Changeset
include/exec/gdbstub.h | 15 +++++++++++++++
gdbstub.c              |  6 ++++++
2 files changed, 21 insertions(+)
Git apply log
Switched to a new branch '20180515181958.25837-1-peter.maydell@linaro.org'
Applying: gdbstub: Clarify what gdb_handlesig() is doing
To https://github.com/patchew-project/qemu
 * [new tag]         patchew/20180515181958.25837-1-peter.maydell@linaro.org -> patchew/20180515181958.25837-1-peter.maydell@linaro.org
Test passed: checkpatch

loading

Test passed: docker-mingw@fedora

loading

Test passed: docker-quick@centos7

loading

Test passed: s390x

loading

[Qemu-devel] [PATCH] gdbstub: Clarify what gdb_handlesig() is doing
Posted by Peter Maydell, 1 week ago
gdb_handlesig()'s behaviour is not entirely obvious at first
glance. Add a doc comment for it, and also add a comment
explaining why it's ok for gdb_do_syscallv() to ignore
gdb_handlesig()'s return value. (Coverity complains about
this: CID 1390850.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
This took me a little while to figure out, so we might as
well write it down.

Incidentally, a lot of the code in the per-target main
loops doesn't really use gdb_handlesig() correctly either:
for instance in the arm main loop we (a) forget to tell
gdb about SIGSEGV and (b) assume that if we tell gdb about
a SIGTRAP then the signal we get back on resume is either
0 or SIGTRAP, when it could really be anything.

Ideally we'd push the gdb_handlesig calls into
target-independent code, ie queue_signal(). I'm not
sure what sort of fake siginfo we need to generate
for the "generate a different signal" codepath, though:
probably need to look at eg what the gdb gdbstub does.
---
 include/exec/gdbstub.h | 15 +++++++++++++++
 gdbstub.c              |  6 ++++++
 2 files changed, 21 insertions(+)

diff --git a/include/exec/gdbstub.h b/include/exec/gdbstub.h
index 2e8a4b83b9..08363969c1 100644
--- a/include/exec/gdbstub.h
+++ b/include/exec/gdbstub.h
@@ -48,6 +48,21 @@ int use_gdb_syscalls(void);
 void gdb_set_stop_cpu(CPUState *cpu);
 void gdb_exit(CPUArchState *, int);
 #ifdef CONFIG_USER_ONLY
+/**
+ * gdb_handlesig: yield control to gdb
+ * @cpu: CPU
+ * @sig: if non-zero, the signal number which caused us to stop
+ *
+ * This function yields control to gdb, when a user-mode-only target
+ * needs to stop execution. If @sig is non-zero, then we will send a
+ * stop packet to tell gdb that we have stopped because of this signal.
+ *
+ * This function will block (handling protocol requests from gdb)
+ * until gdb tells us to continue target execution. When it does
+ * return, the return value is a signal to deliver to the target,
+ * or 0 if no signal should be delivered, ie the signal that caused
+ * us to stop should be ignored.
+ */
 int gdb_handlesig(CPUState *, int);
 void gdb_signalled(CPUArchState *, int);
 void gdbserver_fork(CPUState *);
diff --git a/gdbstub.c b/gdbstub.c
index 3c3807358c..c9a63090ea 100644
--- a/gdbstub.c
+++ b/gdbstub.c
@@ -1548,6 +1548,12 @@ void gdb_do_syscallv(gdb_syscall_complete_cb cb, const char *fmt, va_list va)
     *p = 0;
 #ifdef CONFIG_USER_ONLY
     put_packet(s, s->syscall_buf);
+    /* 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).
+     */
     gdb_handlesig(s->c_cpu, 0);
 #else
     /* In this case wait to send the syscall packet until notification that
-- 
2.17.0