mirror of
https://github.com/tangerineturtle1/NX2navi.git
synced 2026-11-16 13:59:56 -04:00
Every report adds a line for video memory outside VMA (the driver's usage and budget per heap against VMA's blocks, and counts of pipelines by kind and by last use, stage libraries, compute pipelines, layouts and descriptor pools, with the swapchain, DLSS and FSR sizes), a line that splits the process's commit by owner (the game's heap and maps, device-local memory, VMA's host blocks, the C runtime heap with the shader caches, pipeline tables, texture cache, log queue and flight recorder in it, the driver's pipeline cache data and the rest), and a line naming what grew by more than 64 MiB since the report before and since play began. Every ten minutes of play and at exit a ranking line orders the categories by growth and marks those that only climbed; tests/mem_growth_check.py checks that bookkeeping. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
261 lines
11 KiB
C
261 lines
11 KiB
C
// Contains code Copyright (c) 2026 MacNativeGames contributors (MIT; see THIRD_PARTY_NOTICES.md).
|
|
// The SVC entry of the lifted code: preemption checkpoint, trace, then the thread and
|
|
// synchronization SVCs (kernel.c), IPC (ipc.c), memory (memory.c), Break, OutputDebugString, GetInfo and ExitProcess. The 21 SVCs
|
|
// NX2navi has but never reaches on its boot path stop with a named fatal, the caller and the guest backtrace.
|
|
#include <stdio.h>
|
|
#include <string.h>
|
|
|
|
#include "coop.h"
|
|
#include "fs.h"
|
|
#include "host.h"
|
|
#include "kernel.h"
|
|
#include "vk.h"
|
|
#include "platform/platform.h"
|
|
|
|
#define R_OK 0u
|
|
#define ST_FREE 0u
|
|
#define ST_STACK 0xBu
|
|
#define PERM_RW 3u
|
|
#define PROGRAM_ID 0x0400486024dec000ull // the game's title id (GetInfo 18); fs_host.c NX2NAVI_APPLICATION_ID is the same
|
|
|
|
static uint64_t g_svcs;
|
|
void (*volatile svc_test_exit_hook)(unsigned code);
|
|
|
|
RtStats rt_stats(void) {
|
|
return (RtStats){__atomic_load_n(&g_svcs, __ATOMIC_RELAXED), kern_threads_created(), hostfn_calls()};
|
|
}
|
|
|
|
static int readable(uint64_t a, uint64_t n) { return mem_host_readable(a, n); }
|
|
|
|
void kern_backtrace(const Cpu *c) {
|
|
char d[160];
|
|
rt_logf(" lr %s\n", ldr_describe(c->x[30], d, sizeof d));
|
|
uint64_t fp = c->x[29];
|
|
for (int n = 0; n < 32 && fp && !(fp & 7) && readable(fp, 16); n++) {
|
|
uint64_t next = LD64(fp), lr = LD64(fp + 8);
|
|
rt_logf(" guest frame %2d: %s\n", n, ldr_describe(lr, d, sizeof d));
|
|
if (next <= fp || next - fp > (64u << 20)) break;
|
|
fp = next;
|
|
}
|
|
}
|
|
|
|
static const struct { uint32_t imm; const char *name; } unimplemented[] = {
|
|
{0x13, "MapSharedMemory"}, {0x14, "UnmapSharedMemory"},
|
|
{0x23, "SendAsyncRequestWithUserBuffer"}, {0x28, "ReturnFromException"}, {0x2A, "FlushEntireDataCache"},
|
|
{0x2F, "GetLastThreadInfo"}, {0x32, "SetThreadActivity"}, {0x33, "GetThreadContext3"},
|
|
{0x36, "SynchronizePreemptionState"}, {0x40, "CreateSession"}, {0x41, "AcceptSession"},
|
|
{0x43, "ReplyAndReceive"}, {0x44, "ReplyAndReceiveWithUserBuffer"}, {0x48, "MapPhysicalMemoryUnsafe"},
|
|
{0x49, "UnmapPhysicalMemoryUnsafe"}, {0x4B, "CreateCodeMemory"}, {0x50, "CreateSharedMemory"},
|
|
{0x51, "MapTransferMemory"}, {0x52, "UnmapTransferMemory"}, {0x70, "CreatePort"}, {0x72, "ConnectToPort"},
|
|
};
|
|
|
|
static __attribute__((noreturn)) void svc_fatal(Cpu *c, uint32_t imm, uint64_t pc, const char *why) {
|
|
const char *name = "unknown";
|
|
for (unsigned i = 0; i < sizeof unimplemented / sizeof *unimplemented; i++)
|
|
if (unimplemented[i].imm == imm) name = unimplemented[i].name;
|
|
char at[160];
|
|
fflush(stdout);
|
|
rt_logf("FATAL: svc #%#x %s at %s: %s\n x0 %#llx x1 %#llx x2 %#llx x3 %#llx\n", imm, name,
|
|
ldr_describe(pc, at, sizeof at), why, (unsigned long long)c->x[0], (unsigned long long)c->x[1],
|
|
(unsigned long long)c->x[2], (unsigned long long)c->x[3]);
|
|
kern_backtrace(c);
|
|
rt_fatal("svc #%#x %s is not implemented", imm, name);
|
|
}
|
|
|
|
// QueryMemory sees the guest main thread's stack, which memory.c does not know: Stack inside it, free ranges
|
|
// around it clipped (the way memory.c clips moved stacks).
|
|
static void main_stack_query(uint64_t a, RtMemoryInfo *r) {
|
|
uint64_t mb, ms;
|
|
kern_main_stack(&mb, &ms);
|
|
if (!ms) return;
|
|
uint64_t me = mb + ms;
|
|
if (a >= mb && a < me) {
|
|
*r = (RtMemoryInfo){.base = mb, .size = ms, .state = ST_STACK, .perm = PERM_RW};
|
|
return;
|
|
}
|
|
if (r->state != ST_FREE) return;
|
|
uint64_t lo = r->base, hi = r->base + r->size;
|
|
if (me <= a && me > lo) lo = me;
|
|
if (mb > a && mb < hi) hi = mb;
|
|
r->base = lo;
|
|
r->size = hi - lo;
|
|
}
|
|
|
|
static int on_main_stack(uint64_t a, uint64_t n) {
|
|
uint64_t mb, ms;
|
|
kern_main_stack(&mb, &ms);
|
|
return ms && a < mb + ms && mb < a + n;
|
|
}
|
|
|
|
static uint64_t user_exception_context(void) {
|
|
static void *page;
|
|
void *p = __atomic_load_n(&page, __ATOMIC_ACQUIRE);
|
|
if (!p) {
|
|
void *mine = mem_kernel_alloc(0x1000, 0x1000), *expected = NULL;
|
|
if (__atomic_compare_exchange_n(&page, &expected, mine, 0, __ATOMIC_ACQ_REL, __ATOMIC_ACQUIRE)) p = mine;
|
|
else {
|
|
mem_kernel_free(mine, 0x1000);
|
|
p = expected;
|
|
}
|
|
}
|
|
return (uint64_t)(uintptr_t)p;
|
|
}
|
|
|
|
// GetInfo(w1 id, w2 handle, x3 sub id -> x1)
|
|
static void get_info(Cpu *c, uint64_t pc) {
|
|
uint32_t id = (uint32_t)c->x[1];
|
|
uint64_t v = 0;
|
|
if (mem_info(id, &v)) {
|
|
c->x[0] = R_OK;
|
|
c->x[1] = v;
|
|
return;
|
|
}
|
|
switch (id) {
|
|
case 0: v = 0x3f; break; // core mask: cores 0-5 (Switch 2 application cores; see kernel.c NCORE)
|
|
case 1: v = 0x0ffffffff0000000ull; break; // priority mask
|
|
case 8: v = 0; break; // debugger attached
|
|
case 10: v = 0; break; // idle tick count
|
|
case 11: // random entropy
|
|
if (!plat_random(&v, sizeof v)) rt_fatal("GetInfo 11: no random bytes from the system");
|
|
break;
|
|
case 18: v = PROGRAM_ID; break;
|
|
case 20: v = user_exception_context(); break; // the SDK keeps e.g. the dying message at +0x1C0
|
|
case 23: v = 1; break;
|
|
case 29: // region the SDK records at nnosInitialize (address, then
|
|
case 30: v = 0; break;
|
|
case 33: v = 0x1000; break; // VAMM control region size: mapped at 0x300000000 and made into a unit heap
|
|
default: {
|
|
char why[64];
|
|
snprintf(why, sizeof why, "GetInfo id %u (handle %#x, sub id %#llx)", id, (uint32_t)c->x[2],
|
|
(unsigned long long)c->x[3]);
|
|
svc_fatal(c, 0x29, pc, why);
|
|
}
|
|
}
|
|
c->x[0] = R_OK;
|
|
c->x[1] = v;
|
|
}
|
|
|
|
static void svc_body(Cpu *c, uint32_t imm, uint64_t pc);
|
|
|
|
void rt_svc(Cpu *c, uint32_t imm, uint64_t pc) {
|
|
kern_checkpoint();
|
|
if (__builtin_expect(!kern_host_track || !c->thread, 1)) {
|
|
svc_body(c, imm, pc);
|
|
return;
|
|
}
|
|
void *t = c->thread; // kern_host_track: the SVC counted in the thread's host depth (hostfn.c hostfn_tracked)
|
|
int *d = kern_host_depth(t);
|
|
__atomic_store_n(d, *d + 1, __ATOMIC_RELAXED);
|
|
svc_body(c, imm, pc);
|
|
int n = *d - 1;
|
|
__atomic_store_n(d, n, __ATOMIC_RELAXED);
|
|
if (!n && kern_preempt_pending(t)) kern_host_left(t);
|
|
}
|
|
|
|
static void svc_body(Cpu *c, uint32_t imm, uint64_t pc) {
|
|
__atomic_fetch_add(&g_svcs, 1, __ATOMIC_RELAXED);
|
|
if (rt_trace & RT_TRACE_SVC) {
|
|
char at[160];
|
|
rt_log("svc %#x at %s x0=%#llx x1=%#llx x2=%#llx x3=%#llx", imm, ldr_describe(pc, at, sizeof at),
|
|
(unsigned long long)c->x[0], (unsigned long long)c->x[1], (unsigned long long)c->x[2],
|
|
(unsigned long long)c->x[3]);
|
|
}
|
|
if (kern_svc(c, imm)) return;
|
|
uint64_t *x = c->x;
|
|
switch (imm) {
|
|
case 0x15: case 0x1F: case 0x21: case 0x22:
|
|
if (ipc_svc(c, imm)) return;
|
|
svc_fatal(c, imm, pc, "not handled by ipc.c");
|
|
case 0x01: { // SetHeapSize(x1 size -> x1 address)
|
|
uint64_t a = 0;
|
|
uint32_t r = mem_set_heap_size(x[1], &a);
|
|
x[0] = r;
|
|
if (!r) x[1] = a;
|
|
return;
|
|
}
|
|
case 0x02: x[0] = mem_set_permission(x[0], x[1], (uint32_t)x[2]); return;
|
|
case 0x03: x[0] = mem_set_attribute(x[0], x[1], (uint32_t)x[2], (uint32_t)x[3]); return;
|
|
case 0x04: // MapMemory(x0 dst, x1 src, x2 size)
|
|
x[0] = on_main_stack(x[0], x[2]) ? KERN_R_INVALID_STATE_MEM : mem_map(x[0], x[1], x[2]);
|
|
return;
|
|
case 0x05: x[0] = mem_unmap(x[0], x[1], x[2]); return;
|
|
case 0x06: { // QueryMemory(x0 MemoryInfo*, x2 address -> w1 page info)
|
|
RtMemoryInfo mi;
|
|
uint32_t r = mem_query(x[2], &mi);
|
|
main_stack_query(x[2], &mi);
|
|
if (!r && x[0]) memcpy((void *)(uintptr_t)x[0], &mi, sizeof mi);
|
|
x[0] = r;
|
|
x[1] = 0;
|
|
return;
|
|
}
|
|
case 0x2C: x[0] = mem_map_physical(x[0], x[1]); return;
|
|
case 0x2D: x[0] = mem_unmap_physical(x[0], x[1]); return;
|
|
case 0x26: { // Break(x0 reason, x1 info, x2 info size); bit 31 of the reason: notification only
|
|
uint32_t result = 0;
|
|
if (x[1] && x[2] >= 4 && readable(x[1], 4)) memcpy(&result, (const void *)(uintptr_t)x[1], 4);
|
|
char at[160];
|
|
rt_log("guest Break reason %#llx result %#x at %s", (unsigned long long)x[0], result,
|
|
ldr_describe(pc, at, sizeof at));
|
|
if (x[0] & 0x80000000ull) {
|
|
x[0] = R_OK;
|
|
return;
|
|
}
|
|
{ // the SDK keeps the abort text in the user exception context: show any readable run
|
|
const char *ctx = (const char *)(uintptr_t)user_exception_context();
|
|
char run[161];
|
|
int n = 0;
|
|
for (int i = 0; i < 0x400; i++) {
|
|
unsigned char ch = (unsigned char)ctx[i];
|
|
if (ch >= 0x20 && ch < 0x7f) {
|
|
if (n < 160) run[n++] = (char)ch;
|
|
} else if (n >= 4) {
|
|
run[n] = 0;
|
|
rt_log("guest abort text at ctx+%#x: %s", i - n, run);
|
|
n = 0;
|
|
} else {
|
|
n = 0;
|
|
}
|
|
}
|
|
}
|
|
{ // the object the aborting code was working on: its first bytes carry the state machine's byte
|
|
uint64_t obj = c->x[19];
|
|
if (obj && readable(obj, 32)) {
|
|
const unsigned char *p = (const unsigned char *)(uintptr_t)obj;
|
|
rt_log("guest abort object %#llx: %02x %02x %02x %02x %02x %02x %02x %02x "
|
|
"%02x %02x %02x %02x %02x %02x %02x %02x", (unsigned long long)obj,
|
|
p[0], p[1], p[2], p[3], p[4], p[5], p[6], p[7],
|
|
p[8], p[9], p[10], p[11], p[12], p[13], p[14], p[15]);
|
|
}
|
|
}
|
|
svc_fatal(c, imm, pc, "Break (the SDK aborted)");
|
|
}
|
|
case 0x27: { // OutputDebugString(x0 text, x1 size)
|
|
uint64_t n = x[1];
|
|
if (n > 65536) svc_fatal(c, imm, pc, "debug string longer than 64 KiB");
|
|
if (n) {
|
|
fflush(stdout);
|
|
rt_log_raw((const char *)(uintptr_t)x[0], n);
|
|
}
|
|
x[0] = R_OK;
|
|
return;
|
|
}
|
|
case 0x29: get_info(c, pc); return;
|
|
case 0x07: { // ExitProcess, as closing the window (gfx_window_closed): pipeline cache, shader records and
|
|
// profile saved, logs flushed, exit code 0
|
|
char at[160];
|
|
rt_log("guest ExitProcess at %s; exiting", ldr_describe(pc, at, sizeof at));
|
|
coop_shutdown(); // leaves the room, before the process ends
|
|
fs_save_flush();
|
|
vk_pipeline_prewarm_marker_clear(); // as the window's close (gfx_present.c gfx_window_closed)
|
|
vk_mem_exit_log();
|
|
vk_pipeline_save_exit();
|
|
prof_write();
|
|
rtss_limit_restore();
|
|
fflush(NULL);
|
|
if (svc_test_exit_hook) return svc_test_exit_hook(0);
|
|
plat_exit(0);
|
|
}
|
|
}
|
|
svc_fatal(c, imm, pc, "not implemented");
|
|
}
|