Previous code used the request ids to do ordering. This worked until a
client came along that used un-ordered request ids. This reworks the
ordering to use an internal counter (per session) to order all packets
ensuring that responses are sent in the same order as the requests were
received.
Fixes#260
Instead of sendPacket/sendError being sprayed all over the place, this
change has all those places instead return a responsePacket (eventually)
back to the main handling function which then calls sendPacket in one
place.
Behaviour of the code should remain exactly the same.
This makes it much easier to work with the response packets (eg. for the
packet ordering issue I'm working on).
sendError takes a requestPacket but was simplifying it to an ider
interface. Future work needed it to be requestPacket but I wanted to fix
didn't up type usage in its own commit.
Instead of accepting a more general type and then asserting it to the
proper type, just take the proper type as the argument.
Also clean up some of the use of it where it checked old direct sending
code's return error (error is now always nil).
The initial Opendir packet is supposed to repond with an error status if
the directory wasn't found. It was just returning a handle without
checking, now it does a Stat on the path and only returns the handle if
the Stat is successful and it indicates it is a directory, otherwise it
returns an error.
Add errors for all the SSH_FXP_STATUS codes to give the developer
implementing request server handlers greater control over the returned
codes. Most helpful in cases where nothing currently would work (eg.
unsupported).
Fixes#223
Splitted cleanPath into cleanPacketPath and cleanPath for better handling of slashes in file paths
Added test for cleanPath func
Removed code duplication => filepath.ToSlash(filepath.Clean(...)) => cleanPath(...)
Fixed tests for runLs to match year or time
Renamed constants to fit hound rules
There is a data race with the waitgroup (wg) object used to synchronize
the workers with the server exit. The workers called wg.Add()
asynchronously and it was possible for the Wait() to get hit before any
of the Add() calls were made in certain conditions. I only ever saw this
sporatically in the travis tests.
This fixes it by making the wg.Add() calls synchronous.
The worker/packet mangement code needs to be in the packet manager so
the request-server can utilize it as well. This also improves the
encapsulation of the method as it relied on internal data that should be
better isolated inside the file/struct.
File operations that happen after the open packet has been received,
like reading/writing, can be done with the pool as the order they are
run in doesn't matter (the packets contain the file offsets).
Command operations, on the other hand, need to be serialized.
This flips between a pool of workers for file operations and a single
worker for everything else. It flips on Open and Close packets.
Server struct contains Mutex which might be copied in inconsistent
state. Avoid this by declaring methods on pointer receiver.
This calms down go race detector.
Signed-off-by: Pavel Borzenkov <pavel.borzenkov@gmail.com>
This means we are now passing full packets to the workers. This makes
the request id's available which is required for fixing packet ordering.
Benchmark tests weren't affected at all.
Reading and Writing are less coupled in the sftp client and server than
they may be over a traditional ReadWriter. This mirrors the full duplex
/ half close nature of a TCP connection.
In essence conn.Reader cannot be closed locally, it can only close in
response to the remote side using conn.WriteCloser.Close(), or a network
error, which is effectively the same thing.
Make this behaviour super clear by no longer smooshing the type into a
ReadWriteCloser.
* statusFromError now takes a type with an id() uint32 method
* Introduce sendError
Refactor all the calls to sendPacket(statusFromError(...)) to a
sendError helper.
- no need to initalise mutexes, their zero value is valid
- make NewServer simpler, 90% of use cases already had an
io.ReadWriteCloser, in the single case that it it is easy to provide a
simple wrapper.
John Eikenberry
Allan Feid
John Eikenberry
andreas
Pavel Borzenkov
George Xie
Dave Cheney
Mark Sheahan
Matt Layher