ed/raddebugger
1
0
Fork 0
mirror of https://github.com/Ed94/raddebugger.git synced 2026-06-15 08:32:22 -07:00
Code Issues Packages Projects Releases Wiki Activity

float run/mem/reg generations out of demon, into core; fix issue of demon access synchronization w/ mem gens, causing failure-to-update-stale-parts-of-process-memory-cache

Browse Source
This commit is contained in:
Ryan Fleury
2025-06-28 10:02:51 -07:00
parent e34e8adbaa
commit 709bc2cfab
8 changed files with 87 additions and 73 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/ctrl/ctrl_core.c
+69 -10
View File
@@ -1684,7 +1684,7 @@ ctrl_key_from_process_vaddr_range(CTRL_Handle process, Rng1U64 vaddr_range, B32
{
id_exists = 1;
id_stale = (n->mem_gen < mem_gen);
id_working = (ins_atomic_u64_eval(&n->working_count) != 0);
id_working = (n->working_count != 0);
goto end_fast_lookup;
}
}
@@ -1718,7 +1718,6 @@ ctrl_key_from_process_vaddr_range(CTRL_Handle process, Rng1U64 vaddr_range, B32
if(!id_exists || (id_exists && id_stale && !id_working))
{
B32 node_needs_stream = 0;
U64 *node_working_count = 0;
OS_MutexScopeW(process_stripe->rw_mutex)
{
for(CTRL_ProcessMemoryCacheNode *process_n = process_slot->first; process_n != 0; process_n = process_n->next)
@@ -1751,9 +1750,8 @@ ctrl_key_from_process_vaddr_range(CTRL_Handle process, Rng1U64 vaddr_range, B32
}
if(node_needs_stream)
{
ins_atomic_u64_inc_eval(&range_n->working_count);
range_n->working_count += 1;
}
node_working_count = &range_n->working_count;
break;
}
}
@@ -1762,9 +1760,33 @@ ctrl_key_from_process_vaddr_range(CTRL_Handle process, Rng1U64 vaddr_range, B32
{
if(ctrl_u2ms_enqueue_req(key, process, vaddr_range, zero_terminated, endt_us))
{
// NOTE(rjf): debugging
#if 0
raddbg_log("[0x%I64x, 0x%I64x) push: (gen: %I64u)\n", vaddr_range.min, vaddr_range.max, mem_gen);
#endif
async_push_work(ctrl_mem_stream_work);
requested = 1;
}
else OS_MutexScopeW(process_stripe->rw_mutex)
{
for(CTRL_ProcessMemoryCacheNode *process_n = process_slot->first; process_n != 0; process_n = process_n->next)
{
if(ctrl_handle_match(process_n->handle, process))
{
U64 range_slot_idx = range_hash%process_n->range_hash_slots_count;
CTRL_ProcessMemoryRangeHashSlot *range_slot = &process_n->range_hash_slots[range_slot_idx];
CTRL_ProcessMemoryRangeHashNode *range_n = 0;
for(CTRL_ProcessMemoryRangeHashNode *n = range_slot->first; n != 0; n = n->next)
{
if(hs_id_match(n->id, id))
{
n->working_count -= 1;
break;
}
}
}
}
}
}
}
@@ -1960,6 +1982,12 @@ ctrl_process_write(CTRL_Handle process, Rng1U64 range, void *src)
ProfBeginFunction();
B32 result = dmn_process_write(process.dmn_handle, range, src);
//- rjf: success -> bump generation
if(result)
{
ins_atomic_u64_inc_eval(&ctrl_state->mem_gen);
}
//- rjf: success -> wait for cache updates, for small regions - prefer relatively seamless
// writes within calling frame's "view" of the memory, at the expense of a small amount of
// time.
@@ -2066,7 +2094,7 @@ ctrl_reg_block_from_thread(Arena *arena, CTRL_EntityCtx *ctx, CTRL_Handle handle
// rjf: copy from node
if(node)
{
U64 current_reg_gen = dmn_reg_gen();
U64 current_reg_gen = ctrl_reg_gen();
B32 need_stale = 1;
if(node->reg_gen != current_reg_gen && dmn_thread_read_reg_block(handle.dmn_handle, result))
{
@@ -2124,6 +2152,10 @@ ctrl_thread_write_reg_block(CTRL_Handle thread, void *block)
{
// TODO(rjf): @callstacks immediately reflect this in the call stack cache
B32 good = dmn_thread_write_reg_block(thread.dmn_handle, block);
if(good)
{
ins_atomic_u64_inc_eval(&ctrl_state->reg_gen);
}
return good;
}
@@ -3492,26 +3524,26 @@ ctrl_halt(void)
////////////////////////////////
//~ rjf: Shared Accessor Functions
//- rjf: run generation counter
//- rjf: generation counters
internal U64
ctrl_run_gen(void)
{
U64 result = dmn_run_gen();
U64 result = ins_atomic_u64_eval(&ctrl_state->run_gen);
return result;
}
internal U64
ctrl_mem_gen(void)
{
U64 result = dmn_mem_gen();
U64 result = ins_atomic_u64_eval(&ctrl_state->mem_gen);
return result;
}
internal U64
ctrl_reg_gen(void)
{
U64 result = dmn_reg_gen();
U64 result = ins_atomic_u64_eval(&ctrl_state->reg_gen);
return result;
}
@@ -3831,6 +3863,9 @@ ctrl_thread__entry_point(void *p)
}
ins_atomic_u64_eval_assign(&ctrl_state->ctrl_thread_run_state, 0);
}
ins_atomic_u64_inc_eval(&ctrl_state->run_gen);
ins_atomic_u64_inc_eval(&ctrl_state->mem_gen);
ins_atomic_u64_inc_eval(&ctrl_state->reg_gen);
//- rjf: gather & output logs
LogScopeResult log = log_scope_end(scratch.arena);
@@ -6777,6 +6812,7 @@ ctrl_u2ms_dequeue_req(HS_Key *out_key, CTRL_Handle *out_process, Rng1U64 *out_va
ASYNC_WORK_DEF(ctrl_mem_stream_work)
{
#define CTRL_MEM_STREAM_WORK_DEBUG 0
ProfBeginFunction();
CTRL_ProcessMemoryCache *cache = &ctrl_state->process_memory_cache;
@@ -6812,6 +6848,12 @@ ASYNC_WORK_DEF(ctrl_mem_stream_work)
void *range_base = 0;
U64 zero_terminated_size = 0;
U64 pre_read_mem_gen = ctrl_mem_gen();
B32 pre_run_state = ins_atomic_u64_eval(&ctrl_state->ctrl_thread_run_state);
#if CTRL_MEM_STREAM_WORK_DEBUG
Log *log = log_alloc();
log_select(log);
log_scope_begin();
#endif
{
range_size = dim_1u64(vaddr_range_clamped);
U64 page_size = os_get_system_info()->page_size;
@@ -6873,6 +6915,23 @@ ASYNC_WORK_DEF(ctrl_mem_stream_work)
}
}
U64 post_read_mem_gen = ctrl_mem_gen();
B32 post_run_state = ins_atomic_u64_eval(&ctrl_state->ctrl_thread_run_state);
// NOTE(rjf): debugging
#if CTRL_MEM_STREAM_WORK_DEBUG
{
Temp scratch = scratch_begin(0, 0);
String8 sample_data_str = str8_lit("no data");
if(range_base != 0)
{
String8 sample_data = str8((U8*)range_base + 0x100, 16);
String8List sample_data_strs = numeric_str8_list_from_data(scratch.arena, 16, sample_data, 1);
sample_data_str = str8_list_join(scratch.arena, &sample_data_strs, &(StringJoin){.sep = str8_lit(", ")});
}
LogScopeResult log = log_scope_end(scratch.arena);
raddbg_log("[0x%I64x, 0x%I64x) { pre_gen: %I64u, post_gen: %I64u, pre_run: %i, post_run: %i, bytes: [%S], info:```%S``` }\n", vaddr_range.min, vaddr_range.max, pre_read_mem_gen, post_read_mem_gen, pre_run_state, post_run_state, sample_data_str, log.strings[LogMsgKind_Info]);
scratch_end(scratch);
}
#endif
//- rjf: read successful -> submit to hash store
U128 hash = {0};
@@ -6902,7 +6961,7 @@ ASYNC_WORK_DEF(ctrl_mem_stream_work)
{
range_n->mem_gen = post_read_mem_gen;
}
ins_atomic_u64_dec_eval(&range_n->working_count);
range_n->working_count -= 1;
goto commit__break_all;
}
}