Find file
Fetching contributors…
Cannot retrieve contributors at this time
184 lines (125 sloc) 6.46 KB
Things TODO
* Fix the numerous malloc+copy operations for sending data, see "Buffering
Improvements" below for details
* make sure the windowing code adapts better to slow situations so that it
doesn't then use as much memory as today. Possibly by an app-controllable
"Window mode"?
* Introduce a function that can free data that is returned allocated from the
lib, like when libssh2_session_last_error() is asked to return the string
with 'want_buf' set to non-zero.
* Add more info to the man pages.
* Decrease the number of mallocs. Everywhere. Will get easier once the
buffering improvements have been done.
* Use SO_NOSIGPIPE for Mac OS/BSD systems where MSG_NOSIGNAL doesn't
* Extend the test suite to actually test lots of aspects of libssh2
* Fix all compiler warnings (some can't be done without API changes)
* Expose error messages sent by the server
* Make SFTP transfers ask for and send several packages at once so that it
doesn't have to send-waitforack-send-waitforack as much. This will be
easier/smoother if we implement the "New SFTP API" described below.
* select() is troublesome with libssh2 when using multiple channels over
the same session. See "New Transport API" below for more details.
At next SONAME bump
* stop using #defined macros as part of the official API. The macros should
either be turned into real functions or discarded from the API.
* fix the parts of the API where object pointers and function pointers are
mixed like libssh2_session_callback_set()
* remove the following functions from the API/ABI
libssh2_session_startup() (libssh2_session_handshake() is the replacement)
* Rename a few function:
libssh2_hostkey_hash => libssh2_session_hostkey_hash
libssh2_banner_set => libssh2_session_banner_set
* change 'int' to 'libssh2_socket_t' in the public API for sockets.
* Use 'size_t' for string lengths in all functions.
* Add a comment field to struct libssh2_knownhost.
* remove the existing libssh2_knownhost_add() function and rename
libssh2_knownhost_addc to become the new libssh2_knownhost_add instead
* remove the existing libssh2_scp_send_ex() function and rename
libssh2_scp_send64 to become the new libssh2_scp_send instead.
* remove the existing libssh2_knownhost_check() functin and rename
libssh2_knownhost_checkp() to become the new libssh2_knownhost_check instead
Buffering Improvements
- If this function gets called with a total packet size that is larger than
32K, it should create more than one SSH packet so that it keeps the largest
one below 32K
- should not copy/allocate anything for the data, only create a header chunk
and pass on the payload data to channel_write "pointed to"
New Transport API
The problem in a nutshell is that when an application opens up multiple
channels over a single session, those are all using the same socket. If the
application is then using select() to wait for traffic (like any sensible app
does) and wants to act on the data when select() tells there is something to
for example read, what does an application do?
With our current API, you have to loop over all the channels and read from
them to see if they have data. This effectively makes blocking reads
impossible. If the app has many channels in a setup like this, it even becomes
slow. (The original API had the libssh2_poll_channel_read() and libssh2_poll()
to somewhat overcome this hurdle, but they too have pretty much the same
problems plus a few others.)
Traffic in the other direction is similarly limited: the app has to try
sending to all channels, even though some of them may very well not accept any
data at that point.
I suggest we introduce two new helper functions:
- Read "a bunch" of data from the given socket and returns information to the
app about what channels that are now readable (ie they will not block when
read from). The function can be called over and over and it will repeatedly
return info about what channels that are readable at that moment.
- Returns information about what channels that are writable, in the sense
that they have windows set from the remote side that allows data to get
sent. Writing to one of those channels will not block. Of course, the
underlying socket may only accept a certain amount of data, so at the first
short return, nothing more should be attempted to get sent until select()
(or equivalent) has been used on the master socket again.
I haven't yet figured out a sensible API for how these functions should return
that info, but if we agree on the general principles I guess we can work that
I wanted to mention that these two helper functions would not be mandatory
in any way. They would just be there for those who want them, and existing
programs can remain using the old functions only if they prefer to.
Provide API functions that explicitly tells at once that a (full) SFTP file
transfer is wanted, to allow libssh2 to leverage on that knowledge to speed
up things internally. It can for example do read ahead, buffer writes (merge
small writes into larger chunks), better tune the SSH window and more. This
sort of API is already provided for SCP transfers.
New functions:
LIBSSH2_SFTP_HANDLE *libssh2_sftp_send(SFTP_SESSION *sftp,
uint64_t filesize,
char *remote_path,
size_t remote_path_len,
long mode);
Tell libssh2 that a local file with a given size is about to get sent to
the SFTP server.
LIBSSH2_SFTP_HANDLE *libssh2_sftp_recv();
Tell libssh2 that a remote file is requested to get downloaded from the SFTP
Only the setup of the file transfer is different from an application's point
of view. Depending on direction of the transfer(s), the following already
existing functions should then be used until the transfer is complete:
1. Setup the transfer using one of the two new functions.
2. Loop through the reading or writing of data.
3. Cleanup the transfer