The lsattr tests currently pass on blueline but fail on cuttlefish.
Too... many... file systems...
Bug: http://b/147769529
Test: treehugger
Change-Id: I1de08609df55887d5d25bddeae2772fb5b39a649
This reverts commit 0751987715.
Reason for revert: date tests fixed by 5854c801a7.
Bug: http://b/147071480
Change-Id: I7d0170c97c59857ae6605d742052821119b7e810
The accompanying bionic change makes sysconf(_SC_ARG_MAX) report the
exact value the kernel is using again.
Test: treehugger
Change-Id: Ie5a370757477e98838b5f1c7aa133734f798fe7c
This is just blkid and unzip for now. Will probably work on blkid in my
spare time. Might make sense to move the unzip tests back into unzip...
Test: manual
Change-Id: I16ab6a190670c6dcf829fb1f81390ae11b4dbdb8
Suppress these messages when using `adb shell -t` from a build server:
Remote PTY will not be allocated because stdin is not a terminal.
Use multiple -t options to force remote PTY allocation.
Also avoid this incorrect assumption:
error: device offline
-- mktemp not present
This should count as a failure instead, not "this is a test for
something that isn't on the device".
Test: still see escapes run interactively.
Bug: http://b/135744229
Change-Id: I1e6ad7fc6d5cbf4b208ad9e8ecbb51048caa3fbb
We don't want to see escape sequences in the logs in sponge.
Test: still see escapes run interactively.
Bug: http://b/135744229
Change-Id: Ie27355c56eb8c4a09cf3c05c482336f30ea7adb4
Show a summary at the end, and return a meaningful exit status.
This doesn't actually work yet because it needs an upstream fix to use
$(()) rather than $[] for shell arithmetic --- mksh only supports the
former (and bash calls the latter deprecate and threatens to remove it).
Also run each test in its own subdirectory. The cp and mkdir tests, for
example, currently tread on each others' toes, even when run
sequentially, because cp leaves around a `walrus` that mkdir doesn't
expect to exist.
Bug: http://b/135744229
Test: run this script
Change-Id: I5f28340c9ea1f9cebb35175801197715f10700a4
