Pondini wrote:
jasiakman wrote:
. . .
Given that it was only released a month ago, how many people have lost disks and done a time machine restore? Not many I'd suspect.
Apple does their own testing, and plenty of users did beta testing before the release.
Umm... not sure what you're trying to say. Of course they test - in a former life I was a QA engineer. And of course, I've seen them miss plenty of things before - that's why I'm here.
<grin>
Since quite a few people (50-100) rely on Time Machine quite heavily for backups around me, I've been tasked with digging into this further. I'll keep any further data I find going.
Good. If/when you find something, post it here and see [Reporting a Problem to Apple|
http://web.me.com/pondini/AppleTips/BugReport.html
].
Here's what I've got - I grabbed symbol dumps from both a current up version of 10.6.6, and mine.
10.6.6, working - ran:
<pre>
nm /System/Library/Frameworks/Security.framework/Versions/A/Security | sed -e 's/^\[0-9\].* //' | sed -e 's/^[ ].* //' | sort > /tmp/SecSymbolDump1.txt
</pre>
10.6.6 restored hard drive - ran:
<pre>
nm /Volumes/Macintosh HD/System/Library/Frameworks/Security.framework/Versions/A/Security | sed -e 's/^\[0-9\].* //' | sed -e 's/^[ ].* //' | sort > /tmp/SecSymbolDump2.txt
</pre>
(in short, dumped out the symbol tables of both Security framework libraries, stripped out addresses and leading spaces, and sorted the entries).
Running a diff of these two produced the following
On the "new" 10.6.6 update:
APPLE_TP_MACAPPSTORERECEIPT
CSSMOID_APPLE_EXTENSION_MACAPPSTORERECEIPT
CSSMOID_APPLE_TP_MACAPPSTORERECEIPT
CSSMOID_MACAPPSTORE_CERTPOLICY
CSSMOID_MACAPPSTORE_RECEIPT_CERTPOLICY
_Z30certificatePoliciesContainsOIDPK15CE_CertPoliciesPK9cssmdata
_ZL33OID_APPLE_CERT_POLICYMACAPPSTORE
_ZL39OID_APPLE_EXTENSION_MACAPPSTORERECEIPT
_ZL41OID_APPLE_CERT_POLICY_MACAPPSTORERECEIPT
__ZN8Security12KeychainCore5Trust8evaluateEb
_isRevocationServerMetaError
randomfd.22458
On my machine:
__ZN8Security12KeychainCore5Trust8evaluateEv
randomfd.22457
The "new" 10.6.6 machine has the missing symbols, while the old machine does not. Also, the old machine has a slightly different version of the keystore trust library, and is apparently one rev older (22457 vs 22458).
I'm trying to dump the symbol table of a 10.6.5 Security library now - unfortunately, they don't seem to include nm(??) on the machine I'm looking at.
I'll also just try copying the updated 10.6.6 library over my old one - the only reason I haven't yet is that I'll lose any more testing data if it does work. I'd rather try to grab the value of
randomfd out of a 10.6.5 Security library before I do that (any takers out there? Anyone with a 10.6.5 symbol dumper before I boot one into target mode and check it out?)