This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH] Fix out of boundary access in pass_in_v
- From: Yao Qi <qiyaoltc at gmail dot com>
- To: Yao Qi <qiyaoltc at gmail dot com>
- Cc: gdb-patches at sourceware dot org
- Date: Wed, 18 Nov 2015 11:49:13 +0000
- Subject: Re: [PATCH] Fix out of boundary access in pass_in_v
- Authentication-results: sourceware.org; auth=none
- References: <1447686238-9554-1-git-send-email-yao dot qi at linaro dot org>
Yao Qi <qiyaoltc@gmail.com> writes:
> - regcache_cooked_write (regcache, regnum, buf);
> + memset (reg, 0, sizeof (reg));
> + if (gdbarch_byte_order (gdbarch) == BFD_ENDIAN_BIG)
> + memcpy (reg + V_REGISTER_SIZE - len, buf, len);
> + else
> + memcpy (reg, buf, len);
> + regcache_cooked_write (regcache, regnum, reg);
> +
After reading AArch64 procedure call standard again, I find that I don't
have to worry about the endianess here, because arguments are always
allocated at the LSB of SIMD/FP registers.
Patch is updated as below, and is pushed in.
--
Yao (éå)
From: Yao Qi <yao.qi@linaro.org>
Subject: [PATCH] Fix out of boundary access in pass_in_v
Hi,
I build GDB with -fsanitize=address, and run testsuite. In
gdb.base/callfuncs.exp, I see the following error,
p t_float_values(0.0,0.0)
=================================================================
==8088==ERROR: AddressSanitizer: heap-buffer-overflow on address 0x6020000cb650 at pc 0x6e195c bp 0x7fff164f9770 sp 0x7fff164f9768
READ of size 16 at 0x6020000cb650 thread T0^
#0 0x6e195b in regcache_raw_write /home/yao/SourceCode/gnu/gdb/git/gdb/regcache.c:912
#1 0x6e1e52 in regcache_cooked_write /home/yao/SourceCode/gnu/gdb/git/gdb/regcache.c:945
#2 0x466d69 in pass_in_v /home/yao/SourceCode/gnu/gdb/git/gdb/aarch64-tdep.c:1101
#3 0x467512 in pass_in_v_or_stack /home/yao/SourceCode/gnu/gdb/git/gdb/aarch64-tdep.c:1196
#4 0x467d7d in aarch64_push_dummy_call /home/yao/SourceCode/gnu/gdb/git/gdb/aarch64-tdep.c:1335
The code in pass_in_v read contents from V registers (128 bit), but the
data passed through V registers can be less than 128 bit. In this case,
float is passed. So writing V registers contents into contents buff
will cause overflow. In this patch, we add an array reg[V_REGISTER_SIZE],
which is to hold the contents from V registers, and then copy useful
bits to buf.
gdb:
2015-11-18 Yao Qi <yao.qi@linaro.org>
* aarch64-tdep.c (pass_in_v): Add argument len. Add local array
reg. Callers updated.
diff --git a/gdb/ChangeLog b/gdb/ChangeLog
index 1f7aed0..a7b44fe 100644
--- a/gdb/ChangeLog
+++ b/gdb/ChangeLog
@@ -1,3 +1,8 @@
+2015-11-18 Yao Qi <yao.qi@linaro.org>
+
+ * aarch64-tdep.c (pass_in_v): Add argument len. Add local array
+ reg. Callers updated.
+
2015-11-17 Yao Qi <yao.qi@linaro.org>
* infrun.c (resume): Check control.trap_expected only
diff --git a/gdb/aarch64-tdep.c b/gdb/aarch64-tdep.c
index de045e6..de85cb0 100644
--- a/gdb/aarch64-tdep.c
+++ b/gdb/aarch64-tdep.c
@@ -1034,17 +1034,23 @@ static int
pass_in_v (struct gdbarch *gdbarch,
struct regcache *regcache,
struct aarch64_call_info *info,
- const bfd_byte *buf)
+ int len, const bfd_byte *buf)
{
if (info->nsrn < 8)
{
enum bfd_endian byte_order = gdbarch_byte_order (gdbarch);
int regnum = AARCH64_V0_REGNUM + info->nsrn;
+ gdb_byte reg[V_REGISTER_SIZE];
info->argnum++;
info->nsrn++;
- regcache_cooked_write (regcache, regnum, buf);
+ memset (reg, 0, sizeof (reg));
+ /* PCS C.1, the argument is allocated to the least significant
+ bits of V register. */
+ memcpy (reg, buf, len);
+ regcache_cooked_write (regcache, regnum, reg);
+
if (aarch64_debug)
{
debug_printf ("arg %d in %s\n", info->argnum,
@@ -1138,7 +1144,8 @@ pass_in_v_or_stack (struct gdbarch *gdbarch,
struct type *type,
struct value *arg)
{
- if (!pass_in_v (gdbarch, regcache, info, value_contents (arg)))
+ if (!pass_in_v (gdbarch, regcache, info, TYPE_LENGTH (type),
+ value_contents (arg)))
pass_on_stack (info, type, arg);
}
@@ -1263,8 +1270,10 @@ aarch64_push_dummy_call (struct gdbarch *gdbarch, struct value *function,
struct type *target_type =
check_typedef (TYPE_TARGET_TYPE (arg_type));
- pass_in_v (gdbarch, regcache, &info, buf);
pass_in_v (gdbarch, regcache, &info,
+ TYPE_LENGTH (target_type), buf);
+ pass_in_v (gdbarch, regcache, &info,
+ TYPE_LENGTH (target_type),
buf + TYPE_LENGTH (target_type));
}
else