xref: /freebsd/crypto/openssl/doc/man3/BIO_s_bio.pod (revision 02e9120893770924227138ba49df1edb3896112a)
1=pod
2
3=head1 NAME
4
5BIO_s_bio, BIO_make_bio_pair, BIO_destroy_bio_pair, BIO_shutdown_wr,
6BIO_set_write_buf_size, BIO_get_write_buf_size, BIO_new_bio_pair,
7BIO_get_write_guarantee, BIO_ctrl_get_write_guarantee, BIO_get_read_request,
8BIO_ctrl_get_read_request, BIO_ctrl_reset_read_request - BIO pair BIO
9
10=head1 SYNOPSIS
11
12 #include <openssl/bio.h>
13
14 const BIO_METHOD *BIO_s_bio(void);
15
16 int BIO_make_bio_pair(BIO *b1, BIO *b2);
17 int BIO_destroy_bio_pair(BIO *b);
18 int BIO_shutdown_wr(BIO *b);
19
20 int BIO_set_write_buf_size(BIO *b, long size);
21 size_t BIO_get_write_buf_size(BIO *b, long size);
22
23 int BIO_new_bio_pair(BIO **bio1, size_t writebuf1, BIO **bio2, size_t writebuf2);
24
25 int BIO_get_write_guarantee(BIO *b);
26 size_t BIO_ctrl_get_write_guarantee(BIO *b);
27 int BIO_get_read_request(BIO *b);
28 size_t BIO_ctrl_get_read_request(BIO *b);
29 int BIO_ctrl_reset_read_request(BIO *b);
30
31=head1 DESCRIPTION
32
33BIO_s_bio() returns the method for a BIO pair. A BIO pair is a pair of source/sink
34BIOs where data written to either half of the pair is buffered and can be read from
35the other half. Both halves must usually by handled by the same application thread
36since no locking is done on the internal data structures.
37
38Since BIO chains typically end in a source/sink BIO it is possible to make this
39one half of a BIO pair and have all the data processed by the chain under application
40control.
41
42One typical use of BIO pairs is to place TLS/SSL I/O under application control, this
43can be used when the application wishes to use a non standard transport for
44TLS/SSL or the normal socket routines are inappropriate.
45
46Calls to BIO_read_ex() will read data from the buffer or request a retry if no
47data is available.
48
49Calls to BIO_write_ex() will place data in the buffer or request a retry if the
50buffer is full.
51
52The standard calls BIO_ctrl_pending() and BIO_ctrl_wpending() can be used to
53determine the amount of pending data in the read or write buffer.
54
55BIO_reset() clears any data in the write buffer.
56
57BIO_make_bio_pair() joins two separate BIOs into a connected pair.
58
59BIO_destroy_pair() destroys the association between two connected BIOs. Freeing
60up any half of the pair will automatically destroy the association.
61
62BIO_shutdown_wr() is used to close down a BIO B<b>. After this call no further
63writes on BIO B<b> are allowed (they will return an error). Reads on the other
64half of the pair will return any pending data or EOF when all pending data has
65been read.
66
67BIO_set_write_buf_size() sets the write buffer size of BIO B<b> to B<size>.
68If the size is not initialized a default value is used. This is currently
6917K, sufficient for a maximum size TLS record.
70
71BIO_get_write_buf_size() returns the size of the write buffer.
72
73BIO_new_bio_pair() combines the calls to BIO_new(), BIO_make_bio_pair() and
74BIO_set_write_buf_size() to create a connected pair of BIOs B<bio1>, B<bio2>
75with write buffer sizes B<writebuf1> and B<writebuf2>. If either size is
76zero then the default size is used.  BIO_new_bio_pair() does not check whether
77B<bio1> or B<bio2> do point to some other BIO, the values are overwritten,
78BIO_free() is not called.
79
80BIO_get_write_guarantee() and BIO_ctrl_get_write_guarantee() return the maximum
81length of data that can be currently written to the BIO. Writes larger than this
82value will return a value from BIO_write_ex() less than the amount requested or
83if the buffer is full request a retry. BIO_ctrl_get_write_guarantee() is a
84function whereas BIO_get_write_guarantee() is a macro.
85
86BIO_get_read_request() and BIO_ctrl_get_read_request() return the
87amount of data requested, or the buffer size if it is less, if the
88last read attempt at the other half of the BIO pair failed due to an
89empty buffer.  This can be used to determine how much data should be
90written to the BIO so the next read will succeed: this is most useful
91in TLS/SSL applications where the amount of data read is usually
92meaningful rather than just a buffer size. After a successful read
93this call will return zero.  It also will return zero once new data
94has been written satisfying the read request or part of it.
95Note that BIO_get_read_request() never returns an amount larger
96than that returned by BIO_get_write_guarantee().
97
98BIO_ctrl_reset_read_request() can also be used to reset the value returned by
99BIO_get_read_request() to zero.
100
101=head1 NOTES
102
103Both halves of a BIO pair should be freed. That is even if one half is implicit
104freed due to a BIO_free_all() or SSL_free() call the other half needs to be freed.
105
106When used in bidirectional applications (such as TLS/SSL) care should be taken to
107flush any data in the write buffer. This can be done by calling BIO_pending()
108on the other half of the pair and, if any data is pending, reading it and sending
109it to the underlying transport. This must be done before any normal processing
110(such as calling select() ) due to a request and BIO_should_read() being true.
111
112To see why this is important consider a case where a request is sent using
113BIO_write_ex() and a response read with BIO_read_ex(), this can occur during an
114TLS/SSL handshake for example. BIO_write_ex() will succeed and place data in the
115write buffer. BIO_read_ex() will initially fail and BIO_should_read() will be
116true. If the application then waits for data to be available on the underlying
117transport before flushing the write buffer it will never succeed because the
118request was never sent!
119
120BIO_eof() is true if no data is in the peer BIO and the peer BIO has been
121shutdown.
122
123BIO_make_bio_pair(), BIO_destroy_bio_pair(), BIO_shutdown_wr(),
124BIO_set_write_buf_size(), BIO_get_write_buf_size(),
125BIO_get_write_guarantee(), and BIO_get_read_request() are implemented
126as macros.
127
128=head1 RETURN VALUES
129
130BIO_new_bio_pair() returns 1 on success, with the new BIOs available in
131B<bio1> and B<bio2>, or 0 on failure, with NULL pointers stored into the
132locations for B<bio1> and B<bio2>. Check the error stack for more information.
133
134[XXXXX: More return values need to be added here]
135
136=head1 EXAMPLES
137
138The BIO pair can be used to have full control over the network access of an
139application. The application can call select() on the socket as required
140without having to go through the SSL-interface.
141
142 BIO *internal_bio, *network_bio;
143
144 ...
145 BIO_new_bio_pair(&internal_bio, 0, &network_bio, 0);
146 SSL_set_bio(ssl, internal_bio, internal_bio);
147 SSL_operations(); /* e.g. SSL_read and SSL_write */
148 ...
149
150 application |   TLS-engine
151    |        |
152    +----------> SSL_operations()
153             |     /\    ||
154             |     ||    \/
155             |   BIO-pair (internal_bio)
156             |   BIO-pair (network_bio)
157             |     ||     /\
158             |     \/     ||
159    +-----------< BIO_operations()
160    |        |
161    |        |
162   socket
163
164  ...
165  SSL_free(ssl);                /* implicitly frees internal_bio */
166  BIO_free(network_bio);
167  ...
168
169As the BIO pair will only buffer the data and never directly access the
170connection, it behaves nonblocking and will return as soon as the write
171buffer is full or the read buffer is drained. Then the application has to
172flush the write buffer and/or fill the read buffer.
173
174Use the BIO_ctrl_pending(), to find out whether data is buffered in the BIO
175and must be transferred to the network. Use BIO_ctrl_get_read_request() to
176find out, how many bytes must be written into the buffer before the
177SSL_operation() can successfully be continued.
178
179=head1 WARNINGS
180
181As the data is buffered, SSL_operation() may return with an ERROR_SSL_WANT_READ
182condition, but there is still data in the write buffer. An application must
183not rely on the error value of SSL_operation() but must assure that the
184write buffer is always flushed first. Otherwise a deadlock may occur as
185the peer might be waiting for the data before being able to continue.
186
187=head1 SEE ALSO
188
189L<SSL_set_bio(3)>, L<ssl(7)>, L<bio(7)>,
190L<BIO_should_retry(3)>, L<BIO_read_ex(3)>
191
192=head1 COPYRIGHT
193
194Copyright 2000-2020 The OpenSSL Project Authors. All Rights Reserved.
195
196Licensed under the Apache License 2.0 (the "License").  You may not use
197this file except in compliance with the License.  You can obtain a copy
198in the file LICENSE in the source distribution or at
199L<https://www.openssl.org/source/license.html>.
200
201=cut
202