mirror of
https://github.com/mozilla/gecko-dev.git
synced 2024-11-05 08:35:26 +00:00
23a0cee1a8
nsPSMUITracker was problematic. Apparently it was originally intended to prevent NSS shutdown while NSS-related UI operations were going on (such as choosing a client certificate). However, when nsNSSComponent would receive the event that told it to shutdown NSS, it would attempt to call mShutdownObjectList->evaporateAllNSSResources(), which would call mActivityState.restrictActivityToCurrentThread(), which failed if such a UI operation was in progress. This actually prevented the important part of evaporateAllNSSResources, which is the releasing of all NSS objects in use by PSM objects. Importantly, nsNSSComponent didn't check for or handle this failure and proceeded to call NSS_Shutdown(), leaving PSM in an inconsistent state where it thought it was okay to keep using the NSS objects it had when in fact it wasn't. In any case, nsPSMUITracker isn't really necessary as long as we have the nsNSSShutDownPreventionLock mechanism, which mostly works and is what we should use instead (or not at all, if no such lock is needed for the operation being performed (for example, if no NSS functions are being called)). |
||
|---|---|---|
| .. | ||
| apps | ||
| certverifier | ||
| manager | ||
| nss | ||
| patches | ||
| pkix | ||
| sandbox | ||