On 06/08/13 14:20, Colin Percival wrote:
I just found a struct padding bug which affects ARM and possibly other similar
platforms (but not x86). Is anyone running Tarsnap on such systems?
I just realized that the answer here is guaranteed to be no, since any attempt
to use tarsnap on such systems is guaranteed to fail immediately due to said
padding bug. (The precise error is "Unexpected EOF of key data" / "Key cache
initialization failed".)
Fortunately for me, this means I can fix this bug without worrying about running
into backwards compatibility issues with key files generated on bug-present
systems, since it's impossible to generate a key file on such a system. ;-)