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.

4 files changed, 63 insertions, 41 deletions

diff --git a/sysdeps/x86_64/multiarch/memcpy-ssse3-back.S b/sysdeps/x86_64/multiarch/memcpy-ssse3-back.S
index bcfb21e75a..b9188cbcfb 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 7d74dbd2f5..8f68315418 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 b39324df0a..27df2baab5 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 6e4959f104..2e9d86bd33 100644
--- a/sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S
+++ b/sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S
@@ -106,20 +106,20 @@
 	.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
@@ -127,9 +127,13 @@ 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):
@@ -149,38 +153,38 @@ 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
@@ -200,20 +204,20 @@ 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
@@ -221,9 +225,13 @@ 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. */
@@ -250,7 +258,7 @@ L(movsb):
 # endif
 	jb	L(more_8x_vec_backward)
 1:
-	movq	%rdx, %rcx
+	mov	%RDX_LP, %RCX_LP
 	rep movsb
 L(nop):
 	ret