20240718-BIO_DGRAM-memory-leak by douzzer · Pull Request #7760 · wolfSSL/wolfssl

douzzer · 2024-07-18T16:44:34Z

src/bio.c and related:

refactor WOLFSSL_BIO.num and WOLFSSL_BIO.ptr as unions, for clarity and bug resistance (no functional changes).
in wolfSSL_BIO_free(), add WOLFSSL_BIO_DGRAM to the test for closing bio->num.fd, fixing a descriptor leak.
use SOCKET_INVALID consistently as the invalid value for WOLFSSL_BIO.num.fd, and use SOCKET_T consistently as the internal type for file descriptors.
move the definitions for SOCKET_T and SOCKET_INVALID from wolfio.h to the filesystem section of wc_port.h, and allow override definitions of SOCKET_T.

detected and tested with wolfssl-multi-test.sh ... pq-hybrid-all-rpk-valgrind-unittest. also tested with wolfssl-multi-test.sh ... super-quick-check.

douzzer · 2024-07-18T19:28:52Z

retest this please

SparkiDev · 2024-07-19T01:00:12Z

wolfssl/ssl.h

 WOLFSSL_API WOLFSSL_BIO_METHOD* wolfSSL_BIO_f_ssl(void);
-WOLFSSL_API WOLFSSL_BIO*        wolfSSL_BIO_new_socket(int sfd, int flag);
-WOLFSSL_API WOLFSSL_BIO*        wolfSSL_BIO_new_dgram(int fd, int closeF);
+WOLFSSL_API WOLFSSL_BIO*        wolfSSL_BIO_new_socket(SOCKET_T sfd, int flag);


Changing public API. And it doesn't match OpenSSL. (Yes OpenSSL is wrong.)
Make a new one that is SOCKET_T is my suggestion.

not bothering with the new APIs for now -- may not in practice ever be needed.

in any case I've reverted all the changes to public APIs. now, all substantive header changes in the PR are in internal.h.

SparkiDev · 2024-07-19T01:01:54Z

wolfssl/internal.h

    int          rdIdx;         /* current read index */
    int          readRq;        /* read request */
-    int          num;           /* socket num or length */
+    SOCKET_T     num;           /* socket num or length */


Any reason we have to use num for socket?

num is used for the size in some BIO's and the file descriptors too. Not sure if SOCKET_T is the correct type to use for that. I would remove num and create new members for each use case to make it easier to identify what needs fixing.

after comparing notes with Juliusz, I've opted to refactor .num, and .ptr for completeness, as unions. it looks very nice now.

julek-wolfssl · 2024-07-19T10:14:05Z

src/bio.c

 * @return New BIO object or NULL on failure
 */
-WOLFSSL_BIO *wolfSSL_BIO_new_fd(int fd, int close_flag)
+WOLFSSL_BIO *wolfSSL_BIO_new_fd(SOCKET_T fd, int close_flag)


OpenSSL API takes int. Leave it as is and cast inside.

julek-wolfssl · 2024-07-19T10:14:34Z

src/bio.c



-    WOLFSSL_BIO* wolfSSL_BIO_new_socket(int sfd, int closeF)
+    WOLFSSL_BIO* wolfSSL_BIO_new_socket(SOCKET_T sfd, int closeF)


julek-wolfssl · 2024-07-19T10:15:01Z

src/bio.c



-    WOLFSSL_BIO* wolfSSL_BIO_new_dgram(int fd, int closeF)
+    WOLFSSL_BIO* wolfSSL_BIO_new_dgram(SOCKET_T fd, int closeF)


julek-wolfssl · 2024-07-19T10:16:25Z

src/bio.c


 #ifndef NO_FILESYSTEM
-    long wolfSSL_BIO_set_fd(WOLFSSL_BIO* b, int fd, int closeF)
+    long wolfSSL_BIO_set_fd(WOLFSSL_BIO* b, SOCKET_T fd, int closeF)


julek-wolfssl · 2024-07-19T10:16:35Z

src/bio.c

 }

-int wolfSSL_BIO_get_fd(WOLFSSL_BIO *bio, int* fd)
+SOCKET_T wolfSSL_BIO_get_fd(WOLFSSL_BIO *bio, SOCKET_T* fd)


julek-wolfssl · 2024-07-19T10:19:22Z

wolfssl/internal.h

    int          rdIdx;         /* current read index */
    int          readRq;        /* read request */
-    int          num;           /* socket num or length */
+    SOCKET_T     num;           /* socket num or length */


num is used for the size in some BIO's and the file descriptors too. Not sure if SOCKET_T is the correct type to use for that. I would remove num and create new members for each use case to make it easier to identify what needs fixing.

* refactor WOLFSSL_BIO.num and WOLFSSL_BIO.ptr as unions, for clarity and bug resistance (no functional changes). * in wolfSSL_BIO_free(), add WOLFSSL_BIO_DGRAM to the test for closing bio->num.fd, fixing a descriptor leak. * use SOCKET_INVALID consistently as the invalid value for WOLFSSL_BIO.num.fd, and use SOCKET_T consistently as the internal type for file descriptors. * move the definitions for SOCKET_T and SOCKET_INVALID from wolfio.h to the filesystem section of wc_port.h, and allow override definitions of SOCKET_T. detected and tested with wolfssl-multi-test.sh ... pq-hybrid-all-rpk-valgrind-unittest. also tested with wolfssl-multi-test.sh ... super-quick-check.

Fixed

douzzer requested review from julek-wolfssl and wolfSSL-Bot July 18, 2024 16:44

douzzer assigned wolfSSL-Bot Jul 18, 2024

douzzer force-pushed the 20240718-BIO_DGRAM-memory-leak branch 2 times, most recently from d6b4313 to 0582de7 Compare July 18, 2024 18:22

douzzer requested a review from dgarske July 18, 2024 18:26

douzzer force-pushed the 20240718-BIO_DGRAM-memory-leak branch from 0582de7 to 63d6f11 Compare July 18, 2024 18:50

SparkiDev self-assigned this Jul 19, 2024

SparkiDev previously requested changes Jul 19, 2024

View reviewed changes

SparkiDev assigned douzzer and unassigned SparkiDev Jul 19, 2024

julek-wolfssl previously requested changes Jul 19, 2024

View reviewed changes

julek-wolfssl unassigned wolfSSL-Bot Jul 19, 2024

douzzer force-pushed the 20240718-BIO_DGRAM-memory-leak branch from 63d6f11 to 3192d19 Compare July 19, 2024 18:59

douzzer requested review from SparkiDev and julek-wolfssl July 19, 2024 19:10

douzzer assigned wolfSSL-Bot and unassigned douzzer Jul 19, 2024

douzzer force-pushed the 20240718-BIO_DGRAM-memory-leak branch from 3192d19 to 3d26fd2 Compare July 19, 2024 19:18

douzzer force-pushed the 20240718-BIO_DGRAM-memory-leak branch from 3d26fd2 to 787397b Compare July 19, 2024 19:50

dgarske approved these changes Jul 19, 2024

View reviewed changes

dgarske merged commit 4d8a6b8 into wolfSSL:master Jul 19, 2024



		WOLFSSL_BIO* wolfSSL_BIO_new_socket(int sfd, int closeF)
		WOLFSSL_BIO* wolfSSL_BIO_new_socket(SOCKET_T sfd, int closeF)



		WOLFSSL_BIO* wolfSSL_BIO_new_dgram(int fd, int closeF)
		WOLFSSL_BIO* wolfSSL_BIO_new_dgram(SOCKET_T fd, int closeF)

Conversation

douzzer commented Jul 18, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

douzzer commented Jul 18, 2024

Uh oh!

SparkiDev Jul 19, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

douzzer commented Jul 18, 2024 •

edited

Loading

SparkiDev Jul 19, 2024 •

edited

Loading