Add pipelining documentation

=pod

=head1 NAME

SSL_CTX_set_max_send_fragment, SSL_set_max_send_fragment,

SSL_CTX_set_split_send_fragment, SSL_set_split_send_fragment,

SSL_CTX_set_max_pipelines, SSL_set_max_pipelines,

SSL_CTX_set_default_read_buffer_len, SSL_set_default_read_buffer_len - Control

fragment sizes and pipelining operations

=head1 SYNOPSIS

 #include <openssl/ssl.h>

 # define SSL_CTX_set_max_send_fragment(ctx,m) \

         SSL_CTX_ctrl(ctx,SSL_CTRL_SET_MAX_SEND_FRAGMENT,m,NULL)

 # define SSL_set_max_send_fragment(ssl,m) \

         SSL_ctrl(ssl,SSL_CTRL_SET_MAX_SEND_FRAGMENT,m,NULL)

 # define SSL_CTX_set_max_pipelines(ctx,m) \

         SSL_CTX_ctrl(ctx,SSL_CTRL_SET_MAX_PIPELINES,m,NULL)

 # define SSL_set_max_pipelines(ssl,m) \

         SSL_ctrl(ssl,SSL_CTRL_SET_MAX_PIPELINES,m,NULL)

 # define SSL_CTX_set_split_send_fragment(ctx,m) \

         SSL_CTX_ctrl(ctx,SSL_CTRL_SET_SPLIT_SEND_FRAGMENT,m,NULL)

 # define SSL_set_split_send_fragment(ssl,m) \

         SSL_ctrl(ssl,SSL_CTRL_SET_SPLIT_SEND_FRAGMENT,m,NULL)

 void SSL_CTX_set_default_read_buffer_len(SSL_CTX *ctx, size_t len);

 void SSL_set_default_read_buffer_len(SSL *s, size_t len);

=head1 DESCRIPTION

Some engines are able to process multiple simultaneous crypto operations. This

capability could be utilised to parallelise the processing of a single

connection. For example a single write can be split into multiple records and

each one encrypted independently and in parallel. Note: this will only work in

TLS1.1+. There is no support in SSLv3, TLSv1.0 or DTLS (any version). This

capability is known as "pipelining" within OpenSSL.

In order to benefit from the pipelining capability. You need to have an engine

that provides ciphers that support this. The OpenSSL "dasync" engine provides

AES128-SHA based ciphers that have this capability. However these are for

development and test purposes only.

SSL_CTX_set_max_send_fragment() and SSL_set_max_send_fragment() set the

B<max_send_fragment> parameter for SSL_CTX and SSL objects respectively. This

value restricts the amount of plaintext bytes that will be sent in any one

SSL/TLS record. By default its value is SSL3_RT_MAX_PLAIN_LENGTH (16384). These

functions will only accept a value in the range 512 - SSL3_RT_MAX_PLAIN_LENGTH.

SSL_CTX_set_max_pipelines() and SSL_set_max_pipelines() set the maximum number

of pipelines that will be used at any one time. This value applies to both

"read" pipelining and "write" pipelining. By default only one pipeline will be

used (i.e. normal non-parallel operation). The number of pipelines set must be

in the range 1 - SSL_MAX_PIPELINES (32). Setting this to a value > 1 will also

automatically turn on "read_ahead" (see L<SSL_CTX_set_read_ahead(3)>). This is

explained further below. OpenSSL will only every use more than one pipeline if

a ciphersuite is negotiated that uses a pipeline capable cipher provided by an

engine.

Pipelining operates slighly differently for reading encrypted data compared to

writing encrypted data. SSL_CTX_set_split_send_fragment() and

SSL_set_split_send_fragment() define how data is split up into pipelines when

writing encrypted data. The number of pipelines used will be determined by the

amount of data provided to the SSL_write() call divided by

B<split_send_fragment>.

For example if B<split_send_fragment> is set to 2000 and B<max_pipelines> is 4

then:

SSL_write called with 0-2000 bytes == 1 pipeline used

SSL_write called with 2001-4000 bytes == 2 pipelines used

SSL_write called with 4001-6000 bytes == 3 pipelines used

SSL_write called with 6001+ bytes == 4 pipelines used

B<split_send_fragment> must always be less than or equal to

B<max_send_fragment>. By default it is set to be equal to B<max_send_fragment>.

This will mean that the same number of records will always be created as would

have been created in the non-parallel case, although the data will be

apportioned differently. In the parallel case data will be spread equally

between the pipelines.

Read pipelining is controlled in a slightly different way than with write

pipelining. While reading we are constrained by the number of records that the

peer (and the network) can provide to us in one go. The more records we can get

in one go the more opportunity we have to parallelise the processing. As noted

above when setting B<max_pipelines> to a value greater than one, B<read_ahead>

is automatically set. The B<read_ahead> parameter causes OpenSSL to attempt to

read as much data into the read buffer as the network can provide and will fit

into the buffer. Without this set data is read into the read buffer one record

at a time. The more data that can be read, the more opportunity there is for

parallelising the processing at the cost of increased memory overhead per

connection.

The SSL_CTX_set_default_read_buffer_len() and SSL_set_default_read_buffer_len()

functions control the size of the read buffer that will be used. The B<len>

parameter sets the size of the buffer. The value will only be used if it is

greater than the default that would have been used anyway. The normal default

value depends on a number of factors but it will be at least

SSL3_RT_MAX_PLAIN_LENGTH + SSL3_RT_MAX_ENCRYPTED_OVERHEAD (16704) bytes.

=head1 RETURN VALUES

All non-void functions return 1 on success and 0 on failure.

=head1 NOTES

With the exception of SSL_CTX_set_default_read_buffer_len() and

SSL_set_default_read_buffer_len() all these functions are implemented using

macros.

=head1 HISTORY

The SSL_CTX_set_max_pipelines(), SSL_set_max_pipelines(),

SSL_CTX_set_split_send_fragment(), SSL_set_split_send_fragment(),

SSL_CTX_set_default_read_buffer_len() and  SSL_set_default_read_buffer_len()

functions were added in OpenSSL 1.1.0.

=head1 SEE ALSO

L<SSL_CTX_set_read_ahead(3)>

=cut