diff options
author | Morgan Gangwere <morgan.gangwere@gmail.com> | 2015-03-03 00:51:44 -0700 |
---|---|---|
committer | Morgan Gangwere <morgan.gangwere@gmail.com> | 2015-03-03 00:51:44 -0700 |
commit | 73a1ceb7323f951e92cdb5ce577689adf880d799 (patch) | |
tree | 1ba55b62a7218fd9c18e32215d2f859d7e1df34d /OpenKeychain/src/main/java/org | |
parent | 1a12eb1a7af9f7fc80846f1925aeab72a752eb6f (diff) | |
download | open-keychain-73a1ceb7323f951e92cdb5ce577689adf880d799.tar.gz open-keychain-73a1ceb7323f951e92cdb5ce577689adf880d799.tar.bz2 open-keychain-73a1ceb7323f951e92cdb5ce577689adf880d799.zip |
Fix root cause of #1077
When importing lots of keys, lots of messages about the status of keys
is generated, including some debug slime and general fluff that
isn't really needed a lot of the time. As a result, a serious bug can come
along after key imports or certain operations which cause a log to become
parceled.
This commit implements a pool to "dehydrate" logs into: they are placed
into the pool (a ConcurrentHashMap) and a UUID assigned to them, which
is parceled along. When the OperationResult is un-parceled, it reads in
the appropriate UUID bits and rehydrates the appropriate log.
In order to avoid any memory leaks, the log pool removes a reference to
the log itself, allowing the log to die a natural death at the hands
of the GC..
Diffstat (limited to 'OpenKeychain/src/main/java/org')
-rw-r--r-- | OpenKeychain/src/main/java/org/sufficientlysecure/keychain/operations/results/OperationResult.java | 76 |
1 files changed, 70 insertions, 6 deletions
diff --git a/OpenKeychain/src/main/java/org/sufficientlysecure/keychain/operations/results/OperationResult.java b/OpenKeychain/src/main/java/org/sufficientlysecure/keychain/operations/results/OperationResult.java index f79900aab..02379e917 100644 --- a/OpenKeychain/src/main/java/org/sufficientlysecure/keychain/operations/results/OperationResult.java +++ b/OpenKeychain/src/main/java/org/sufficientlysecure/keychain/operations/results/OperationResult.java @@ -38,6 +38,8 @@ import java.util.ArrayList; import java.util.Arrays; import java.util.Iterator; import java.util.List; +import java.util.UUID; +import java.util.concurrent.ConcurrentHashMap; /** Represent the result of an operation. * @@ -52,6 +54,63 @@ public abstract class OperationResult implements Parcelable { public static final String EXTRA_RESULT = "operation_result"; + + /** + * A HashMap of UUID:OperationLog which contains logs that we don't need + * to care about. This is used such that when we become parceled, we are + * well below the 1Mbit boundary that is specified. + */ + private static ConcurrentHashMap<UUID, OperationLog> dehydratedLogs; + static { + // Static initializer for ConcurrentHashMap + dehydratedLogs = new ConcurrentHashMap<UUID,OperationLog>(); + } + + /** + * Dehydrate a log (such that it is available after deparcelization) + * + * Returns the NULL uuid (0) if you hand it null. + * @param log An OperationLog to dehydrate + * @return a UUID, the ticket for your dehydrated log + * + */ + private static UUID dehydrateLog(OperationLog log) { + if(log == null) { + return new UUID(0,0); + } + else { + UUID ticket = UUID.randomUUID(); + dehydratedLogs.put(ticket, log); + return ticket; + } + } + + /*** + * Rehydrate a log after going through parcelization, invalidating its place in the + * dehydration pool. + * This is used such that when parcelized, the parcel is no larger than 1mbit. + * @param ticket A UUID ticket that identifies the log in question. + * @return An OperationLog. + */ + private static OperationLog rehydrateLog(UUID ticket) { + if(ticket.getMostSignificantBits() == 0 && ticket.getLeastSignificantBits() == 0) { + return null; + } + else + { + OperationLog log = dehydratedLogs.get(ticket); + invalidateDehydrateTicket(ticket); + return log; + } + } + private static void invalidateDehydrateTicket(UUID ticket) { + if(ticket.getLeastSignificantBits() != 0 && ticket.getMostSignificantBits() != 0 + && dehydratedLogs.containsKey(ticket)) + { + dehydratedLogs.remove(ticket); + } + } + /** Holds the overall result, the number specifying varying degrees of success: * - The first bit is 0 on overall success, 1 on overall failure * - The second bit indicates if the action was cancelled - may still be an error or success! @@ -65,7 +124,7 @@ public abstract class OperationResult implements Parcelable { public static final int RESULT_WARNINGS = 4; /// A list of log entries tied to the operation result. - final OperationLog mLog; + protected OperationLog mLog; public OperationResult(int result, OperationLog log) { mResult = result; @@ -74,8 +133,11 @@ public abstract class OperationResult implements Parcelable { public OperationResult(Parcel source) { mResult = source.readInt(); - mLog = new OperationLog(); - mLog.addAll(source.createTypedArrayList(LogEntryParcel.CREATOR)); + long mostSig = source.readLong(); + long leastSig = source.readLong(); + UUID mTicket = new UUID(mostSig, leastSig); + // fetch the dehydrated log out of storage (this removes it from the dehydration pool) + mLog = rehydrateLog(mTicket); } public int getResult() { @@ -739,9 +801,11 @@ public abstract class OperationResult implements Parcelable { @Override public void writeToParcel(Parcel dest, int flags) { dest.writeInt(mResult); - if (mLog != null) { - dest.writeTypedList(mLog.toList()); - } + // Get a ticket for our log. + UUID mTicket = dehydrateLog(mLog); + // And write out the UUID most and least significant bits. + dest.writeLong(mTicket.getMostSignificantBits()); + dest.writeLong(mTicket.getLeastSignificantBits()); } public static class OperationLog implements Iterable<LogEntryParcel> { |