When PTRACE_O_TRACEEXEC is used, a ptrace call to fetch the registers at
the PTRACE_EVENT_EXEC stop (PTRACE_PEEKUSR) will oops in CHECK_FULL_REGS.
With recent versions, "gdb --args /bin/sh -c 'exec /bin/true'" and "run" at
the (gdb) prompt is sufficient to produce this. I also have written an
isolated test case, see https://bugzilla.redhat.com/show_bug.cgi?id=301791#c15.
This change fixes the problem by clearing the low bit of pt_regs.trap in
start_thread so that FULL_REGS is true again. This is correct since all of
the GPRs that "full" refers to are cleared in start_thread.
Signed-off-by: Roland McGrath <roland@redhat.com>
---
arch/powerpc/kernel/process.c | 7 +++++++
1 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
index e477c9d..fd799d2 100644
--- a/arch/powerpc/kernel/process.c
+++ b/arch/powerpc/kernel/process.c
@@ -605,6 +605,13 @@ void start_thread(struct pt_regs *regs, unsigned long start, unsigned long sp)
regs->ccr = 0;
regs->gpr[1] = sp;
+ /*
+ * We have just cleared all the nonvolatile GPRs, so make
+ * FULL_REGS(regs) return true. This is necessary to allow
+ * ptrace to examine the thread immediately after exec.
+ */
+ regs->trap &= ~1UL;
+
#ifdef CONFIG_PPC32
regs->mq = 0;
regs->nip = start;
-
