diff --git a/src/panfrost/genxml/cs_builder.h b/src/panfrost/genxml/cs_builder.h
index 537122eef5f..acdbb3a3e30 100644
--- a/src/panfrost/genxml/cs_builder.h
+++ b/src/panfrost/genxml/cs_builder.h
@@ -2512,6 +2512,69 @@ cs_trace_run_fragment(struct cs_builder *b, const struct cs_tracing_ctx *ctx,
    cs_flush_stores(b);
 }
 
+#if PAN_ARCH >= 13
+#define CS_RUN_FULLSCREEN_SR_MASK \
+   (BITFIELD64_RANGE(40, 4) | BITFIELD64_RANGE(56, 4) | BITFIELD64_RANGE(61, 3))
+#define CS_RUN_FULLSCREEN_SR_COUNT 11
+#elif PAN_ARCH >= 11
+#define CS_RUN_FULLSCREEN_SR_MASK \
+   (BITFIELD64_RANGE(40, 4) | BITFIELD64_BIT(56) | BITFIELD64_RANGE(61, 3))
+#define CS_RUN_FULLSCREEN_SR_COUNT 8
+#else
+#define CS_RUN_FULLSCREEN_SR_MASK \
+   (BITFIELD64_RANGE(40, 4) | BITFIELD64_BIT(56))
+#define CS_RUN_FULLSCREEN_SR_COUNT 5
+#endif
+
+struct cs_run_fullscreen_trace {
+   uint64_t ip;
+   uint64_t dcd;
+   uint32_t sr[CS_RUN_FULLSCREEN_SR_COUNT];
+} __attribute__((aligned(64)));
+
+static inline void
+cs_trace_run_fullscreen(struct cs_builder *b, const struct cs_tracing_ctx *ctx,
+                        struct cs_index scratch_regs, uint32_t flags_override,
+                        struct cs_index dcd)
+{
+   if (likely(!ctx->enabled)) {
+      cs_run_fullscreen(b, flags_override, dcd);
+      return;
+   }
+
+   struct cs_index tracebuf_addr = cs_reg64(b, scratch_regs.reg);
+   struct cs_index data = cs_reg64(b, scratch_regs.reg + 2);
+
+   cs_trace_preamble(b, ctx, scratch_regs,
+                     sizeof(struct cs_run_fullscreen_trace));
+
+   /* cs_run_xx() must immediately follow cs_load_ip_to() otherwise the IP
+    * won't point to the right instruction. */
+   cs_load_ip_to(b, data);
+   cs_run_fullscreen(b, flags_override, dcd);
+   cs_store64(b, data, tracebuf_addr,
+              cs_trace_field_offset(run_fullscreen, ip));
+
+   cs_store64(b, dcd, tracebuf_addr,
+              cs_trace_field_offset(run_fullscreen, dcd));
+
+   ASSERTED unsigned sr_count = 0;
+   unsigned sr_offset = cs_trace_field_offset(run_fullscreen, sr);
+   for (unsigned i = 0; i < 64; i += 16) {
+      unsigned mask = (CS_RUN_FULLSCREEN_SR_MASK >> i) & BITFIELD_MASK(16);
+      if (!mask)
+         continue;
+
+      cs_store(b, cs_reg_tuple(b, i, util_last_bit(mask)),
+               tracebuf_addr, mask, sr_offset);
+      sr_offset += util_bitcount(mask) * sizeof(uint32_t);
+      sr_count += util_bitcount(mask);
+   }
+   assert(sr_count == CS_RUN_FULLSCREEN_SR_COUNT);
+
+   cs_flush_stores(b);
+}
+
 #if PAN_ARCH >= 12
 struct cs_run_idvs2_trace {
    uint64_t ip;
diff --git a/src/panfrost/genxml/decode_csf.c b/src/panfrost/genxml/decode_csf.c
index 453ea850b78..0b7dce24982 100644
--- a/src/panfrost/genxml/decode_csf.c
+++ b/src/panfrost/genxml/decode_csf.c
@@ -2450,6 +2450,7 @@ print_cs_binary(struct pandecode_context *ctx, uint64_t bin,
       case MALI_CS_OPCODE_RUN_IDVS:
 #endif
       case MALI_CS_OPCODE_RUN_FRAGMENT:
+      case MALI_CS_OPCODE_RUN_FULLSCREEN:
       case MALI_CS_OPCODE_RUN_COMPUTE:
       case MALI_CS_OPCODE_RUN_COMPUTE_INDIRECT:
          fprintf(ctx->dump_stream, " // tracepoint_%" PRIx64,
@@ -2569,6 +2570,22 @@ GENX(pandecode_cs_trace)(struct pandecode_context *ctx, uint64_t trace,
          break;
       }
 
+      case MALI_CS_OPCODE_RUN_FULLSCREEN: {
+         struct cs_run_fullscreen_trace *fs_trace = trace_data;
+
+         assert(trace_size >= sizeof(*fs_trace));
+         cs_unpack(instr, CS_RUN_FULLSCREEN, I);
+         regs[I.dcd + 0] = (uint32_t)(fs_trace->dcd);
+         regs[I.dcd + 1] = (uint32_t)(fs_trace->dcd >> 32);
+         uint32_t sr_idx = 0;
+         u_foreach_bit64(b, CS_RUN_FULLSCREEN_SR_MASK)
+            regs[b] = fs_trace->sr[sr_idx++];
+         pandecode_run_fullscreen(ctx, ctx->dump_stream, &qctx, &I);
+         trace_data = fs_trace + 1;
+         trace_size -= sizeof(*fs_trace);
+         break;
+      }
+
       case MALI_CS_OPCODE_RUN_COMPUTE: {
          struct cs_run_compute_trace *comp_trace = trace_data;