From e6e209eb05545a36283335b07a1db9471cbccd1a Mon Sep 17 00:00:00 2001

From: "H.J. Lu" <hjl.tools@gmail.com>

Date: Wed, 2 Mar 2022 13:56:00 -0800

Subject: [PATCH] x86-64 memcpy: Properly handle the length parameter [BZ

 #24097]

On x32, the size_t parameter may be passed in the lower 32 bits of a

64-bit register with the non-zero upper 32 bits.  The string/memory

functions written in assembly can only use the lower 32 bits of a

64-bit register as length or must clear the upper 32 bits before using

the full 64-bit register for length.

This pach fixes memcpy for x32.  Tested on x86-64 and x32.  On x86-64,

libc.so is the same with and withou the fix.

	[BZ #24097]

	CVE-2019-6488

	* sysdeps/x86_64/multiarch/memcpy-ssse3-back.S: Use RDX_LP for

	length.  Clear the upper 32 bits of RDX register.

	* sysdeps/x86_64/multiarch/memcpy-ssse3.S: Likewise.

	* sysdeps/x86_64/multiarch/memmove-avx512-no-vzeroupper.S:

	Likewise.

	* sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:

	Likewise.

	* sysdeps/x86_64/x32/Makefile (tests): Add tst-size_t-memcpy.

	tst-size_t-wmemchr.

	* sysdeps/x86_64/x32/tst-size_t-memcpy.c: New file.

(cherry picked from commit 231c56760c1e2ded21ad96bbb860b1f08c556c7a)

---

 sysdeps/x86_64/multiarch/memcpy-ssse3-back.S  | 17 ++++--

 sysdeps/x86_64/multiarch/memcpy-ssse3.S       | 17 ++++--

 .../multiarch/memmove-avx512-no-vzeroupper.S  | 16 +++--

 .../multiarch/memmove-vec-unaligned-erms.S    | 54 +++++++++--------

 sysdeps/x86_64/x32/Makefile                   |  2 +-

 sysdeps/x86_64/x32/tst-size_t-memcpy.c        | 58 +++++++++++++++++++

 6 files changed, 122 insertions(+), 42 deletions(-)

 create mode 100644 sysdeps/x86_64/x32/tst-size_t-memcpy.c

diff --git a/sysdeps/x86_64/multiarch/memcpy-ssse3-back.S b/sysdeps/x86_64/multiarch/memcpy-ssse3-back.S

index 3cd11233..568eebd3 100644

--- a/sysdeps/x86_64/multiarch/memcpy-ssse3-back.S

+++ b/sysdeps/x86_64/multiarch/memcpy-ssse3-back.S

@@ -45,28 +45,33 @@

 	.section .text.ssse3,"ax",@progbits

 #if !defined USE_AS_MEMPCPY && !defined USE_AS_MEMMOVE

 ENTRY (MEMPCPY_CHK)

-	cmpq	%rdx, %rcx

+	cmp	%RDX_LP, %RCX_LP

 	jb	HIDDEN_JUMPTARGET (__chk_fail)

 END (MEMPCPY_CHK)

 ENTRY (MEMPCPY)

-	movq	%rdi, %rax

-	addq	%rdx, %rax

+	mov	%RDI_LP, %RAX_LP

+	add	%RDX_LP, %RAX_LP

 	jmp	L(start)

 END (MEMPCPY)

 #endif

 #if !defined USE_AS_BCOPY

 ENTRY (MEMCPY_CHK)

-	cmpq	%rdx, %rcx

+	cmp	%RDX_LP, %RCX_LP

 	jb	HIDDEN_JUMPTARGET (__chk_fail)

 END (MEMCPY_CHK)

 #endif

 ENTRY (MEMCPY)

-	mov	%rdi, %rax

+	mov	%RDI_LP, %RAX_LP

 #ifdef USE_AS_MEMPCPY

-	add	%rdx, %rax

+	add	%RDX_LP, %RAX_LP

+#endif

+

+#ifdef __ILP32__

+	/* Clear the upper 32 bits.  */

+	mov	%edx, %edx

 #endif

 #ifdef USE_AS_MEMMOVE

diff --git a/sysdeps/x86_64/multiarch/memcpy-ssse3.S b/sysdeps/x86_64/multiarch/memcpy-ssse3.S

index 0240bfa3..0bd5ee99 100644

--- a/sysdeps/x86_64/multiarch/memcpy-ssse3.S

+++ b/sysdeps/x86_64/multiarch/memcpy-ssse3.S

@@ -45,28 +45,33 @@

 	.section .text.ssse3,"ax",@progbits

 #if !defined USE_AS_MEMPCPY && !defined USE_AS_MEMMOVE

 ENTRY (MEMPCPY_CHK)

-	cmpq	%rdx, %rcx

+	cmp	%RDX_LP, %RCX_LP

 	jb	HIDDEN_JUMPTARGET (__chk_fail)

 END (MEMPCPY_CHK)

 ENTRY (MEMPCPY)

-	movq	%rdi, %rax

-	addq	%rdx, %rax

+	mov	%RDI_LP, %RAX_LP

+	add	%RDX_LP, %RAX_LP

 	jmp	L(start)

 END (MEMPCPY)

 #endif

 #if !defined USE_AS_BCOPY

 ENTRY (MEMCPY_CHK)

-	cmpq	%rdx, %rcx

+	cmp	%RDX_LP, %RCX_LP

 	jb	HIDDEN_JUMPTARGET (__chk_fail)

 END (MEMCPY_CHK)

 #endif

 ENTRY (MEMCPY)

-	mov	%rdi, %rax

+	mov	%RDI_LP, %RAX_LP

 #ifdef USE_AS_MEMPCPY

-	add	%rdx, %rax

+	add	%RDX_LP, %RAX_LP

+#endif

+

+#ifdef __ILP32__

+	/* Clear the upper 32 bits.  */

+	mov	%edx, %edx

 #endif

 #ifdef USE_AS_MEMMOVE

diff --git a/sysdeps/x86_64/multiarch/memmove-avx512-no-vzeroupper.S b/sysdeps/x86_64/multiarch/memmove-avx512-no-vzeroupper.S

index effc3ac2..6ca2bbc9 100644

--- a/sysdeps/x86_64/multiarch/memmove-avx512-no-vzeroupper.S

+++ b/sysdeps/x86_64/multiarch/memmove-avx512-no-vzeroupper.S

@@ -24,27 +24,31 @@

 	.section .text.avx512,"ax",@progbits

 ENTRY (__mempcpy_chk_avx512_no_vzeroupper)

-	cmpq	%rdx, %rcx

+	cmp	%RDX_LP, %RCX_LP

 	jb	HIDDEN_JUMPTARGET (__chk_fail)

 END (__mempcpy_chk_avx512_no_vzeroupper)

 ENTRY (__mempcpy_avx512_no_vzeroupper)

-	movq	%rdi, %rax

-	addq	%rdx, %rax

+	mov	%RDI_LP, %RAX_LP

+	add	%RDX_LP, %RAX_LP

 	jmp	L(start)

 END (__mempcpy_avx512_no_vzeroupper)

 ENTRY (__memmove_chk_avx512_no_vzeroupper)

-	cmpq	%rdx, %rcx

+	cmp	%RDX_LP, %RCX_LP

 	jb	HIDDEN_JUMPTARGET (__chk_fail)

 END (__memmove_chk_avx512_no_vzeroupper)

 ENTRY (__memmove_avx512_no_vzeroupper)

-	mov	%rdi, %rax

+	mov	%RDI_LP, %RAX_LP

 # ifdef USE_AS_MEMPCPY

-	add	%rdx, %rax

+	add	%RDX_LP, %RAX_LP

 # endif

 L(start):

+# ifdef __ILP32__

+	/* Clear the upper 32 bits.  */

+	mov	%edx, %edx

+# endif

 	lea	(%rsi, %rdx), %rcx

 	lea	(%rdi, %rdx), %r9

 	cmp	$512, %rdx

diff --git a/sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S b/sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S

index c952576c..274aa1c7 100644

--- a/sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S

+++ b/sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S

@@ -95,20 +95,20 @@

 	.section SECTION(.text),"ax",@progbits

 #if defined SHARED && IS_IN (libc)

 ENTRY (MEMMOVE_CHK_SYMBOL (__mempcpy_chk, unaligned))

-	cmpq	%rdx, %rcx

+	cmp	%RDX_LP, %RCX_LP

 	jb	HIDDEN_JUMPTARGET (__chk_fail)

 END (MEMMOVE_CHK_SYMBOL (__mempcpy_chk, unaligned))

 #endif

 ENTRY (MEMPCPY_SYMBOL (__mempcpy, unaligned))

-	movq	%rdi, %rax

-	addq	%rdx, %rax

+	mov	%RDI_LP, %RAX_LP

+	add	%RDX_LP, %RAX_LP

 	jmp	L(start)

 END (MEMPCPY_SYMBOL (__mempcpy, unaligned))

 #if defined SHARED && IS_IN (libc)

 ENTRY (MEMMOVE_CHK_SYMBOL (__memmove_chk, unaligned))

-	cmpq	%rdx, %rcx

+	cmp	%RDX_LP, %RCX_LP

 	jb	HIDDEN_JUMPTARGET (__chk_fail)

 END (MEMMOVE_CHK_SYMBOL (__memmove_chk, unaligned))

 #endif

@@ -116,9 +116,13 @@ END (MEMMOVE_CHK_SYMBOL (__memmove_chk, unaligned))

 ENTRY (MEMMOVE_SYMBOL (__memmove, unaligned))

 	movq	%rdi, %rax

 L(start):

-	cmpq	$VEC_SIZE, %rdx

+# ifdef __ILP32__

+	/* Clear the upper 32 bits.  */

+	movl	%edx, %edx

+# endif

+	cmp	$VEC_SIZE, %RDX_LP

 	jb	L(less_vec)

-	cmpq	$(VEC_SIZE * 2), %rdx

+	cmp	$(VEC_SIZE * 2), %RDX_LP

 	ja	L(more_2x_vec)

 #if !defined USE_MULTIARCH || !IS_IN (libc)

 L(last_2x_vec):

@@ -138,38 +142,38 @@ END (MEMMOVE_SYMBOL (__memmove, unaligned))

 # if VEC_SIZE == 16

 ENTRY (__mempcpy_chk_erms)

-	cmpq	%rdx, %rcx

+	cmp	%RDX_LP, %RCX_LP

 	jb	HIDDEN_JUMPTARGET (__chk_fail)

 END (__mempcpy_chk_erms)

 /* Only used to measure performance of REP MOVSB.  */

 ENTRY (__mempcpy_erms)

-	movq	%rdi, %rax

+	mov	%RDI_LP, %RAX_LP

 	/* Skip zero length.  */

-	testq	%rdx, %rdx

+	test	%RDX_LP, %RDX_LP

 	jz	2f

-	addq	%rdx, %rax

+	add	%RDX_LP, %RAX_LP

 	jmp	L(start_movsb)

 END (__mempcpy_erms)

 ENTRY (__memmove_chk_erms)

-	cmpq	%rdx, %rcx

+	cmp	%RDX_LP, %RCX_LP

 	jb	HIDDEN_JUMPTARGET (__chk_fail)

 END (__memmove_chk_erms)

 ENTRY (__memmove_erms)

 	movq	%rdi, %rax

 	/* Skip zero length.  */

-	testq	%rdx, %rdx

+	test	%RDX_LP, %RDX_LP

 	jz	2f

 L(start_movsb):

-	movq	%rdx, %rcx

-	cmpq	%rsi, %rdi

+	mov	%RDX_LP, %RCX_LP

+	cmp	%RSI_LP, %RDI_LP

 	jb	1f

 	/* Source == destination is less common.  */

 	je	2f

-	leaq	(%rsi,%rcx), %rdx

-	cmpq	%rdx, %rdi

+	lea	(%rsi,%rcx), %RDX_LP

+	cmp	%RDX_LP, %RDI_LP

 	jb	L(movsb_backward)

 1:

 	rep movsb

@@ -189,20 +193,20 @@ strong_alias (__memmove_chk_erms, __memcpy_chk_erms)

 # ifdef SHARED

 ENTRY (MEMMOVE_CHK_SYMBOL (__mempcpy_chk, unaligned_erms))

-	cmpq	%rdx, %rcx

+	cmp	%RDX_LP, %RCX_LP

 	jb	HIDDEN_JUMPTARGET (__chk_fail)

 END (MEMMOVE_CHK_SYMBOL (__mempcpy_chk, unaligned_erms))

 # endif

 ENTRY (MEMMOVE_SYMBOL (__mempcpy, unaligned_erms))

-	movq	%rdi, %rax

-	addq	%rdx, %rax

+	mov	%RDI_LP, %RAX_LP

+	add	%RDX_LP, %RAX_LP

 	jmp	L(start_erms)

 END (MEMMOVE_SYMBOL (__mempcpy, unaligned_erms))

 # ifdef SHARED

 ENTRY (MEMMOVE_CHK_SYMBOL (__memmove_chk, unaligned_erms))

-	cmpq	%rdx, %rcx

+	cmp	%RDX_LP, %RCX_LP

 	jb	HIDDEN_JUMPTARGET (__chk_fail)

 END (MEMMOVE_CHK_SYMBOL (__memmove_chk, unaligned_erms))

 # endif

@@ -210,9 +214,13 @@ END (MEMMOVE_CHK_SYMBOL (__memmove_chk, unaligned_erms))

 ENTRY (MEMMOVE_SYMBOL (__memmove, unaligned_erms))

 	movq	%rdi, %rax

 L(start_erms):

-	cmpq	$VEC_SIZE, %rdx

+# ifdef __ILP32__

+	/* Clear the upper 32 bits.  */

+	movl	%edx, %edx

+# endif

+	cmp	$VEC_SIZE, %RDX_LP

 	jb	L(less_vec)

-	cmpq	$(VEC_SIZE * 2), %rdx

+	cmp	$(VEC_SIZE * 2), %RDX_LP

 	ja	L(movsb_more_2x_vec)

 L(last_2x_vec):

 	/* From VEC and to 2 * VEC.  No branch when size == VEC_SIZE. */

@@ -236,7 +244,7 @@ L(movsb):

 	/* Avoid slow backward REP MOVSB.  */

 	jb	L(more_8x_vec_backward)

 1:

-	movq	%rdx, %rcx

+	mov	%RDX_LP, %RCX_LP

 	rep movsb

 L(nop):

 	ret

diff --git a/sysdeps/x86_64/x32/Makefile b/sysdeps/x86_64/x32/Makefile

index ddec7f04..2fe1e5ac 100644

--- a/sysdeps/x86_64/x32/Makefile

+++ b/sysdeps/x86_64/x32/Makefile

@@ -6,7 +6,7 @@ CFLAGS-s_llround.c += -fno-builtin-lround

 endif

 ifeq ($(subdir),string)

-tests += tst-size_t-memchr tst-size_t-memcmp

+tests += tst-size_t-memchr tst-size_t-memcmp tst-size_t-memcpy

 endif

 ifeq ($(subdir),wcsmbs)

diff --git a/sysdeps/x86_64/x32/tst-size_t-memcpy.c b/sysdeps/x86_64/x32/tst-size_t-memcpy.c

new file mode 100644

index 00000000..66b71e17

--- /dev/null

+++ b/sysdeps/x86_64/x32/tst-size_t-memcpy.c

@@ -0,0 +1,58 @@

+/* Test memcpy with size_t in the lower 32 bits of 64-bit register.

+   Copyright (C) 2019 Free Software Foundation, Inc.

+   This file is part of the GNU C Library.

+

+   The GNU C Library is free software; you can redistribute it and/or

+   modify it under the terms of the GNU Lesser General Public

+   License as published by the Free Software Foundation; either

+   version 2.1 of the License, or (at your option) any later version.

+

+   The GNU C Library is distributed in the hope that it will be useful,

+   but WITHOUT ANY WARRANTY; without even the implied warranty of

+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU

+   Lesser General Public License for more details.

+

+   You should have received a copy of the GNU Lesser General Public

+   License along with the GNU C Library; if not, see

+   <http://www.gnu.org/licenses/>.  */

+

+#define TEST_NAME "memcpy"

+#include "test-size_t.h"

+

+IMPL (memcpy, 1)

+

+typedef void *(*proto_t) (void *, const void *, size_t);

+

+static void *

+__attribute__ ((noinline, noclone))

+do_memcpy (parameter_t a, parameter_t b)

+{

+  return CALL (&b, a.p, b.p, a.len);

+}

+

+static int

+test_main (void)

+{

+  test_init ();

+

+  parameter_t dest = { { page_size }, buf1 };

+  parameter_t src = { { 0 }, buf2 };

+

+  int ret = 0;

+  FOR_EACH_IMPL (impl, 0)

+    {

+      src.fn = impl->fn;

+      do_memcpy (dest, src);

+      int res = memcmp (dest.p, src.p, dest.len);

+      if (res)

+	{

+	  error (0, 0, "Wrong result in function %s: %i != 0",

+		 impl->name, res);

+	  ret = 1;

+	}

+    }

+

+  return ret ? EXIT_FAILURE : EXIT_SUCCESS;

+}

+

+#include <support/test-driver.c>

-- 

GitLab