arm64: ptrace: fix compat hardware watchpoint reporting
authorWill Deacon <will.deacon@arm.com>
Fri, 22 Aug 2014 13:13:24 +0000 (14:13 +0100)
committerWill Deacon <will.deacon@arm.com>
Thu, 28 Aug 2014 19:01:36 +0000 (20:01 +0100)
I'm not sure what I was on when I wrote this, but when iterating over
the hardware watchpoint array (hbp_watch_array), our index is off by
ARM_MAX_BRP, so we walk off the end of our thread_struct...

... except, a dodgy condition in the loop means that it never executes
at all (bp cannot be NULL).

This patch fixes the code so that we remove the bp check and use the
correct index for accessing the watchpoint structures.

Cc: <stable@vger.kernel.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
arch/arm64/include/asm/hw_breakpoint.h
arch/arm64/kernel/ptrace.c

index d064047612b12acb2668927c4b3cedf504d22c11..52b484b6aa1a7fec251a72b028f655523c74236a 100644 (file)
@@ -79,7 +79,6 @@ static inline void decode_ctrl_reg(u32 reg,
  */
 #define ARM_MAX_BRP            16
 #define ARM_MAX_WRP            16
-#define ARM_MAX_HBP_SLOTS      (ARM_MAX_BRP + ARM_MAX_WRP)
 
 /* Virtual debug register bases. */
 #define AARCH64_DBG_REG_BVR    0
index 70526cfda056e7216ef7f773b059d31b6ca4f88a..2ac99887800142a5670e0e55bea89ecc020bfe57 100644 (file)
@@ -87,7 +87,8 @@ static void ptrace_hbptriggered(struct perf_event *bp,
                        break;
                }
        }
-       for (i = ARM_MAX_BRP; i < ARM_MAX_HBP_SLOTS && !bp; ++i) {
+
+       for (i = 0; i < ARM_MAX_WRP; ++i) {
                if (current->thread.debug.hbp_watch[i] == bp) {
                        info.si_errno = -((i << 1) + 1);
                        break;