1=pod 2 3=head1 NAME 4 5SSL_CTX_set_mode, SSL_CTX_clear_mode, SSL_set_mode, SSL_clear_mode, SSL_CTX_get_mode, SSL_get_mode - manipulate SSL engine mode 6 7=head1 SYNOPSIS 8 9 #include <openssl/ssl.h> 10 11 long SSL_CTX_set_mode(SSL_CTX *ctx, long mode); 12 long SSL_CTX_clear_mode(SSL_CTX *ctx, long mode); 13 long SSL_set_mode(SSL *ssl, long mode); 14 long SSL_clear_mode(SSL *ssl, long mode); 15 16 long SSL_CTX_get_mode(SSL_CTX *ctx); 17 long SSL_get_mode(SSL *ssl); 18 19=head1 DESCRIPTION 20 21SSL_CTX_set_mode() adds the mode set via bit mask in B<mode> to B<ctx>. 22Options already set before are not cleared. 23SSL_CTX_clear_mode() removes the mode set via bit mask in B<mode> from B<ctx>. 24 25SSL_set_mode() adds the mode set via bit mask in B<mode> to B<ssl>. 26Options already set before are not cleared. 27SSL_clear_mode() removes the mode set via bit mask in B<mode> from B<ssl>. 28 29SSL_CTX_get_mode() returns the mode set for B<ctx>. 30 31SSL_get_mode() returns the mode set for B<ssl>. 32 33=head1 NOTES 34 35The following mode changes are available: 36 37=over 4 38 39=item SSL_MODE_ENABLE_PARTIAL_WRITE 40 41Allow SSL_write_ex(..., n, &r) to return with 0 < r < n (i.e. report success 42when just a single record has been written). This works in a similar way for 43SSL_write(). When not set (the default), SSL_write_ex() or SSL_write() will only 44report success once the complete chunk was written. Once SSL_write_ex() or 45SSL_write() returns successful, B<r> bytes have been written and the next call 46to SSL_write_ex() or SSL_write() must only send the n-r bytes left, imitating 47the behaviour of write(). 48 49=item SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER 50 51Make it possible to retry SSL_write_ex() or SSL_write() with changed buffer 52location (the buffer contents must stay the same). This is not the default to 53avoid the misconception that nonblocking SSL_write() behaves like 54nonblocking write(). 55 56=item SSL_MODE_AUTO_RETRY 57 58During normal operations, non-application data records might need to be sent or 59received that the application is not aware of. 60If a non-application data record was processed, 61L<SSL_read_ex(3)> and L<SSL_read(3)> can return with a failure and indicate the 62need to retry with B<SSL_ERROR_WANT_READ>. 63If such a non-application data record was processed, the flag 64B<SSL_MODE_AUTO_RETRY> causes it to try to process the next record instead of 65returning. 66 67In a nonblocking environment applications must be prepared to handle 68incomplete read/write operations. 69Setting B<SSL_MODE_AUTO_RETRY> for a nonblocking B<BIO> will process 70non-application data records until either no more data is available or 71an application data record has been processed. 72 73In a blocking environment, applications are not always prepared to 74deal with the functions returning intermediate reports such as retry 75requests, and setting the B<SSL_MODE_AUTO_RETRY> flag will cause the functions 76to only return after successfully processing an application data record or a 77failure. 78 79Turning off B<SSL_MODE_AUTO_RETRY> can be useful with blocking B<BIO>s in case 80they are used in combination with something like select() or poll(). 81Otherwise the call to SSL_read() or SSL_read_ex() might hang when a 82non-application record was sent and no application data was sent. 83 84=item SSL_MODE_RELEASE_BUFFERS 85 86When we no longer need a read buffer or a write buffer for a given SSL, 87then release the memory we were using to hold it. 88Using this flag can 89save around 34k per idle SSL connection. 90This flag has no effect on SSL v2 connections, or on DTLS connections. 91 92=item SSL_MODE_SEND_FALLBACK_SCSV 93 94Send TLS_FALLBACK_SCSV in the ClientHello. 95To be set only by applications that reconnect with a downgraded protocol 96version; see draft-ietf-tls-downgrade-scsv-00 for details. 97 98DO NOT ENABLE THIS if your application attempts a normal handshake. 99Only use this in explicit fallback retries, following the guidance 100in draft-ietf-tls-downgrade-scsv-00. 101 102=item SSL_MODE_ASYNC 103 104Enable asynchronous processing. TLS I/O operations may indicate a retry with 105SSL_ERROR_WANT_ASYNC with this mode set if an asynchronous capable engine is 106used to perform cryptographic operations. See L<SSL_get_error(3)>. 107 108=item SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG 109 110Older versions of OpenSSL had a bug in the computation of the label length 111used for computing the endpoint-pair shared secret. The bug was that the 112terminating zero was included in the length of the label. Setting this option 113enables this behaviour to allow interoperability with such broken 114implementations. Please note that setting this option breaks interoperability 115with correct implementations. This option only applies to DTLS over SCTP. 116 117=back 118 119All modes are off by default except for SSL_MODE_AUTO_RETRY which is on by 120default since 1.1.1. 121 122=head1 RETURN VALUES 123 124SSL_CTX_set_mode() and SSL_set_mode() return the new mode bit mask 125after adding B<mode>. 126 127SSL_CTX_get_mode() and SSL_get_mode() return the current bit mask. 128 129=head1 SEE ALSO 130 131L<ssl(7)>, L<SSL_read_ex(3)>, L<SSL_read(3)>, L<SSL_write_ex(3)> or 132L<SSL_write(3)>, L<SSL_get_error(3)> 133 134=head1 HISTORY 135 136SSL_MODE_ASYNC was added in OpenSSL 1.1.0. 137 138=head1 COPYRIGHT 139 140Copyright 2001-2020 The OpenSSL Project Authors. All Rights Reserved. 141 142Licensed under the OpenSSL license (the "License"). You may not use 143this file except in compliance with the License. You can obtain a copy 144in the file LICENSE in the source distribution or at 145L<https://www.openssl.org/source/license.html>. 146 147=cut 148