diff options
| -rw-r--r-- | VERSION | 2 | ||||
| -rw-r--r-- | vm.c | 251 |
2 files changed, 188 insertions, 65 deletions
@@ -1 +1 @@ -0.1.2905 +0.1.2907 @@ -34,6 +34,7 @@ typedef union _vm_port_t vm_port_t; typedef union _vm_const_port_t vm_const_port_t; typedef struct _vm_stack_t vm_stack_t; typedef struct _plughandle_t plughandle_t; +typedef struct _forge_t forge_t; typedef double num_t; @@ -70,8 +71,12 @@ struct _plughandle_t { vm_const_port_t in [CTRL_MAX]; vm_port_t out [CTRL_MAX]; - num_t in0 [CTRL_MAX]; + float in0 [CTRL_MAX]; num_t out0 [CTRL_MAX]; + float inm [CTRL_MAX]; + float outm [CTRL_MAX]; + bool inf [CTRL_MAX]; + bool outf [CTRL_MAX]; PROPS_T(props, MAX_NPROPS); plugstate_t state; @@ -92,6 +97,12 @@ struct _plughandle_t { timely_t timely; }; +struct _forge_t { + LV2_Atom_Forge forge; + LV2_Atom_Forge_Frame frame; + LV2_Atom_Forge_Ref ref; +}; + static inline void _stack_clear(vm_stack_t *stack) { @@ -268,8 +279,59 @@ connect_port(LV2_Handle instance, uint32_t port, void *data) #define CLIP(a, v, b) fmin(fmax(a, v), b) static void -run_internal(plughandle_t *handle, uint32_t frames, bool notify, - const float *in [CTRL_MAX], float *out [CTRL_MAX]) +run_pre(plughandle_t *handle) +{ + // reset maximum values and notification flags + for(unsigned i = 0; i < CTRL_MAX; i++) + { + handle->inm[i] = 0.f; + handle->inf[i] = false; + + handle->outm[i] = 0.f; + handle->outf[i] = false; + } +} + +static void +run_post(plughandle_t *handle, uint32_t frames) +{ + for(unsigned i = 0; i < CTRL_MAX; i++) + { + if(handle->inf[i]) // port needs notification + { + LV2_Atom_Forge_Frame tup_frame; + if(handle->ref) + handle->ref = lv2_atom_forge_frame_time(&handle->forge, frames); + if(handle->ref) + handle->ref = lv2_atom_forge_tuple(&handle->forge, &tup_frame); + if(handle->ref) + handle->ref = lv2_atom_forge_int(&handle->forge, i + 2); + if(handle->ref) + handle->ref = lv2_atom_forge_float(&handle->forge, handle->inm[i]); + if(handle->ref) + lv2_atom_forge_pop(&handle->forge, &tup_frame); + } + + if(handle->outf[i]) // port needs notification + { + LV2_Atom_Forge_Frame tup_frame; + if(handle->ref) + handle->ref = lv2_atom_forge_frame_time(&handle->forge, frames); + if(handle->ref) + handle->ref = lv2_atom_forge_tuple(&handle->forge, &tup_frame); + if(handle->ref) + handle->ref = lv2_atom_forge_int(&handle->forge, i + 10); + if(handle->ref) + handle->ref = lv2_atom_forge_float(&handle->forge, handle->outm[i]); + if(handle->ref) + lv2_atom_forge_pop(&handle->forge, &tup_frame); + } + } +} + +static void +run_internal(plughandle_t *handle, uint32_t frames, + const float *in [CTRL_MAX], float *out [CTRL_MAX], forge_t forgs [CTRL_MAX]) { for(unsigned i = 0; i < CTRL_MAX; i++) { @@ -282,19 +344,10 @@ run_internal(plughandle_t *handle, uint32_t frames, bool notify, handle->needs_recalc = true; handle->in0[i] = in1; - if(notify) //FIXME better report e.g. maximum + if(fabsf(in1) > fabsf(handle->inm[i])) { - LV2_Atom_Forge_Frame tup_frame; - if(handle->ref) - handle->ref = lv2_atom_forge_frame_time(&handle->forge, frames); - if(handle->ref) - handle->ref = lv2_atom_forge_tuple(&handle->forge, &tup_frame); - if(handle->ref) - handle->ref = lv2_atom_forge_int(&handle->forge, i + 2); - if(handle->ref) - handle->ref = lv2_atom_forge_float(&handle->forge, in1); - if(handle->ref) - lv2_atom_forge_pop(&handle->forge, &tup_frame); + handle->inm[i] = in1; + handle->inf[i] = true; // notify in run_post } } } @@ -839,19 +892,19 @@ loop: { { *out[i] = out1; - if(notify) //FIXME better report e.g. maximum + if(forgs && (handle->vm_plug == VM_PLUG_ATOM) ) { - LV2_Atom_Forge_Frame tup_frame; - if(handle->ref) - handle->ref = lv2_atom_forge_frame_time(&handle->forge, frames); - if(handle->ref) - handle->ref = lv2_atom_forge_tuple(&handle->forge, &tup_frame); - if(handle->ref) - handle->ref = lv2_atom_forge_int(&handle->forge, i + 10); + // send changes on atom output ports + if(forgs[i].ref) + forgs[i].ref = lv2_atom_forge_frame_time(&forgs[i].forge, frames); if(handle->ref) - handle->ref = lv2_atom_forge_float(&handle->forge, out1); - if(handle->ref) - lv2_atom_forge_pop(&handle->forge, &tup_frame); + forgs[i].ref = lv2_atom_forge_float(&forgs[i].forge, out1); + } + + if(fabsf(out1) > fabsf(handle->outm[i])) + { + handle->outm[i] = out1; + handle->outf[i] = true; // notify out run_post } } } @@ -867,6 +920,8 @@ run_control(LV2_Handle instance, uint32_t nsamples) lv2_atom_forge_set_buffer(&handle->forge, (uint8_t *)handle->notify, capacity); handle->ref = lv2_atom_forge_sequence_head(&handle->forge, &frame, 0); + run_pre(handle); + int64_t last_t = 0; LV2_ATOM_SEQUENCE_FOREACH(handle->control, ev) { @@ -910,10 +965,11 @@ run_control(LV2_Handle instance, uint32_t nsamples) handle->out[7].flt }; - const bool notify = true; - run_internal(handle, nsamples -1, notify, in, out); + run_internal(handle, nsamples -1, in, out, NULL); } + run_post(handle, nsamples - 1); + if(handle->ref) handle->ref = lv2_atom_forge_frame_time(&handle->forge, nsamples - 1); if(handle->ref) @@ -922,13 +978,8 @@ run_control(LV2_Handle instance, uint32_t nsamples) if(handle->ref) lv2_atom_forge_pop(&handle->forge, &frame); else - { lv2_atom_sequence_clear(handle->notify); - if(handle->log) - lv2_log_trace(&handle->logger, "notify sequence overflowed\n"); - } - handle->off += nsamples; } @@ -969,8 +1020,7 @@ run_cv_audio_advance(plughandle_t *handle, const LV2_Atom_Object *obj, &handle->out[7].flt[i] }; - const bool notify = (i == 0); // FIXME - run_internal(handle, i, notify, in, out); + run_internal(handle, i, in, out, NULL); } } } @@ -985,6 +1035,8 @@ run_cv_audio(LV2_Handle instance, uint32_t nsamples) lv2_atom_forge_set_buffer(&handle->forge, (uint8_t *)handle->notify, capacity); handle->ref = lv2_atom_forge_sequence_head(&handle->forge, &frame, 0); + run_pre(handle); + int64_t last_t = 0; LV2_ATOM_SEQUENCE_FOREACH(handle->control, ev) { @@ -1004,6 +1056,8 @@ run_cv_audio(LV2_Handle instance, uint32_t nsamples) handle->needs_sync = false; } + run_post(handle, nsamples - 1); + if(handle->ref) handle->ref = lv2_atom_forge_frame_time(&handle->forge, nsamples - 1); if(handle->ref) @@ -1018,6 +1072,27 @@ run_cv_audio(LV2_Handle instance, uint32_t nsamples) } static void +run_atom_advance(plughandle_t *handle, const LV2_Atom_Object *obj, + uint32_t from, uint32_t to, const float *in [CTRL_MAX], float *out [CTRL_MAX], + forge_t forgs [CTRL_MAX]) +{ + if(from == to) // just run timely_advance for void range + { + timely_advance(&handle->timely, obj, from, to); + } + else + { + for(unsigned i = from; i < to; i++) + { + if(timely_advance(&handle->timely, obj, i, i + 1)) + obj = NULL; // invalidate obj for further steps if handled + + run_internal(handle, i, in, out, forgs); + } + } +} + +static void run_atom(LV2_Handle instance, uint32_t nsamples) { plughandle_t *handle = instance; @@ -1027,30 +1102,18 @@ run_atom(LV2_Handle instance, uint32_t nsamples) lv2_atom_forge_set_buffer(&handle->forge, (uint8_t *)handle->notify, capacity); handle->ref = lv2_atom_forge_sequence_head(&handle->forge, &frame, 0); - int64_t last_t = 0; - LV2_ATOM_SEQUENCE_FOREACH(handle->control, ev) - { - const LV2_Atom *atom= &ev->body; - const LV2_Atom_Object *obj = (const LV2_Atom_Object *)&ev->body; - - props_advance(&handle->props, &handle->forge, ev->time.frames, obj, &handle->ref); - timely_advance(&handle->timely, obj, last_t, ev->time.frames); - - last_t = ev->time.frames; - } - timely_advance(&handle->timely, NULL, last_t, nsamples); - - if(handle->needs_sync) - { - props_set(&handle->props, &handle->forge, nsamples - 1, handle->vm_graph, &handle->ref); - handle->needs_sync = false; - } + run_pre(handle); + forge_t forgs [CTRL_MAX]; //FIXME make part of plughandle_t float pin [CTRL_MAX]; float pout [CTRL_MAX]; for(unsigned i = 0; i < CTRL_MAX; i++) { + memcpy(&forgs[i].forge, &handle->forge, sizeof(LV2_Atom_Forge)); //FIXME do at instantiation time + lv2_atom_forge_set_buffer(&forgs[i].forge, (uint8_t *)handle->out[i].seq, handle->out[i].seq->atom.size); + forgs[i].ref = lv2_atom_forge_sequence_head(&forgs[i].forge, &forgs[i].frame, 0); + pin[i] = handle->in0[i]; pout[i] = handle->out0[i]; } @@ -1077,27 +1140,79 @@ run_atom(LV2_Handle instance, uint32_t nsamples) &pout[7] }; - //FIXME needs to be muxed with control - for(unsigned i = 0; i < CTRL_MAX; i++) + const unsigned nseqs = CTRL_MAX + 1; + const LV2_Atom_Sequence *seqs [nseqs]; + const LV2_Atom_Event *evs [nseqs]; + + for(unsigned i = 0; i < nseqs; i++) { - LV2_ATOM_SEQUENCE_FOREACH(handle->in[i].seq, ev) - { - const LV2_Atom_Float *f32 = (const LV2_Atom_Float *)&ev->body; + seqs[i] = (i == 0) + ? handle->control + : handle->in[i-1].seq; - //printf("got atom:Float: %u %f\n", i, f32->body); + evs[i] = lv2_atom_sequence_begin(&seqs[i]->body); + } - if(f32->atom.type == handle->forge.Float) + int64_t last_t = 0; + while(true) + { + int nxt = -1; + int64_t frames = nsamples; + + // search next event + for(unsigned i = 0; i < nseqs; i++) + { + if(!evs[i] || lv2_atom_sequence_is_end(&seqs[i]->body, seqs[i]->atom.size, evs[i])) { - pin[i] = f32->body; + evs[i] = NULL; // invalidate, sequence has been drained + continue; + } + + if(evs[i]->time.frames < frames) + { + frames = evs[i]->time.frames; + nxt = i; + } + } + + if(nxt == -1) + break; // no events anymore, exit loop - const bool notify = true; //FIXME - run_internal(handle, nsamples - 1, notify, in, out); //FIXME use ev.time.frames + // handle event + { + const bool is_control = (nxt == 0); // is event from control port? + const LV2_Atom_Event *ev = evs[nxt]; + const LV2_Atom *atom= &ev->body; + const LV2_Atom_Object *obj = (const LV2_Atom_Object *)&ev->body; + const LV2_Atom_Float *f32 = (const LV2_Atom_Float *)&ev->body; - //FIXME send floats to output sequences + if(is_control) + { + props_advance(&handle->props, &handle->forge, ev->time.frames, obj, &handle->ref); + } + else if(f32->atom.type == handle->forge.Float) + { + pin[nxt-1] = f32->body; } + + run_atom_advance(handle, is_control ? obj : NULL, last_t, ev->time.frames, in, out, forgs); + + last_t = ev->time.frames; } + + // advance event iterator on active sequence + evs[nxt] = lv2_atom_sequence_next(evs[nxt]); + } + run_atom_advance(handle, NULL, last_t, nsamples, in, out, forgs); + + if(handle->needs_sync) + { + props_set(&handle->props, &handle->forge, nsamples - 1, handle->vm_graph, &handle->ref); + handle->needs_sync = false; } + run_post(handle, nsamples - 1); + if(handle->ref) handle->ref = lv2_atom_forge_frame_time(&handle->forge, nsamples - 1); if(handle->ref) @@ -1108,6 +1223,14 @@ run_atom(LV2_Handle instance, uint32_t nsamples) else lv2_atom_sequence_clear(handle->notify); + for(unsigned i = 0; i < CTRL_MAX; i++) + { + if(forgs[i].ref) + lv2_atom_forge_pop(&forgs[i].forge, &forgs[i].frame); + else + lv2_atom_sequence_clear(handle->out[i].seq); + } + handle->off += nsamples; } |