Remove usage of singleton software_slot_ in nss on ChromeOS

crypto/nss_util_internal.h

Index: crypto/nss_util_internal.h
diff --git a/crypto/nss_util_internal.h b/crypto/nss_util_internal.h
index 262a59a1ba93fa059a0fd8bb2bb61bb49fcbfaf5..072ed795304a9b9970ae14fe1db5778ce0bf6a98 100644
--- a/crypto/nss_util_internal.h
+++ b/crypto/nss_util_internal.h
@@ -51,8 +51,14 @@ class CRYPTO_EXPORT AutoSECMODListReadLock {
 CRYPTO_EXPORT bool InitializeNSSForChromeOSUser(
     const std::string& email,
     const std::string& username_hash,
-    bool is_primary_user,
-    const base::FilePath& path) WARN_UNUSED_RESULT;
+    const base::FilePath& path);
+
+// Returns whether TPM for ChromeOS user still needs initialization. If
+// true is returned, the caller must proceed to initialize TPM slot for the
+// user. All subsequent calls will return false.
+// |InitializeNSSForChromeOSUser| must have been called first.
+CRYPTO_EXPORT bool GetAndSetStartedInitializeTPMForChromeOSUser(

[Ryan Sleevi, 2014/07/09 19:05:40]

Naming still confused the hell out of me.

ShouldInitializeTPMForChromeOSUser?

I'm guessing the issue is that once this API returns true, it expects the caller
to initialize, because it's going to start returning false for everyone else?
That seems like a dangerous sort of API.

IsResponsibleToInitializeTPMForChromeOSUser?

Alternatively, split the ShouldInit and SetWillInit into two methods.

[tbarzic, 2014/07/09 20:45:34]

Yeah, we had the same expectation with old API, but somewhat hidden in
InitializeNSSForChromeOSUser.
OK, I'll go with this. Should be less confusing.

+    const std::string& username_hash) WARN_UNUSED_RESULT;
 
 // Use TPM slot |slot_id| for user.  InitializeNSSForChromeOSUser must have been
 // called first.