| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
encrypted file without a header
|
| |
|
| |
|
|\
| |
| | |
Fix collision on temporary files + adjust permissions
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This changeset hardens the temporary file and directory creation to address multiple problems that may lead to exposure of files to other users, data loss or other unexpected behaviour that is impossible to debug.
**[CWE-668: Exposure of Resource to Wrong Sphere](https://cwe.mitre.org/data/definitions/668.html)**
The temporary file and folder handling as implemented in ownCloud is performed using a MD5 hash over `time()` concatenated with `rand()`. This is insufficiently and leads to the following security problems:
The generated filename could already be used by another user. It is not verified whether the file is already used and thus temporary files might be used for another user as well resulting in all possible stuff such as "user has file of other user".
Effectively this leaves us with:
1. A timestamp based on seconds (no entropy at all)
2. `rand()` which returns usually a number between 0 and 2,147,483,647
Considering the birthday paradox and that we use this method quite often (especially when handling external storage) this is quite error prone and needs to get addressed.
This behaviour has been fixed by using `tempnam` instead for single temporary files. For creating temporary directories an additional postfix will be appended, the solution is for directories still not absolutely bulletproof but the best I can think about at the moment. Improvement suggestions are welcome.
**[CWE-378: Creation of Temporary File With Insecure Permissions](https://cwe.mitre.org/data/definitions/378.html)**
Files were created using `touch()` which defaults to a permission of 0644. Thus other users on the machine may read potentially sensitive information as `/tmp/` is world-readable. However, ownCloud always encourages users to use a dedicated machine to run the ownCloud instance and thus this is no a high severe issue. Permissions have been adjusted to 0600.
**[CWE-379: Creation of Temporary File in Directory with Incorrect Permissions](https://cwe.mitre.org/data/definitions/379.html)**
Files were created using `mkdir()` which defaults to a permission of 0777. Thus other users on the machine may read potentially sensitive information as `/tmp/` is world-readable. However, ownCloud always encourages users to use a dedicated machine to run the ownCloud instance and thus this is no a high severe issue. Permissions have been adjusted to 0700.Please enter the commit message for your changes.
|
| | |
|
|\ \
| | |
| | | |
Block old legacy clients
|
| | |
| | |
| | |
| | | |
In case of an not sent UA header consider the client as valid
|
| | | |
|
| | | |
|
| | | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This Pull Request introduces a SabreDAV plugin that will block all older clients than 1.6.1 to connect and sync with the ownCloud instance.
This has multiple reasons:
1. Old ownCloud client versions before 1.6.0 are not properly working with sticky cookies for load balancers and thus generating sessions en masse
2. Old ownCloud client versions tend to be horrible buggy
In some cases we had in 80minutes about 10'000 sessions created by a single user. While this change set does not really "fix" the problem as 3rdparty legacy clients are affected as well, it is a good work-around and hopefully should force users to update their client
|
|\ \ \
| | | |
| | | | |
fix encryption header error
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | |
| | | |
| | | | |
testRewind tests reading and writing after rewind on an encrypted stream; testWriteWriteRead tests r+ mode
|
| | | | |
|
|/ / /
| | |
| | | |
When moving back the pointer to position 0 (using stream_seek), the pointer on the encrypted stream will be moved to the position immediately after the header. Reading the header again (invoked by stream_read) will cause an error, writing the header again (invoked by stream_write) will corrupt the file. Reading/writing the header should therefore happen when opening the file rather than upon read or write. Note that a side-effect of this PR is that empty files will still get an encryption header; I think that is OK, but it is different from how it was originally implemented.
|
|\ \ \
| | | |
| | | | |
[encryption] fix moving files to a shared folder
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | |/
| |/| |
|
|\ \ \
| | | |
| | | | |
Ignore test folders when checking the code for compliance
|
| | | | |
|
|\ \ \ \
| | | | |
| | | | | |
Tear down FS after running trash expire
|
| | | | | |
|
|\ \ \ \ \
| |/ / / /
|/| | | | |
Show hint if there is no delete permission
|
| | |/ /
| |/| |
| | | |
| | | |
| | | |
| | | | |
* add class .no-permission which shows the default delete icon
* fixes #15172
* add unit test for no permission and for delete permission
|
| | | | |
|
|\ \ \ \
| | | | |
| | | | | |
Add missing loading indicator for enabled apps
|
| | |/ /
| |/| |
| | | |
| | | |
| | | |
| | | |
| | | | |
* happened when an app gets activated, because the
new appended HTML doesn't contain the loading
image
* fixes #15806
|
|\ \ \ \
| |/ / /
|/| | | |
Issue 15804 occ user delete exception
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| |/ / |
|
|\ \ \
| | | |
| | | | |
Fix webdav access for public reshare
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|