VM: Use constant pool also for leaf runtime calls on x64, arm, arm64 and mips.

VM: Use constant pool also for leaf runtime calls on x64, arm, arm64 and mips.

Remove unused branch/call macro instructions from the assembler.

Setup a frame and pool pointer when invoking leaf runtime functions (in the
deoptimzation stub and store-buffer update stub).

BUG=
R=rmacnak@google.com

Committed: https://github.com/dart-lang/sdk/commit/7e082c108210e4401160e5c4b382d0ed4a3e4405

[Florian Schneider, 2015-08-25 10:55:33]

fschneider@google.com changed reviewers:
+ rmacnak@google.com

[rmacnak, 2015-08-26 23:55:24]

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/assembler_ar...
File runtime/vm/assembler_arm.cc (right):

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/assembler_ar...
runtime/vm/assembler_arm.cc:2683: LoadImmediate(IP, label.address(), cond);  //
Address is never patched.
Shouldn't this be LoadExternalLabel? We still need to relocate it, no?

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/assembler_ar...
runtime/vm/assembler_arm.cc:2694: LoadPatchableImmediate(IP, label.address());
Ditto

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/flow_graph_c...
File runtime/vm/flow_graph_compiler.cc (right):

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/flow_graph_c...
runtime/vm/flow_graph_compiler.cc:70: DECLARE_FLAG(bool, print_stop_message);
I don't see any use of this flag.

[Florian Schneider, 2015-08-27 08:27:20]

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/assembler_ar...
File runtime/vm/assembler_arm.cc (right):

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/assembler_ar...
runtime/vm/assembler_arm.cc:2683: LoadImmediate(IP, label.address(), cond);  //
Address is never patched.
On 2015/08/26 23:55:24, rmacnak wrote:
> Shouldn't this be LoadExternalLabel? We still need to relocate it, no?

Yes, using LoadExternalLabel is necessary here and will work without any further
changes since this branch won't be patched.

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/assembler_ar...
runtime/vm/assembler_arm.cc:2694: LoadPatchableImmediate(IP, label.address());
On 2015/08/26 23:55:24, rmacnak wrote:
> Ditto

Yes, but not the purpose of this CL (see description): The instruction patterns
in the patching code need to change as well if I use LoadExternalLabel. I just
changed code generation for leaf runtime calls plus I just removed unused
variants of Branch/BranchLink here without changing the generated code.

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/flow_graph_c...
File runtime/vm/flow_graph_compiler.cc (right):

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/flow_graph_c...
runtime/vm/flow_graph_compiler.cc:70: DECLARE_FLAG(bool, print_stop_message);
On 2015/08/26 23:55:24, rmacnak wrote:
> I don't see any use of this flag.

It's right below.

[Florian Schneider, 2015-08-27 08:38:10]

If there are no further comments, I consider this as ready for landing.

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/assembler_ar...
File runtime/vm/assembler_arm.cc (right):

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/assembler_ar...
runtime/vm/assembler_arm.cc:2683: LoadImmediate(IP, label.address(), cond);  //
Address is never patched.
On 2015/08/27 08:27:19, Florian Schneider wrote:
> On 2015/08/26 23:55:24, rmacnak wrote:
> > Shouldn't this be LoadExternalLabel? We still need to relocate it, no?
> 
> Yes, using LoadExternalLabel is necessary here and will work without any
further
> changes since this branch won't be patched.

Correction: Actually, it won't work right away, since this sequence is still
used in places where there is no constant pool loaded.
(FlowGraphCompiler::EmitFrameEntry)

[rmacnak, 2015-08-27 20:14:05]

LGTM

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/flow_graph_c...
File runtime/vm/flow_graph_compiler.cc (right):

https://codereview.chromium.org/1314883002/diff/20001/runtime/vm/flow_graph_c...
runtime/vm/flow_graph_compiler.cc:70: DECLARE_FLAG(bool, print_stop_message);
On 2015/08/27 08:27:20, Florian Schneider wrote:
> On 2015/08/26 23:55:24, rmacnak wrote:
> > I don't see any use of this flag.
> 
> It's right below.

Nvm, I thought this was defining a new flag.

[Florian Schneider, 2015-08-28 08:46:08]

Committed patchset #3 (id:40001) manually as
7e082c108210e4401160e5c4b382d0ed4a3e4405 (presubmit successful).