[debugserver] Fix handling of the 'g' packet

LLDB doesn't use this packet so we never hit this, but it looks like
some other projects talk to debugserver and are hitting an assert
(https://github.com/derekparker/delve/issues/1015).

We had an off by 1 in the accounting of the FPU structure sizes.
I added a test that basically just check that 'g' doesn't return
an error (currently it assert in debug builds). I didn't make
it an lldb-server test because it looks like lldb-server doesn't
implement the g packet.

llvm-svn: 331004

lldb/packages/Python/lldbsuite/test/tools/lldb-server/TestGdbRemoteGPacket.py

@@ -0,0 +1,30 @@
+from __future__ import print_function
+
+
+import gdbremote_testcase
+from lldbsuite.test.decorators import *
+from lldbsuite.test.lldbtest import *
+from lldbsuite.test import lldbutil
+
+
+class TestGdbRemoteGPacket(gdbremote_testcase.GdbRemoteTestCaseBase):
+
+    mydir = TestBase.compute_mydir(__file__)
+
+    def run_test_g_packet(self):
+        self.build()
+        self.prep_debug_monitor_and_inferior()
+        self.test_sequence.add_log_lines(
+            ["read packet: $g#67",
+             {"direction": "send", "regex": r"^\$(.+)#[0-9a-fA-F]{2}$",
+              "capture": {1: "register_bank"}}],
+            True)
+        self.connect_to_debug_monitor()
+        context = self.expect_gdbremote_sequence()
+        register_bank = context.get("register_bank")
+        self.assertTrue(register_bank[0] != 'E')
+
+    @debugserver_test
+    def test_g_packet_debugserver(self):
+        self.init_debugserver_test()
+        self.run_test_g_packet()

lldb/tools/debugserver/source/MacOSX/x86_64/DNBArchImplX86_64.cpp

@@ -2633,7 +2633,9 @@ nub_size_t DNBArchImplX86_64::GetRegisterContext(void *buf,
 
       // Walk around the gaps in the FPU regs
       memcpy(p, &m_state.context.fpu.no_avx.__fpu_fcw, 5);
-      p += 5;
+      // We read 5 bytes, but we skip 6 to account for __fpu_rsrv1
+      // to match the g_fpu_registers_* tables.
+      p += 6;
       memcpy(p, &m_state.context.fpu.no_avx.__fpu_fop, 8);
       p += 8;
       memcpy(p, &m_state.context.fpu.no_avx.__fpu_dp, 6);