Manual replacement of all strlcpy in favor of strscpy

strlcpy() copy a C-String into a sized buffer, the result is always a
valid NULL-terminated that fits in the buffer, howerver it has severals
issues. It reads the source buffer first, which is dangerous if it is non
NULL-terminated or if the corresponding buffer is unbounded. Its safe
replacement is strscpy(), as suggested in the deprecated interface [1].

We plan to make this contribution in two steps:
- Firsly all cases of strlcpy's return value are manually replaced by the
  corresponding calls of strscpy() with the new handling of the return
  value (as the return code is different in case of error).
- Then all other cases are automatically replaced by using coccinelle.

This series covers manual replacements.

[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy

Romain Perier (20):
  cgroup: Manual replacement of the deprecated strlcpy() with return
    values
  crypto: Manual replacement of the deprecated strlcpy() with return
    values
  devlink: Manual replacement of the deprecated strlcpy() with return
    values
  dma-buf: Manual replacement of the deprecated strlcpy() with return
    values
  kobject: Manual replacement of the deprecated strlcpy() with return
    values
  ima: Manual replacement of the deprecated strlcpy() with return values
  SUNRPC: Manual replacement of the deprecated strlcpy() with return
    values
  kernfs: Manual replacement of the deprecated strlcpy() with return
    values
  m68k/atari: Manual replacement of the deprecated strlcpy() with return
    values
  module: Manual replacement of the deprecated strlcpy() with return
    values
  hwmon: Manual replacement of the deprecated strlcpy() with return
    values
  s390/hmcdrv: Manual replacement of the deprecated strlcpy() with
    return values

44 more lines

The strlcpy() reads the entire source buffer first, it is dangerous if
the source buffer lenght is unbounded or possibility non NULL-terminated.
It can lead to linear read overflows, crashes, etc...

As recommended in the deprecated interfaces [1], it should be replaced
by strscpy.

This commit replaces all calls to strlcpy that handle the return values
by the corresponding strscpy calls with new handling of the return
values (as it is quite different between the two functions).

[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy

Signed-off-by: Romain Perier <romain.perier@gmail.com>
---
 crypto/lrw.c |    6 +++---
 crypto/xts.c |    6 +++---
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/crypto/lrw.c b/crypto/lrw.c
index bcf09fbc750a..4d35f4439012 100644
--- a/crypto/lrw.c
+++ b/crypto/lrw.c

@@ -357,10 +357,10 @@ static int lrw_create(struct crypto_template *tmpl, struct rtattr **tb)
 	 * cipher name.
 	 */
 	if (!strncmp(cipher_name, "ecb(", 4)) {
-		unsigned len;
+		ssize_t len;
 
-		len = strlcpy(ecb_name, cipher_name + 4, sizeof(ecb_name));
-		if (len < 2 || len >= sizeof(ecb_name))
+		len = strscpy(ecb_name, cipher_name + 4, sizeof(ecb_name));
+		if (len == -E2BIG || len < 2)
 			goto err_free_inst;
 
 		if (ecb_name[len - 1] != ')')

diff --git a/crypto/xts.c b/crypto/xts.c
index 6c12f30dbdd6..1dfe39d61418 100644
--- a/crypto/xts.c
+++ b/crypto/xts.c

@@ -396,10 +396,10 @@ static int xts_create(struct crypto_template *tmpl, struct rtattr **tb)
 	 * cipher name.
 	 */
 	if (!strncmp(cipher_name, "ecb(", 4)) {
-		unsigned len;
+		ssize_t len;
 
-		len = strlcpy(ctx->name, cipher_name + 4, sizeof(ctx->name));
-		if (len < 2 || len >= sizeof(ctx->name))
+		len = strscpy(ctx->name, cipher_name + 4, sizeof(ctx->name));
+		if (len == -E2BIG || len < 2)
 			goto err_free_inst;
 
 		if (ctx->name[len - 1] != ')')

On 2/22/21 8:12 AM, Romain Perier wrote:

strlcpy() copy a C-String into a sized buffer, the result is always a
valid NULL-terminated that fits in the buffer, howerver it has severals
issues. It reads the source buffer first, which is dangerous if it is non
NULL-terminated or if the corresponding buffer is unbounded. Its safe
replacement is strscpy(), as suggested in the deprecated interface [1].

We plan to make this contribution in two steps:
- Firsly all cases of strlcpy's return value are manually replaced by the
   corresponding calls of strscpy() with the new handling of the return
   value (as the return code is different in case of error).
- Then all other cases are automatically replaced by using coccinelle.

Cool. A quick check shows me 1031 strscpy() calls with no return
checks. All or some of these probably need to be reviewed and add
return checks. Is this something that is in the plan to address as
part of this work?

thanks,
-- Shuah

On Mon, Feb 22, 2021 at 04:12:13PM +0100, Romain Perier wrote:

len == -E2BIG is superfluous as len < 2 will catch it anyway.

Thanks,
-- 
Email: Herbert Xu [off-list ref]
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt