Implement 'dart:typeddata' directly in the VM instead of using 'dart:scalarlist'

Implement 'dart:typeddata' directly in the VM instead of using 'dart:scalarlist'
Committed: https://code.google.com/p/dart/source/detail?r=19656

[Ivan Posva, 2013-03-07 08:53:31]

LGTM with comments.

-Ivan

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
File runtime/lib/typeddata.cc (right):

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:52: ASSERT(instance.IsExternalTypedData());
I am not sure the ASSERT here is good enough. As this is under user control.

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:63: intptr_t max = (kSmiMax /
TypedData::ElementSizeInBytes(cid));               \
Please comment why you have this restriction here.

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:75: intptr_t max = (kSmiMax /
ExternalTypedData::ElementSizeInBytes(cid));       \
ditto

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:104: ASSERT(instance.IsExternalTypedData());           
                          \
ditto

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:123: ASSERT(instance.IsExternalTypedData());           
                        \
ditto

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:144: ASSERT(instance.IsExternalTypedData());           
                        \
ditto

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:170: ASSERT(BigintOperations::FitsIntoUint64(bigint)); 
                        \
This assert is strange here. Shouldn't this just truncate the Bigint to 64-bits?

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:182: ASSERT(instance.IsExternalTypedData());           
                        \
ditto

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/vm/object....
runtime/vm/object.cc:12378: intptr_t lengthInBytes = len *
ElementSizeInBytes(class_id);
How are we assuring this multiplication does not overflow?

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/vm/raw_obj...
File runtime/vm/raw_object.cc (right):

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/vm/raw_obj...
runtime/vm/raw_object.cc:219: case kExternalTypedData##clazz##Cid:
This code should never be reached, as the size of the RawExternalTypeData object
does not depend on the length.

[siva, 2013-03-07 21:47:43]

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
File runtime/lib/typeddata.cc (right):

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:52: ASSERT(instance.IsExternalTypedData());
Changed the ASSERT to thrown an ArgumentError exception if the type is not a
TypedData or ExternalTypedData type.

On 2013/03/07 08:53:31, Ivan Posva wrote:
> I am not sure the ASSERT here is good enough. As this is under user control.

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:63: intptr_t max = (kSmiMax /
TypedData::ElementSizeInBytes(cid));               \
On 2013/03/07 08:53:31, Ivan Posva wrote:
> Please comment why you have this restriction here.

Done.

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:75: intptr_t max = (kSmiMax /
ExternalTypedData::ElementSizeInBytes(cid));       \
On 2013/03/07 08:53:31, Ivan Posva wrote:
> ditto

Done.

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:104: ASSERT(instance.IsExternalTypedData());           
                          \
On 2013/03/07 08:53:31, Ivan Posva wrote:
> ditto

Done.

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:123: ASSERT(instance.IsExternalTypedData());           
                        \
On 2013/03/07 08:53:31, Ivan Posva wrote:
> ditto

Done.

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:144: ASSERT(instance.IsExternalTypedData());           
                        \
On 2013/03/07 08:53:31, Ivan Posva wrote:
> ditto

Done.

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:170: ASSERT(BigintOperations::FitsIntoUint64(bigint)); 
                        \
I don't see a truncate operation in our current bigint_operations file. I added
a TODO here to consider truncating the value instead of asserting here.

On 2013/03/07 08:53:31, Ivan Posva wrote:
> This assert is strange here. Shouldn't this just truncate the Bigint to
64-bits?

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/lib/typedd...
runtime/lib/typeddata.cc:182: ASSERT(instance.IsExternalTypedData());           
                        \
On 2013/03/07 08:53:31, Ivan Posva wrote:
> ditto

Done.

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/vm/object....
runtime/vm/object.cc:12378: intptr_t lengthInBytes = len *
ElementSizeInBytes(class_id);
The native code in typeddata.c has a lengthCheck against the max which ensures
it should not overflow.

I will add a comment and an assert here.

On 2013/03/07 08:53:31, Ivan Posva wrote:
> How are we assuring this multiplication does not overflow?

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/vm/raw_obj...
File runtime/vm/raw_object.cc (right):

https://chromiumcodereview.appspot.com/12544015/diff/13003/runtime/vm/raw_obj...
runtime/vm/raw_object.cc:219: case kExternalTypedData##clazz##Cid:
True, I removed it.

On 2013/03/07 08:53:31, Ivan Posva wrote:
> This code should never be reached, as the size of the RawExternalTypeData
object
> does not depend on the length.

[siva, 2013-03-07 21:50:41]

Committed patchset #6 manually as r19656 (presubmit successful).