Update NSS to NSS 3.14 Beta 1.

Update NSS to NSS 3.14 pre-release snapshot 2012-09-25.

R=rsleevi@chromium.org
BUG=151692
TEST=none
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=158748

[wtc, 2012-09-22 15:38:47]

rsleevi: please review. The NSS_3_14_BETA1 tag hasn't
been created yet, which is why the nss-checkout.sh script
is pulling from -r HEAD. That will be fixed before I commit
this. Thanks.

[Ryan Sleevi, 2012-09-25 21:56:55]

One thing I don't see here is updating the .def file, which I assume will be
needed in order to properly link, now that the softoken bits have moved to
nssutil and been renamed.

Can you confirm that this roll works as expected when linked on a Windows shared
build?

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/f...
File mozilla/security/nss/lib/freebl/blapi.h (right):

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/f...
mozilla/security/nss/lib/freebl/blapi.h:1264: * seedBbytes must be in the range
[N..L*2] or an error will result.
typo: seedBbytes -> seedBytes

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/f...
File mozilla/security/nss/lib/freebl/pqg.c (right):

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/f...
mozilla/security/nss/lib/freebl/pqg.c:38: * numbers.There are 3 ways to counter
this: 1) use one of the cool provably
typo: numbers.There -> numbers. There

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/f...
mozilla/security/nss/lib/freebl/pqg.c:41: * or 3) use a larger M-R test count. I
chose the latter. It increases the time
nit: "latter" in lists with > 2 items = weird way to say it.

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/p...
File mozilla/security/nss/lib/pk11wrap/pk11obj.c (right):

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/p...
mozilla/security/nss/lib/pk11wrap/pk11obj.c:544: }
The lack of commenting here is unfortunate. It required me consulting the spec
to see that CKA_SUBPRIME is a Big Integer type, hence the stripping of the
leading zero. The *2 I presume is because the signature format is concatenation
of r and s.

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/s...
File mozilla/security/nss/lib/softoken/sdb.c (right):

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/s...
mozilla/security/nss/lib/softoken/sdb.c:1881: /*sdb->sdb_type = SDB_SQL; */
Looks like you already fixed this yesterday. I was going to suggest it be
removed if not used :)

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/u...
File mozilla/security/nss/lib/util/utilpars.h (right):

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/u...
mozilla/security/nss/lib/util/utilpars.h:4: #include "utilparst.h"
no header guards?

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/u...
mozilla/security/nss/lib/util/utilpars.h:45: 
Unfortunate whitespace

[Ryan Sleevi, 2012-09-25 21:56:59]

lgtm

[wtc, 2012-09-26 00:21:09]

rsleevi: thank you for the review. I will create upstream
NSS patches to make all the changes you suggested.

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/f...
File mozilla/security/nss/lib/freebl/pqg.c (right):

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/f...
mozilla/security/nss/lib/freebl/pqg.c:37: * iterations of Miller-Rabin than the
for your normal randomly selected

This sentence also has typos. I don't know how to fix them.
Do you? I suspect Bob meant to say:
  It may be possible to purposefully construct composites with more
  iterations of Miller-Rabin than for your normal randomly selected
  numbers.

But I don't understand what this means.

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/f...
mozilla/security/nss/lib/freebl/pqg.c:41: * or 3) use a larger M-R test count. I
chose the latter. It increases the time

On 2012/09/25 21:56:55, Ryan Sleevi wrote:
> nit: "latter" in lists with > 2 items = weird way to say it.

Should I change this to "the last"?

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/p...
File mozilla/security/nss/lib/pk11wrap/pk11obj.c (right):

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/p...
mozilla/security/nss/lib/pk11wrap/pk11obj.c:544: }

On 2012/09/25 21:56:55, Ryan Sleevi wrote:
> The lack of commenting here is unfortunate. It required me consulting the spec
> to see that CKA_SUBPRIME is a Big Integer type, hence the stripping of the
> leading zero.

That was the intention. But I found that the Big Integer
type in PKCS #11 is an unsigned integer, so it does not
need an extra leading byte if the most significant bit is
1. (The ASN.1 INTEGER type is a signed integer.)

So I believe this code is not necessary, unless some
tokens are known to store Big Integers with leading zero
bytes incorrectly.

> The *2 I presume is because the signature format is concatenation
> of r and s.

Yes.

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/u...
File mozilla/security/nss/lib/util/utilpars.h (right):

http://codereview.chromium.org/10961060/diff/14006/mozilla/security/nss/lib/u...
mozilla/security/nss/lib/util/utilpars.h:4: #include "utilparst.h"

I fixed both of these problems yesterday :-)