Re: Code formatting [Re: [patch] s390*: watchpoints regression [repost]]

[Pedro Alves <alves.ped@gmail.com>]

On Sunday 18 December 2011 18:02:43, Eli Zaretskii wrote:
> > Date: Sun, 18 Dec 2011 18:24:18 +0100
> > From: Jan Kratochvil <jan.kratochvil@redhat.com>
> > Cc: Mark Kettenis <mark.kettenis@xs4all.nl>, brobecker@adacore.com,
> >         gdb-patches@sourceware.org, pedro@codesourcery.com
> > 
> > You dropped the important part about "still going through the shell", that was
> > the surprising fact to note there.
> > 
> > If you do not drop that shell part of the text the Pedro's text becomes
> > shorter, therefore more clear.
> 
> Then don't drop it.  I think I dropped it by mistake.  Anyway, it was
> just an example of how to reword a comment to avoid the problem that
> started this thread.

Thanks.  Here's what I checked in.

2011-12-20  Pedro Alves  <alves.ped@gmail.com>
	    Jan Kratochvil  <jan.kratochvil@redhat.com>

        * linux-nat.c (add_lwp): Don't call linux_nat_new_thread on the
	first LWP.
	* amd64-linux-nat.c (update_debug_registers_callback): Instantiate
	`lwp->arch_private' if NULL.
	(amd64_linux_prepare_to_resume): Do nothing if `lwp->arch_private'
	is NULL.
	* i386-linux-nat.c (update_debug_registers_callback): Instantiate
	`lwp->arch_private' if NULL.
	(i386_linux_prepare_to_resume): Do nothing if `lwp->arch_private'
	is NULL.
---
 gdb/amd64-linux-nat.c |    9 +++++++++
 gdb/i386-linux-nat.c  |    9 +++++++++
 gdb/linux-nat.c       |   10 +++++++++-
 3 files changed, 27 insertions(+), 1 deletions(-)

diff --git a/gdb/amd64-linux-nat.c b/gdb/amd64-linux-nat.c
index 288160b..865b971 100644
--- a/gdb/amd64-linux-nat.c
+++ b/gdb/amd64-linux-nat.c
@@ -343,6 +343,9 @@ amd64_linux_dr_get_status (void)
 static int
 update_debug_registers_callback (struct lwp_info *lwp, void *arg)
 {
+  if (lwp->arch_private == NULL)
+    lwp->arch_private = XCNEW (struct arch_lwp_info);
+
   /* The actual update is done later just before resuming the lwp, we
      just mark that the registers need updating.  */
   lwp->arch_private->debug_registers_changed = 1;
@@ -386,6 +389,12 @@ amd64_linux_prepare_to_resume (struct lwp_info *lwp)
 {
   int clear_status = 0;
 
+  /* NULL means this is the main thread still going through the shell,
+     or, no watchpoint has been set yet.  In that case, there's
+     nothing to do.  */
+  if (lwp->arch_private == NULL)
+    return;
+
   if (lwp->arch_private->debug_registers_changed)
     {
       struct i386_debug_reg_state *state = i386_debug_reg_state ();
diff --git a/gdb/i386-linux-nat.c b/gdb/i386-linux-nat.c
index 190979b..4527913 100644
--- a/gdb/i386-linux-nat.c
+++ b/gdb/i386-linux-nat.c
@@ -715,6 +715,9 @@ i386_linux_dr_get_status (void)
 static int
 update_debug_registers_callback (struct lwp_info *lwp, void *arg)
 {
+  if (lwp->arch_private == NULL)
+    lwp->arch_private = XCNEW (struct arch_lwp_info);
+
   /* The actual update is done later just before resuming the lwp, we
      just mark that the registers need updating.  */
   lwp->arch_private->debug_registers_changed = 1;
@@ -758,6 +761,12 @@ i386_linux_prepare_to_resume (struct lwp_info *lwp)
 {
   int clear_status = 0;
 
+  /* NULL means this is the main thread still going through the shell,
+     or, no watchpoint has been set yet.  In that case, there's
+     nothing to do.  */
+  if (lwp->arch_private == NULL)
+    return;
+
   if (lwp->arch_private->debug_registers_changed)
     {
       struct i386_debug_reg_state *state = i386_debug_reg_state ();
diff --git a/gdb/linux-nat.c b/gdb/linux-nat.c
index 1cbfc44..e5f7c3e 100644
--- a/gdb/linux-nat.c
+++ b/gdb/linux-nat.c
@@ -1151,7 +1151,15 @@ add_lwp (ptid_t ptid)
   lp->next = lwp_list;
   lwp_list = lp;
 
-  if (linux_nat_new_thread != NULL)
+  /* Let the arch specific bits know about this new thread.  Current
+     clients of this callback take the opportunity to install
+     watchpoints in the new thread.  Don't do this for the first
+     thread though.  If we're spawning a child ("run"), the thread
+     executes the shell wrapper first, and we shouldn't touch it until
+     it execs the program we want to debug.  For "attach", it'd be
+     okay to call the callback, but it's not necessary, because
+     watchpoints can't yet have been inserted into the inferior.  */
+  if (num_lwps (GET_PID (ptid)) > 1 && linux_nat_new_thread != NULL)
     linux_nat_new_thread (lp);
 
   return lp;