| Age | Commit message (Collapse) | Author | Files | Lines | 
|---|
|  |  | 
|  | instead of Unix one | 
|  | When a .synced file is encountered, the .synced should be stripped off the
local filename. However the strncpy doesn't 0-terminate the string by itself
so the output filename usually contains some garbage characters at the end.
This commit properly 0-terminates the local filename to avoid this. | 
|  | ping message | 
|  | This bug caused it to never wait for the 'ping' message from
the crashreportmover service | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  | As the debugserver exits after a detach or disconnect, we need to ensure to
start it again upon a new client connection. | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  | A device would report "ErrorCode 1: Error copying keychain data
(MBErrorDomain/1)" if a device entered lock state and had a passcode enabled
during the backup or restore process. To prevent this, an escrow bag must be
sent when starting the service to keep the operation "alive". | 
|  |  | 
|  |  | 
|  |  | 
|  | Under certain circumstances a device could request an empty array
of filenames to be transferred with DLMessageDownloadFiles.
This was not handled correctly in idevicebackup2 by just doing
nothing, resulting in a 'retry loop' waiting endlessly for new
messages from the device.
The solution is to send a termination packet and a success
status response.
Thanks to pondahai for reporting. | 
|  | for reading/writing plist files | 
|  | symlink() | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  |