2401   /***

  2402    *  Arguments:

  2403    *

  2404    *  Inputs:

  2405    *   c_rarg0   - int   adler

  2406    *   c_rarg1   - byte* buff

  2407    *   c_rarg2   - int   len

  2408    *

  2409    * Output:

  2410    *   c_rarg0   - int adler result

  2411    */

  2412   address generate_updateBytesAdler32() {

  2413     __ align(CodeEntryAlignment);

  2414     StubCodeMark mark(this, "StubRoutines", "updateBytesAdler32");

  2415     address start = __ pc();

  2416 

  2417     Label L_simple_by1_loop, L_nmax, L_nmax_loop, L_by16, L_by16_loop, L_by1_loop, L_do_mod, L_combine, L_by1;

  2418 

  2419     // Aliases

  2420     Register adler  = c_rarg0;

  2421     Register s1     = c_rarg0;

  2422     Register s2     = c_rarg3;

  2423     Register buff   = c_rarg1;

  2424     Register len    = c_rarg2;

  2425     Register nmax  = r4;

  2426     Register base = r5;

  2427     Register count = r6;

  2428     Register temp0 = rscratch1;

  2429     Register temp1 = rscratch2;

  2430     Register temp2 = r7;

  2431 

  2432     // Max number of bytes we can process before having to take the mod

  2433     // 0x15B0 is 5552 in decimal, the largest n such that 255n(n+1)/2 + (n+1)(BASE-1) <= 2^32-1

  2434     unsigned long BASE = 0xfff1;

  2435     unsigned long NMAX = 0x15B0;

  2436 

  2437     __ mov(base, BASE);

  2438     __ mov(nmax, NMAX);

  2439 

  2440     // s1 is initialized to the lower 16 bits of adler

  2441     // s2 is initialized to the upper 16 bits of adler

  2442     __ ubfx(s2, adler, 16, 16);  // s2 = ((adler >> 16) & 0xffff)

  2443     __ uxth(s1, adler);          // s1 = (adler & 0xffff)

  2444 

  2445     // The pipelined loop needs at least 16 elements for 1 iteration

  2446     // It does check this, but it is more effective to skip to the cleanup loop

  2447     __ cmp(len, 16);

  2448     __ br(Assembler::HS, L_nmax);

  2449     __ cbz(len, L_combine);

  2450 

  2451     __ bind(L_simple_by1_loop);

  2452     __ ldrb(temp0, Address(__ post(buff, 1)));

  2453     __ add(s1, s1, temp0);

  2454     __ add(s2, s2, s1);

  2455     __ subs(len, len, 1);

  2456     __ br(Assembler::HI, L_simple_by1_loop);

  2457 

  2458     // s1 = s1 % BASE

  2459     __ subs(temp0, s1, base);

  2460     __ csel(s1, temp0, s1, Assembler::HS);

  2461 

  2462     // s2 = s2 % BASE

  2463     __ lsr(temp0, s2, 16);

  2464     __ lsl(temp1, temp0, 4);

  2465     __ sub(temp1, temp1, temp0);

  2466     __ add(s2, temp1, s2, ext::uxth);

  2467 

  2468     __ subs(temp0, s2, base);

  2469     __ csel(s2, temp0, s2, Assembler::HS);

  2470 

  2471     __ b(L_combine);

  2472 

  2473     __ bind(L_nmax);

  2474     __ subs(len, len, nmax);

  2475     __ sub(count, nmax, 16);

  2476     __ br(Assembler::LO, L_by16);

  2477 

  2478     __ bind(L_nmax_loop);

  2479 

  2480     __ ldp(temp0, temp1, Address(__ post(buff, 16)));

  2481 

  2482     __ add(s1, s1, temp0, ext::uxtb);

  2483     __ ubfx(temp2, temp0, 8, 8);

  2484     __ add(s2, s2, s1);

  2485     __ add(s1, s1, temp2);

  2486     __ ubfx(temp2, temp0, 16, 8);

  2487     __ add(s2, s2, s1);

  2488     __ add(s1, s1, temp2);

  2489     __ ubfx(temp2, temp0, 24, 8);

  2490     __ add(s2, s2, s1);

  2491     __ add(s1, s1, temp2);

  2492     __ ubfx(temp2, temp0, 32, 8);

  2493     __ add(s2, s2, s1);

  2494     __ add(s1, s1, temp2);

  2495     __ ubfx(temp2, temp0, 40, 8);

  2496     __ add(s2, s2, s1);

  2497     __ add(s1, s1, temp2);

  2498     __ ubfx(temp2, temp0, 48, 8);

  2499     __ add(s2, s2, s1);

  2500     __ add(s1, s1, temp2);

  2501     __ add(s2, s2, s1);

  2502     __ add(s1, s1, temp0, Assembler::LSR, 56);

  2503     __ add(s2, s2, s1);

  2504 

  2505     __ add(s1, s1, temp1, ext::uxtb);

  2506     __ ubfx(temp2, temp1, 8, 8);

  2507     __ add(s2, s2, s1);

  2508     __ add(s1, s1, temp2);

  2509     __ ubfx(temp2, temp1, 16, 8);

  2510     __ add(s2, s2, s1);

  2511     __ add(s1, s1, temp2);

  2512     __ ubfx(temp2, temp1, 24, 8);

  2513     __ add(s2, s2, s1);

  2514     __ add(s1, s1, temp2);

  2515     __ ubfx(temp2, temp1, 32, 8);

  2516     __ add(s2, s2, s1);

  2517     __ add(s1, s1, temp2);

  2518     __ ubfx(temp2, temp1, 40, 8);

  2519     __ add(s2, s2, s1);

  2520     __ add(s1, s1, temp2);

  2521     __ ubfx(temp2, temp1, 48, 8);

  2522     __ add(s2, s2, s1);

  2523     __ add(s1, s1, temp2);

  2524     __ add(s2, s2, s1);

  2525     __ add(s1, s1, temp1, Assembler::LSR, 56);

  2526     __ add(s2, s2, s1);

  2527 

  2528     __ subs(count, count, 16);

  2529     __ br(Assembler::HS, L_nmax_loop);

  2530 

  2531     // s1 = s1 % BASE

  2532     __ lsr(temp0, s1, 16);

  2533     __ lsl(temp1, temp0, 4);

  2534     __ sub(temp1, temp1, temp0);

  2535     __ add(temp1, temp1, s1, ext::uxth);

  2536 

  2537     __ lsr(temp0, temp1, 16);

  2538     __ lsl(s1, temp0, 4);

  2539     __ sub(s1, s1, temp0);

  2540     __ add(s1, s1, temp1, ext:: uxth);

  2541 

  2542     __ subs(temp0, s1, base);

  2543     __ csel(s1, temp0, s1, Assembler::HS);

  2544 

  2545     // s2 = s2 % BASE

  2546     __ lsr(temp0, s2, 16);

  2547     __ lsl(temp1, temp0, 4);

  2548     __ sub(temp1, temp1, temp0);

  2549     __ add(temp1, temp1, s2, ext::uxth);

  2550 

  2551     __ lsr(temp0, temp1, 16);

  2552     __ lsl(s2, temp0, 4);

  2553     __ sub(s2, s2, temp0);

  2554     __ add(s2, s2, temp1, ext:: uxth);

  2555 

  2556     __ subs(temp0, s2, base);

  2557     __ csel(s2, temp0, s2, Assembler::HS);

  2558 

  2559     __ subs(len, len, nmax);

  2560     __ sub(count, nmax, 16);

  2561     __ br(Assembler::HS, L_nmax_loop);

  2562 

  2563     __ bind(L_by16);

  2564     __ adds(len, len, count);

  2565     __ br(Assembler::LO, L_by1);

  2566 

  2567     __ bind(L_by16_loop);

  2568 

  2569     __ ldp(temp0, temp1, Address(__ post(buff, 16)));

  2570 

  2571     __ add(s1, s1, temp0, ext::uxtb);

  2572     __ ubfx(temp2, temp0, 8, 8);

  2573     __ add(s2, s2, s1);

  2574     __ add(s1, s1, temp2);

  2575     __ ubfx(temp2, temp0, 16, 8);

  2576     __ add(s2, s2, s1);

  2577     __ add(s1, s1, temp2);

  2578     __ ubfx(temp2, temp0, 24, 8);

  2579     __ add(s2, s2, s1);

  2580     __ add(s1, s1, temp2);

  2581     __ ubfx(temp2, temp0, 32, 8);

  2582     __ add(s2, s2, s1);

  2583     __ add(s1, s1, temp2);

  2584     __ ubfx(temp2, temp0, 40, 8);

  2585     __ add(s2, s2, s1);

  2586     __ add(s1, s1, temp2);

  2587     __ ubfx(temp2, temp0, 48, 8);

  2588     __ add(s2, s2, s1);

  2589     __ add(s1, s1, temp2);

  2590     __ add(s2, s2, s1);

  2591     __ add(s1, s1, temp0, Assembler::LSR, 56);

  2592     __ add(s2, s2, s1);

  2593 

  2594     __ add(s1, s1, temp1, ext::uxtb);

  2595     __ ubfx(temp2, temp1, 8, 8);

  2596     __ add(s2, s2, s1);

  2597     __ add(s1, s1, temp2);

  2598     __ ubfx(temp2, temp1, 16, 8);

  2599     __ add(s2, s2, s1);

  2600     __ add(s1, s1, temp2);

  2601     __ ubfx(temp2, temp1, 24, 8);

  2602     __ add(s2, s2, s1);

  2603     __ add(s1, s1, temp2);

  2604     __ ubfx(temp2, temp1, 32, 8);

  2605     __ add(s2, s2, s1);

  2606     __ add(s1, s1, temp2);

  2607     __ ubfx(temp2, temp1, 40, 8);

  2608     __ add(s2, s2, s1);

  2609     __ add(s1, s1, temp2);

  2610     __ ubfx(temp2, temp1, 48, 8);

  2611     __ add(s2, s2, s1);

  2612     __ add(s1, s1, temp2);

  2613     __ add(s2, s2, s1);

  2614     __ add(s1, s1, temp1, Assembler::LSR, 56);

  2615     __ add(s2, s2, s1);

  2616 

  2617     __ subs(len, len, 16);

  2618     __ br(Assembler::HS, L_by16_loop);

  2619 

  2620     __ bind(L_by1);

  2621     __ adds(len, len, 15);

  2622     __ br(Assembler::LO, L_do_mod);

  2623 

  2624     __ bind(L_by1_loop);

  2625     __ ldrb(temp0, Address(__ post(buff, 1)));

  2626     __ add(s1, temp0, s1);

  2627     __ add(s2, s2, s1);

  2628     __ subs(len, len, 1);

  2629     __ br(Assembler::HS, L_by1_loop);

  2630 

  2631     __ bind(L_do_mod);

  2632     // s1 = s1 % BASE

  2633     __ lsr(temp0, s1, 16);

  2634     __ lsl(temp1, temp0, 4);

  2635     __ sub(temp1, temp1, temp0);

  2636     __ add(temp1, temp1, s1, ext::uxth);

  2637 

  2638     __ lsr(temp0, temp1, 16);

  2639     __ lsl(s1, temp0, 4);

  2640     __ sub(s1, s1, temp0);

  2641     __ add(s1, s1, temp1, ext:: uxth);

  2642 

  2643     __ subs(temp0, s1, base);

  2644     __ csel(s1, temp0, s1, Assembler::HS);

  2645 

  2646     // s2 = s2 % BASE

  2647     __ lsr(temp0, s2, 16);

  2648     __ lsl(temp1, temp0, 4);

  2649     __ sub(temp1, temp1, temp0);

  2650     __ add(temp1, temp1, s2, ext::uxth);

  2651 

  2652     __ lsr(temp0, temp1, 16);

  2653     __ lsl(s2, temp0, 4);

  2654     __ sub(s2, s2, temp0);

  2655     __ add(s2, s2, temp1, ext:: uxth);

  2656 

  2657     __ subs(temp0, s2, base);

  2658     __ csel(s2, temp0, s2, Assembler::HS);

  2659 

  2660     // Combine lower bits and higher bits

  2661     __ bind(L_combine);

  2662     __ orr(s1, s1, s2, Assembler::LSL, 16); // adler = s1 | (s2 << 16)

  2663 

  2664     __ ret(lr);

  2665 

  2666     return start;

  2667   }

  2668 

  3887     // generate Adler32 intrinsics code

  3888     if (UseAdler32Intrinsics) {

  3889       StubRoutines::_updateBytesAdler32 = generate_updateBytesAdler32();

  3890     }

  3891