powerpc: Fix booke user_disable_single_step()
authorDave Kleikamp <shaggy@linux.vnet.ibm.com>
Wed, 8 Jul 2009 13:46:18 +0000 (13:46 +0000)
committerBenjamin Herrenschmidt <benh@kernel.crashing.org>
Wed, 15 Jul 2009 07:41:45 +0000 (17:41 +1000)
On booke processors, gdb is seeing spurious SIGTRAPs when setting a
watchpoint.

user_disable_single_step() simply quits when the DAC is non-zero.  It should
be clearing the DBCR0_IC and DBCR0_BT bits from the dbcr0 register and
TIF_SINGLESTEP from the thread flag.

Signed-off-by: Dave Kleikamp <shaggy@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
arch/powerpc/kernel/ptrace.c

index 9fa2c7d..ef14988 100644 (file)
@@ -736,15 +736,16 @@ void user_disable_single_step(struct task_struct *task)
 {
        struct pt_regs *regs = task->thread.regs;
 
-
-#if defined(CONFIG_BOOKE)
-       /* If DAC then do not single step, skip */
-       if (task->thread.dabr)
-               return;
-#endif
-
        if (regs != NULL) {
-#if defined(CONFIG_40x) || defined(CONFIG_BOOKE)
+#if defined(CONFIG_BOOKE)
+               /* If DAC don't clear DBCRO_IDM or MSR_DE */
+               if (task->thread.dabr)
+                       task->thread.dbcr0 &= ~(DBCR0_IC | DBCR0_BT);
+               else {
+                       task->thread.dbcr0 &= ~(DBCR0_IC | DBCR0_BT | DBCR0_IDM);
+                       regs->msr &= ~MSR_DE;
+               }
+#elif defined(CONFIG_40x)
                task->thread.dbcr0 &= ~(DBCR0_IC | DBCR0_BT | DBCR0_IDM);
                regs->msr &= ~MSR_DE;
 #else