Skip to content

Enhance extra user data value and external cookie length max size#10236

Open
Roy-Carter wants to merge 5 commits intowolfSSL:masterfrom
Roy-Carter:feature/enhance_conf_and_max_size
Open

Enhance extra user data value and external cookie length max size#10236
Roy-Carter wants to merge 5 commits intowolfSSL:masterfrom
Roy-Carter:feature/enhance_conf_and_max_size

Conversation

@Roy-Carter
Copy link
Copy Markdown

Description

Enhance configuration limits and fix max size constants to align with RFCs and large-scale deployment needs.

  1. SSL_get_ex_new_index limit raised - --enable-context-extra-user-data now accepts values up to 9999 (was 99). Large platforms with high-scale operations need more than 99 ex_data indices. I've encountered it since my code uses :
    SSL_EX_DATA_IND_DTLS_SESSION = SSL_get_ex_new_index(0, NULL, NULL, NULL, NULL);
    SSL_EX_DATA_IND_PSK = SSL_get_ex_new_index(0, NULL, NULL, NULL, NULL);
    and on "Strong" machines in which i had 50+ cores running it which means (2x50) I failed to initialize an index for a DTLS session.

  2. DTLS MAX_COOKIE_LEN raised to 254 - RFC 6347 defines cookie as opaque<0..2^8-1>, so max valid length is 255. Set to 254 to prevent buffer overflow attempts at boundary. Previous value of 32 was too restrictive for legitimate external cookie use. I've encountered it while trying to inject an external cookie which had valid length of more than 32 .

Testing

Build configuration tested with --enable-context-extra-user-data values: 1, 99, 100, 999, 9999
Verified configure.ac pattern matching rejects invalid inputs (0, 10000, strings)
DTLS cookie handling reviewed for buffer safety with new MAX_COOKIE_LEN

…hich use high scale of operations require more than 99 index options back from SSL_get_ex_new_index
…r hijacking) can be more than 32 in size based on RFC6347
…hat we define maximum copy of <=254 to avoid buffer overflow attempts upon exactly 255..
@Roy-Carter
Copy link
Copy Markdown
Author

@julek-wolfssl if you can please take a look , as part of the integration i've come across these 2 minor issues :)

@wolfSSL-Bot
Copy link
Copy Markdown

Can one of the admins verify this patch?

@julek-wolfssl
Copy link
Copy Markdown
Member

@Roy-Carter is an approved contributor.

@julek-wolfssl
Copy link
Copy Markdown
Member

ok to test

@github-actions
Copy link
Copy Markdown

MemBrowse Memory Report

No memory changes detected for:

Comment thread wolfssl/internal.h
ID_LEN = 32, /* session id length */
COOKIE_SECRET_SZ = 14, /* dtls cookie secret size */
MAX_COOKIE_LEN = 32, /* max dtls cookie size */
MAX_COOKIE_LEN = 254, /* max dtls cookie size per RFC 6347 (opaque<0..2^8-1>) more than 254 can be malformed / malicious */
Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Since this is an increase in the buffer used. Please add a WOLFSSL_COOKIE_LEN macro that is defined by default to 32 but can be increased when needed.

Comment thread configure.ac
AM_CFLAGS="$AM_CFLAGS -DHAVE_EX_DATA -DMAX_EX_DATA=$ENABLED_EX_DATA"
;;
*) AC_MSG_ERROR([Invalid argument to --enable-context-extra-user-data -- must be yes, no, or a number from 1 to 99])
*) AC_MSG_ERROR([Invalid argument to --enable-context-extra-user-data -- must be yes, no, or a number from 1 to 9999])
Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do you really need 9k pointers in the external data?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants