The original logic had maxSdkVersion=0 meaning infinity. That was changed
to be a very large value SDK_VERSION_MAX_VALUE, but getMinMaxSdkVersion()
was still returning 0 for APKs where maxSdkVersion was not set.
This is a follow up on fc0df0dcf4dd0d5f13de82d7cd9254b2b48cb62d
Apparently, the CREATOR field is not (yet?) needed in the tests, since
they work without it. This gets us closer to making lint errors fail
the CI builds.
closes#580
This is the recommended way to deal with updating Notifications. Each new
update should be a new Notification, but the same Builder instance should
be used to generate each new Notification instance.
New installs where being caught up in the logic to check whether a download
is still in progress after InstallManagerService got killed. Also checking
whether Intent was just redelivered lets the new installs through while
screening out the inactive Intents that were redelivered. This logic also
cancels the notification for any download that was in progress when the
InstallManagerService was killed.
#660
Check repo index timestamps
The Update Framework documents provide a [nice discussion of possible attacks](https://github.com/theupdateframework/tuf/blob/develop/SECURITY.md) against update systems. One example is a "rollback attack", where the attacker just serves the old signed `index.jar` to keep all clients from updating their apps. That allows the attacker to exploit known vulnerabilities in those un-updated apps.
While this is a reasonably hard attack, this fix is an important step towards removing the requirement for trusting the web server operator. Ultimately, it should be able trusting the index signing key only. Then it doesn't matter were the files come from, it just matters that they are verifiably signed by the index signing key.
This does not address "freeze attacks" since it allows an index update with the same timestamp. I did that deliberately to slowly ramp up the security checks in order to avoid problems along the way. Code-wise, blocking freeze-attacks is mostly a matter of changing the timestamp check from `<` to `<=`.
See merge request !302
Installer instances always copy the APK to a safe place to run the install
from. That copy needs to be deleted. Until we have the whole lifecycle in
InstallManagerService, we need this hack. It should be handled on the
broadcast from InstallerService to say that its complete.
#611!300
Including the App and Apk instances in the Intent that starts
InstallManagerService ensures that the needed data is present in the
Service no matter what happens outside of the Service. For example, if the
index is updated or cleared while an install is in progress, the install
process still needs to know the name and packageName of the app to update
the Notification.
A cleaner but more labor-intensive way to implement this would be to make
App and Apk properly implement the full Parcelable interface. That would
require tests to check that the Parcelable methods have all the same fields
as toContentValues() and the database.
closes#660https://gitlab.com/fdroid/fdroidclient/issues/660
A hacked fdroid server could "replay" old index.jar files known to have
apps with vulnerabilities in it. That provides a long window of time for
exploiting that vulnerability. By checking that the timestamp of an update
is never older than the current index, this attack is prevented.
Its officially possible to have the ROM's filesystem with any name. While I
have never seen that in practice, Android does provide an easy method to
get the real name. Plus this should help avoid typos and the like, and
make it easy to track things that rely on that filesystem path.
This makes it easy to tell which debug build a device is running, since the
versionName now automatically describes the exact commit that was built,
based on `git describe`, e.g.: 0.100-alpha7-33-gc2e8e8a
For release builds, i.e. builds from commits that are tagged as releases,
the versionName will be just the tag name: 0.100-alpha8
closes#664https://gitlab.com/fdroid/fdroidclient/issues/664
For some odd reason, something is sending a URL to be downloaded that then
results in a null Apk instance. My first guess was because it was being
canceled, but the interrupted receiver is not even registered yet. My
second thought is that something is sending a download and cancel Intent at
the same time. In any case, its something to keep in mind when reworking
InstallManagerService once InstallerService comes along.
closes#660https://gitlab.com/fdroid/fdroidclient/issues/660
The App(Context context, PackageManager pm, String packageName) constructor
was not setting App.icon, which is required for lots of things. This makes
it always get set, since its just a standard file name, and it does not
have to even exist yet.
When reworking this in 7f10be18c6dd0b69e2fdbae98d09b197e60af443, I confused
the "Processing" with the "Downloading", probably because I thought those
steps were combined, but they are not. Also, I forgot that Downloader
instances do not broadcast status. So its just a matter of setting up the
right ProgressListeners.
https://gitlab.com/fdroid/fdroidclient/issues/633
Since Intents can come in any time, whether WifiInfoThread is running or
not, the global static vars for storing the WiFi settings info should only
be updated from the WifiInfoThread. Otherwise, the WiFi settings could be
nulled out between the time of the null guard and the execution in code
like this:
if (!TextUtils.isEmpty(FDroidApp.ipAddressString) && netmask != null) {
FDroidApp.subnetInfo = new SubnetUtils(FDroidApp.ipAddressString, netmask).getInfo();
fixes#589https://gitlab.com/fdroid/fdroidclient/issues/589
java.lang.RuntimeException: An error occured while executing doInBackground()
at android.os.AsyncTask$3.done(AsyncTask.java:304)
at java.util.concurrent.FutureTask.finishCompletion(FutureTask.java:355)
at java.util.concurrent.FutureTask.setException(FutureTask.java:222)
at java.util.concurrent.FutureTask.run(FutureTask.java:242)
at android.os.AsyncTask$SerialExecutor$1.run(AsyncTask.java:231)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1112)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:587)
at java.lang.Thread.run(Thread.java:818)
Caused by: java.lang.IllegalArgumentException: Could not parse [null/24]
at org.apache.commons.net.util.SubnetUtils.calculate(SubnetUtils.java:275)
at org.apache.commons.net.util.SubnetUtils.<init>(SubnetUtils.java:62)
at org.fdroid.fdroid.net.WifiStateChangeService$WaitForWifiAsyncTask.doInBackground(WifiStateChangeService.java:89)
at org.fdroid.fdroid.net.WifiStateChangeService$WaitForWifiAsyncTask.doInBackground(WifiStateChangeService.java:70)
at android.os.AsyncTask$2.call(AsyncTask.java:292)
at java.util.concurrent.FutureTask.run(FutureTask.java:237)
... 4 more
java.lang.IllegalArgumentException: Could not parse [null/24]
at org.apache.commons.net.util.SubnetUtils.calculate(SubnetUtils.java:275)
at org.apache.commons.net.util.SubnetUtils.<init>(SubnetUtils.java:62)
at org.fdroid.fdroid.net.WifiStateChangeService$WaitForWifiAsyncTask.doInBackground(WifiStateChangeService.java:89)
at org.fdroid.fdroid.net.WifiStateChangeService$WaitForWifiAsyncTask.doInBackground(WifiStateChangeService.java:70)
at android.os.AsyncTask$2.call(AsyncTask.java:292)
at java.util.concurrent.FutureTask.run(FutureTask.java:237)
at android.os.AsyncTask$SerialExecutor$1.run(AsyncTask.java:231)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1112)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:587)
at java.lang.Thread.run(Thread.java:818)
If the device is small, then the "Conenct and trade apps with people near you"
header takes up too much space and we end up not being able to see any nearby
people at all, even if they are in the list. As such, this also removes that
header for "small" and "ldpi" devices. During testing I found that "small" was
not enough, because a 240x400 screen is considered "medium" and there is not
enough space. ldpi seems to be a reasonable metric for "that header is going to
be taking valuable space and should not be shown then".
All larger devices retain the header and seem to look nice.
This also pushes the "Can't find what you're looking for?" message and associated
buttons right to the bottom of the screen. This is more in line with the original
design.
Fixes#604.
Although this is deprecated, and will result in queries being run on the
main thread, it is required to fix#606. In the future, `LoaderManagers`
should be used to address the concerns of querying on the main thread.
Don't allow download notifications to be cancelled, other than our "Cancel" action on them.
*NOTE: To be cherry-picked into stable-v0.100*
As a user, I expect removing a notification of a pending action to cancel that pending action. This change makes the pending notifications uncancellable, unless users use our cancel action added to the events by @paresh. If the user is on an older device that doesn't support these type of rich notifications, then they will need to touch the pending download notification, which will take them to the `AppDetails` activity for that app. From there, they can cancel the install, which will remove the app from the pending download queue.
Until the notifications are reworked, this is the simplest solution. In the near future, the notifications will likely be combined into one more intelligent notification with better defined semantics.
See merge request !292
Until the notifications are reworked, this is the simplest solution.
In the near future, the notifications will likely be combined into one
more intelligent notification with better defined semantics.
Type parameters can be ommited if defined and declared in same statement.
`onStart()` is deprecated and not required, as we target APIs > 5.
`Intent.FLAG_ACTIVITY_CLEAR_TASK` is not supported on APIs < 11 but we target 8.
Before, it would change fields in a final Repo instance, which means that
things could be out of sync when accessed. Now it swaps out the old one
with a new Repo instance in one step.
The local repo variables are now declared volatile so that they are more
predictable when accessed from various threads (WifiStateChangeService,
SwapService, etc.)
askServerToSwapWithUs(NewRepoConfig) was unused, so I removed it.
The IntentService provides the nice incoming Intent queue. It also runs
the Intent in a thread, so even the initial check is now in a very low
priority thread. The queuing prevents the incoming Intents from competing.
This also simplifies the code since the lifecycle is more automatic now.
The `android:process` statement in AndroidManifest.xml causes another
process to be created to run CrashReportActivity. This was causing lots of
things to be started/run twice including CleanCacheService and
WifiStateChangeService.
Lint says that only GET_META_DATA and GET_SHARED_LIBRARY_FILES are allowed.
This contradicts Android's documentation where GET_UNINSTALLED_PACKAGES
is also allowed.
Fixes#605
This was crashing when coming to SwapAppsView because some of the flow
changed related to the new DownloaderService and InstallManagerService.
Also, this lazy loading is a tiny optimization that we cannot afford right
now, there are far too many lifecycle bugs with swap.
This should fix the PMD error:
"Check the value returned by the skip() method of an InputStream to see if
the requested number of bytes has been skipped."
The check was set up to only cancel when the `AppDetails` for that app
was shown. This is the correct behaviour for the 'complete' event, but
not the cancel. The cancel event should always result in the relevant
notification being removed.
