, it's used only by one of them at any given time, which is coordinated through a separate socketpair. The processes can read from the gdbserver socket only one byte at a time, which is not great for performance, but, fortunately, the follow-fork-mode handling involves only a few messages. Advertise the fork-events support, and remember whether GDB has it as well. Implement the state machine that is initialized on fork(), decides the current owner of the gdbserver socket, and is terminated when one of the two processes is detached. The logic for the parent and the child is the same, only the initial state is different. Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> Message-Id: <20240219141628.246823-12-iii@linux.ibm.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Message-Id: <20240305121005.3528075-13-alex.bennee@linaro.org>
2024-03-05 15:09:48 +03:00
if (strstr(gdb_supported, "fork-events+")) {
gdbserver_user_state.fork_events = true;
}
g_string_append(gdbserver_state.str_buf, ";fork-events+");
}
bool gdb_handle_set_thread_user(uint32_t pid, uint32_t tid)
{
gdbstub: Implement follow-fork-mode child Currently it's not possible to use gdbstub for debugging linux-user code that runs in a forked child, which is normally done using the `set follow-fork-mode child` GDB command. Purely on the protocol level, the missing piece is the fork-events feature. However, a deeper problem is supporting $Hg switching between different processes - right now it can do only threads. Implementing this for the general case would be quite complicated, but, fortunately, for the follow-fork-mode case there are a few factors that greatly simplify things: fork() happens in the exclusive section, there are only two processes involved, and before one of them is resumed, the second one is detached. This makes it possible to implement a simplified scheme: the parent and the child share the gdbserver socket, it's used only by one of them at any given time, which is coordinated through a separate socketpair. The processes can read from the gdbserver socket only one byte at a time, which is not great for performance, but, fortunately, the follow-fork-mode handling involves only a few messages. Advertise the fork-events support, and remember whether GDB has it as well. Implement the state machine that is initialized on fork(), decides the current owner of the gdbserver socket, and is terminated when one of the two processes is detached. The logic for the parent and the child is the same, only the initial state is different. Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> Message-Id: <20240219141628.246823-12-iii@linux.ibm.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Message-Id: <20240305121005.3528075-13-alex.bennee@linaro.org>
2024-03-05 15:09:48 +03:00
if (gdbserver_user_state.fork_state == GDB_FORK_ACTIVE &&
pid == gdbserver_user_state.fork_peer_pid &&
tid == gdbserver_user_state.fork_peer_tid) {
gdbserver_user_state.fork_state = GDB_FORK_DEACTIVATING;
gdb_put_packet("OK");
return true;
}
return false;
}
bool gdb_handle_detach_user(uint32_t pid)
{
gdbstub: Implement follow-fork-mode child Currently it's not possible to use gdbstub for debugging linux-user code that runs in a forked child, which is normally done using the `set follow-fork-mode child` GDB command. Purely on the protocol level, the missing piece is the fork-events feature. However, a deeper problem is supporting $Hg switching between different processes - right now it can do only threads. Implementing this for the general case would be quite complicated, but, fortunately, for the follow-fork-mode case there are a few factors that greatly simplify things: fork() happens in the exclusive section, there are only two processes involved, and before one of them is resumed, the second one is detached. This makes it possible to implement a simplified scheme: the parent and the child share the gdbserver socket, it's used only by one of them at any given time, which is coordinated through a separate socketpair. The processes can read from the gdbserver socket only one byte at a time, which is not great for performance, but, fortunately, the follow-fork-mode handling involves only a few messages. Advertise the fork-events support, and remember whether GDB has it as well. Implement the state machine that is initialized on fork(), decides the current owner of the gdbserver socket, and is terminated when one of the two processes is detached. The logic for the parent and the child is the same, only the initial state is different. Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> Message-Id: <20240219141628.246823-12-iii@linux.ibm.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Message-Id: <20240305121005.3528075-13-alex.bennee@linaro.org>
2024-03-05 15:09:48 +03:00
bool enable;
if (gdbserver_user_state.fork_state == GDB_FORK_ACTIVE) {
enable = pid == gdbserver_user_state.fork_peer_pid;
if (enable || pid == getpid()) {
gdbserver_user_state.fork_state = enable ? GDB_FORK_ENABLING :
GDB_FORK_DISABLING;
gdb_put_packet("OK");
return true;
}
}
return false;
}
/*
* Execution state helpers
*/
void gdb_handle_query_attached(GArray *params, void *user_ctx)
{
gdb_put_packet("0");
}
void gdb_continue(void)
{
gdbserver_user_state.running_state = 1;
trace_gdbstub_op_continue();
}
/*
* Resume execution, for user-mode emulation it's equivalent to
* gdb_continue.
*/
int gdb_continue_partial(char *newstates)
{
CPUState *cpu;
int res = 0;
/*
* This is not exactly accurate, but it's an improvement compared to the
* previous situation, where only one CPU would be single-stepped.
*/
CPU_FOREACH(cpu) {
if (newstates[cpu->cpu_index] == 's') {
trace_gdbstub_op_stepping(cpu->cpu_index);
cpu_single_step(cpu, gdbserver_state.sstep_flags);
}
}
gdbserver_user_state.running_state = 1;
return res;
}
/*
* Memory access helpers
*/
int gdb_target_memory_rw_debug(CPUState *cpu, hwaddr addr,
uint8_t *buf, int len, bool is_write)
{
CPUClass *cc;
cc = CPU_GET_CLASS(cpu);
if (cc->memory_rw_debug) {
return cc->memory_rw_debug(cpu, addr, buf, len, is_write);
}
return cpu_memory_rw_debug(cpu, addr, buf, len, is_write);
}
/*
* cpu helpers
*/
unsigned int gdb_get_max_cpus(void)
{
CPUState *cpu;
unsigned int max_cpus = 1;
CPU_FOREACH(cpu) {
max_cpus = max_cpus <= cpu->cpu_index ? cpu->cpu_index + 1 : max_cpus;
}
return max_cpus;
}
/* replay not supported for user-mode */
bool gdb_can_reverse(void)
{
return false;
}
/*
* Break/Watch point helpers
*/
bool gdb_supports_guest_debug(void)
{
/* user-mode == TCG == supported */
return true;
}
int gdb_breakpoint_insert(CPUState *cs, int type, vaddr addr, vaddr len)
{
CPUState *cpu;
int err = 0;
switch (type) {
case GDB_BREAKPOINT_SW:
case GDB_BREAKPOINT_HW:
CPU_FOREACH(cpu) {
err = cpu_breakpoint_insert(cpu, addr, BP_GDB, NULL);
if (err) {
break;
}
}
return err;
default:
/* user-mode doesn't support watchpoints */
return -ENOSYS;
}
}
int gdb_breakpoint_remove(CPUState *cs, int type, vaddr addr, vaddr len)
{
CPUState *cpu;
int err = 0;
switch (type) {
case GDB_BREAKPOINT_SW:
case GDB_BREAKPOINT_HW:
CPU_FOREACH(cpu) {
err = cpu_breakpoint_remove(cpu, addr, BP_GDB);
if (err) {
break;
}
}
return err;
default:
/* user-mode doesn't support watchpoints */
return -ENOSYS;
}
}
void gdb_breakpoint_remove_all(CPUState *cs)
{
cpu_breakpoint_remove_all(cs, BP_GDB);
}
/*
* For user-mode syscall support we send the system call immediately
* and then return control to gdb for it to process the syscall request.
* Since the protocol requires that gdb hands control back to us
* using a "here are the results" F packet, we don't need to check
* gdb_handlesig's return value (which is the signal to deliver if
* execution was resumed via a continue packet).
*/
void gdb_syscall_handling(const char *syscall_packet)
{
gdb_put_packet(syscall_packet);
gdb_handlesig(gdbserver_state.c_cpu, 0, NULL, NULL, 0);
}
static bool should_catch_syscall(int num)
{
if (gdbserver_user_state.catch_all_syscalls) {
return true;
}
if (num < 0 || num >= GDB_NR_SYSCALLS) {
return false;
}
return test_bit(num, gdbserver_user_state.catch_syscalls_mask);
}
void gdb_syscall_entry(CPUState *cs, int num)
{
if (should_catch_syscall(num)) {
g_autofree char *reason = g_strdup_printf("syscall_entry:%x;", num);
gdb_handlesig(cs, gdb_target_sigtrap(), reason, NULL, 0);
}
}
void gdb_syscall_return(CPUState *cs, int num)
{
if (should_catch_syscall(num)) {
g_autofree char *reason = g_strdup_printf("syscall_return:%x;", num);
gdb_handlesig(cs, gdb_target_sigtrap(), reason, NULL, 0);
}
}
void gdb_handle_set_catch_syscalls(GArray *params, void *user_ctx)
{
const char *param = get_param(params, 0)->data;
GDBSyscallsMask catch_syscalls_mask;
bool catch_all_syscalls;
unsigned int num;
const char *p;
/* "0" means not catching any syscalls. */
if (strcmp(param, "0") == 0) {
gdbserver_user_state.catch_all_syscalls = false;
memset(gdbserver_user_state.catch_syscalls_mask, 0,
sizeof(gdbserver_user_state.catch_syscalls_mask));
gdb_put_packet("OK");
return;
}
/* "1" means catching all syscalls. */
if (strcmp(param, "1") == 0) {
gdbserver_user_state.catch_all_syscalls = true;
gdb_put_packet("OK");
return;
}
/*
* "1;..." means catching only the specified syscalls.
* The syscall list must not be empty.
*/
if (param[0] == '1' && param[1] == ';') {
catch_all_syscalls = false;
memset(catch_syscalls_mask, 0, sizeof(catch_syscalls_mask));
for (p = &param[2];; p++) {
if (qemu_strtoui(p, &p, 16, &num) || (*p && *p != ';')) {
goto err;
}
if (num >= GDB_NR_SYSCALLS) {
/*
* Fall back to reporting all syscalls. Reporting extra
* syscalls is inefficient, but the spec explicitly allows it.
* Keep parsing in case there is a syntax error ahead.
*/
catch_all_syscalls = true;
} else {
set_bit(num, catch_syscalls_mask);
}
if (!*p) {
break;
}
}
gdbserver_user_state.catch_all_syscalls = catch_all_syscalls;
if (!catch_all_syscalls) {
memcpy(gdbserver_user_state.catch_syscalls_mask,
catch_syscalls_mask, sizeof(catch_syscalls_mask));
}
gdb_put_packet("OK");
return;
}
err:
gdb_put_packet("E00");
}
void gdb_handle_query_xfer_siginfo(GArray *params, void *user_ctx)
{
unsigned long offset, len;
uint8_t *siginfo_offset;
offset = get_param(params, 0)->val_ul;
len = get_param(params, 1)->val_ul;
if (offset + len > gdbserver_user_state.siginfo_len) {
/* Invalid offset and/or requested length. */
gdb_put_packet("E01");
return;
}
siginfo_offset = (uint8_t *)gdbserver_user_state.siginfo + offset;
/* Reply */
g_string_assign(gdbserver_state.str_buf, "l");
gdb_memtox(gdbserver_state.str_buf, (const char *)siginfo_offset, len);
gdb_put_packet_binary(gdbserver_state.str_buf->str,
gdbserver_state.str_buf->len, true);
}