vanja/neural-amp-modeler-lv2
1
0
Fork 0
mirror of https://github.com/mikeoliphant/neural-amp-modeler-lv2.git synced 2026-05-06 19:50:11 +02:00
Code Issues Packages Projects Releases Wiki Activity

rework worker implementation, dont rely on class variables

Browse Source 
previous implementation was racy and bound to issues when more
than 1 file change request happened before worker was triggered.

using C++ move assignment is nice, but LV2 worker is a C API
that does not fit non-POD types very well, leading to awkward
implementations alike before with current + staged + deleted models.

let us "downgrade" to raw pointers, which are C compatible.
since LV2 worker rules are well defined, any crashes or racy
behaviour can be considered host-side bugs.

Signed-off-by: falkTX <falktx@falktx.com>
This commit is contained in:
falkTX
2023-06-15 17:00:02 +02:00
parent 0d7954cf2a
commit 4e9b7d6b49
2 changed files with 90 additions and 83 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/nam_plugin.cpp
+73 -73
View File
@@ -15,6 +15,11 @@ namespace NAM {
currentModelPath.reserve(MAX_FILE_NAME+1);
}
Plugin::~Plugin()
{
delete currentModel;
}
bool Plugin::initialize(double rate, const LV2_Feature* const* features) noexcept
{
for (size_t i = 0; features[i]; ++i) {
@@ -61,41 +66,41 @@ namespace NAM {
return true;
}
// runs on non-RT, can block or use [de]allocations
LV2_Worker_Status Plugin::work(LV2_Handle instance, LV2_Worker_Respond_Function respond, LV2_Worker_Respond_Handle handle,
uint32_t size, const void* data)
{
switch (*((const uint32_t*)data))
switch (*(const LV2WorkType*)data)
{
case kWorkTypeLoad:
auto msg = reinterpret_cast<const LV2LoadModelMsg*>(data);
{
auto msg = static_cast<const LV2LoadModelMsg*>(data);
auto nam = static_cast<NAM::Plugin*>(instance);
try
{
// If we had a previous model, delete it
if (nam->deleteModel)
{
nam->deleteModel.reset();
}
// load model from path
const size_t pathlen = strlen(msg->path);
::DSP* model;
if (strlen(msg->path) == 0)
{
if (pathlen == 0 || pathlen >= MAX_FILE_NAME)
{
// avoid logging an error on an empty path.
// but do clear the model.
nam->stagedModel = nullptr;
nam->stagedModelPath = msg->path;
} else
model = nullptr;
}
else
{
lv2_log_trace(&nam->logger, "Staging model change: `%s`\n", msg->path);
nam->stagedModel = get_dsp(msg->path);
nam->stagedModelPath = msg->path;
model = get_dsp(msg->path).release();
// Enable model loudness normalization
nam->stagedModel->SetNormalize(true);
model->SetNormalize(true);
}
LV2WorkType response = kWorkTypeSwitch;
LV2SwitchModelMsg response = { kWorkTypeSwitch, {}, model };
memcpy(response.path, msg->path, pathlen);
respond(handle, sizeof(response), &response);
return LV2_WORKER_SUCCESS;
@@ -106,41 +111,49 @@ namespace NAM {
}
break;
}
}
return LV2_WORKER_ERR_UNKNOWN;
}
LV2_Worker_Status Plugin::work_response(LV2_Handle instance, uint32_t size, const void* data)
{
switch (*((const uint32_t*)data))
{
case kWorkTypeSwitch:
try
case kWorkTypeFree:
{
auto nam = static_cast<NAM::Plugin*>(instance);
std::swap(nam->currentModel, nam->stagedModel);
nam->currentModelPath = nam->stagedModelPath;
assert(nam->currentModelPath.capacity() >= MAX_FILE_NAME + 1);
nam->stateChanged = true;
nam->deleteModel = std::move(nam->stagedModel);
nam->write_set_patch(nam->currentModelPath);
auto msg = static_cast<const LV2FreeModelMsg*>(data);
delete msg->model;
return LV2_WORKER_SUCCESS;
}
catch (std::exception& e)
{
}
break;
case kWorkTypeSwitch:
// should not happen!
break;
}
return LV2_WORKER_ERR_UNKNOWN;
}
// runs on RT, right after process(), must not block or [de]allocate memory
LV2_Worker_Status Plugin::work_response(LV2_Handle instance, uint32_t size, const void* data)
{
if (*(const LV2WorkType*)data != kWorkTypeSwitch)
return LV2_WORKER_ERR_UNKNOWN;
auto msg = static_cast<const LV2SwitchModelMsg*>(data);
auto nam = static_cast<NAM::Plugin*>(instance);
// prepare reply for deleting old model
LV2FreeModelMsg reply = { kWorkTypeFree, nam->currentModel };
// swap current model with new one
nam->currentModel = msg->model;
nam->currentModelPath = msg->path;
assert(nam->currentModelPath.capacity() >= MAX_FILE_NAME + 1);
// send reply
nam->schedule->schedule_work(nam->schedule->handle, sizeof(reply), &reply);
// report change to host/ui
nam->write_current_path();
nam->write_state_changed();
return LV2_WORKER_SUCCESS;
}
void Plugin::process(uint32_t n_samples) noexcept
{
@@ -152,34 +165,28 @@ namespace NAM {
if (event->body.type == uris.atom_Object)
{
const auto obj = reinterpret_cast<LV2_Atom_Object*>(&event->body);
if (obj->body.otype == uris.patch_Get) {
lv2_atom_forge_frame_time(&atom_forge, 0);
write_set_patch(currentModelPath);
if (obj->body.otype == uris.patch_Get)
{
write_current_path();
}
if (obj->body.otype == uris.patch_Set)
else if (obj->body.otype == uris.patch_Set)
{
const LV2_Atom* property = NULL;
const LV2_Atom* file_path = NULL;
lv2_atom_object_get(obj, uris.patch_property, &property, 0);
lv2_atom_object_get(obj,
uris.patch_property, &property,
uris.patch_value, &file_path,
0);
if (property && (property->type == uris.atom_URID))
if (property && property->type == uris.atom_URID &&
((const LV2_Atom_URID*)property)->body == uris.model_Path &&
file_path && file_path->type == uris.atom_Path &&
file_path->size > 0 && file_path->size < MAX_FILE_NAME)
{
if (((const LV2_Atom_URID*)property)->body == uris.model_Path)
{
lv2_atom_object_get(obj, uris.patch_value, &file_path, 0);
if (file_path && (file_path->size > 0) && (file_path->size < MAX_FILE_NAME))
{
LV2LoadModelMsg msg = { kWorkTypeLoad, {} };
memcpy(msg.path, (const char*)LV2_ATOM_BODY_CONST(file_path), file_path->size);
schedule->schedule_work(schedule->handle, sizeof(msg), &msg);
}
}
LV2LoadModelMsg msg = { kWorkTypeLoad, {} };
memcpy(msg.path, file_path + 1, file_path->size);
schedule->schedule_work(schedule->handle, sizeof(msg), &msg);
}
}
}
@@ -236,15 +243,6 @@ namespace NAM {
ports.audio_out[i] *= outputLevel;
}
}
if (stateChanged)
{
stateChanged = false;
lv2_atom_forge_frame_time(&atom_forge, 0);
write_state_changed();
}
lv2_atom_forge_pop(&atom_forge,&sequence_frame);
}
LV2_State_Status Plugin::save(LV2_Handle instance, LV2_State_Store_Function store, LV2_State_Handle handle,
@@ -361,16 +359,17 @@ namespace NAM {
return result;
}
void Plugin::write_set_patch( std::string filename)
void Plugin::write_current_path()
{
LV2_Atom_Forge_Frame frame;
lv2_atom_forge_frame_time(&atom_forge, 0);
lv2_atom_forge_object(&atom_forge, &frame, 0, uris.patch_Set);
lv2_atom_forge_key(&atom_forge, uris.patch_property);
lv2_atom_forge_urid(&atom_forge, uris.model_Path);
lv2_atom_forge_key(&atom_forge, uris.patch_value);
lv2_atom_forge_path(&atom_forge, filename.c_str(), filename.length());
lv2_atom_forge_path(&atom_forge, currentModelPath.c_str(), currentModelPath.length() + 1);
lv2_atom_forge_pop(&atom_forge, &frame);
}
@@ -379,6 +378,7 @@ namespace NAM {
{
LV2_Atom_Forge_Frame frame;
lv2_atom_forge_frame_time(&atom_forge, 0);
lv2_atom_forge_object(&atom_forge, &frame, 0, uris.state_StateChanged);
/* object with no properties */
src/nam_plugin.h
+17 -10
View File
@@ -29,7 +29,8 @@ namespace NAM {
enum LV2WorkType {
kWorkTypeLoad,
kWorkTypeSwitch
kWorkTypeSwitch,
kWorkTypeFree
};
struct LV2LoadModelMsg {
@@ -37,6 +38,17 @@ namespace NAM {
char path[MAX_FILE_NAME];
};
struct LV2SwitchModelMsg {
LV2WorkType type;
char path[MAX_FILE_NAME];
::DSP* model;
};
struct LV2FreeModelMsg {
LV2WorkType type;
::DSP* model;
};
class Plugin {
public:
struct Ports {
@@ -54,23 +66,18 @@ namespace NAM {
LV2_Log_Logger logger = {};
LV2_Worker_Schedule* schedule = nullptr;
std::unique_ptr<::DSP> currentModel;
std::unique_ptr<::DSP> stagedModel;
std::unique_ptr<::DSP> deleteModel;
bool stateChanged = false;
::DSP* currentModel = nullptr;
std::string currentModelPath;
std::string stagedModelPath;
std::unordered_map<std::string, double> mNAMParams = {};
Plugin();
~Plugin() = default;
~Plugin();
bool initialize(double rate, const LV2_Feature* const* features) noexcept;
void process(uint32_t n_samples) noexcept;
void write_set_patch(std::string filename);
void write_current_path();
void write_state_changed();
static LV2_Worker_Status work(LV2_Handle instance, LV2_Worker_Respond_Function respond, LV2_Worker_Respond_Handle handle,