https://bugs.webkit.org/show_bug.cgi?id=241905

Reviewed by Yusuke Suzuki.

offlineasm used to emit this LLInt code:
    ".loc 1 996\n"        "ldr x19, [x0] \n"                   // LowLevelInterpreter.asm:996
    ".loc 1 997\n"        "ldr x20, [x0, WebKit#8] \n"               // LowLevelInterpreter.asm:997
    ".loc 1 998\n"        "ldr x21, [x0, WebKit#16] \n"              // LowLevelInterpreter.asm:998
    ".loc 1 999\n"        "ldr x22, [x0, WebKit#24] \n"              // LowLevelInterpreter.asm:999
    ...
    ".loc 1 1006\n"       "ldr d8, [x0, WebKit#80] \n"               // LowLevelInterpreter.asm:1006
    ".loc 1 1007\n"       "ldr d9, [x0, WebKit#88] \n"               // LowLevelInterpreter.asm:1007
    ".loc 1 1008\n"       "ldr d10, [x0, WebKit#96] \n"              // LowLevelInterpreter.asm:1008
    ".loc 1 1009\n"       "ldr d11, [x0, WebKit#104] \n"             // LowLevelInterpreter.asm:1009
    ...

Now, it can emit this instead:
    ".loc 1 996\n"        "ldp x19, x20, [x0, #0] \n"          // LowLevelInterpreter.asm:996
    ".loc 1 997\n"        "ldp x21, x22, [x0, WebKit#16] \n"         // LowLevelInterpreter.asm:997
    ...
    ".loc 1 1001\n"       "ldp d8, d9, [x0, WebKit#80] \n"           // LowLevelInterpreter.asm:1001
    ".loc 1 1002\n"       "ldp d10, d11, [x0, WebKit#96] \n"         // LowLevelInterpreter.asm:1002
    ...

Also, there was some code that kept recomputing the base address of a sequence of load/store
instructions.  For example,
    ".loc 6 902\n"        "add x13, sp, x10, lsl WebKit#3 \n"        // WebAssembly.asm:902
                          "ldr x0, [x13, WebKit#48] \n"
                          "add x13, sp, x10, lsl WebKit#3 \n"
                          "ldr x1, [x13, WebKit#56] \n"
                          "add x13, sp, x10, lsl WebKit#3 \n"
                          "ldr x2, [x13, WebKit#64] \n"
                          "add x13, sp, x10, lsl WebKit#3 \n"
                          "ldr x3, [x13, WebKit#72] \n"
    ...

For such places, we observe that the base address is the same for every load/store instruction
in the sequence, and precompute it in the LLInt asm code to help out the offline asm.  This
allows the offlineasm to now emit this more efficient code instead:
    ".loc 6 896\n"        "add x10, sp, x10, lsl WebKit#3 \n"        // WebAssembly.asm:896
    ".loc 6 898\n"        "ldp x0, x1, [x10, WebKit#48] \n"          // WebAssembly.asm:898
                          "ldp x2, x3, [x10, WebKit#64] \n"
    ...

* Source/JavaScriptCore/llint/LowLevelInterpreter.asm:
* Source/JavaScriptCore/llint/WebAssembly.asm:
* Source/JavaScriptCore/offlineasm/arm64.rb:
* Source/JavaScriptCore/offlineasm/instructions.rb:

Canonical link: https://commits.webkit.org/251799@main