mirror/openssl
2
0
Fork 0
mirror of https://github.com/openssl/openssl.git synced 2024-12-15 06:01:37 +08:00
Code Issues Packages Projects Releases Wiki Activity
61e033308b
openssl/doc/man3/ASYNC_WAIT_CTX_new.pod

145 lines
7.2 KiB
Plaintext
Raw Normal View History

Refactor the async wait fd logic Implementation experience has shown that the original plan for async wait fds was too simplistic. Originally the async logic created a pipe internally and user/engine code could then get access to it via API calls. It is more flexible if the engine is able to create its own fd and provide it to the async code. Another issue is that there can be a lot of churn in the fd value within the context of (say) a single SSL connection leading to continually adding and removing fds from (say) epoll. It is better if we can provide some stability of the fd value across a whole SSL connection. This is problematic because an engine has no concept of an SSL connection. This commit refactors things to introduce an ASYNC_WAIT_CTX which acts as a proxy for an SSL connection down at the engine layer. Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-01-25 23:28:57 +08:00
=pod
=head1 NAME
ASYNC_WAIT_CTX_new, ASYNC_WAIT_CTX_free, ASYNC_WAIT_CTX_set_wait_fd,
ASYNC_WAIT_CTX_get_fd, ASYNC_WAIT_CTX_get_all_fds,
ASYNC_WAIT_CTX_get_changed_fds, ASYNC_WAIT_CTX_clear_fd - functions to manage
waiting for asynchronous jobs to complete
=head1 SYNOPSIS
#include <openssl/async.h>
ASYNC_WAIT_CTX *ASYNC_WAIT_CTX_new(void);
void ASYNC_WAIT_CTX_free(ASYNC_WAIT_CTX *ctx);
int ASYNC_WAIT_CTX_set_wait_fd(ASYNC_WAIT_CTX *ctx, const void *key,
OSSL_ASYNC_FD fd,
void *custom_data,
void (*cleanup)(ASYNC_WAIT_CTX *, const void *,
doc/man3: reformat the function prototypes in the synopses I tried hard to keep the lines at 80 characters or less, but in a few cases I had to punt and just indented the subsequent lines by 4 spaces. A few well-placed typedefs for callback functions would really help, but these would be part of the API, so that's probably for later. I also took the liberty of inserting empty lines in overlong blocks to provide some visual space. Reviewed-by: Rich Salz <rsalz@openssl.org> Reviewed-by: Matt Caswell <matt@openssl.org> (Merged from https://github.com/openssl/openssl/pull/1956)
2017-01-21 02:58:49 +08:00
OSSL_ASYNC_FD, void *));
Refactor the async wait fd logic Implementation experience has shown that the original plan for async wait fds was too simplistic. Originally the async logic created a pipe internally and user/engine code could then get access to it via API calls. It is more flexible if the engine is able to create its own fd and provide it to the async code. Another issue is that there can be a lot of churn in the fd value within the context of (say) a single SSL connection leading to continually adding and removing fds from (say) epoll. It is better if we can provide some stability of the fd value across a whole SSL connection. This is problematic because an engine has no concept of an SSL connection. This commit refactors things to introduce an ASYNC_WAIT_CTX which acts as a proxy for an SSL connection down at the engine layer. Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-01-25 23:28:57 +08:00
int ASYNC_WAIT_CTX_get_fd(ASYNC_WAIT_CTX *ctx, const void *key,
OSSL_ASYNC_FD *fd, void **custom_data);
int ASYNC_WAIT_CTX_get_all_fds(ASYNC_WAIT_CTX *ctx, OSSL_ASYNC_FD *fd,
size_t *numfds);
int ASYNC_WAIT_CTX_get_changed_fds(ASYNC_WAIT_CTX *ctx, OSSL_ASYNC_FD *addfd,
size_t *numaddfds, OSSL_ASYNC_FD *delfd,
size_t *numdelfds);
int ASYNC_WAIT_CTX_clear_fd(ASYNC_WAIT_CTX *ctx, const void *key);
=head1 DESCRIPTION
For an overview of how asynchronous operations are implemented in OpenSSL see
L<ASYNC_start_job(3)>. An ASYNC_WAIT_CTX object represents an asynchronous
"session", i.e. a related set of crypto operations. For example in SSL terms
this would have a one-to-one correspondence with an SSL connection.
Application code must create an ASYNC_WAIT_CTX using the ASYNC_WAIT_CTX_new()
function prior to calling ASYNC_start_job() (see L<ASYNC_start_job(3)>). When
the job is started it is associated with the ASYNC_WAIT_CTX for the duration of
that job. An ASYNC_WAIT_CTX should only be used for one ASYNC_JOB at any one
time, but can be reused after an ASYNC_JOB has finished for a subsequent
ASYNC_JOB. When the session is complete (e.g. the SSL connection is closed),
application code cleans up with ASYNC_WAIT_CTX_free().
ASYNC_WAIT_CTXs can have "wait" file descriptors associated with them. Calling
ASYNC_WAIT_CTX_get_all_fds() and passing in a pointer to an ASYNC_WAIT_CTX in
the B<ctx> parameter will return the wait file descriptors associated with that
job in B<*fd>. The number of file descriptors returned will be stored in
B<*numfds>. It is the caller's responsibility to ensure that sufficient memory
has been allocated in B<*fd> to receive all the file descriptors. Calling
ASYNC_WAIT_CTX_get_all_fds() with a NULL B<fd> value will return no file
descriptors but will still populate B<*numfds>. Therefore application code is
typically expected to call this function twice: once to get the number of fds,
and then again when sufficient memory has been allocated. If only one
Fix spelling in pod files Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Rich Salz <rsalz@openssl.org>
2016-05-02 01:52:58 +08:00
asynchronous engine is being used then normally this call will only ever return
Refactor the async wait fd logic Implementation experience has shown that the original plan for async wait fds was too simplistic. Originally the async logic created a pipe internally and user/engine code could then get access to it via API calls. It is more flexible if the engine is able to create its own fd and provide it to the async code. Another issue is that there can be a lot of churn in the fd value within the context of (say) a single SSL connection leading to continually adding and removing fds from (say) epoll. It is better if we can provide some stability of the fd value across a whole SSL connection. This is problematic because an engine has no concept of an SSL connection. This commit refactors things to introduce an ASYNC_WAIT_CTX which acts as a proxy for an SSL connection down at the engine layer. Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-01-25 23:28:57 +08:00
one fd. If multiple asynchronous engines are being used then more could be
returned.
Fix typo in ASYNC_WAIT_CTX_new.pod doc For the function that get the changed fds, it should be 'ASYNC_WAIT_CTX_get_changed_fds()' instead of 'ASYNC_WAIT_CTX_fds_have_changed()'. CLA: trivial Signed-off-by: Paul Yang <paulyang.inf@gmail.com> Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Matt Caswell <matt@openssl.org> (Merged from https://github.com/openssl/openssl/pull/2966)
2017-03-16 16:58:30 +08:00
The function ASYNC_WAIT_CTX_get_changed_fds() can be used to detect if any fds
Refactor the async wait fd logic Implementation experience has shown that the original plan for async wait fds was too simplistic. Originally the async logic created a pipe internally and user/engine code could then get access to it via API calls. It is more flexible if the engine is able to create its own fd and provide it to the async code. Another issue is that there can be a lot of churn in the fd value within the context of (say) a single SSL connection leading to continually adding and removing fds from (say) epoll. It is better if we can provide some stability of the fd value across a whole SSL connection. This is problematic because an engine has no concept of an SSL connection. This commit refactors things to introduce an ASYNC_WAIT_CTX which acts as a proxy for an SSL connection down at the engine layer. Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-01-25 23:28:57 +08:00
have changed since the last call time ASYNC_start_job() returned an ASYNC_PAUSE
result (or since the ASYNC_WAIT_CTX was created if no ASYNC_PAUSE result has
been received). The B<numaddfds> and B<numdelfds> parameters will be populated
with the number of fds added or deleted respectively. B<*addfd> and B<*delfd>
will be populated with the list of added and deleted fds respectively. Similarly
to ASYNC_WAIT_CTX_get_all_fds() either of these can be NULL, but if they are not
NULL then the caller is responsible for ensuring sufficient memory is allocated.
Implementors of async aware code (e.g. engines) are encouraged to return a
stable fd for the lifetime of the ASYNC_WAIT_CTX in order to reduce the "churn"
of regularly changing fds - although no guarantees of this are provided to
applications.
Applications can wait for the file descriptor to be ready for "read" using a
system function call such as select or poll (being ready for "read" indicates
that the job should be resumed). If no file descriptor is made available then an
application will have to periodically "poll" the job by attempting to restart it
to see if it is ready to continue.
Async aware code (e.g. engines) can get the current ASYNC_WAIT_CTX from the job
Fix broken link to ASYNC_get_wait_ctx and rewrap the paragraph Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Rich Salz <rsalz@openssl.org> (Merged from https://github.com/openssl/openssl/pull/1745)
2016-10-18 17:53:14 +08:00
via L<ASYNC_get_wait_ctx(3)> and provide a file descriptor to use for waiting
on by calling ASYNC_WAIT_CTX_set_wait_fd(). Typically this would be done by an
engine immediately prior to calling ASYNC_pause_job() and not by end user code.
An existing association with a file descriptor can be obtained using
Refactor the async wait fd logic Implementation experience has shown that the original plan for async wait fds was too simplistic. Originally the async logic created a pipe internally and user/engine code could then get access to it via API calls. It is more flexible if the engine is able to create its own fd and provide it to the async code. Another issue is that there can be a lot of churn in the fd value within the context of (say) a single SSL connection leading to continually adding and removing fds from (say) epoll. It is better if we can provide some stability of the fd value across a whole SSL connection. This is problematic because an engine has no concept of an SSL connection. This commit refactors things to introduce an ASYNC_WAIT_CTX which acts as a proxy for an SSL connection down at the engine layer. Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-01-25 23:28:57 +08:00
ASYNC_WAIT_CTX_get_fd() and cleared using ASYNC_WAIT_CTX_clear_fd(). Both of
Fix broken link to ASYNC_get_wait_ctx and rewrap the paragraph Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Rich Salz <rsalz@openssl.org> (Merged from https://github.com/openssl/openssl/pull/1745)
2016-10-18 17:53:14 +08:00
these functions requires a B<key> value which is unique to the async aware
code. This could be any unique value but a good candidate might be the
B<ENGINE *> for the engine. The B<custom_data> parameter can be any value, and
will be returned in a subsequent call to ASYNC_WAIT_CTX_get_fd(). The
Refactor the async wait fd logic Implementation experience has shown that the original plan for async wait fds was too simplistic. Originally the async logic created a pipe internally and user/engine code could then get access to it via API calls. It is more flexible if the engine is able to create its own fd and provide it to the async code. Another issue is that there can be a lot of churn in the fd value within the context of (say) a single SSL connection leading to continually adding and removing fds from (say) epoll. It is better if we can provide some stability of the fd value across a whole SSL connection. This is problematic because an engine has no concept of an SSL connection. This commit refactors things to introduce an ASYNC_WAIT_CTX which acts as a proxy for an SSL connection down at the engine layer. Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-01-25 23:28:57 +08:00
ASYNC_WAIT_CTX_set_wait_fd() function also expects a pointer to a "cleanup"
Fix broken link to ASYNC_get_wait_ctx and rewrap the paragraph Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Rich Salz <rsalz@openssl.org> (Merged from https://github.com/openssl/openssl/pull/1745)
2016-10-18 17:53:14 +08:00
routine. This can be NULL but if provided will automatically get called when
the ASYNC_WAIT_CTX is freed, and gives the engine the opportunity to close the
fd or any other resources. Note: The "cleanup" routine does not get called if
the fd is cleared directly via a call to ASYNC_WAIT_CTX_clear_fd().
Refactor the async wait fd logic Implementation experience has shown that the original plan for async wait fds was too simplistic. Originally the async logic created a pipe internally and user/engine code could then get access to it via API calls. It is more flexible if the engine is able to create its own fd and provide it to the async code. Another issue is that there can be a lot of churn in the fd value within the context of (say) a single SSL connection leading to continually adding and removing fds from (say) epoll. It is better if we can provide some stability of the fd value across a whole SSL connection. This is problematic because an engine has no concept of an SSL connection. This commit refactors things to introduce an ASYNC_WAIT_CTX which acts as a proxy for an SSL connection down at the engine layer. Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-01-25 23:28:57 +08:00
An example of typical usage might be an async capable engine. User code would
initiate cryptographic operations. The engine would initiate those operations
asynchronously and then call ASYNC_WAIT_CTX_set_wait_fd() followed by
ASYNC_pause_job() to return control to the user code. The user code can then
perform other tasks or wait for the job to be ready by calling "select" or other
similar function on the wait file descriptor. The engine can signal to the user
code that the job should be resumed by making the wait file descriptor
"readable". Once resumed the engine should clear the wake signal on the wait
file descriptor.
=head1 RETURN VALUES
ASYNC_WAIT_CTX_new() returns a pointer to the newly allocated ASYNC_WAIT_CTX or
NULL on error.
ASYNC_WAIT_CTX_set_wait_fd, ASYNC_WAIT_CTX_get_fd, ASYNC_WAIT_CTX_get_all_fds,
ASYNC_WAIT_CTX_get_changed_fds and ASYNC_WAIT_CTX_clear_fd all return 1 on
success or 0 on error.
include/openssl: don't include <windows.h> in public headers. If application uses any of Windows-specific interfaces, make it application developer's respondibility to include <windows.h>. Rationale is that <windows.h> is quite "toxic" and is sensitive to inclusion order (most notably in relation to <winsock2.h>). It's only natural to give complete control to the application developer. Reviewed-by: Rich Salz <rsalz@openssl.org> Reviewed-by: Matt Caswell <matt@openssl.org>
2016-06-26 19:40:15 +08:00
=head1 NOTES
On Windows platforms the openssl/async.h header is dependent on some
of the types customarily made available by including windows.h. The
application developer is likely to require control over when the latter
is included, commonly as one of the first included headers. Therefore
it is defined as an application developer's responsibility to include
windows.h prior to async.h.
Refactor the async wait fd logic Implementation experience has shown that the original plan for async wait fds was too simplistic. Originally the async logic created a pipe internally and user/engine code could then get access to it via API calls. It is more flexible if the engine is able to create its own fd and provide it to the async code. Another issue is that there can be a lot of churn in the fd value within the context of (say) a single SSL connection leading to continually adding and removing fds from (say) epoll. It is better if we can provide some stability of the fd value across a whole SSL connection. This is problematic because an engine has no concept of an SSL connection. This commit refactors things to introduce an ASYNC_WAIT_CTX which acts as a proxy for an SSL connection down at the engine layer. Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-01-25 23:28:57 +08:00
=head1 SEE ALSO
Fix referenses in section 3 manuals Reviewed-by: Rich Salz <rsalz@openssl.org> (Merged from https://github.com/openssl/openssl/pull/1900)
2016-11-11 16:33:09 +08:00
L<crypto(7)>, L<ASYNC_start_job(3)>
Refactor the async wait fd logic Implementation experience has shown that the original plan for async wait fds was too simplistic. Originally the async logic created a pipe internally and user/engine code could then get access to it via API calls. It is more flexible if the engine is able to create its own fd and provide it to the async code. Another issue is that there can be a lot of churn in the fd value within the context of (say) a single SSL connection leading to continually adding and removing fds from (say) epoll. It is better if we can provide some stability of the fd value across a whole SSL connection. This is problematic because an engine has no concept of an SSL connection. This commit refactors things to introduce an ASYNC_WAIT_CTX which acts as a proxy for an SSL connection down at the engine layer. Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-01-25 23:28:57 +08:00
=head1 HISTORY
man: harmonize the various formulations in the HISTORY sections While stereotyped repetitions are frowned upon in literature, they serve a useful purpose in manual pages, because it is easier for the user to find certain information if it is always presented in the same way. For that reason, this commit harmonizes the varying formulations in the HISTORY section about which functions, flags, etc. were added in which OpenSSL version. It also attempts to make the pod files more grep friendly by avoiding to insert line breaks between the symbol names and the corresponding version number in which they were introduced (wherever possible). Some punctuation and typographical errors were fixed on the way. Reviewed-by: Tim Hudson <tjh@openssl.org> (Merged from https://github.com/openssl/openssl/pull/7854)
2018-12-09 08:02:36 +08:00
ASYNC_WAIT_CTX_new(), ASYNC_WAIT_CTX_free(), ASYNC_WAIT_CTX_set_wait_fd(),
ASYNC_WAIT_CTX_get_fd(), ASYNC_WAIT_CTX_get_all_fds(),
ASYNC_WAIT_CTX_get_changed_fds() and ASYNC_WAIT_CTX_clear_fd()
were added in OpenSSL 1.1.0.
Refactor the async wait fd logic Implementation experience has shown that the original plan for async wait fds was too simplistic. Originally the async logic created a pipe internally and user/engine code could then get access to it via API calls. It is more flexible if the engine is able to create its own fd and provide it to the async code. Another issue is that there can be a lot of churn in the fd value within the context of (say) a single SSL connection leading to continually adding and removing fds from (say) epoll. It is better if we can provide some stability of the fd value across a whole SSL connection. This is problematic because an engine has no concept of an SSL connection. This commit refactors things to introduce an ASYNC_WAIT_CTX which acts as a proxy for an SSL connection down at the engine layer. Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-01-25 23:28:57 +08:00
Add copyright to manpages Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-05-18 23:44:05 +08:00
=head1 COPYRIGHT
Copyright 2016 The OpenSSL Project Authors. All Rights Reserved.
Following the license change, modify the boilerplates in doc/man3/ [skip ci] Reviewed-by: Matt Caswell <matt@openssl.org> (Merged from https://github.com/openssl/openssl/pull/7829)
2018-12-06 21:04:44 +08:00
Licensed under the Apache License 2.0 (the "License"). You may not use
Add copyright to manpages Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-05-18 23:44:05 +08:00
this file except in compliance with the License. You can obtain a copy
in the file LICENSE in the source distribution or at
L<https://www.openssl.org/source/license.html>.
=cut
Reference in New Issue Copy Permalink