uprobes: Teach handle_swbp() to rely on "is_swbp" rather than uprobes_srcu
authorOleg Nesterov <oleg@redhat.com>
Tue, 29 May 2012 19:29:47 +0000 (21:29 +0200)
committerIngo Molnar <mingo@kernel.org>
Wed, 6 Jun 2012 15:22:03 +0000 (17:22 +0200)
Currently handle_swbp() assumes that it can't race with
unregister, so it roughly does:

if (find_uprobe(vaddr))
process_uprobe();
else
send_sig(SIGTRAP);

This relies on the not-really-working uprobes_srcu code we are
going to remove, see the next patch.

With this patch we rely on the result of
is_swbp_at_addr(bp_vaddr) if find_uprobe() fails.

If is_swbp == 1, then we hit the normal int3, we should send
SIGTRAP.

If is_swbp == 0, we raced with uprobe_unregister(), we simply
restart this insn again.

The "difficult" case is is_swbp == -EFAULT, when we can't read
this memory. In this case I think we should restart too, and
this is more correct compared to the current code which sends
SIGTRAP.

Ignoring ENOMEM/etc from get_user_pages(), this can only happen
if another thread unmaps this memory before find_active_uprobe()
takes mmap_sem. It would be better to pretend it was unmapped
before this insn was executed, restart, and get SIGSEGV.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Anton Arapov <anton@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20120529192947.GF8057@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/events/uprobes.c

index a2ed82b4808c6c017519f3b727abe8883c9bd8a8..1f02e3bbfc1dafd6333e2789058b04c8feb0eb39 100644 (file)
@@ -1530,14 +1530,26 @@ static void handle_swbp(struct pt_regs *regs)
        struct uprobe_task *utask;
        struct uprobe *uprobe;
        unsigned long bp_vaddr;
-       int is_swbp;
+       int uninitialized_var(is_swbp);
 
        bp_vaddr = uprobe_get_swbp_addr(regs);
        uprobe = find_active_uprobe(bp_vaddr, &is_swbp);
 
        if (!uprobe) {
-               /* No matching uprobe; signal SIGTRAP. */
-               send_sig(SIGTRAP, current, 0);
+               if (is_swbp > 0) {
+                       /* No matching uprobe; signal SIGTRAP. */
+                       send_sig(SIGTRAP, current, 0);
+               } else {
+                       /*
+                        * Either we raced with uprobe_unregister() or we can't
+                        * access this memory. The latter is only possible if
+                        * another thread plays with our ->mm. In both cases
+                        * we can simply restart. If this vma was unmapped we
+                        * can pretend this insn was not executed yet and get
+                        * the (correct) SIGSEGV after restart.
+                        */
+                       instruction_pointer_set(regs, bp_vaddr);
+               }
                return;
        }