mbedtls/library/mps_reader.h

359 lines
17 KiB
C
Raw Normal View History

/*
* Copyright The Mbed TLS Contributors
* SPDX-License-Identifier: Apache-2.0
*
* Licensed under the Apache License, Version 2.0 (the "License"); you may
* not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*
* This file is part of mbed TLS (https://tls.mbed.org)
*/
/**
* \file reader.h
*
* \brief This file defines reader objects, which together with their
* sibling writer objects form the basis for the communication
* between the various layers of the Mbed TLS messaging stack,
* as well as the communication between the messaging stack and
* the (D)TLS handshake protocol implementation.
*
* Readers provide a means of transferring incoming data from
* a 'producer' providing it in chunks of arbitrary size, to
* a 'consumer' which fetches and processes it in chunks of
* again arbitrary, and potentially different, size.
*
Don't invalidate MPS reader buffers upon commit call Previously, the semantics of mbedtls_mps_reader_commit() was to invalidate all buffers previously fetched via mbedtls_mps_reader_get(), forbidding any further use by the 'consumer'. This was in fact a necessary constraint for the current implementation, which did some memory moving in mbedtls_mps_reader_commit(). This commit simplifies the reader's semantics and implementation in the following way: - API: A call to mbedtls_mps_reader_commit() does no longer invalidate the buffers previously obtained via mbedtls_mps_reader_get(). Instead, they can continue to be used until mbedtls_mps_reader_reclaim() is called. Calling mbedtls_mps_reader_commit() now only sets a marker indicating which parts of the data received through mbedtls_mps_reader_get() need not be backed up once mbedtls_mps_reader_reclaim() is called. Allowing the user to call mbedtls_mbedtls_reader_commit() multiple times before mbedtls_mps_reader_reclaim() is mere convenience: We'd get exactly the same functionality if instead of mbedtls_mps_reader_commit(), there was an additional argument to mbedtls_mps_reader_reclaim() indicating how much data to retain. However, the present design is more convenient for the user and doesn't appear to introduce any unnecessary complexity (anymore), so we stick with it for now. - Implementation: mbedtls_mps_reader_commit() is now a 1-liner, setting the 'commit-marker', but doing nothing else. Instead, the complexity of mbedtls_mp_reader_reclaim() slightly increases because it has to deal with creating backups from both the accumulator and the current fragment. In the previous implementation, which shifted the accumulator content with every call to mbedtls_mps_reader_commit(), only the backup from the fragment was necessary; with the new implementation which doesn't shift anything in mbedtls_mps_reader_commit(), we need to do the accumulator shift in mbedtls_mps_reader_reclaim(). Signed-off-by: Hanno Becker <hanno.becker@arm.com>
2021-02-08 07:54:30 +01:00
* Readers can thus be seen as datagram-to-stream converters,
* and they abstract away the following two tasks from the user:
* 1. The pointer arithmetic of stepping through a producer-
* provided chunk in smaller chunks.
* 2. The merging of incoming data chunks in case the
* consumer requests data in larger chunks than what the
* producer provides.
*
* The basic abstract flow of operation is the following:
* - Initially, the reader is in 'producing mode'.
* - The producer hands an incoming data buffer to the reader,
* moving it from 'producing' to 'consuming' mode.
* - The consumer subsequently fetches and processes the buffer
* content. Once that's done -- or partially done and a consumer's
* requests can't be fulfilled -- the producer revokes the reader's
* access to the incoming data buffer, putting the reader back to
* producing mode.
* - The producer subsequently gathers more incoming data and hands
* it to reader until the latter switches back to consuming mode
* if enough data is available for the last consumer request to
* be satisfiable.
* - Repeat the above.
*
Don't invalidate MPS reader buffers upon commit call Previously, the semantics of mbedtls_mps_reader_commit() was to invalidate all buffers previously fetched via mbedtls_mps_reader_get(), forbidding any further use by the 'consumer'. This was in fact a necessary constraint for the current implementation, which did some memory moving in mbedtls_mps_reader_commit(). This commit simplifies the reader's semantics and implementation in the following way: - API: A call to mbedtls_mps_reader_commit() does no longer invalidate the buffers previously obtained via mbedtls_mps_reader_get(). Instead, they can continue to be used until mbedtls_mps_reader_reclaim() is called. Calling mbedtls_mps_reader_commit() now only sets a marker indicating which parts of the data received through mbedtls_mps_reader_get() need not be backed up once mbedtls_mps_reader_reclaim() is called. Allowing the user to call mbedtls_mbedtls_reader_commit() multiple times before mbedtls_mps_reader_reclaim() is mere convenience: We'd get exactly the same functionality if instead of mbedtls_mps_reader_commit(), there was an additional argument to mbedtls_mps_reader_reclaim() indicating how much data to retain. However, the present design is more convenient for the user and doesn't appear to introduce any unnecessary complexity (anymore), so we stick with it for now. - Implementation: mbedtls_mps_reader_commit() is now a 1-liner, setting the 'commit-marker', but doing nothing else. Instead, the complexity of mbedtls_mp_reader_reclaim() slightly increases because it has to deal with creating backups from both the accumulator and the current fragment. In the previous implementation, which shifted the accumulator content with every call to mbedtls_mps_reader_commit(), only the backup from the fragment was necessary; with the new implementation which doesn't shift anything in mbedtls_mps_reader_commit(), we need to do the accumulator shift in mbedtls_mps_reader_reclaim(). Signed-off-by: Hanno Becker <hanno.becker@arm.com>
2021-02-08 07:54:30 +01:00
* The abstract states of the reader from the producer's and
* consumer's perspective are as follows:
*
* - From the perspective of the consumer, the state of the
* reader consists of the following:
* - A byte stream representing (concatenation of) the data
* received through calls to mbedtls_mps_reader_get(),
* - A marker within that byte stream indicating which data
* need not be retained when the reader is passed back to
* the producer via mbedtls_mps_reader_reclaim().
* The marker can be set via mbedtls_mps_reader_commit()
* which places it at the end of the current byte stream.
* The consumer need not be aware of the distinction between consumer
* and producer mode, because he only interfaces with the reader
* when the latter is in consuming mode.
*
* - From the perspective of the producer, the reader's state is one of:
* - Attached: The reader is in consuming mode.
* - Unset: No incoming data buffer is currently managed by the reader,
* and all previously handed incoming data buffers have been
* fully processed. More data needs to be fed into the reader
* via mbedtls_mps_reader_feed().
*
* - Accumulating: No incoming data buffer is currently managed by the
* reader, but some data from the previous incoming data
* buffer hasn't been processed yet and is internally
* held back.
* The Attached state belongs to consuming mode, while the Unset and
* Accumulating states belong to producing mode.
*
* Transitioning from the Unset or Accumulating state to Attached is
* done via successful calls to mbedtls_mps_reader_feed(), while
* transitioning from Consuming to either Unset or Accumulating (depending
* on what has been processed) is done via mbedtls_mps_reader_reclaim().
*
* The following diagram depicts the producer-state progression:
*
* +------------------+ reclaim
* | Unset +<-------------------------------------+ get
* +--------|---------+ | +------+
* | | | |
* | | | |
* | feed +---------+---+--+ |
Don't invalidate MPS reader buffers upon commit call Previously, the semantics of mbedtls_mps_reader_commit() was to invalidate all buffers previously fetched via mbedtls_mps_reader_get(), forbidding any further use by the 'consumer'. This was in fact a necessary constraint for the current implementation, which did some memory moving in mbedtls_mps_reader_commit(). This commit simplifies the reader's semantics and implementation in the following way: - API: A call to mbedtls_mps_reader_commit() does no longer invalidate the buffers previously obtained via mbedtls_mps_reader_get(). Instead, they can continue to be used until mbedtls_mps_reader_reclaim() is called. Calling mbedtls_mps_reader_commit() now only sets a marker indicating which parts of the data received through mbedtls_mps_reader_get() need not be backed up once mbedtls_mps_reader_reclaim() is called. Allowing the user to call mbedtls_mbedtls_reader_commit() multiple times before mbedtls_mps_reader_reclaim() is mere convenience: We'd get exactly the same functionality if instead of mbedtls_mps_reader_commit(), there was an additional argument to mbedtls_mps_reader_reclaim() indicating how much data to retain. However, the present design is more convenient for the user and doesn't appear to introduce any unnecessary complexity (anymore), so we stick with it for now. - Implementation: mbedtls_mps_reader_commit() is now a 1-liner, setting the 'commit-marker', but doing nothing else. Instead, the complexity of mbedtls_mp_reader_reclaim() slightly increases because it has to deal with creating backups from both the accumulator and the current fragment. In the previous implementation, which shifted the accumulator content with every call to mbedtls_mps_reader_commit(), only the backup from the fragment was necessary; with the new implementation which doesn't shift anything in mbedtls_mps_reader_commit(), we need to do the accumulator shift in mbedtls_mps_reader_reclaim(). Signed-off-by: Hanno Becker <hanno.becker@arm.com>
2021-02-08 07:54:30 +01:00
* +--------------------------------------> <---+
* | Attached |
* +--------------------------------------> <---+
* | feed, enough data available +---------+---+--+ |
* | to serve previous consumer request | | |
* | | | |
* +--------+---------+ | +------+
* +----> Accumulating |<-------------------------------------+ commit
* | +---+--------------+ reclaim, previous read request
* | | couldn't be fulfilled
* | |
* +--------+
* feed, need more data to serve
* previous consumer request
Don't invalidate MPS reader buffers upon commit call Previously, the semantics of mbedtls_mps_reader_commit() was to invalidate all buffers previously fetched via mbedtls_mps_reader_get(), forbidding any further use by the 'consumer'. This was in fact a necessary constraint for the current implementation, which did some memory moving in mbedtls_mps_reader_commit(). This commit simplifies the reader's semantics and implementation in the following way: - API: A call to mbedtls_mps_reader_commit() does no longer invalidate the buffers previously obtained via mbedtls_mps_reader_get(). Instead, they can continue to be used until mbedtls_mps_reader_reclaim() is called. Calling mbedtls_mps_reader_commit() now only sets a marker indicating which parts of the data received through mbedtls_mps_reader_get() need not be backed up once mbedtls_mps_reader_reclaim() is called. Allowing the user to call mbedtls_mbedtls_reader_commit() multiple times before mbedtls_mps_reader_reclaim() is mere convenience: We'd get exactly the same functionality if instead of mbedtls_mps_reader_commit(), there was an additional argument to mbedtls_mps_reader_reclaim() indicating how much data to retain. However, the present design is more convenient for the user and doesn't appear to introduce any unnecessary complexity (anymore), so we stick with it for now. - Implementation: mbedtls_mps_reader_commit() is now a 1-liner, setting the 'commit-marker', but doing nothing else. Instead, the complexity of mbedtls_mp_reader_reclaim() slightly increases because it has to deal with creating backups from both the accumulator and the current fragment. In the previous implementation, which shifted the accumulator content with every call to mbedtls_mps_reader_commit(), only the backup from the fragment was necessary; with the new implementation which doesn't shift anything in mbedtls_mps_reader_commit(), we need to do the accumulator shift in mbedtls_mps_reader_reclaim(). Signed-off-by: Hanno Becker <hanno.becker@arm.com>
2021-02-08 07:54:30 +01:00
* |
* |
* producing mode | consuming mode
* |
*
*/
#ifndef MBEDTLS_READER_H
#define MBEDTLS_READER_H
#include <stdio.h>
#include "mps_common.h"
#include "mps_error.h"
struct mbedtls_mps_reader;
typedef struct mbedtls_mps_reader mbedtls_mps_reader;
/*
* Structure definitions
*/
struct mbedtls_mps_reader
{
unsigned char *frag; /*!< The fragment of incoming data managed by
* the reader; it is provided to the reader
* through mbedtls_mps_reader_feed(). The reader
* does not own the fragment and does not
* perform any allocation operations on it,
* but does have read and write access to it. */
mbedtls_mps_stored_size_t frag_len;
/*!< The length of the current fragment.
* Must be 0 if \c frag == \c NULL. */
mbedtls_mps_stored_size_t commit;
/*!< The offset of the last commit, relative
* to the first byte in the accumulator.
* This is only used when the reader is in
Don't invalidate MPS reader buffers upon commit call Previously, the semantics of mbedtls_mps_reader_commit() was to invalidate all buffers previously fetched via mbedtls_mps_reader_get(), forbidding any further use by the 'consumer'. This was in fact a necessary constraint for the current implementation, which did some memory moving in mbedtls_mps_reader_commit(). This commit simplifies the reader's semantics and implementation in the following way: - API: A call to mbedtls_mps_reader_commit() does no longer invalidate the buffers previously obtained via mbedtls_mps_reader_get(). Instead, they can continue to be used until mbedtls_mps_reader_reclaim() is called. Calling mbedtls_mps_reader_commit() now only sets a marker indicating which parts of the data received through mbedtls_mps_reader_get() need not be backed up once mbedtls_mps_reader_reclaim() is called. Allowing the user to call mbedtls_mbedtls_reader_commit() multiple times before mbedtls_mps_reader_reclaim() is mere convenience: We'd get exactly the same functionality if instead of mbedtls_mps_reader_commit(), there was an additional argument to mbedtls_mps_reader_reclaim() indicating how much data to retain. However, the present design is more convenient for the user and doesn't appear to introduce any unnecessary complexity (anymore), so we stick with it for now. - Implementation: mbedtls_mps_reader_commit() is now a 1-liner, setting the 'commit-marker', but doing nothing else. Instead, the complexity of mbedtls_mp_reader_reclaim() slightly increases because it has to deal with creating backups from both the accumulator and the current fragment. In the previous implementation, which shifted the accumulator content with every call to mbedtls_mps_reader_commit(), only the backup from the fragment was necessary; with the new implementation which doesn't shift anything in mbedtls_mps_reader_commit(), we need to do the accumulator shift in mbedtls_mps_reader_reclaim(). Signed-off-by: Hanno Becker <hanno.becker@arm.com>
2021-02-08 07:54:30 +01:00
* consuming mode, i.e. \c frag != \c NULL;
* otherwise, its value is \c 0. */
mbedtls_mps_stored_size_t end;
/*!< The offset of the end of the last chunk
* passed to the user through a call to
* mbedtls_mps_reader_get(), relative to the first
* byte in the accumulator.
* This is only used when the reader is in
* consuming mode, i.e. \c frag != \c NULL;
* otherwise, its value is \c 0. */
mbedtls_mps_stored_size_t pending;
/*!< The amount of incoming data missing on the
* last call to mbedtls_mps_reader_get().
* In particular, it is \c 0 if the last call
* was successful.
* If a reader is reclaimed after an
* unsuccessful call to mbedtls_mps_reader_get(),
* this variable is used to have the reader
* remember how much data should be accumulated
* before the reader can be passed back to
* the user again.
* This is only used when the reader is in
* consuming mode, i.e. \c frag != \c NULL;
* otherwise, its value is \c 0. */
/* The accumulator is only needed if we need to be able to pause
* the reader. A few bytes could be saved by moving this to a
* separate struct and using a pointer here. */
unsigned char *acc; /*!< The accumulator is used to gather incoming
* data if a read-request via mbedtls_mps_reader_get()
* cannot be served from the current fragment. */
mbedtls_mps_stored_size_t acc_len;
/*!< The total size of the accumulator. */
mbedtls_mps_stored_size_t acc_available;
/*!< The number of bytes currently gathered in
* the accumulator. This is both used in
* producing and in consuming mode:
* While producing, it is increased until
* it reaches the value of \c acc_remaining below.
* While consuming, it is used to judge if a
* read request can be served from the
* accumulator or not.
* Must not be larger than acc_len. */
union
{
mbedtls_mps_stored_size_t acc_remaining;
/*!< This indicates the amount of data still
* to be gathered in the accumulator. It is
* only used in producing mode.
* Must be at most acc_len - acc_available. */
mbedtls_mps_stored_size_t frag_offset;
/*!< This indicates the offset of the current
* fragment from the beginning of the
* accumulator.
* It is only used in consuming mode.
* Must not be larger than \c acc_available. */
} acc_share;
};
/*
* API organization:
* A reader object is usually prepared and maintained
* by some lower layer and passed for usage to an upper
* layer, and the API naturally splits according to which
* layer is supposed to use the respective functions.
*/
/*
* Maintenance API (Lower layer)
*/
/**
* \brief Initialize a reader object
*
* \param reader The reader to be initialized.
* \param acc The buffer to be used as a temporary accumulator
* in case read requests through mbedtls_mps_reader_get()
* exceed the buffer provided by mbedtls_mps_reader_feed().
* This buffer is owned by the caller and exclusive use
* for reading and writing is given to the reade for the
* duration of the reader's lifetime. It is thus the caller's
* responsibility to maintain (and not touch) the buffer for
* the lifetime of the reader, and to properly zeroize and
* free the memory after the reader has been destroyed.
* \param acc_len The size in Bytes of \p acc.
*
* \return \c 0 on success.
* \return A negative \c MBEDTLS_ERR_READER_XXX error code on failure.
*/
int mbedtls_mps_reader_init( mbedtls_mps_reader *reader,
unsigned char *acc,
mbedtls_mps_size_t acc_len );
/**
* \brief Free a reader object
*
* \param reader The reader to be freed.
*
* \return \c 0 on success.
* \return A negative \c MBEDTLS_ERR_READER_XXX error code on failure.
*/
int mbedtls_mps_reader_free( mbedtls_mps_reader *reader );
/**
* \brief Pass chunk of data for the reader to manage.
*
* \param reader The reader context to use. The reader must be
* in producing state.
* \param buf The buffer to be managed by the reader.
* \param buflen The size in Bytes of \p buffer.
*
* \return \c 0 on success. In this case, the reader will be
* moved to consuming state, and ownership of \p buf
* will be passed to the reader until mbedtls_mps_reader_reclaim()
* is called.
* \return \c MBEDTLS_ERR_MPS_READER_NEED_MORE if more input data is
* required to fulfill a previous request to mbedtls_mps_reader_get().
* In this case, the reader remains in producing state and
* takes no ownership of the provided buffer (an internal copy
* is made instead).
* \return Another negative \c MBEDTLS_ERR_READER_XXX error code on
* different kinds of failures.
*/
int mbedtls_mps_reader_feed( mbedtls_mps_reader *reader,
unsigned char *buf,
mbedtls_mps_size_t buflen );
/**
* \brief Reclaim reader's access to the current input buffer.
*
* \param reader The reader context to use. The reader must be
* in producing state.
* \param paused If not \c NULL, the integer at address \p paused will be
* modified to indicate whether the reader has been paused
* (value \c 1) or not (value \c 0). Pausing happens if there
* is uncommitted data and a previous request to
* mbedtls_mps_reader_get() has exceeded the bounds of the
* input buffer.
*
* \return \c 0 on success.
* \return A negative \c MBEDTLS_ERR_READER_XXX error code on failure.
*/
int mbedtls_mps_reader_reclaim( mbedtls_mps_reader *reader,
mbedtls_mps_size_t *paused );
/*
* Usage API (Upper layer)
*/
/**
* \brief Request data from the reader.
*
* \param reader The reader context to use. The reader must
* in consuming state.
* \param desired The desired amount of data to be read, in Bytes.
* \param buffer The address to store the buffer pointer in.
* This must not be \c NULL.
* \param buflen The address to store the actual buffer
* length in, or \c NULL.
*
* \return \c 0 on success. In this case, \c *buf holds the
* address of a buffer of size \c *buflen
* (if \c buflen != \c NULL) or \c desired
* (if \c buflen == \c NULL). The user hass ownership
Don't invalidate MPS reader buffers upon commit call Previously, the semantics of mbedtls_mps_reader_commit() was to invalidate all buffers previously fetched via mbedtls_mps_reader_get(), forbidding any further use by the 'consumer'. This was in fact a necessary constraint for the current implementation, which did some memory moving in mbedtls_mps_reader_commit(). This commit simplifies the reader's semantics and implementation in the following way: - API: A call to mbedtls_mps_reader_commit() does no longer invalidate the buffers previously obtained via mbedtls_mps_reader_get(). Instead, they can continue to be used until mbedtls_mps_reader_reclaim() is called. Calling mbedtls_mps_reader_commit() now only sets a marker indicating which parts of the data received through mbedtls_mps_reader_get() need not be backed up once mbedtls_mps_reader_reclaim() is called. Allowing the user to call mbedtls_mbedtls_reader_commit() multiple times before mbedtls_mps_reader_reclaim() is mere convenience: We'd get exactly the same functionality if instead of mbedtls_mps_reader_commit(), there was an additional argument to mbedtls_mps_reader_reclaim() indicating how much data to retain. However, the present design is more convenient for the user and doesn't appear to introduce any unnecessary complexity (anymore), so we stick with it for now. - Implementation: mbedtls_mps_reader_commit() is now a 1-liner, setting the 'commit-marker', but doing nothing else. Instead, the complexity of mbedtls_mp_reader_reclaim() slightly increases because it has to deal with creating backups from both the accumulator and the current fragment. In the previous implementation, which shifted the accumulator content with every call to mbedtls_mps_reader_commit(), only the backup from the fragment was necessary; with the new implementation which doesn't shift anything in mbedtls_mps_reader_commit(), we need to do the accumulator shift in mbedtls_mps_reader_reclaim(). Signed-off-by: Hanno Becker <hanno.becker@arm.com>
2021-02-08 07:54:30 +01:00
* of the buffer until the next call mbedtls_mps_reader_reclaim().
* \return #MBEDTLS_ERR_MPS_READER_OUT_OF_DATA if there is not enough
* data available to serve the read request. In this case,
* the reader remains intact, and additional data can be
* provided by reclaiming the current input buffer via
* mbedtls_mps_reader_reclaim() and feeding a new one via
* mbedtls_mps_reader_feed().
* \return Another negative \c MBEDTLS_ERR_READER_XXX error
* code for different kinds of failure.
*
* \note Passing \c NULL as \p buflen is a convenient way to
* indicate that fragmentation is not tolerated.
* It's functionally equivalent to passing a valid
* address as buflen and checking \c *buflen == \c desired
* afterwards.
*/
int mbedtls_mps_reader_get( mbedtls_mps_reader *reader,
mbedtls_mps_size_t desired,
unsigned char **buffer,
mbedtls_mps_size_t *buflen );
/**
Don't invalidate MPS reader buffers upon commit call Previously, the semantics of mbedtls_mps_reader_commit() was to invalidate all buffers previously fetched via mbedtls_mps_reader_get(), forbidding any further use by the 'consumer'. This was in fact a necessary constraint for the current implementation, which did some memory moving in mbedtls_mps_reader_commit(). This commit simplifies the reader's semantics and implementation in the following way: - API: A call to mbedtls_mps_reader_commit() does no longer invalidate the buffers previously obtained via mbedtls_mps_reader_get(). Instead, they can continue to be used until mbedtls_mps_reader_reclaim() is called. Calling mbedtls_mps_reader_commit() now only sets a marker indicating which parts of the data received through mbedtls_mps_reader_get() need not be backed up once mbedtls_mps_reader_reclaim() is called. Allowing the user to call mbedtls_mbedtls_reader_commit() multiple times before mbedtls_mps_reader_reclaim() is mere convenience: We'd get exactly the same functionality if instead of mbedtls_mps_reader_commit(), there was an additional argument to mbedtls_mps_reader_reclaim() indicating how much data to retain. However, the present design is more convenient for the user and doesn't appear to introduce any unnecessary complexity (anymore), so we stick with it for now. - Implementation: mbedtls_mps_reader_commit() is now a 1-liner, setting the 'commit-marker', but doing nothing else. Instead, the complexity of mbedtls_mp_reader_reclaim() slightly increases because it has to deal with creating backups from both the accumulator and the current fragment. In the previous implementation, which shifted the accumulator content with every call to mbedtls_mps_reader_commit(), only the backup from the fragment was necessary; with the new implementation which doesn't shift anything in mbedtls_mps_reader_commit(), we need to do the accumulator shift in mbedtls_mps_reader_reclaim(). Signed-off-by: Hanno Becker <hanno.becker@arm.com>
2021-02-08 07:54:30 +01:00
* \brief Mark data obtained from mbedtls_writer_get() as processed.
*
Don't invalidate MPS reader buffers upon commit call Previously, the semantics of mbedtls_mps_reader_commit() was to invalidate all buffers previously fetched via mbedtls_mps_reader_get(), forbidding any further use by the 'consumer'. This was in fact a necessary constraint for the current implementation, which did some memory moving in mbedtls_mps_reader_commit(). This commit simplifies the reader's semantics and implementation in the following way: - API: A call to mbedtls_mps_reader_commit() does no longer invalidate the buffers previously obtained via mbedtls_mps_reader_get(). Instead, they can continue to be used until mbedtls_mps_reader_reclaim() is called. Calling mbedtls_mps_reader_commit() now only sets a marker indicating which parts of the data received through mbedtls_mps_reader_get() need not be backed up once mbedtls_mps_reader_reclaim() is called. Allowing the user to call mbedtls_mbedtls_reader_commit() multiple times before mbedtls_mps_reader_reclaim() is mere convenience: We'd get exactly the same functionality if instead of mbedtls_mps_reader_commit(), there was an additional argument to mbedtls_mps_reader_reclaim() indicating how much data to retain. However, the present design is more convenient for the user and doesn't appear to introduce any unnecessary complexity (anymore), so we stick with it for now. - Implementation: mbedtls_mps_reader_commit() is now a 1-liner, setting the 'commit-marker', but doing nothing else. Instead, the complexity of mbedtls_mp_reader_reclaim() slightly increases because it has to deal with creating backups from both the accumulator and the current fragment. In the previous implementation, which shifted the accumulator content with every call to mbedtls_mps_reader_commit(), only the backup from the fragment was necessary; with the new implementation which doesn't shift anything in mbedtls_mps_reader_commit(), we need to do the accumulator shift in mbedtls_mps_reader_reclaim(). Signed-off-by: Hanno Becker <hanno.becker@arm.com>
2021-02-08 07:54:30 +01:00
* This call indicates that all data received from prior calls to
* mbedtls_mps_reader_fetch() has been or will have been
* processed when mbedtls_mps_reader_reclaim() is called,
* and thus need not be backed up.
*
Don't invalidate MPS reader buffers upon commit call Previously, the semantics of mbedtls_mps_reader_commit() was to invalidate all buffers previously fetched via mbedtls_mps_reader_get(), forbidding any further use by the 'consumer'. This was in fact a necessary constraint for the current implementation, which did some memory moving in mbedtls_mps_reader_commit(). This commit simplifies the reader's semantics and implementation in the following way: - API: A call to mbedtls_mps_reader_commit() does no longer invalidate the buffers previously obtained via mbedtls_mps_reader_get(). Instead, they can continue to be used until mbedtls_mps_reader_reclaim() is called. Calling mbedtls_mps_reader_commit() now only sets a marker indicating which parts of the data received through mbedtls_mps_reader_get() need not be backed up once mbedtls_mps_reader_reclaim() is called. Allowing the user to call mbedtls_mbedtls_reader_commit() multiple times before mbedtls_mps_reader_reclaim() is mere convenience: We'd get exactly the same functionality if instead of mbedtls_mps_reader_commit(), there was an additional argument to mbedtls_mps_reader_reclaim() indicating how much data to retain. However, the present design is more convenient for the user and doesn't appear to introduce any unnecessary complexity (anymore), so we stick with it for now. - Implementation: mbedtls_mps_reader_commit() is now a 1-liner, setting the 'commit-marker', but doing nothing else. Instead, the complexity of mbedtls_mp_reader_reclaim() slightly increases because it has to deal with creating backups from both the accumulator and the current fragment. In the previous implementation, which shifted the accumulator content with every call to mbedtls_mps_reader_commit(), only the backup from the fragment was necessary; with the new implementation which doesn't shift anything in mbedtls_mps_reader_commit(), we need to do the accumulator shift in mbedtls_mps_reader_reclaim(). Signed-off-by: Hanno Becker <hanno.becker@arm.com>
2021-02-08 07:54:30 +01:00
* This function has no user observable effect until
* mbedtls_mps_reader_reclaim() is called. In particular,
* buffers received from mbedtls_mps_reader_fetch() remain
* valid until mbedtls_mps_reader_reclaim() is called.
*
Don't invalidate MPS reader buffers upon commit call Previously, the semantics of mbedtls_mps_reader_commit() was to invalidate all buffers previously fetched via mbedtls_mps_reader_get(), forbidding any further use by the 'consumer'. This was in fact a necessary constraint for the current implementation, which did some memory moving in mbedtls_mps_reader_commit(). This commit simplifies the reader's semantics and implementation in the following way: - API: A call to mbedtls_mps_reader_commit() does no longer invalidate the buffers previously obtained via mbedtls_mps_reader_get(). Instead, they can continue to be used until mbedtls_mps_reader_reclaim() is called. Calling mbedtls_mps_reader_commit() now only sets a marker indicating which parts of the data received through mbedtls_mps_reader_get() need not be backed up once mbedtls_mps_reader_reclaim() is called. Allowing the user to call mbedtls_mbedtls_reader_commit() multiple times before mbedtls_mps_reader_reclaim() is mere convenience: We'd get exactly the same functionality if instead of mbedtls_mps_reader_commit(), there was an additional argument to mbedtls_mps_reader_reclaim() indicating how much data to retain. However, the present design is more convenient for the user and doesn't appear to introduce any unnecessary complexity (anymore), so we stick with it for now. - Implementation: mbedtls_mps_reader_commit() is now a 1-liner, setting the 'commit-marker', but doing nothing else. Instead, the complexity of mbedtls_mp_reader_reclaim() slightly increases because it has to deal with creating backups from both the accumulator and the current fragment. In the previous implementation, which shifted the accumulator content with every call to mbedtls_mps_reader_commit(), only the backup from the fragment was necessary; with the new implementation which doesn't shift anything in mbedtls_mps_reader_commit(), we need to do the accumulator shift in mbedtls_mps_reader_reclaim(). Signed-off-by: Hanno Becker <hanno.becker@arm.com>
2021-02-08 07:54:30 +01:00
* \param reader The reader context to use.
*
Don't invalidate MPS reader buffers upon commit call Previously, the semantics of mbedtls_mps_reader_commit() was to invalidate all buffers previously fetched via mbedtls_mps_reader_get(), forbidding any further use by the 'consumer'. This was in fact a necessary constraint for the current implementation, which did some memory moving in mbedtls_mps_reader_commit(). This commit simplifies the reader's semantics and implementation in the following way: - API: A call to mbedtls_mps_reader_commit() does no longer invalidate the buffers previously obtained via mbedtls_mps_reader_get(). Instead, they can continue to be used until mbedtls_mps_reader_reclaim() is called. Calling mbedtls_mps_reader_commit() now only sets a marker indicating which parts of the data received through mbedtls_mps_reader_get() need not be backed up once mbedtls_mps_reader_reclaim() is called. Allowing the user to call mbedtls_mbedtls_reader_commit() multiple times before mbedtls_mps_reader_reclaim() is mere convenience: We'd get exactly the same functionality if instead of mbedtls_mps_reader_commit(), there was an additional argument to mbedtls_mps_reader_reclaim() indicating how much data to retain. However, the present design is more convenient for the user and doesn't appear to introduce any unnecessary complexity (anymore), so we stick with it for now. - Implementation: mbedtls_mps_reader_commit() is now a 1-liner, setting the 'commit-marker', but doing nothing else. Instead, the complexity of mbedtls_mp_reader_reclaim() slightly increases because it has to deal with creating backups from both the accumulator and the current fragment. In the previous implementation, which shifted the accumulator content with every call to mbedtls_mps_reader_commit(), only the backup from the fragment was necessary; with the new implementation which doesn't shift anything in mbedtls_mps_reader_commit(), we need to do the accumulator shift in mbedtls_mps_reader_reclaim(). Signed-off-by: Hanno Becker <hanno.becker@arm.com>
2021-02-08 07:54:30 +01:00
* \return \c 0 on success.
* \return A negative \c MBEDTLS_ERR_READER_XXX error code on failure.
*
*/
int mbedtls_mps_reader_commit( mbedtls_mps_reader *reader );
#endif /* MBEDTLS_READER_H */