[PATCH 0/8] replace memset by memzero_explicit

[PATCH 0/8] replace memset by memzero_explicit

[Julia Lawall]

Memset on a local variable may be removed when it is called just before the
variable goes out of scope.  Using memzero_explicit defeats this
optimization.  The complete semantic patch that makes this change is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
identifier x;
local idexpression e;
type T,T1;
@@

{
... when any
T x[...];
... when any
    when exists
(
e = (T1)x
|
e = (T1)&x[0]
)
... when any
    when exists
- memset
+ memzero_explicit
  (x,
-0,
  ...)
... when != x
    when != e
    when strict
}

@@
identifier i,x;
local idexpression e;
type T;
@@

{
... when any
struct i x;
... when any
    when exists
e = (T)&x
... when any
    when exists
- memset
+ memzero_explicit
  (&x,
-0,
  ...)
... when != x
    when != e
    when strict
}

// ------------------------------------------------------------------------

@@
identifier x;
type T,T1;
expression e;
@@

{
... when any
T x[...];
... when any
    when exists
    when != e = (T1)x
    when != e = (T1)&x[0]
- memset
+ memzero_explicit
  (x,
-0,
  ...)
... when != x
    when strict
}

@@
identifier i,x;
expression e;
type T;
@@

{
... when any
struct i x;
... when any
    when exists
    when != e = (T)&x
- memset
+ memzero_explicit
  (&x,
-0,
  ...)
... when != x
    when strict
}
// </smpl>

[PATCH 1/8] purgatory/sha256: replace memset by memzero_explicit

[Julia Lawall]

From: Julia Lawall <julia@diku.dk>

Memset on a local variable may be removed when it is called just before the
variable goes out of scope.  Using memzero_explicit defeats this
optimization.  A simplified version of the semantic patch that makes this
change is as follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
identifier x;
type T;
@@

{
... when any
T x[...];
... when any
    when exists
- memset
+ memzero_explicit
  (x,
-0,
  ...)
... when != x
    when strict
}
// </smpl>

This change was suggested by Daniel Borkmann <dborkman@redhat.com>

Signed-off-by: Julia Lawall <julia@diku.dk>

---
Daniel Borkmann suggested that these patches could go through Herbert Xu's
cryptodev tree.

 arch/x86/purgatory/sha256.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/purgatory/sha256.c b/arch/x86/purgatory/sha256.c
index 548ca67..8271ca5 100644
--- a/arch/x86/purgatory/sha256.c
+++ b/arch/x86/purgatory/sha256.c
@@ -205,7 +205,7 @@ static void sha256_transform(u32 *state, const u8 *input)
 
 	/* clear any sensitive info... */
 	a = b = c = d = e = f = g = h = t1 = t2 = 0;
-	memset(W, 0, 64 * sizeof(u32));
+	memzero_explicit(W, 64 * sizeof(u32));
 }
 
 int sha256_init(struct sha256_state *sctx)

Re: [PATCH 1/8] purgatory/sha256: replace memset by memzero_explicit

[David Sterba]

On Sun, Nov 30, 2014 at 05:59:27PM +0100, Julia Lawall wrote:
> @@ -205,7 +205,7 @@ static void sha256_transform(u32 *state, const u8 *input)
>  
>  	/* clear any sensitive info... */
>  	a = b = c = d = e = f = g = h = t1 = t2 = 0;

All of these get optimized away.

How to clear them effectively is another question.  Setting them one by
one casting to volatile would generate 10 instructions. Although this
might be quick as there are no dependencies, the function
sha256_transform is called repeatedly and the small overhead may add up.

Other attempts to fix it require restructuring the code so all the
variables are eg. stored in an array that gets cleared in one go. Here
the indirect access could be hidden into the macros (e0, e1, s0, s1).

> -	memset(W, 0, 64 * sizeof(u32));
> +	memzero_explicit(W, 64 * sizeof(u32));

[PATCH 2/8] sparc64: replace memset by memzero_explicit

[Julia Lawall]

From: Julia Lawall <julia@diku.dk>

Memset on a local variable may be removed when it is called just before the
variable goes out of scope.  Using memzero_explicit defeats this
optimization.  A simplified version of the semantic patch that makes this
change is as follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
identifier x;
type T;
@@

{
... when any
T x[...];
... when any
    when exists
- memset
+ memzero_explicit
  (x,
-0,
  ...)
... when != x
    when strict
}
// </smpl>

This change was suggested by Daniel Borkmann <dborkman@redhat.com>

Signed-off-by: Julia Lawall <julia@diku.dk>

---
Daniel Borkmann suggested that these patches could go through Herbert Xu's
cryptodev tree.

I was not able to compile test these.

 arch/sparc/crypto/sha256_glue.c |    2 +-
 arch/sparc/crypto/sha512_glue.c |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/sparc/crypto/sha512_glue.c b/arch/sparc/crypto/sha512_glue.c
index 9fff885..11eb36c 100644
--- a/arch/sparc/crypto/sha512_glue.c
+++ b/arch/sparc/crypto/sha512_glue.c
@@ -139,7 +139,7 @@ static int sha384_sparc64_final(struct shash_desc *desc, u8 *hash)
 	sha512_sparc64_final(desc, D);
 
 	memcpy(hash, D, 48);
-	memset(D, 0, 64);
+	memzero_explicit(D, 64);
 
 	return 0;
 }
diff --git a/arch/sparc/crypto/sha256_glue.c b/arch/sparc/crypto/sha256_glue.c
index 41f27cc..285268c 100644
--- a/arch/sparc/crypto/sha256_glue.c
+++ b/arch/sparc/crypto/sha256_glue.c
@@ -135,7 +135,7 @@ static int sha224_sparc64_final(struct shash_desc *desc, u8 *hash)
 	sha256_sparc64_final(desc, D);
 
 	memcpy(hash, D, SHA224_DIGEST_SIZE);
-	memset(D, 0, SHA256_DIGEST_SIZE);
+	memzero_explicit(D, SHA256_DIGEST_SIZE);
 
 	return 0;
 }

[PATCH 3/8] crypto: replace memset by memzero_explicit

[Julia Lawall]

From: Julia Lawall <julia@diku.dk>

Memset on a local variable may be removed when it is called just before the
variable goes out of scope.  Using memzero_explicit defeats this
optimization.  A simplified version of the semantic patch that makes this
change is as follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
identifier x;
type T;
@@

{
... when any
T x[...];
... when any
    when exists
- memset
+ memzero_explicit
  (x,
-0,
  ...)
... when != x
    when strict
}
// </smpl>

This change was suggested by Daniel Borkmann <dborkman@redhat.com>

Signed-off-by: Julia Lawall <julia@diku.dk>

---
Daniel Borkmann suggested that these patches could go through Herbert Xu's
cryptodev tree.

 arch/x86/crypto/sha256_ssse3_glue.c |    2 +-
 arch/x86/crypto/sha512_ssse3_glue.c |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/crypto/sha512_ssse3_glue.c b/arch/x86/crypto/sha512_ssse3_glue.c
index 26a5898..0b6af26 100644
--- a/arch/x86/crypto/sha512_ssse3_glue.c
+++ b/arch/x86/crypto/sha512_ssse3_glue.c
@@ -219,7 +219,7 @@ static int sha384_ssse3_final(struct shash_desc *desc, u8 *hash)
 	sha512_ssse3_final(desc, D);
 
 	memcpy(hash, D, SHA384_DIGEST_SIZE);
-	memset(D, 0, SHA512_DIGEST_SIZE);
+	memzero_explicit(D, SHA512_DIGEST_SIZE);
 
 	return 0;
 }
diff --git a/arch/x86/crypto/sha256_ssse3_glue.c b/arch/x86/crypto/sha256_ssse3_glue.c
index 4dc100d..8fad72f 100644
--- a/arch/x86/crypto/sha256_ssse3_glue.c
+++ b/arch/x86/crypto/sha256_ssse3_glue.c
@@ -211,7 +211,7 @@ static int sha224_ssse3_final(struct shash_desc *desc, u8 *hash)
 	sha256_ssse3_final(desc, D);
 
 	memcpy(hash, D, SHA224_DIGEST_SIZE);
-	memset(D, 0, SHA256_DIGEST_SIZE);
+	memzero_explicit(D, SHA256_DIGEST_SIZE);
 
 	return 0;
 }

[PATCH 4/8] cifs: replace memset by memzero_explicit

[Julia Lawall]

From: Julia Lawall <julia@diku.dk>

Memset on a local variable may be removed when it is called just before the
variable goes out of scope.  Using memzero_explicit defeats this
optimization.  A simplified version of the semantic patch that makes this
change is as follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
identifier x;
type T;
@@

{
... when any
T x[...];
... when any
    when exists
- memset
+ memzero_explicit
  (x,
-0,
  ...)
... when != x
    when strict
}
// </smpl>

This change was suggested by Daniel Borkmann <dborkman@redhat.com>

Signed-off-by: Julia Lawall <julia@diku.dk>

---
Daniel Borkmann suggested that these patches could go through Herbert Xu's
cryptodev tree.

 fs/cifs/smbencrypt.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/cifs/smbencrypt.c b/fs/cifs/smbencrypt.c
index 6c15663..6512c19 100644
--- a/fs/cifs/smbencrypt.c
+++ b/fs/cifs/smbencrypt.c
@@ -221,7 +221,7 @@ E_md4hash(const unsigned char *passwd, unsigned char *p16,
 	}
 
 	rc = mdfour(p16, (unsigned char *) wpwd, len * sizeof(__le16));
-	memset(wpwd, 0, 129 * sizeof(__le16));
+	memzero_explicit(wpwd, 129 * sizeof(__le16));
 
 	return rc;
 }

[PATCH 5/8] dm: replace memset by memzero_explicit

[Julia Lawall]

From: Julia Lawall <julia@diku.dk>

Memset on a local variable may be removed when it is called just before the
variable goes out of scope.  Using memzero_explicit defeats this
optimization.  A simplified version of the semantic patch that makes this
change is as follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
identifier x;
type T;
@@

{
... when any
T x[...];
... when any
    when exists
- memset
+ memzero_explicit
  (x,
-0,
  ...)
... when != x
    when strict
}
// </smpl>

This change was suggested by Daniel Borkmann <dborkman@redhat.com>

Signed-off-by: Julia Lawall <julia@diku.dk>

---
Daniel Borkmann suggested that these patches could go through Herbert Xu's
cryptodev tree.

 drivers/md/dm-crypt.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
index fc93b93..08981be 100644
--- a/drivers/md/dm-crypt.c
+++ b/drivers/md/dm-crypt.c
@@ -705,7 +705,7 @@ static int crypt_iv_tcw_whitening(struct crypt_config *cc,
 	for (i = 0; i < ((1 << SECTOR_SHIFT) / 8); i++)
 		crypto_xor(data + i * 8, buf, 8);
 out:
-	memset(buf, 0, sizeof(buf));
+	memzero_explicit(buf, sizeof(buf));
 	return r;
 }
 

[PATCH 6/8] crypto: replace memset by memzero_explicit

[Julia Lawall]

From: Julia Lawall <julia@diku.dk>

Memset on a local variable may be removed when it is called just before the
variable goes out of scope.  Using memzero_explicit defeats this
optimization.  A simplified version of the semantic patch that makes this
change is as follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
identifier x;
type T;
@@

{
... when any
T x[...];
... when any
    when exists
- memset
+ memzero_explicit
  (x,
-0,
  ...)
... when != x
    when strict
}
// </smpl>

This change was suggested by Daniel Borkmann <dborkman@redhat.com>

Signed-off-by: Julia Lawall <julia@diku.dk>

---
Daniel Borkmann suggested that these patches could go through Herbert Xu's
cryptodev tree.

 arch/powerpc/crypto/sha1.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/powerpc/crypto/sha1.c b/arch/powerpc/crypto/sha1.c
index 0f88c7b..d3feba5 100644
--- a/arch/powerpc/crypto/sha1.c
+++ b/arch/powerpc/crypto/sha1.c
@@ -66,7 +66,7 @@ static int sha1_update(struct shash_desc *desc, const u8 *data,
 			src = data + done;
 		} while (done + 63 < len);
 
-		memset(temp, 0, sizeof(temp));
+		memzero_explicit(temp, sizeof(temp));
 		partial = 0;
 	}
 	memcpy(sctx->buffer + partial, src, len - done);

[PATCH 7/8] crypto: replace memset by memzero_explicit

[Julia Lawall]

From: Julia Lawall <julia@diku.dk>

Memset on a local variable may be removed when it is called just before the
variable goes out of scope.  Using memzero_explicit defeats this
optimization.  A simplified version of the semantic patch that makes this
change is as follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
identifier x;
type T;
@@

{
... when any
T x[...];
... when any
    when exists
- memset
+ memzero_explicit
  (x,
-0,
  ...)
... when != x
    when strict
}
// </smpl>

This change was suggested by Daniel Borkmann <dborkman@redhat.com>

Signed-off-by: Julia Lawall <julia@diku.dk>

---
Daniel Borkmann suggested that these patches could go through Herbert Xu's
cryptodev tree.

I was not able to compile test this one.

 arch/arm/crypto/sha512_neon_glue.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm/crypto/sha512_neon_glue.c b/arch/arm/crypto/sha512_neon_glue.c
index f3452c6..b124dce 100644
--- a/arch/arm/crypto/sha512_neon_glue.c
+++ b/arch/arm/crypto/sha512_neon_glue.c
@@ -241,7 +241,7 @@ static int sha384_neon_final(struct shash_desc *desc, u8 *hash)
 	sha512_neon_final(desc, D);
 
 	memcpy(hash, D, SHA384_DIGEST_SIZE);
-	memset(D, 0, SHA512_DIGEST_SIZE);
+	memzero_explicit(D, SHA512_DIGEST_SIZE);
 
 	return 0;
 }

[PATCH 8/8] wusb: replace memset by memzero_explicit

[Julia Lawall]

From: Julia Lawall <julia@diku.dk>

Memset on a local variable may be removed when it is called just before the
variable goes out of scope.  Using memzero_explicit defeats this
optimization.  A simplified version of the semantic patch that makes this
change is as follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
identifier x;
type T;
@@

{
... when any
T x[...];
... when any
    when exists
- memset
+ memzero_explicit
  (x,
-0,
  ...)
... when != x
    when strict
}
// </smpl>

This change was suggested by Daniel Borkmann <dborkman@redhat.com>

Signed-off-by: Julia Lawall <julia@diku.dk>

---
Daniel Borkmann suggested that these patches could go through Herbert Xu's
cryptodev tree.

 drivers/usb/wusbcore/dev-sysfs.c |    2 +-
 drivers/usb/wusbcore/security.c  |    8 ++++----
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/usb/wusbcore/dev-sysfs.c b/drivers/usb/wusbcore/dev-sysfs.c
index 1018345..415b140 100644
--- a/drivers/usb/wusbcore/dev-sysfs.c
+++ b/drivers/usb/wusbcore/dev-sysfs.c
@@ -101,7 +101,7 @@ static ssize_t wusb_ck_store(struct device *dev,
 	if (wusbhc == NULL)
 		return -ENODEV;
 	result = wusb_dev_4way_handshake(wusbhc, usb_dev->wusb_dev, &ck);
-	memset(&ck, 0, sizeof(ck));
+	memzero_explicit(&ck, sizeof(ck));
 	wusbhc_put(wusbhc);
 	return result < 0 ? result : size;
 }
diff --git a/drivers/usb/wusbcore/security.c b/drivers/usb/wusbcore/security.c
index cc74d66..b66faaf 100644
--- a/drivers/usb/wusbcore/security.c
+++ b/drivers/usb/wusbcore/security.c
@@ -522,10 +522,10 @@ error_hs3:
 error_hs2:
 error_hs1:
 	memset(hs, 0, 3*sizeof(hs[0]));
-	memset(&keydvt_out, 0, sizeof(keydvt_out));
-	memset(&keydvt_in, 0, sizeof(keydvt_in));
-	memset(&ccm_n, 0, sizeof(ccm_n));
-	memset(mic, 0, sizeof(mic));
+	memzero_explicit(&keydvt_out, sizeof(keydvt_out));
+	memzero_explicit(&keydvt_in, sizeof(keydvt_in));
+	memzero_explicit(&ccm_n, sizeof(ccm_n));
+	memzero_explicit(mic, sizeof(mic));
 	if (result < 0)
 		wusb_dev_set_encryption(usb_dev, 0);
 error_dev_set_encryption:

Re: [PATCH 8/8] wusb: replace memset by memzero_explicit

[Greg Kroah-Hartman]

On Sun, Nov 30, 2014 at 05:59:34PM +0100, Julia Lawall wrote:
> From: Julia Lawall <julia@diku.dk>
> 
> Memset on a local variable may be removed when it is called just before the
> variable goes out of scope.  Using memzero_explicit defeats this
> optimization.  A simplified version of the semantic patch that makes this
> change is as follows: (http://coccinelle.lip6.fr/)
> 
> // <smpl>
> @@
> identifier x;
> type T;
> @@
> 
> {
> ... when any
> T x[...];
> ... when any
>     when exists
> - memset
> + memzero_explicit
>   (x,
> -0,
>   ...)
> ... when != x
>     when strict
> }
> // </smpl>
> 
> This change was suggested by Daniel Borkmann <dborkman@redhat.com>
> 
> Signed-off-by: Julia Lawall <julia@diku.dk>
> 
> ---
> Daniel Borkmann suggested that these patches could go through Herbert Xu's
> cryptodev tree.

Why?  There's no dependancy on anything in the cryptodev tree,
memzero_explicit is in Linus's tree now.

thanks,

greg k-h

Re: [PATCH 8/8] wusb: replace memset by memzero_explicit

[Daniel Borkmann]

On 12/03/2014 01:14 AM, Greg Kroah-Hartman wrote:
> On Sun, Nov 30, 2014 at 05:59:34PM +0100, Julia Lawall wrote:
>> From: Julia Lawall <julia@diku.dk>
>>
>> Memset on a local variable may be removed when it is called just before the
>> variable goes out of scope.  Using memzero_explicit defeats this
>> optimization.  A simplified version of the semantic patch that makes this
>> change is as follows: (http://coccinelle.lip6.fr/)
>>
>> // <smpl>
>> @@
>> identifier x;
>> type T;
>> @@
>>
>> {
>> ... when any
>> T x[...];
>> ... when any
>>      when exists
>> - memset
>> + memzero_explicit
>>    (x,
>> -0,
>>    ...)
>> ... when != x
>>      when strict
>> }
>> // </smpl>
>>
>> This change was suggested by Daniel Borkmann <dborkman@redhat.com>
>>
>> Signed-off-by: Julia Lawall <julia@diku.dk>
>>
>> ---
>> Daniel Borkmann suggested that these patches could go through Herbert Xu's
>> cryptodev tree.
>
> Why?  There's no dependancy on anything in the cryptodev tree,
> memzero_explicit is in Linus's tree now.

Sorry, I guess this was not really clear, that comment actually only
refers to the arch/*/crypto/ patches.

Anyway, thanks for queueing this up, Greg.

Cheers,
Daniel