> On 19 Nov 2017, at 16:32, John Hening <gocio...@gmail.com> wrote:
>
> I am analyzing the following Java code:
>
> class Main {
>
> private static class C{
> private C ref;
> }
>
>
> private static void foo(C c, C c2){
> c.ref = c2;
> }
> }
>
>
>
>
>
>
> And JITed version of it:
>
>
> [Entry Point]
> [Verified Entry Point]
> [Constants]
> # {method} {0x00007ff5b3c02af8} 'foo' '(LMain$C;LMain$C;)V' in 'Main'
> # parm0: rsi:rsi = 'Main$C'
> # parm1: rdx:rdx = 'Main$C'
> # [sp+0x50] (sp of caller)
> 0x00007ff5c110cc60: mov %eax,-0x14000(%rsp)
> 0x00007ff5c110cc67: push %rbp
> 0x00007ff5c110cc68: sub $0x40,%rsp
> 0x00007ff5c110cc6c: mov $0x7ff5b3c03200,%rdi ; What is value the
> $0x7ff5b3c03200? (1)
> 0x00007ff5c110cc76: mov 0x8(%rdi),%ebx
> 0x00007ff5c110cc79: add $0x8,%ebx
> 0x00007ff5c110cc7c: mov %ebx,0x8(%rdi)
> 0x00007ff5c110cc7f: mov $0x7ff5b3c02af8,%rdi ; {metadata({method}
> {0x00007ff5b3c02af8} 'foo' '(LMain$C;LMain$C;)V' in 'Main')}
> 0x00007ff5c110cc89: and $0x3ff8,%ebx ; What is the value
> $0x3ff8 (2)
> 0x00007ff5c110cc8f: cmp $0x0,%ebx
> 0x00007ff5c110cc92: je 0x00007ff5c110ccc1 ;*aload_0
> ; - Main::foo@0 (line 11)
>
> 0x00007ff5c110cc98: mov %rdx,%r10
> 0x00007ff5c110cc9b: shr $0x3,%r10 ; Why a param1(reference
> to c2) is shifted right by 3? (3)
> 0x00007ff5c110cc9f: mov %r10d,0xc(%rsi) ; implicit exception:
> dispatches to 0x00007ff5c110ccd5
> 0x00007ff5c110cca3: shr $0x9,%rsi
> 0x00007ff5c110cca7: mov $0x7ff5d13bd000,%rdi
> 0x00007ff5c110ccb1: movb $0x0,(%rsi,%rdi,1) ;*putfield ref
> ; - Main$C::access$002@3
> (line 6)
> ; - Main::foo@2 (line 11)
>
> 0x00007ff5c110ccb5: add $0x40,%rsp
> 0x00007ff5c110ccb9: pop %rbp
> 0x00007ff5c110ccba: test %eax,0x17d2c440(%rip) #
> 0x00007ff5d8e39100 What is a constant 0x17d2c440?
> ; {poll_return}
> 0x00007ff5c110ccc0: retq
> 0x00007ff5c110ccc1: mov %rdi,0x8(%rsp)
> 0x00007ff5c110ccc6: movq $0xffffffffffffffff,(%rsp)
> 0x00007ff5c110ccce: callq 0x00007ff5c1101560 ; OopMap{rsi=Oop rdx=Oop
> off=115}
> ;*synchronization entry
> ; - Main::foo@-1 (line 11)
> ; {runtime_call}
> 0x00007ff5c110ccd3: jmp 0x00007ff5c110cc98
> 0x00007ff5c110ccd5: callq 0x00007ff5c106f7c0 ; OopMap{rsi=Oop off=122}
> ;*putfield ref
> ; - Main$C::access$002@3
> (line 6)
> ; - Main::foo@2 (line 11)
> ; {runtime_call}
> 0x00007ff5c110ccda: nop
> 0x00007ff5c110ccdb: nop
> 0x00007ff5c110ccdc: mov 0x290(%r15),%rax
> 0x00007ff5c110cce3: mov $0x0,%r10
> 0x00007ff5c110cced: mov %r10,0x290(%r15)
> 0x00007ff5c110ccf4: mov $0x0,%r10
> 0x00007ff5c110ccfe: mov %r10,0x298(%r15)
> 0x00007ff5c110cd05: add $0x40,%rsp
> 0x00007ff5c110cd09: pop %rbp
> 0x00007ff5c110cd0a: jmpq 0x00007ff5c106e3e0 ; {runtime_call}
>
>
>
>
> I don't understand some issues, and I would like to someone ask for it:
>
> I've annotated line in the assembly code with numbers. For example,
> The first question (1.) refers to the line annotated with (1). And so on.
>
> 1. What is value the $0x7ff5b3c03200?
> 2. What is the value $0x3ff8?
> 3. Why a param1(reference to c2) is shifted right by 3?
>
I have a presentation on Hotspot under the Hood at JavaOne in 2016 which talks
a little bit about how some of this works (including right shifting
references). You might find it instructive to look through that deck.
I also recorded a voice-over of the presentation at Vimeo, in case you want to
watch it as a video. Unfortunately it wasn’t recorded live at JavaOne itself.
http://alblue.bandlem.com/2016/09/javaone-hotspot.html
--
You received this message because you are subscribed to the Google Groups
"mechanical-sympathy" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to mechanical-sympathy+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
