Eliminate an extra allocation in encryptor_win when en/decrypting.

Eliminate an extra allocation in encryptor_win when en/decrypting.

R=wtc
BUG=none
TEST=crypto_unittests


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=91803

[Ryan Sleevi, 2011-06-28 04:44:49]

wtc: PTAL. I kept the existing ->resize(0) style since that matched
encryptor_mac.cc, but I'm perfectly happy to swap that with a ->clear() if you'd
prefer.

Just a small clean-up that came up when looking through this file again. 

It's dependent on http://codereview.chromium.org/7230037/, but only because
StringPiece doesn't implicitly cast to an std::string, so it would otherwise
have to change after that CL.

[wtc, 2011-06-29 00:46:41]

http://codereview.chromium.org/7272022/diff/3001/crypto/encryptor_win.cc
File crypto/encryptor_win.cc (right):

http://codereview.chromium.org/7272022/diff/3001/crypto/encryptor_win.cc#newc...
crypto/encryptor_win.cc:91: reinterpret_cast<BYTE*>(WriteInto(ciphertext,
total_len+1)),
Add spaces around '+'.

It is safer to do things in this order:

1. Call WriteInto(ciphertext, total_len + 1) to get a pointer
to an output buffer of the right size.

2. Copy plaintext into the output buffer.

3. Pass the output buffer to CryptEncrypt.

The reason to call WriteInto before copying is that I
am not sure if WriteInto preserves the existing data in
the std::string.

I prefer ciphertext->clear() to ciphertext->resize(0).

[Ryan Sleevi, 2011-07-01 06:20:42]

wtc: PTAL.

WriteInto is simply implemented in terms of string->reserve() +
string->resize(), so in practice it doesn't modify the contents. To be on the
safe side though, I went with your suggested ordering, in the event of some
future optimization to WriteInto.

[wtc, 2011-07-07 01:03:12]

LGTM.

http://codereview.chromium.org/7272022/diff/6001/crypto/encryptor_win.cc
File crypto/encryptor_win.cc (right):

http://codereview.chromium.org/7272022/diff/6001/crypto/encryptor_win.cc#newc...
crypto/encryptor_win.cc:89: plaintext.copy(ciphertext_data, plaintext.size(),
0);
Note: This comment also applies to the ciphertext.copy() call below.

plaintext.size() => data_len

Nit: it seems better to just call memcpy() here.

[commit-bot: I haz the power, 2011-07-08 03:32:34]

Change committed as 91803