| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
| |
disk cache interface (async)
BUG=26730
TEST=unit tests
Review URL: http://codereview.chromium.org/1135002
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@42031 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
| |
loop when there are (active) doomed entries that we want to
remove from the "deleted" list.
BUG=32119
TEST=unittests
Review URL: http://codereview.chromium.org/969002
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@41657 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
works as expected with synchronous responses.
As an added bonus, now changing the default test mode
of kRangeGET_TransactionOK doesn't end up with some tests
hanging, so it is easy to manually verify that all tests
work as expected.
BUG=26729
TEST=unittests
Review URL: http://codereview.chromium.org/501099
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@34914 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
| |
When a test doesn't hang or crash, it can run on the bots
and on developers' machines.
BUG=none
TEST=none
Review URL: http://codereview.chromium.org/269104
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@29279 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
| |
wstring. After this patch, I'm able to start chrome in a
user data dir with non-ascii characters on non-utf8 systems.
BUG=24444
Review URL: http://codereview.chromium.org/267085
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@29136 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
| |
This converts BackendImpl to using FilePath.
BUG=24444
Review URL: http://codereview.chromium.org/261045
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@28711 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
| |
This is #9 flaky test and rising (#5 in tests failing recently).
TBR=rvargas
Review URL: http://codereview.chromium.org/193127
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@26345 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
| |
BUG=16507
TEST=Unittests
Review URL: http://codereview.chromium.org/173170
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@23990 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
| |
TBR=willchan
TEST=none
BUG=none
Review URL: http://codereview.chromium.org/174205
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@23926 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
| |
BUG=16507
TEST=Unittests
Review URL: http://codereview.chromium.org/171085
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@23919 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
be able to return the key of an open entry.
Whenever a critical corruption is detected by the disk cache,
the backend disables itself and starts failing all requests
until it's able to re-create the backing store.
Key's longer than 928 bytes are not stored inside the entry
itself, so a file object is required to access them. The backend
will reject any request for a file object after it is disabled,
so a user's request for the key of an open entry will also
fail.
Now we keep a pointer to the related file object (if needed) so
that we don't have to ask the backend for it when the user
requests the current key.
BUG=9952
TEST=unittest
Review URL: http://codereview.chromium.org/165030
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@22637 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We now have a map of open entries so we don't need to
do a lookup through the rankings node anymore. This
simplifies the 64 bit version of the code.
BUG=17881
TEST=unittests
Review URL: http://codereview.chromium.org/159643
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@22074 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Now the code that releases resources runs from the message loop
so that methods can cause the cache to disable itself while
still being able to touch internal state.
BUG=17604
TEST=unittests.
Review URL: http://codereview.chromium.org/159327
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@21550 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
when doing evictions or enumerations.
This cl fixes an issue with dirty entries being deleted
twice from disk, and improves the whole thing so that
different flavors or corrupt entries are removed
gracefully. We started deleting things twice when we
created the map of currently-open entries, because
we don't want to have dirty entries on that map, so it is
possible to have multiple EntryImpl objects that refer
to the same entry (for corrupt entries).
BUG=17474
TEST=Unittests, stress_cache.exe
Review URL: http://codereview.chromium.org/155951
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@21360 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We still have a few crashes when for some reason we believe
an entry is not dirty and we follow the pointer stored by
its rankings node only to crash while accessing the memory.
I have no explanation to why the dirty id matches the current
one (a page boundary issue maybe?), but having a map with all
open entries solves the issue of having to follow pointers
from disk.
BUG=15596, b/1120346
TEST=unittests
Review URL: http://codereview.chromium.org/149218
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@20067 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Since the code now does range request without any caching
the code path for passing file handle is not used any more.
Changes:
1. Remove response_data_file in webkit_glue::ResourceResponseHead
2. Remove response_data_file in net::ResourceInfo
3. Remove code that passes file handle using IPC
4. Remove code that passes file hadnle from network layer to ResourceDispatcherHost
5. Remove MediaResourceHandler
6. Remove code in disk_cache that expose the file handle
7. Remove ChromeURLRequestContext::CreateOffTheRecordForMedia() so no more OTR request context for media, in OTR mode simply memory cache is used
8. Reset cache size for media cache to default
BUG=12249
BUG=12256
Review URL: http://codereview.chromium.org/113931
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@17227 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It is possible that the size for the index table is reduced when
the cache is restarted (the available disk space could be lower
than what it was when the cache was originally created). If that
is the case, not resetting mask_ to 0 will prevent the mask to
be set correctly for the new table size, resulting in a crash.
BUG=12378
TEST=unittests
Review URL: http://codereview.chromium.org/113930
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@17052 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
| |
Review URL: http://codereview.chromium.org/100240
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@15088 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
| |
eviction algorithms without having to recompile the code.
Review URL: http://codereview.chromium.org/79031
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@13909 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
| |
(still limiting the ceiling).
Review URL: http://codereview.chromium.org/68007
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@13684 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
"clean" is not really dirty.
If for some reason an entry is left on disk with a pointer
on the rankings node but without the dirty flag set, we now
recognize it as dirty the next time we read it from disk.
BUG=3987
TEST=DiskCacheTest.Backend_NotMarkedButDirty
Review URL: http://codereview.chromium.org/57024
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@12820 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
| |
at the same time (for media files).
Review URL: http://codereview.chromium.org/45061
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@12561 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
| |
multiple instances of BackendImpl. We need multiple
objects to be able to support media files on the cache.
After this change, histograms will be the only thing that
get messed up by multiple disk caches.
Review URL: http://codereview.chromium.org/49027
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@12520 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
| |
and copy Purify's list of leaky net tests for Valgrind's benefit.
Review URL: http://codereview.chromium.org/42309
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@12042 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
| |
Review URL: http://codereview.chromium.org/48155
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@12028 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
| |
Normalize end of file newlines in net/. All files end in a single newline.
Review URL: http://codereview.chromium.org/43079
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@11442 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Disabled by a #def. When enabled, file format 2.1 is used,
with a possible update from ver 2.0. If a version with this
code disabled is run after the upgrade, it will just
discard the file and go back to 2.0.
We now put some of those extra list to use! Entries are
separated into various lists depending on how often are
used, and we keep track of previously evicted entries.
If the new algorithm is not enabled, most of the code just
goes through a different path (with the old code instead of
the new one). One notable exception is OpenFollowingEntry,
used to enumerate the entries on the cache; the code changed
significantly to support the new version of the "cache
iterator", but functionally should do the same as the current
code.
Review URL: http://codereview.chromium.org/27345
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@11145 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
| |
Review URL: http://codereview.chromium.org/21284
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@9641 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
|
|
| |
This is cleanup from bug 5325.
Original code review: http://codereview.chromium.org/20134/show
Review URL: http://codereview.chromium.org/20251
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@9626 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
| |
Review URL: http://codereview.chromium.org/21236
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@9532 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
| |
This is cleanup from bug 5325.
Review URL: http://codereview.chromium.org/20134
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@9528 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
|
|
| |
For a char*, sizeof() == arraysize(), so there is nothing wrong with the current code. However I think it's important to be clear that the lcpy() functions work in characters and not bytes. The danger would be accidently using sizeof() with wcslcpy, for example copying some code as an example or modifying old code to use a wchar instead of a char.
Review URL: http://codereview.chromium.org/17019
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@7536 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
| |
Review URL: http://codereview.chromium.org/7995
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@4022 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
| |
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@1475 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
| |
async IO to be performed synchronously.
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@1414 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
| |
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@1287 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
|
| |
TEST=unit test
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@1270 0039d316-1c4b-4281-b951-d872f2087c98
|
|
|
|
|
|
| |
reinitialization is handled properly.
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@137 0039d316-1c4b-4281-b951-d872f2087c98
|
|
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@14 0039d316-1c4b-4281-b951-d872f2087c98
|