Fix debug printing of pointers, and a keyed store with smi index error, in X6...

Fix debug printing of pointers, and a keyed store with smi index error, in X64

Committed: http://code.google.com/p/v8/source/detail?r=2605

[Lasse Reichstein, 2009-07-31 14:31:07]

LGTM

http://codereview.chromium.org/160452/diff/1/5
File src/string-stream.cc (right):

http://codereview.chromium.org/160452/diff/1/5#newcode172
Line 172: EmbeddedVector<char, 28> formatted;
28 seems like overkill. If printed as hex, even with a "0x" prefix, it shouldn't
take more than 18 bytes.

http://codereview.chromium.org/160452/diff/1/4
File src/x64/ic-x64.cc (right):

http://codereview.chromium.org/160452/diff/1/4#newcode428
Line 428: // used as an index in a memory operand.
Where is it checked that it isn't negative?
(On the other hand, a zero extended negative 32-bit number would be too large
for use as an index).

http://codereview.chromium.org/160452/diff/1/4#newcode429
Line 429: __ movl(rbx, rbx);  // Clear the high bits of rbx.
A thought: Would xchgl(rax, rax) also clear the top half?
(being nop, it shouldn't).

[Lasse Reichstein, 2009-07-31 15:17:12]

http://codereview.chromium.org/160452/diff/1/4
File src/x64/ic-x64.cc (right):

http://codereview.chromium.org/160452/diff/1/4#newcode429
Line 429: __ movl(rbx, rbx);  // Clear the high bits of rbx.
Ah, special case: The default size of xchg is 64-bit, you need  a two-byte
sequence to make it 32-bit.

[William Hesse, 2009-08-03 11:52:20]

http://codereview.chromium.org/160452/diff/1/5
File src/string-stream.cc (right):

http://codereview.chromium.org/160452/diff/1/5#newcode172
Line 172: EmbeddedVector<char, 28> formatted;
On 2009/07/31 14:31:07, Lasse Reichstein wrote:
> 28 seems like overkill. If printed as hex, even with a "0x" prefix, it
shouldn't
> take more than 18 bytes.
I was worried that someone would try to print octal values.
I guess we know that won't happen. 20 should be safe: 16 + "0x" + "x" or "-" +
'\0'.

http://codereview.chromium.org/160452/diff/1/4
File src/x64/ic-x64.cc (right):

http://codereview.chromium.org/160452/diff/1/4#newcode428
Line 428: // used as an index in a memory operand.
On 2009/07/31 14:31:07, Lasse Reichstein wrote:
> Where is it checked that it isn't negative?
> (On the other hand, a zero extended negative 32-bit number would be too large
> for use as an index).

It is checked below.

http://codereview.chromium.org/160452/diff/1/4#newcode452
Line 452: __ j(below, &fast);
This test checks that the smi is not negative, and is below the array size, by
using an unsigned comparison.