Background
==========
On Windows, if you run "adb shell exit" in a loop in two windows,
eventually the adb client will be unable to connect to the adb server. I
think connect() is returning WSAEADDRINUSE: "Only one usage of each
socket address (protocol/network address/port) is normally permitted.
(10048)". The Windows System Event Log may also show Event 4227, Tcpip.
Netstat output is filled with:
# for the adb server
TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT
# for the adb client
TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT
The error probably means that the client is running out of free
address:port pairs.
The first netstat line is unavoidable, but the second line exists
because the adb client is not waiting for orderly/graceful shutdown of
the socket, and that is apparently required on Windows to get rid of the
second line. For more info, see
https://github.com/CompareAndSwap/SocketCloseTest .
This is exacerbated by the fact that "adb shell exit" makes 4 socket
connections to the adb server: 1) host:version, 2) host:features, 3)
host:version (again), 4) shell:exit. Also exacerbating is the fact that
the adb protocol is length-prefixed so the client typically does not
have to 'read() until zero' which effectively waits for orderly/graceful
shutdown.
The Fix
=======
Introduce a function, ReadOrderlyShutdown(), that should be called in
the adb client to wait for the server to close its socket, before
closing the client socket.
I reviewed all code where the adb client makes a connection to the adb
server and added ReadOrderlyShutdown() when it made sense. I wasn't able
to add it to the following:
* interactive_shell: this doesn't matter because this is interactive and
thus can't be run fast enough to use up ports.
* adb sideload: I couldn't get enough test coverage and I don't think
this is being called frequently enough to be a problem.
* send_shell_command, backup, adb_connect_command, adb shell, adb
exec-out, install_multiple_app, adb_send_emulator_command: These
already wait for server socket shutdown since they already call
recv() until zero.
* restore, adb exec-in: protocol design can't have the server close
first.
* adb start-server: no fd is actually returned
* create_local_service_socket, local_connect_arbitrary_ports,
connect_device: probably called rarely enough not to be a problem.
Also in this change
===================
* Clarify comments in when adb_shutdown() is called before exit().
* add some missing adb_close() in adb sideload.
* Fixup error handling and comments in adb_send_emulator_command().
* Make SyncConnection::SendQuit return a success boolean.
* Add unittest for adb emu kill command. This gets code coverage over
this very careful piece of code.
Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c
Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
This makes no measurable difference to the sync time; "adb sync" of
everything on /system for a Nexus 9 still takes 20s.
Change-Id: Ifa2626f7453937e43856b9c4ee06e1f5db0aa273
I fixed the server side to send detailed error reports, but the client-side
code for pull was broken because it had already read the "FAIL" header before
calling the general error reporting code that expects to be able to read
the header. This meant we'd always report that we failed to read the failure
message.
Also add a couple of missing "\n"s, make sure every error message is prefixed
by "adb: ", and remove a useless path length check that would silently ignore
over-long paths rather than relying on SendRequest to detect and report the
problem.
Bug: http://b/6205106
Change-Id: I23862ececf03b761115ffa3f7725b7e1cecb48c7
Sending the SEND/DATA/DONE packets all in one write makes small file and
symbolic link transfer orders of magnitude faster. Transferring all the
symbolic links in /system/bin or all the cacerts in /system/etc goes down
from about 6s in each case to less than 0.5s.
Change-Id: Ia1706cef25535d0b685d3744241a0557a52b22ff
adb push was not returning a bad exit code when write_data_file() or
write_data_link() failed. I encountered this when running the unittest
on Windows which can get into situations where stat() succeeds, but
open() fails due to pre-existing exclusive file access (which typically
doesn't exist on unix).
The same code is used by adb install, so this also fixes its error
handling.
Fixed some fd leaks and propagation of errors when reading a file.
Fixed a unittest to close temp files before reading them.
Change-Id: Ieba0026fa4c79eb0484676e4f2faaac9603ef584
Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
We can double the speed of "adb sync" (on N9) if we increase SYNC_DATA_MAX
from 64KiB to 256KiB. This change doesn't do that, because I still haven't
managed to plumb through the information about whether we're a new adb/adbd
to file_sync_client.cpp and file_sync_service.cpp. But this is already a big
change with a lot of cleanup, so let's do the cleanup and worry about the
intended change another day...
This change does improve performance somewhat by halving the number of
lstat(2) calls made on the client side, and ensuring that most packets are
sent with a single write. This has the pleasing result of making the null
sync on an AOSP N9 go from just over 300ms to around 100ms, which means it
now seems instantaneous (https://en.wikipedia.org/wiki/Mental_chronometry).
Change-Id: If9f6d4c1f93ec752b95f71211bbbb1c513045166
This removes adb_dirstart and adb_dirstop. It also fixes a couple of memory
leaks by switching to std::string. This also fixes the bug in the previous
change --- mkdirs is given input like "/system/bin/sh" and only expected to
create "/system/bin". In a later change, we should remove mkdirs and only
expose the intended "unlink && mkdirs && create" functionality.
Change-Id: I30289dc1b3dff575cc1b158d993652178f587552
~ Rewrote mkdirs to be in C++ style.
~ Replaced adb_dir{start,stop} with std::string params and (r)find.
+ Added test for mkdirs.
Also make base/test_utils.h public and support temporary directories
as well as files.
Change-Id: I6fcbdc5e0099f3359d3aac6b00c436f250ca1329
At runtime, vsnprintf (and android::base::StringPrintf which calls it)
call a mingw version of vsnprintf, not the vsnprintf from MSVCRT.DLL.
The mingw version properly understands %zd and PRIu64 (the latter,
provided that you #include <inttypes.h>).
The problem was that android::base::StringPrintf was causing
compile-time errors saying that %zd and PRIu64 were not recognized. It
seems that this was because the attribute on the function prototypes
specified `printf' instead of `gnu_printf'. Once that was fixed to match
vsnprintf's attribute, the warnings went away.
This uses similar preprocessor techniques as <android/log.h>.
Also restore a %zd usage to avoid a static_cast<>, and make
print_transfer_progress()'s format string compile-time checkable (and
tweak some types and %llu => PRIu64).
Change-Id: I80b31b9994858a28cb7c6847143b86108b8ab842
Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
Also remove an sprintf. Also fix various bits of code that were
reporting stale adb_error values when they meant strerror.
Bug: http://b/20666660
Change-Id: Ibeb48b7bc21bb0ec30ba47889d1d671ee480e1b7
I keep trying to clean things up and needing std::strings. Might as
well just do this now.
usb_linux_client.c is going to stay as C because GCC isn't smart
enough to deal with the designated initializers it uses (though for
some reason it is in C mode).
The Darwin files are staying as C because I don't have a way to test
that they build.
The Windows files are staying as C because while I can actually build
for them, it's slow and painful.
Change-Id: I75367d29205a9049d34460032b3bb36384f43941
Josh Gao
Spencer Low
Elliott Hughes
Alex Vallée
Dan Albert