docs/system: clarify limits of using gdbstub in system emulation

It seems some users will try and use the gdbstub to debug userspace
inside a system emulation. While possible clarify the limitations of
this approach and direct the users to a less head scratching way of
debugging user-space.

Clarifies: https://gitlab.com/qemu-project/qemu/-/issues/1274
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20231120150833.2552739-9-alex.bennee@linaro.org>

docs/system/gdb.rst

@@ -60,7 +60,7 @@ As TCG cannot track all memory accesses in user-mode there is no
 support for watchpoints.
 
 Relocating code
----------------
+===============
 
 On modern kernels confusion can be caused by code being relocated by
 features such as address space layout randomisation. To avoid
@@ -68,6 +68,17 @@ confusion when debugging such things you either need to update gdb's
 view of where things are in memory or perhaps more trivially disable
 ASLR when booting the system.
 
+Debugging user-space in system emulation
+========================================
+
+While it is technically possible to debug a user-space program running
+inside a system image, it does present challenges. Kernel preemption
+and execution mode changes between kernel and user mode can make it
+hard to follow what's going on. Unless you are specifically trying to
+debug some interaction between kernel and user-space you are better
+off running your guest program with gdb either in the guest or using
+a gdbserver exposed via a port to the outside world.
+
 Debugging multicore machines
 ============================