[Pdbg] [PATCH 10/10] gdbserver: Add a note about the attn_handler service on the BMC

[Nicholas Piggin]

gdbserver's use of attn for breakpoints is not well integrated into the
rest of the firmware stack at the moment. attn firing can cause the BMC
to perform error handling (e.g., PEL, reboot).

Add a note to the README to disable the attn_handler.service if
breakpoints are to be used with gdbserver. While we're here update
the documentation to reflect that we no longer have to stop all
threads before running gdbserver.

Signed-off-by: Nicholas Piggin <npiggin at gmail.com>
---
 README.md | 15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)

diff --git a/README.md b/README.md
index f97b6824..338dba2e 100644
--- a/README.md
+++ b/README.md
@@ -540,12 +540,15 @@ Memory access can only be performed on kernel memory.
 
 To run a gdbserver on a machine from a BMC running OpenBMC:
 
-Stop all the threads of the core(s) you want to look at. Ideally all
-threads in the machine should be debugged:
+* Read NOTES and set up the BMC and host as recommended.
+
+* (Optional) Stop the threads of the core(s) you want to look at. Ideally
+  all threads in the machine should be debugged:
 
 $ ./pdbg -a stop
 
-Run gdbserver on the target threads, accessible through port 44
+* Run gdbserver on the target threads, accessible through port 44
+
 $ ./pdbg -a gdbserver 44
 
 The thread-id tid is set to the PIR of the corresponding thread, the
@@ -580,6 +583,12 @@ good idea.
 misbehaving, booting Linux with powersave=off is the first thing to try.
 5. attn instructions seem to cause host hangs on POWER9. gdb breakpoints should
 not be used.
+6. attn instructions can cause the service processor to begin error handling.
+   If breakpoints are to be used, the attn handler service on the BMC should
+   be stopped first (don't forget to start it again when done).
+
+   systemctl stop attn_handler.service
+
 
 ## Submitting patches
 
-- 
2.35.1