The shadow route algorithm is extending the route randomly using channels
adjacent to the current destination, in the hope to create a plausible route
extension. However, instead of only retrieving the channels adjacent to the
destination it was retrieving all channels in the entire topology, and
selecting a random channel from there. This resulted in a very large request
for all channels being processed, and then mostly not being used, but also in
shadow extensions to the path which were not plausible (they didn't extend the
real path, just random edges). This is fixed by restricting the call to
`listchannels` to the channels with the current destination as source.
On my laptop retrieving all channels in the current mainnet takes
approximately 1.2 seconds, and given the geometric series expansion of the 50%
extension probability this indeed would result in an overhead of 1.2 seconds
to the `pay` command. In contrast specifying a source results in an overhead
of ~30ms.
So good news everyone, your pay commands just shaved 1.17 seconds off their
runtime.
Changelog-Changed: pay: Improved the performance of the `pay`-plugin by limiting the `listchannels` when computing the shadow route.
Changelog-Fixed: pay: The `pay`-plugin was generating non-contiguous shadow routes
Turns out that unnecessary: all callers can access the feature_set,
so make it much more like a normal primitive.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is a common thing to do, so create a macro.
Unfortunately, it still needs the type arg, because the paramter may
be const, and the return cannot be, and C doesn't have a general
"(-const)" cast.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Before this patch we used to send `double`s over the wire by just
copying them. This is not portable because the internal represenation
of a `double` is implementation specific.
Instead of this, multiply any floating-point numbers that come from
the outside (e.g. JSONs) by 1 million and round them to integers when
handling them.
* Introduce a new param_millionths() that expects a floating-point
number and returns it multipled by 1000000 as an integer.
* Replace param_double() and param_percent() with param_millionths()
* Previously the riskfactor would be allowed to be negative, which must
have been unintentional. This patch changes that to require a
non-negative number.
Changelog-None
This adds helpers to start and send a jsonrpc request using json_stream
in order to benefit from the helpers.
This then simplifies existing plugins RPC requests by using json_stream
helpers.
Before this patch we used `int` for error codes. The problem with
`int` is that we try to pass it to/from wire and the size of `int` is
not defined by the standard. So a sender with 4-byte `int` would write
4 bytes to the wire and a receiver with 2-byte `int` (for example) would
read just 2 bytes from the wire.
To resolve this:
* Introduce an error code type with a known size:
`typedef s32 errcode_t`.
* Change all error code macros to constants of type `errcode_t`.
Constants also play better with gdb - it would visualize the name of
the constant instead of the numeric value.
* Change all functions that take error codes to take the new type
`errcode_t` instead of `int`.
* Introduce towire / fromwire functions to send / receive the newly added
type `errcode_t` and use it instead of `towire_int()`.
In addition:
* Remove the now unneeded `towire_int()`.
* Replace a hardcoded error code `-2` with a new constant
`INVOICE_EXPIRED_DURING_WAIT` (903).
Changelog-Changed: The waitinvoice command would now return error code 903 to designate that the invoice expired during wait, instead of the previous -2
Changelog-Fixed: Detect a previously non-permanent error (`final_cltv_too_soon`) that has been merged into a permanent error (`incorrect_or_unknown_payment_details`), and retry that failure case in `pay`.
This won't usually be visible to the end-user, since the pay plugin doesn't
do multi-part yet (and mpp requires EXPERIMENTAL_FEATURES), but we're ready
once it does.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The pay plugin has been supplying the bolt11 string since 0.7.0, so only
ancient "pay" commands would be omitted by this change.
You can create a no-bolt11 "sendpay" manually, but then you'll find it
in 'listsendpays'.
Changelog-Removed: JSON: `listpays` won't shown payments made via sendpay without a bolt11 string, or before 0.7.0.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
When doing the random walk through the channel, we now add the fees
(both the base and the proportional one) for that channel in addition to
the cltv delta.
Changelog-Added: Payment amount fuzzing is restored, but through shadow routing.
The string cut & paste hack was nasty; make ->failure a json_out
object so we can splice it in properly.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We now hand around struct json_out members, rather than using formatted
strings, so plugins need to construct them properly.
There's no automatic conversion between ' and " any more, so those
are eliminated too. pay still uses some manual construction of elements.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
I tried to just do gossipd, but it was uncontainable, so this ended up being
a complete sweep.
We didn't get much space saving in gossipd, even though we should save
24 bytes per node.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
*best is checked for null before the comparison against the uninitialized
variable ever happens, so this isn't a real issue.
Initialize it to zero so that we don't fail to compile on certain gcc versions.
plugins/pay.c: In function ‘add_shadow_route’:
plugins/pay.c:644:18: error: ‘sample’ may be used uninitialized in this function
if (!best || v > sample) {
~~^~~~~~~~
Signed-off-by: William Casarin <jb55@jb55.com>
New name is less confusing, and most people should be transitioning to
listpays rather than this anyway.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is to future-proof against multi-part-payments: the low-level commands
will start returning multiple results once we have that, so prepare
transition plan now.
Closes: #2372
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
