stat on Windows fails with ENOENT when passed a path with a trailing
slash or backslash, regardless of whether the target is actually a
directory. Emulate the correct POSIX behavior by stripping trailing
path separators and then checking if the target is a directory if
successful.
Bug: http://b/30481559
Bug: https://code.google.com/p/android/issues/detail?id=214633
Change-Id: I1d398d19a9bce1ecb3fdc4aabc31aa98c82c3f93
Test: Relevant adb_tests pass on Linux and Windows 10.
* Use const reference type for parameters to avoid unnecessary copy.
* Suppress warning of not using faster overloaded string find function.
Bug: 30407689
Bug: 30411878
Change-Id: I6cfdbbd50cf5e8f3db6e5263076d3a17a9a791ee
Test: build with WITH_TIDY=1
Merged-In: Ie79dbe21899867bc62031f8618bb1322b8071525
Taking a zip bugreport has 2 phases: generating the bugreport and
pulling the file.
Initially adb was printing 2 messages, but since the latter is almost
instantaneously, it could be confusing to have 2 lines...
Fixes: 30451250
Change-Id: I1c6cc6163492c1fb6064667dfdb7aaf6ed4c4c6f
adb calls bugreportz to generate a bugreport; initially, bugreportz
would only report the final status of the operation (OK or FAIL), but
now it sends intermediate PROGRESS lines reporting its progress (in the
form of current/max).
Similarly, the initial implementation of 'adb bugreport <zip_file>'
would print an initial 'please wait' message and wait for the full
stdout before parsing the result, but now it uses a new callback class
to handle the stdout as it is generated by bugreportz.
BUG: 28609499
Change-Id: I6644fc39a686279e1635f946a47f3847b547d1c1
bugreport() will be soon refactored to track progress, which will
require more comprehensive unit tests.
As such, it's better to move it to its own files, which in turn also
requires moving send_shell_command() and usage() to commandline.h.
Fixes: 30100363
Bug: 30268737
Change-Id: I3cdf114a0b5547293320042ff0749a60886440b0
adb install-multiple makes a minimum of 3 calls to the 'pm' command. This
causes at least 3 separate "pm" processes to be spun up. Instead, use the
'cmd' command which runs in the existing system_server process.
Bug: 27483932
Change-Id: Ia1bed405c3d7e675a1a56ff82c692aaa94388a5c
(cherry picked from commit 4838ae1ab5)
Malformed or corrupt archives may be missing their in-band EOD
content, so make sure that the target sees one regardless rather
than continuing to block on read, expecting in-band signaling.
Bug 28056941
Change-Id: Ic39966d3448787a8c511783d39172032ed9589c3
This keeps regressing, so add a test to keep this from happening.
Bug: http://b/23603716
Bug: http://b/25965770
Bug: http://b/29565233
Test: Ran test with/without commit cd5d737, fails before, passes after
Change-Id: I8c431e10fc76da5a9fd404dd70f17bb8a8df24e6
SIGHUP handling depends on the local socket fd being fully closed in
order to trigger an event on its counterpart. Release the local socket
fd inside of Subprocess when returning it to ensure this.
Functionally, a cherry-pick of internal commit 42afe20.
Bug: http://b/29565233
Change-Id: I13b17bcddf0b396a5f4880f9e410fbbf24d9370d
SIGHUP handling depends on the local socket fd being fully closed in
order to trigger an event on its counterpart. Release the local socket
fd inside of Subprocess when returning it to ensure this.
Bug: http://b/29565233
Change-Id: I13b17bcddf0b396a5f4880f9e410fbbf24d9370d
This commit fixes two somewhat related issues in shell_service.
- The fd returned by StartSubprocess is owned by a unique_fd
contained in the Subprocess object, but also gets closed by the
caller. Resolve this by duping the returned file descriptor.
- A Subprocess object can be destroyed immediately after its initial
construction in StartSubprocess if we're sufficiently unlucky.
Split up the fork/exec and "start management thread" steps, so that
we can safely do everything we need to do on the Subprocess before
handing it over to the thread that'll eventually destroy it.
Also includes squashed patches from AOSP master that allow for use of
unique_fd inside adb.
Bug: http://b/29254462
Change-Id: Id9cf0b7e7a7293bee7176919edc758597691c636
(cherry picked from commit c0e6e40cc9)
(cherry picked from commit 54c72aaccc)
(cherry picked from commit 2c5d1d7cd9)
(cherry picked from commit 2a7b86337f)
(cherry picked from commit 13ea01db45)
(cherry picked from commit 344778da41)
This commit fixes two somewhat related issues in shell_service.
- The fd returned by StartSubprocess is owned by a unique_fd
contained in the Subprocess object, but also gets closed by the
caller. Resolve this by duping the returned file descriptor.
- A Subprocess object can be destroyed immediately after its initial
construction in StartSubprocess if we're sufficiently unlucky.
Split up the fork/exec and "start management thread" steps, so that
we can safely do everything we need to do on the Subprocess before
handing it over to the thread that'll eventually destroy it.
Bug: http://b/29254462
Change-Id: Id9cf0b7e7a7293bee7176919edc758597691c636
The <hardware/qemu_pipe.h> header isn't related to any loadable
hardware module, this patch thus relocates it to <system/qemu_pipe.h>
which is a much more logical location.
Note the following changes:
- The pipe name must begin with the 'pipe:' prefix to avoid an
un-necessary snprintf() and buffer copy.
- Does not probe for the obsolete /dev/goldfish_pipe device, i.e.
only use /dev/qemu_pipe instead.
- Use QEMU_PIPE_DEBUG() instead of D() as the debugging macro.
+ Update ADB to use the newest <system/qemu_pipe.h>
+ Add qemu_pipe_frame_send() and qemu_pipe_frame_recv() utility
functions which replace the obsolete qemud_channel_send() and
qemud_channel_recv() from the defunct <hardware/qemud.h>
header.
BUG=25875346
Change-Id: Ic290a5b79d466c2af64b49bd9134643277c11bfd
`adb root` would hang for devices connected over TCP/IP, because they
never come back. Unfortunately, there doesn't seem to be an easy way to
find out what we're talking to, so replace it with a long sleep for now.
Bug: http://b/27836058
Change-Id: Icd6e6bf817ca9e2124c2961665c37b2f31f7ef29
Josh Gao