before this change, we rely on `using namespace seastar` to use
`seastar::format()` without qualifying the `format()` with its
namespace. this works fine until we changed the parameter type
of format string `seastar::format()` from `const char*` to
`fmt::format_string<...>`. this change practically invited
`seastar::format()` to the club of `std::format()` and `fmt::format()`,
where all members accept a templated parameter as its `fmt`
parameter. and `seastar::format()` is not the best candidate anymore.
despite that argument-dependent lookup (ADT for short) favors the
function which is in the same namespace as its parameter, but
`using namespace` makes `seastar::format()` more competitive,
so both `std::format()` and `seastar::format()` are considered
as the condidates.
that is what is happening scylladb in quite a few caller sites of
`format()`, hence ADT is not able to tell which function the winner
in the name lookup:
```
/__w/scylladb/scylladb/mutation/mutation_fragment_stream_validator.cc:265:12: error: call to 'format' is ambiguous
265 | return format("{} ({}.{} {})", _name_view, s.ks_name(), s.cf_name(), s.id());
| ^~~~~~
/usr/bin/../lib/gcc/x86_64-redhat-linux/14/../../../../include/c++/14/format:4290:5: note: candidate function [with _Args = <const std::basic_string_view<char> &, const seastar::basic_sstring<char, unsigned int, 15> &, const seastar::basic_sstring<char, unsigned int, 15> &, const utils::tagged_uuid<table_id_tag> &>]
4290 | format(format_string<_Args...> __fmt, _Args&&... __args)
| ^
/__w/scylladb/scylladb/seastar/include/seastar/core/print.hh:143:1: note: candidate function [with A = <const std::basic_string_view<char> &, const seastar::basic_sstring<char, unsigned int, 15> &, const seastar::basic_sstring<char, unsigned int, 15> &, const utils::tagged_uuid<table_id_tag> &>]
143 | format(fmt::format_string<A...> fmt, A&&... a) {
| ^
```
in this change, we
change all `format()` to either `fmt::format()` or `seastar::format()`
with following rules:
- if the caller expects an `sstring` or `std::string_view`, change to
`seastar::format()`
- if the caller expects an `std::string`, change to `fmt::format()`.
because, `sstring::operator std::basic_string` would incur a deep
copy.
we will need another change to enable scylladb to compile with the
latest seastar. namely, to pass the format string as a templated
parameter down to helper functions which format their parameters.
to miminize the scope of this change, let's include that change when
bumping up the seastar submodule. as that change will depend on
the seastar change.
Signed-off-by: Kefu Chai <kefu.chai@scylladb.com>
The friendship was needed to get engine and instance cache from manager,
but there's a shorter way to create cotnext with the info it needs.
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
The managerr needs to get two "fuel" configurables from db::config in
order to create context. Instead of carrying db config from callers,
keep the options on existing lang::manager::config and use them.
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
The lang::manager starts with the help of a context because it needs to
have std::shared_ptr<> pointg to cross-shard shared wasm engine and
runner thread. For that a context is created in advance, that then helps
sharing the engine and runner across manager instances.
This patch removes the "context" and replaces it with classical
manager::config. With it, it's lang::manager who's now responsible for
initializing itself.
In order to have cross-shard engine and thread pointers, the start()
method uses invoke_on_others() facility to share the pointer.
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
And, while at it, rename local variable to refer to it to as "manager"
not "wasm". Query processor and database also have getters named
"wasm()", these are not renamed yet to keep patch smaller (and those
getters are going to be reworked further anyway).
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
It's going to become a facade in front of both -- wasm and lua, so keep
it in files with language independent names.
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
get0() dates back from the days where Seastar futures carried tuples, and
get0() was a way to get the first (and usually only) element. Now
it's a distraction, and Seastar is likely to deprecate and remove it.
Replace with seastar::future::get(), which does the same thing.
Add a constructor that builds context out of const manager reference.
The existing one needs to get engine and instance cache and does it via
query_processor. This change lets removing those exports and finally --
drop the wasm::manager -> cql3::query_processor friendship
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
When the q.p. stops it also "stops" the wasm manager. Move this call
into main. The cql test env doesn't need this change, it stops the whole
sharded service which stops instances on its own
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
The wasm::manager is just cql3::wasm_context renamed. It now sits in
lang/wasm* and is started as a sharded service in main (and cql test
env). This move also needs some headers shuffling, but it's not severe
This change is required to make it possible for the wasm::manager to be
shared (by reference) between q.p. and replica::database further
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
Spans are more flexible and can be constructed from any contiguous
container (such as small_vector), or a subrange of such a container.
This can save allocations, so change the signature to accept a span.
Spans cannot be constructed from std::initializer_list, so one such
call site is changed to use construct a span directly from the single
argument.
The wasm engine is moved from replica::database to the query_processor.
The wasm instance cache and compilation thread runner were already there,
but now they're also initialized in the query_processor constructor.
By moving the initialization to the constructor, we can now
be certain that all wasm-related objects (wasm instance cache,
compilation thread runner, and wasm engine, which was already
passed in the constructor) are initialized when we try to use
them because we have to use the query processor to access them
anyway.
The change is also motivated by the fact that we're planning
to take Wasm UDFs out of experimental, after which they should
stop getting special treatment.
Closes#13311
* github.com:scylladb/scylladb:
wasm: move wasm initialization to query_processor constructor
wasm: return wasm instance cache as a reference instead of a pointer
wasm: move wasm engine to query_processor
By moving the initialization to the constructor, we can now
be certain that all wasm-related objects (wasm instance cache,
compilation thread runner, and wasm engine, which was already
passed in the constructor) are initialized when we try to use
them because we have to use the query processor to access them
anyway.
The change is also motivated by the fact that we're planning
to take Wasm UDFs out of experimental, after which they should
stop getting special treatment.
In an incoming change, the wasm instance cache will be modified to be owned
by the query_processor - it will hold an optional instead of a raw
pointer to the cache, so we should stop returning the raw pointer
from the getter as well.
Consequently, the cache is also stored as a reference in wasm::cache,
as it gets the reference from the query_processor.
For consistency with the wasm engine and the wasm alien thread runner,
the name of the getter is also modified to follow the same pattern.
The Wasm compilation is a slow, low priority task, so it should
not compete with reactor threads or the networking core.
To achieve that, we increase the niceness of the thread by 10.
An alternative solution would be to set the priority using
pthread_setschedparam, but it's not currently feasible,
because as long as we're using the SCHED_OTHER policy for our
threads, we cannot select any other priority than 0.
Closes#13307
The compilation of wasm UDFs is performed by a call to a foreign
function, which cannot be divided with yielding points and, as a
result, causes long reactor stalls for big UDFs.
We avoid them by submitting the compilation task to a non-seastar
std::thread, and retrieving the result using seastar::alien.
The thread is created at the start of the program. It executes
tasks from a queue in an infinite loop.
All seastar shards reference the thread through a std::shared_ptr
to a `alien_thread_runner`.
Considering that the compilation takes a long time anyway, the
alien_thread_runner is implemented with focus on simplicity more
than on performance. The tasks are stored in an std::queue, reading
and writing to it is synchronized using an std::mutex for reading/
writing to the queue, and an std::condition_variable waiting until
the queue has elements.
When the destructor of the alien runner is called, an std::nullopt
sentinel is pushed to the queue, and after all remaining tasks are
finished and the sentinel is read, the thread finishes.
After we move the compilation to a alien thread, the completion
of the compilation will be signaled by fulfilling a seastar promise.
As a result, the `precompile` function will return a future, and
because of that, other functions that use the `precompile` functions
will also become futures.
We can do all the neccessary adjustments beforehand, so that the actual
patch that moves the compilation will contain less irrelevant changes.
Currently, because serialize_visitor::operator() is not implemented
for counters, we cannot convert a counter returned by a WASM UDF
to bytes when returning from wasm::run_script().
We could disallow using counters as WASM UDF return types, but an
easier solution which we're already using in Lua UDFs is treating
the returned counters as 64-bit integers when deserializing. This
patch implements the latter approach and adds a test for it.
The wasmtime runtime allocates memory for the executable code of
the WASM programs using mmap and not the seastar allocator. As
a result, the memory that Scylla actually uses becomes not only
the memory preallocated for the seastar allocator but the sum of
that and the memory allocated for executable codes by the WASM
runtime.
To keep limiting the memory used by Scylla, we measure how much
memory do the WASM programs use and if they use too much, compiled
WASM UDFs (modules) that are currently not in use are evicted to
make room.
To evict a module it is required to evict all instances of this
module (the underlying implementation of modules and instances uses
shared pointers to the executable code). For this reason, we add
reference counts to modules. Each instance using a module is a
reference. When an instance is destroyed, a reference is removed.
If all references to a module are removed, the executable code
for this module is deallocated.
The eviction of a module is actually acheved by eviction of all
its references. When we want to free memory for a new module we
repeatedly evict instances from the wasm_instance_cache using its
LRU strategy until some module loses all its instances. This
process may not succeed if the instances currently in use (so not
in the cache) use too much memory - in this case the query also
fails. Otherwise the new module is added to the tracking system.
This strategy may evict some instances unnecessarily, but evicting
modules should not happen frequently, and any more efficient
solution requires an even bigger intervention into the code.
Different users may require different limits for their UDFs. This
patch allows them to configure the size of their cache of wasm,
the maximum size of indivitual instances stored in the cache, the
time after which the instances are evicted, the fuel that all wasm
UDFs are allowed to consume before yielding (for the control of
latency), the fuel that wasm UDFs are allowed to consume in total
(to allow performing longer computations in the UDF without
detecting an infinite loop) and the hard limit of the size of UDFs
that are executed (to avoid large allocations)
This patch replaces all dependencies on the wasmtime
C++ bindings with our new ones.
The wasmtime.hh and wasm_engine.hh files are deleted.
The libwasmtime.a library is no longer required by
configure.py. The SCYLLA_ENABLE_WASMTIME macro is
removed and wasm udfs are now compiled by default
on all architectures.
In terms of implementation, most of code using
wasmtime was moved to the Rust source files. The
remaining code uses names from the new bindings
(which are mostly unchanged). Most of wasmtime objects
are now stored as a rust::Box<>, to make it compatible
with rust lifetime requirements.
Signed-off-by: Wojciech Mitros <wojciech.mitros@scylladb.com>
When executing a wasm UDF, most of the time is spent on
setting up the instance. To minimize its cost, we reuse
the instance using wasm::instance_cache.
This patch adds a wasm instance cache, that stores
a wasmtime instance for each UDF and scheduling group.
The instances are evicted using LRU strategy. The
cache may store some entries for the UDF after evicting
the instance, but they are evicted when the corresponding
UDF is dropped, which greatly limits their number.
The size of stored instances is estimated using the size
of their WASM memories. In order to be able to read the
size of memory, we require that the memory is exported
by the client.
Signed-off-by: Wojciech Mitros <wojciech.mitros@scylladb.com>
Convert most use sites from `co_return coroutine::make_exception`
to `co_await coroutine::return_exception{,_ptr}` where possible.
In cases this is done in a catch clause, convert to
`co_return coroutine::exception`, generating an exception_ptr
if needed.
Signed-off-by: Benny Halevy <bhalevy@scylladb.com>
Closes#10972
To call a UDF that is using WASI, we need to properly
configure the wasmtime instance that it will be called
on. The configuration was missing from udf_cache::load(),
so we add it here.
The free function does not return any value, so we should use
a calling method that does not expect any returns.
This patch adds such a method and uses it.
A test that did not pass without this fix and does pass after
is added.
Signed-off-by: Wojciech Mitros <wojciech.mitros@scylladb.com>
Closes#10935
Because the only available version of wasm ABI did not allow
freeing any allocated memory, a new version of the ABI is
introduced. In this version, the host is required to export
_scylla_malloc and _scylla_free methods, which are later used
for the memory management.
Signed-off-by: Wojciech Mitros <wojciech.mitros@scylladb.com>
One of the issues that comes with compiling programs to WebAssembly
is the lack of a default implementation of a memory allocator. As
a result, the only available solutions to the need of memory allocation
are growing the wasm memory for each new allocated segment, or
implementing one's own memory allocator. To avoid both of these
approaches, for many languages, the user may compile a program to
a WASI target. By doing so, the compiler adds default implementations
of malloc and free methods, and the user can use them for dynamic
memory management.
This patch enables executing programs compiled with WASI by enabling
it in the wasmtime runtime.
Signed-off-by: Wojciech Mitros <wojciech.mitros@scylladb.com>
Different languages may require different ABIs for passing
parameters, etc. This patch adds a requirement for all wasm
UDFs to export an _scylla_abi symbol, that is an 32-bit integer
with a value specifying the ABI version.
Signed-off-by: Wojciech Mitros <wojciech.mitros@scylladb.com>
WebAssembly uses 32-bit address space, while also
having 64-bit integers as it native types. As a result,
when passing size of an object in memory and its address,
it can be combined into one 64-bit value. As a bonus,
if the object is null, we can signal it by passing -1 as
its size.
This patch implements handling of this new ABI and adjusts
expamples in test_wasm.py.
Signed-off-by: Wojciech Mitros <wojciech.mitros@scylladb.com>
Both init_nullable_arg_visitor and, in case
of abstract_type, init_arg_visitor were
the same method with one difference. The
common part was moved to init_abstract_arg,
and the difference remained in the operator()
method.
Signed-off-by: Wojciech Mitros <wojciech.mitros@scylladb.com>
The memory.grow and memory.size wasm methods return
the memory size in pages, and memory.size takes its
argument in the number of pages. A WebAssembly page
has a size of 64KiB, so during memory allocation
we have to divide our desired size in bytes by page
size and round up. Similarly, when reading memory
size we need to multiply the result by 64KiB to
get the size in bytes.
The change affects current naive allocator for
arguments when calling wasm UDFs and the examples
in wasm_test.py - both commented code and compiled
wasm in text representation.
Signed-off-by: Wojciech Mitros <wojciech.mitros@scylladb.com>
Instead of lengthy blurbs, switch to single-line, machine-readable
standardized (https://spdx.dev) license identifiers. The Linux kernel
switched long ago, so there is strong precedent.
Three cases are handled: AGPL-only, Apache-only, and dual licensed.
For the latter case, I chose (AGPL-3.0-or-later and Apache-2.0),
reasoning that our changes are extensive enough to apply our license.
The changes we applied mechanically with a script, except to
licenses/README.md.
Closes#9937
The engine is based on wasmtime and is able to:
- compile wasm text format to bytecode
- run a given compiled function with custom arguments
This implementation is missing crucial features, like running
on any other types than 32-bit integers. It serves as a skeleton
for future full implementation.
