diff mbox

[v2,1/1] gdbserver: Don't send a GDB syscall until the system CPU is stopped

Message ID 1329495681-5187-2-git-send-email-meadori@codesourcery.com
State New
Headers show

Commit Message

Meador Inge Feb. 17, 2012, 4:21 p.m. UTC
Fix an issue where the GDB server implementation was sending GDB syscall
requests while the system CPU was still running.  Syscall requests must
be sent while the CPU is stopped otherwise replies from the GDB client
might get dropped and the GDB server might be incorrectly transitioned
into a 'RUN_STATE_PAUSED' state.

Signed-off-by: Meador Inge <meadori@codesourcery.com>
---
 gdbstub.c |   44 ++++++++++++++++++++++++++++----------------
 1 files changed, 28 insertions(+), 16 deletions(-)

Comments

Peter Maydell Feb. 20, 2012, 4:18 p.m. UTC | #1
On 17 February 2012 16:21, Meador Inge <meadori@codesourcery.com> wrote:
> Fix an issue where the GDB server implementation was sending GDB syscall
> requests while the system CPU was still running.  Syscall requests must
> be sent while the CPU is stopped otherwise replies from the GDB client
> might get dropped and the GDB server might be incorrectly transitioned
> into a 'RUN_STATE_PAUSED' state.
>
> Signed-off-by: Meador Inge <meadori@codesourcery.com>

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

I can confirm that this fixes the "read/write syscalls stall" issue
I was seeing with your previous patch.

-- PMM
Peter Maydell March 6, 2012, 7:05 p.m. UTC | #2
Ping? Who commits gdbstub patches?

-- PMM

On 20 February 2012 16:18, Peter Maydell <peter.maydell@linaro.org> wrote:
> On 17 February 2012 16:21, Meador Inge <meadori@codesourcery.com> wrote:
>> Fix an issue where the GDB server implementation was sending GDB syscall
>> requests while the system CPU was still running.  Syscall requests must
>> be sent while the CPU is stopped otherwise replies from the GDB client
>> might get dropped and the GDB server might be incorrectly transitioned
>> into a 'RUN_STATE_PAUSED' state.
>>
>> Signed-off-by: Meador Inge <meadori@codesourcery.com>
>
> Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
>
> I can confirm that this fixes the "read/write syscalls stall" issue
> I was seeing with your previous patch.
>
> -- PMM
diff mbox

Patch

diff --git a/gdbstub.c b/gdbstub.c
index 7d470b6..a08532f 100644
--- a/gdbstub.c
+++ b/gdbstub.c
@@ -283,8 +283,7 @@  enum RSState {
     RS_IDLE,
     RS_GETLINE,
     RS_CHKSUM1,
-    RS_CHKSUM2,
-    RS_SYSCALL,
+    RS_CHKSUM2
 };
 typedef struct GDBState {
     CPUState *c_cpu; /* current CPU for step/continue ops */
@@ -304,6 +303,8 @@  typedef struct GDBState {
     CharDriverState *chr;
     CharDriverState *mon_chr;
 #endif
+    char syscall_buf[256];
+    gdb_syscall_complete_cb current_syscall_cb;
 } GDBState;
 
 /* By default use no IRQs and no timers while single stepping so as to
@@ -346,8 +347,6 @@  static int get_char(GDBState *s)
 }
 #endif
 
-static gdb_syscall_complete_cb gdb_current_syscall_cb;
-
 static enum {
     GDB_SYS_UNKNOWN,
     GDB_SYS_ENABLED,
@@ -2095,8 +2094,10 @@  static int gdb_handle_packet(GDBState *s, const char *line_buf)
             if (*p == ',')
                 p++;
             type = *p;
-            if (gdb_current_syscall_cb)
-                gdb_current_syscall_cb(s->c_cpu, ret, err);
+            if (s->current_syscall_cb) {
+                s->current_syscall_cb(s->c_cpu, ret, err);
+                s->current_syscall_cb = NULL;
+            }
             if (type == 'C') {
                 put_packet(s, "T02");
             } else {
@@ -2396,7 +2397,12 @@  static void gdb_vm_state_change(void *opaque, int running, RunState state)
     const char *type;
     int ret;
 
-    if (running || s->state == RS_INACTIVE || s->state == RS_SYSCALL) {
+    if (running || s->state == RS_INACTIVE) {
+        return;
+    }
+    /* Is there a GDB syscall waiting to be sent?  */
+    if (s->current_syscall_cb) {
+        put_packet(s, s->syscall_buf);
         return;
     }
     switch (state) {
@@ -2466,8 +2472,8 @@  send_packet:
 void gdb_do_syscall(gdb_syscall_complete_cb cb, const char *fmt, ...)
 {
     va_list va;
-    char buf[256];
     char *p;
+    char *p_end;
     target_ulong addr;
     uint64_t i64;
     GDBState *s;
@@ -2475,14 +2481,13 @@  void gdb_do_syscall(gdb_syscall_complete_cb cb, const char *fmt, ...)
     s = gdbserver_state;
     if (!s)
         return;
-    gdb_current_syscall_cb = cb;
-    s->state = RS_SYSCALL;
+    s->current_syscall_cb = cb;
 #ifndef CONFIG_USER_ONLY
     vm_stop(RUN_STATE_DEBUG);
 #endif
-    s->state = RS_IDLE;
     va_start(va, fmt);
-    p = buf;
+    p = s->syscall_buf;
+    p_end = &s->syscall_buf[sizeof(s->syscall_buf)];
     *(p++) = 'F';
     while (*fmt) {
         if (*fmt == '%') {
@@ -2490,17 +2495,17 @@  void gdb_do_syscall(gdb_syscall_complete_cb cb, const char *fmt, ...)
             switch (*fmt++) {
             case 'x':
                 addr = va_arg(va, target_ulong);
-                p += snprintf(p, &buf[sizeof(buf)] - p, TARGET_FMT_lx, addr);
+                p += snprintf(p, p_end - p, TARGET_FMT_lx, addr);
                 break;
             case 'l':
                 if (*(fmt++) != 'x')
                     goto bad_format;
                 i64 = va_arg(va, uint64_t);
-                p += snprintf(p, &buf[sizeof(buf)] - p, "%" PRIx64, i64);
+                p += snprintf(p, p_end - p, "%" PRIx64, i64);
                 break;
             case 's':
                 addr = va_arg(va, target_ulong);
-                p += snprintf(p, &buf[sizeof(buf)] - p, TARGET_FMT_lx "/%x",
+                p += snprintf(p, p_end - p, TARGET_FMT_lx "/%x",
                               addr, va_arg(va, int));
                 break;
             default:
@@ -2515,10 +2520,16 @@  void gdb_do_syscall(gdb_syscall_complete_cb cb, const char *fmt, ...)
     }
     *p = 0;
     va_end(va);
-    put_packet(s, buf);
 #ifdef CONFIG_USER_ONLY
+    put_packet(s, s->syscall_buf);
     gdb_handlesig(s->c_cpu, 0);
 #else
+    /* In this case wait to send the syscall packet until notification that
+       the CPU has stopped.  This must be done because if the packet is sent
+       now the reply from the syscall request could be received while the CPU
+       is still in the running state, which can cause packets to be dropped
+       and state transition 'T' packets to be sent while the syscall is still
+       being processed.  */
     cpu_exit(s->c_cpu);
 #endif
 }
@@ -2917,6 +2928,7 @@  int gdbserver_start(const char *device)
     s->chr = chr;
     s->state = chr ? RS_IDLE : RS_INACTIVE;
     s->mon_chr = mon_chr;
+    s->current_syscall_cb = NULL;
 
     return 0;
 }