Anna Shaleva
0ceae612c6
vm: improve failed SYSCALL logging
...
For better user experience.
2022-11-28 17:16:27 +03:00
Roman Khimov
8e7f65be17
vm: use proper estack for exception handler
...
v.estack might be some inner invoked contract and its stack must not be used
for exception handler set up by higher-order contract.
2022-11-18 11:36:38 +03:00
Roman Khimov
cb64957af5
vm: don't use Stack for istack
...
We don't use all of the Stack functionality for it, so drop useless methods
and avoid some interface conversions. It increases single-node TPS by about
0.9%, so nothing really important there, but not a bad change either. Maybe it
can be reworked again with generics though.
2022-11-18 11:35:29 +03:00
Roman Khimov
aef01bf663
vm: fix istack marshaling, fix #2799
2022-11-16 00:40:12 +03:00
Roman Khimov
90582faacd
vm: save current stack slice when loading new context
...
v.estack is used throughout the code to work with estack, while ctx.sc.estack
is (theoretically) just a reference to it that is saved on script load and
restored to v.estack on context unload. The problem is that v.estack can grow
as we use it and can be reallocated away from its original slice (saved in the
ctx.sc.estack), so either ctx.sc.estack should be a pointer or we need to
ensure that it's correct when loading a new script. The second approach is a
bit safer for now and it fixes #2798 .
2022-11-15 23:48:02 +03:00
Anna Shaleva
7db9258104
vm: make LoadFileWithFlags actually load with flags provided
2022-10-13 16:07:34 +03:00
Anna Shaleva
f45d8fc08d
vm: remove default syscall handler
...
It's not needed anymore. Close #1075 .
2022-10-07 15:56:34 +03:00
Roman Khimov
e8d2277fe5
contract/vm: only push NULL after call in dynamic contexts
...
And determine the need for Null dynamically. For some reason the only dynamic
context is Contract.Call. CALLT is not dynamic and neither is a call from
native contract, go figure...
2022-08-05 14:35:00 +03:00
Roman Khimov
99e2681d3a
interop/vm: use more robust CalledByEntry check
...
Directly check contexts.
2022-08-05 14:35:00 +03:00
Roman Khimov
13f5fdbe8a
vm: extract shared parts of the Context
...
Local calls reuse them, cross-contract calls create new ones. This allows to
avoid some allocations and use a little less memory.
2022-08-05 14:26:25 +03:00
Roman Khimov
1e62474514
vm: move InvocationTree into a package of its own
...
result.Invoke shouldn't depend on vm.
2022-07-08 23:30:30 +03:00
Roman Khimov
fab8dfb9f8
vm: move State type into a package of its own
...
It's used a lot in other places that need it, but don't need whole VM at the
same time.
2022-07-08 18:34:52 +03:00
Roman Khimov
92c94f265c
interop/vm: make VM reusable and use on VM for all in-block execs
...
Avoid allocating again and again. Increases TPS by about 3%.
2022-06-07 10:05:47 +03:00
Roman Khimov
edb6ca8926
Merge pull request #2531 from nspcc-dev/perf
...
Minor performance improvements
2022-06-03 10:40:56 +03:00
Roman Khimov
3d4076ca36
vm: microoptimize new estack creation
...
Subslice won't reach element -1, but it can reuse the same buffer space more
effectively.
2022-05-31 18:53:05 +03:00
Anna Shaleva
7296f0c913
vm: support immutable compound types
2022-05-31 08:07:50 +03:00
Anna Shaleva
107f5e0793
vm: adjust refcount operations order
...
Perform add/set/remove operations with VM type firstly, and with
refcounter after that. It is needed to be able to extend add/set/remove
operations with extra checks.
2022-05-31 08:02:13 +03:00
Anna Shaleva
a79c80cb8d
vm: wrap cross-contract exceptions
2022-05-26 11:44:26 +03:00
Anna Shaleva
a39b7cc3fd
core, vm: move all isolation-related logic out of VM
...
Keep it inside the interop context.
2022-05-26 09:26:31 +03:00
Anna Shaleva
f79f62dab4
vm: use single function instead of committer/discarder
2022-05-26 09:26:31 +03:00
Anna Shaleva
08b68e9b82
vm, core: push Null return value only if no exception occurs
...
Close https://github.com/nspcc-dev/neo-go/issues/2509 .
2022-05-26 09:26:31 +03:00
Anna Shaleva
ce226f6b76
vm: optimize context wrapping code
...
We can omit DAO wrapping for safe methods and for those methods that are not
wrapped into try-catch block. However, we still need to persist
notificationsCount changes for these methods to the parent context.
2022-05-26 09:26:31 +03:00
Anna Shaleva
0cd19f550e
core, vm: implement snapshot isolation
...
Initial implementation without optimisations, but it should work.
2022-05-26 09:26:25 +03:00
Roman Khimov
f2f66ad36e
vm: clear static slot refs on exit, fix #2501
2022-05-17 16:26:22 +03:00
Roman Khimov
0d93b353e6
vm: count initial Null slot variables as well
...
Match C# behavior. Thanks to @ixje for finding this.
2022-05-16 17:42:57 +03:00
Roman Khimov
18d627e7f7
vm: count map key in the refcounter as well
...
Thanks @ixje for spotting this.
2022-05-16 16:07:25 +03:00
Anna Shaleva
a94ccf9236
vm: add MODMUL and MODPOW opcodes
...
Close #2470 .
2022-05-12 14:25:14 +03:00
Elizaveta Chichindaeva
28908aa3cf
[ #2442 ] English Check
...
Signed-off-by: Elizaveta Chichindaeva <elizaveta@nspcc.ru>
2022-05-04 19:48:27 +03:00
Anna Shaleva
2096ad6e81
*: remove io/ioutil uses
...
Close #1764 .
2022-03-17 19:39:18 +03:00
Evgeniy Stratonikov
32f4404954
vm: allow HASKEY on byte-arrays
...
Current neo-vm master has them https://github.com/neo-project/neo-vm/blob/master/src/neo-vm/ExecutionEngine.cs#L1157
Were silently added in
029466fa9d
.
Signed-off-by: Evgeniy Stratonikov <evgeniy@nspcc.ru>
2022-03-10 10:34:13 +03:00
Evgeniy Stratonikov
39866b8512
vm: fix integer conversions
...
As can be seen in https://dotnetfiddle.net/s7eg21 (int) conversions
result in an exception in C# code.
Signed-off-by: Evgeniy Stratonikov <evgeniy@nspcc.ru>
2022-03-10 10:21:34 +03:00
Evgeniy Stratonikov
10d006195a
vm: properly clear try stack in CALL
...
Signed-off-by: Evgeniy Stratonikov <evgeniy@nspcc.ru>
2022-02-04 11:04:03 +03:00
Evgeniy Stratonikov
ffb6504f67
vm: cut trailing spaces in PrintOps
...
When there is a single big instruction (like PUSHDATA4) in script,
all other instructions are padded to the right with spaces.
This makes it hard to view script in terminal, because long lines
are usually wrapped at the screen boundary and printed as multiple lines.
The culprit is our `cursor` field which is printed in the last column
and causes all previous fields to have the same length for every
instruction. One way to fix this is to omit cursor field if it is empty.
Signed-off-by: Evgeniy Stratonikov <evgeniy@nspcc.ru>
2021-12-28 11:27:31 +03:00
Roman Khimov
53cfde62de
vm: don't allocate another stack for entry scripts
...
vm.New() creates a new stack, then we load an entry script with
LoadScriptWithFlags and it creates another one which doesn't make much
sense. rvcount is -1 for it, so all elements are to be copied down anyway and
it's clear so a new loaded script can't dig down to anything it shouldn't be
able to.
2021-12-02 15:59:51 +03:00
Roman Khimov
aa1da9de5a
vm: preallocate for istack
...
Avoid growing it in runtime.
2021-12-01 21:36:35 +03:00
Roman Khimov
418ba1dbc3
*: use internal variables for simple big.NewInt() values
...
Avoid additional allocations.
2021-12-01 21:36:25 +03:00
Roman Khimov
ee05f73b6f
vm: simplify slot and make it private
...
Hiding refcounter inside a slot is actually a good idea, but it makes the
structure somewhat bigger, especially given that the refcounter is the same
and belongs more to VM or Context. New structure is a bit more efficient:
name old time/op new time/op delta
ScriptFibonacci-8 672µs ± 2% 644µs ± 0% -4.15% (p=0.008 n=5+5)
ScriptNestedRefCount-8 1.08ms ± 1% 1.05ms ± 2% -2.56% (p=0.008 n=5+5)
ScriptPushPop/4-8 1.52µs ± 1% 1.47µs ± 1% -3.14% (p=0.008 n=5+5)
ScriptPushPop/16-8 3.66µs ± 1% 3.54µs ± 1% -3.24% (p=0.008 n=5+5)
ScriptPushPop/128-8 24.7µs ± 1% 23.2µs ± 1% -6.14% (p=0.008 n=5+5)
ScriptPushPop/1024-8 183µs ± 1% 173µs ± 1% -5.01% (p=0.008 n=5+5)
name old alloc/op new alloc/op delta
ScriptFibonacci-8 114kB ± 0% 114kB ± 0% ~ (p=0.079 n=4+5)
ScriptNestedRefCount-8 241kB ± 0% 241kB ± 0% ~ (p=0.333 n=5+4)
ScriptPushPop/4-8 160B ± 0% 160B ± 0% ~ (all equal)
ScriptPushPop/16-8 640B ± 0% 640B ± 0% ~ (all equal)
ScriptPushPop/128-8 8.70kB ± 0% 8.70kB ± 0% ~ (all equal)
ScriptPushPop/1024-8 73.2kB ± 0% 73.2kB ± 0% ~ (all equal)
name old allocs/op new allocs/op delta
ScriptFibonacci-8 3.17k ± 0% 3.17k ± 0% -0.03% (p=0.008 n=5+5)
ScriptNestedRefCount-8 10.7k ± 0% 10.7k ± 0% ~ (all equal)
ScriptPushPop/4-8 8.00 ± 0% 8.00 ± 0% ~ (all equal)
ScriptPushPop/16-8 32.0 ± 0% 32.0 ± 0% ~ (all equal)
ScriptPushPop/128-8 259 ± 0% 259 ± 0% ~ (all equal)
ScriptPushPop/1024-8 2.05k ± 0% 2.05k ± 0% ~ (all equal)
It'd be especially nice to internalize static slot, but as we can't compare
slices it's not possible.
2021-12-01 21:36:25 +03:00
Roman Khimov
d01f9da8f3
vm: add invocation tree tracing
2021-11-20 21:57:41 +03:00
Roman Khimov
b0a8f54776
vm: drop LoadArgs method, it's a Legacy remnant
2021-11-20 21:57:41 +03:00
Roman Khimov
cd3670a199
vm: refactor LoadScript* internals a bit
...
Reduce code duplication, provide a single point where scripts are being loaded
into the VM (and new contexts are created).
2021-11-20 21:57:41 +03:00
Roman Khimov
83d0b2f465
interop: move invocation counter from VM to Context
...
It's created for interop and used by interop, VM doesn't care.
2021-11-20 21:57:41 +03:00
Roman Khimov
29cda5112a
vm: simplify NEF loading by providing a special method
...
Offsets are still handled outside of VM to avoid introducing manifest
dependency (that's likely to be circular).
2021-11-20 21:57:41 +03:00
Roman Khimov
82d2231ea6
vm: make retCount an internal member of Context
...
Doesn't make much sense exposing it, it's only used by vm package itself.
2021-11-20 21:57:41 +03:00
Roman Khimov
c53d978edd
vm: drop ParamCount from Context
...
It's unused and it's useless.
2021-11-20 21:57:41 +03:00
Roman Khimov
765235eca3
vm: omit context parameter from Call()
...
It's not used and should not be used to switch from contract to contract.
2021-11-20 21:57:41 +03:00
Roman Khimov
7b713762be
vm: do not increment invocation counter in Call
...
Everywhere it matters (and that's callExFromNative() now) it's incremented
already, so when we're doing Call() at the same time (and it's done to invoke
`_initialize` method) we're effectively double-incrementing it.
2021-11-20 21:57:41 +03:00
Roman Khimov
a577d40de9
vm: move Jump method to the Context
...
Because it's only relevant for execution context itself.
2021-11-20 21:57:35 +03:00
Roman Khimov
3e6ce3c221
vm: add PACKMAP/PACKSTRUCT, extend UNPACK
...
See neo-project/neo-vm#443 .
2021-11-12 17:19:59 +03:00
Evgeniy Stratonikov
5c470a6e3d
vm: catch out-of-bounds exceptions, fix #2208
...
Signed-off-by: Evgeniy Stratonikov <evgeniy@nspcc.ru>
2021-10-11 17:28:00 +03:00
Evgeniy Stratonikov
435463e620
vm: allow big string keys in PICKITEM
...
Because `MaxKeySize` is bigger than integer size, we fail on integer
cast while retreiving items from map. SETITEM is not affected.
Signed-off-by: Evgeniy Stratonikov <evgeniy@nspcc.ru>
2021-10-11 17:14:28 +03:00