mInputStream holds a reference to the current opened HashStore file.
While resetting the file, mInputStream should be closed first, otherwise,
the file->Remove returns failure code NS_ERROR_FILE_IS_LOCKED.
(This only happens in Windows platform)
Differential Revision: https://phabricator.services.mozilla.com/D42460
This patch does the following:
1. Remove testing files from disk because they are no longer required.
2. Load completions from previous version of HashStore until an update
is applied.
3. Older version of HashStore(.sbstore) & PrefixSet(.vlpset) will be
removed during an update
Differential Revision: https://phabricator.services.mozilla.com/D36002
Completions are now stored in .vlpset, we can remove it from .sbstore
Functions related to optimize reading completions from .sbstore can also
be removed because it is no longer HashStore's responsibility
Differential Revision: https://phabricator.services.mozilla.com/D34574
Here is the flow how prefixes are handled during an V4 update:
1. Prefixes are received from Safe Browsing update, stored in ProtocolBuffer
2. Copy the prefixes from ProtocolBuffer to TableUpdate structure
3. Prefixes in TableUpdate are merged with local prefixes (stored in LookupCacheV4)
4. Merged prefixes are processes by PrefixSet to generate the in-memory prefix
set data structure (MakePrefixSet).
In this patch, we free the prefixes stored in TableUpdate right after step3.
This reduces the peak memory used during an update (peak happens in step 4).
Differential Revision: https://phabricator.services.mozilla.com/D26860
SafeBrowsing V4 protocol use SHA-256 as the checksum to check integrity
of update data and also the integrity of prefix files.
SafeBrowsing V2 HashStore use MD5 as the checksum to check integrity of
.sbstore
Since we are going to use CRC32 as the integrity check of V4 prefix files,
I think rename V4 "checksum" to SHA256 can improve readability.
Differential Revision: https://phabricator.services.mozilla.com/D21460
SafeBrowsing V4 protocol use SHA-256 as the checksum to check integrity
of update data and also the integrity of prefix files.
SafeBrowsing V2 HashStore use MD5 as the checksum to check integrity of
.sbstore
Since we are going to use CRC32 as the integrity check of V4 prefix files,
I think rename V4 "checksum" to SHA256 can improve readability.
Differential Revision: https://phabricator.services.mozilla.com/D21460
I tried to make TableUpdateArray point to const TableUpdate objects
everywhere but there were two problems:
- HashStore::ApplyUpdate() triggers a few Merge() calls which include
sorting the underlying TableUpdate object first.
- LookupCacheV4::ApplyUpdate() calls TableUpdateV4::NewChecksum() when the
checksum is missing and that sets mChecksum.
MozReview-Commit-ID: LIhJcoxo7e7
Manually keeping tabs on the lifetime of these objects is a pain
and is the likely source of some of our crashes. I suspect we might
also be leaking memory.
This change creates an explicit copy of the main array into the
update thread to avoid using a non-thread-safe shared data
structure. This is a shallow copy. Only the pointers to the
TableUpdates are copied, which means one pointer per list (e.g. 5
in total for google4 in a new profile).
MozReview-Commit-ID: 221d6GkKt0M
HashStore::ApplyUpdate() is a V2-only function and so we can be
explicit about that and remove unnecessary casts.
Add a new update error code for when we fail to cast a TableUpdate
object to the expected protocol version.
MozReview-Commit-ID: 65BBwiZJw6J
Given we're no longer using dependent strings in
LookupCacheV4::PrefixString(), we will end up make a copy of the
prefixes at some point. Let's do it early and remove a bunch of
complicated code.
Make the string copies fallible so that we return an error and
fail the update instead of crashing.
MozReview-Commit-ID: 5cZHSDIJSlD
We have a minimum requirement of VS 2015 for Windows builds, which supports
the z length modifier for format specifiers. So we don't need SizePrintfMacros.h
any more, and can just use %zu and friends directly everywhere.
MozReview-Commit-ID: 6s78RvPFMzv
We write a lot of 4-bytes prefixes to file which call many system calls.
We should use a buffer and only write to file if the buffer is full or
finish writing. nsIBufferedOutputStream is a good candidate to do that
MozReview-Commit-ID: CzGOd7iXVTv
In this patch, we will make Safebrowsing V2 caching use the same algorithm as V4.
So we remove "mMissCache" for negative caching and TableFresness check for
positive caching.
But Safebrowsing V2 doesn't contain negative/positive cache duration information in
gethash response. So we hard-code a fixed value, 15 minutes, as cache duration.
In this way, we can sync the mechanism we handle caching for V2 and V4.
An extra effort for V2 here is that we need to manually record prefixes misses
because we won't get any response for those prefixes(implemented in
nsUrlClassifierLookupCallback::CacheMisses).
In Bug 1323953, we always send 4-bytes prefix for completion and the prefix is also
used as the key to store cache result from gethash request.
Since it is always 4-bytes, we could convert it to integer for simplicity.
MozReview-Commit-ID: Lkvrg0wvX5Z
This patch includes following changes:
1. nsUrlClassifierHashCompleter.js
nsUrlClassifierHashCompleter.idl
- Add completionV4 interface for hashCompleter to pass response data to
DB service.
- Process response data includes negative cache duration, matched full
hashes and cache duration for each match. Full matches are passed through
nsIFullHashMatch interface.
- Change _requests.responses from array contains matched fullhashes to
dictionary so that it can store additional information likes negative cache
duration.
2. nsUrlClassifierDBService.cpp
- Implement CompletionV4 interface, store response data to CacheResultV4
object. Expired duration to expired time is handled here.
- Add CacheResultToTableUpdate function to convert V2 & V4 cache result
to TableUpdate object.
3. LookupCache.h
- Extend CacheResult to CacheResultV2 and CacheResultV4 so we can store
response data in CompletionV2 and CompletionV4.
4. HashStore.h
- Add API and member variable in TableUpdateV4 to store response data.
TableUpdate object is used by DB service to pass update data or gethash
response to Classifier, so we need to extend TableUpdateV4 to be able
to store fullHashes.find response.
6. Entry.h
- Define the structure about how we cache fullHashes.find response.
MozReview-Commit-ID: FV4yAl2SAc6
This patch includes following changes:
1. nsUrlClassifierHashCompleter.js
nsUrlClassifierHashCompleter.idl
- Add completionV4 interface for hashCompleter to pass response data to
DB service.
- Process response data includes negative cache duration, matched full
hashes and cache duration for each match. Full matches are passed through
nsIFullHashMatch interface.
- Change _requests.responses from array contains matched fullhashes to
dictionary so that it can store additional information likes negative cache
duration.
2. nsUrlClassifierDBService.cpp
- Implement CompletionV4 interface, store response data to CacheResultV4
object. Expired duration to expired time is handled here.
- Add CacheResultToTableUpdate function to convert V2 & V4 cache result
to TableUpdate object.
3. LookupCache.h
- Extend CacheResult to CacheResultV2 and CacheResultV4 so we can store
response data in CompletionV2 and CompletionV4.
4. HashStore.h
- Add API and member variable in TableUpdateV4 to store response data.
TableUpdate object is used by DB service to pass update data or gethash
response to Classifier, so we need to extend TableUpdateV4 to be able
to store fullHashes.find response.
6. Entry.h
- Define the structure about how we cache fullHashes.find response.
MozReview-Commit-ID: KgR1NASl7GC
A new function Classifier::AsyncApplyUpdates() is implemented for async update.
Besides, all public Classifier interfaces become "worker thread only" and
we remove DBServiceWorker::ApplyUpdatesBackground/Foreground.
In DBServiceWorker::FinishUpdate, instead of calling Classifier::ApplyUpdates,
we call Classifier::AsyncApplyUpdates and install a callback for notifying
the update observer when update is finished. The callback will occur on the
caller thread (i.e. worker thread.)
As for the shutdown issue, when the main thread is notified to shut down,
we at first *synchronously* dispatch an event to the worker thread to
shut down the update thread. After getting synchronized with all other
threads, we send last two events "CancelUpdate" and "CloseDb" to notify
dangling update (i.e. BeginUpdate is called but FinishUpdate isn't)
and do cleanup work.
MozReview-Commit-ID: DXZvA2eFKlc
Switch over to using a single fallible array in ByteSliceWrite. This allows us
to only use 2 transient arrays 1/4 the size of the input array rather than 5
transient arrays.
