Reimplement commit ffc11d3cf3 using
fdevent. The previous attempt was reverted because we were blindly
continuing when revents & POLLIN == 0, which ignored POLLHUP/POLLERR,
leading to spinloops when the opposite end of the file descriptor was
shutdown when we had no data left to read.
This patch reimplements the functionality implemented by that commit
using fdevent, which gets us detection of spin loops for free.
Bug: http://b/74616284
Test: ./test_device.py
Change-Id: I1abd671fef4c29e99dad968aa66bb754ca382578
We were logging "fd.get()" after we had already done a
"std::move(fd)". That won't log the value we were hoping for.
We instead cache the file descriptor value in a local int prior
to the move(), and log that.
Test: TreeHugger
Change-Id: I715874ac63329280ffb55881fb2590fb31dc2457
vsock is a socket address family for communicating into and out of
virtual machines. Addresses have a port and CID. The CID is unique to
each virtual machine on the computer. The VM host always has CID 2.
http://man7.org/linux/man-pages/man7/vsock.7.html
Inside the android guest, the adb daemon hosts a vsock server with
VMADDR_CID_ANY, automatically using the guest CID. The adb server
can now connect to addresses of the form vsock:cid:port, where the CID
must be specified and the port defaults to 5555.
This is a significant speed improvement for ADB connections in
Cuttlefish, with 150-200 MB/s for `adb push` and 100-150 MB/s for
`adb pull`. It also allows removing some proxying steps from Cuttlefish,
simplifying the full connection path, and removes a dependency on the
unstable ivshmem protocol.
Commands tested against a Cuttlefish VM with CID 3:
adb connect vsock:3:5555
adb -s vsock:3:5555 shell
adb disconnect vsock:3:5555
Supporting "adb disconnect" and "adb -s" required modifying some of the
parts that parse addresses / serials.
push/pull trials with native adb vsock support in cuttlefish:
100m: 1 file pushed. 297.6 MB/s (104857600 bytes in 0.336s)
100m: 1 file pushed. 270.3 MB/s (104857600 bytes in 0.370s)
100m: 1 file pushed. 271.7 MB/s (104857600 bytes in 0.368s)
100m: 1 file pushed. 250.5 MB/s (104857600 bytes in 0.399s)
100m: 1 file pushed. 277.1 MB/s (104857600 bytes in 0.361s)
100m: 1 file pushed. 263.5 MB/s (104857600 bytes in 0.379s)
100m: 1 file pushed. 242.6 MB/s (104857600 bytes in 0.412s)
100m: 1 file pushed. 271.8 MB/s (104857600 bytes in 0.368s)
100m: 1 file pushed. 267.1 MB/s (104857600 bytes in 0.374s)
/data/local/tmp/100m: 1 file pulled. 212.8 MB/s (104857600 bytes in 0.470s)
/data/local/tmp/100m: 1 file pulled. 236.7 MB/s (104857600 bytes in 0.423s)
/data/local/tmp/100m: 1 file pulled. 201.2 MB/s (104857600 bytes in 0.497s)
/data/local/tmp/100m: 1 file pulled. 255.6 MB/s (104857600 bytes in 0.391s)
/data/local/tmp/100m: 1 file pulled. 199.6 MB/s (104857600 bytes in 0.501s)
/data/local/tmp/100m: 1 file pulled. 214.6 MB/s (104857600 bytes in 0.466s)
/data/local/tmp/100m: 1 file pulled. 254.2 MB/s (104857600 bytes in 0.393s)
/data/local/tmp/100m: 1 file pulled. 212.5 MB/s (104857600 bytes in 0.471s)
/data/local/tmp/100m: 1 file pulled. 218.9 MB/s (104857600 bytes in 0.457s)
/data/local/tmp/100m: 1 file pulled. 223.6 MB/s (104857600 bytes in 0.447s)
Bug: 121166534
Change-Id: I50f21fb5c9acafb8daa789df4e28c9e1bbbbf2ef
Test: adb connect/shell/disconnect
Rename it to something more appropriate, while we're at it.
Test: test_adb.py
Test: test_device.py
Test: $ANDROID_HOST_OUT/nativetest64/adb_test/adb_test
Test: adb shell /data/nativetest64/adbd_test/adbd_test
Change-Id: I0f95d348eeacb45a810696d748c8340d2068f666
Add some services that skip the service fd to see how much of a benefit
it'll be to eliminate it.
Test: adb raw source:$((300 * 1024 * 1024)) | pv > /dev/null
Test: dd if=/dev/zero bs=1M count=100 | pv | adb raw sink:$((100 * 1024 * 1024))
Change-Id: I042f25f85b16ae9869cb1f1e306d8671b024ed97
Let's use LOG(FATAL)/PLOG(FATAL) for actual fatal stuff.
Add a Windows error(3) and move folks who didn't really mean "abort"
fatal over to it. Also get rid of syntax_error which wasn't adding a
lot of value, and most of the places it was adding "usage: " didn't seem
entirely appropriate anyway.
In particular, we seemed to have confused fastdeploy.cpp into aborting
in most user error cases, and none of the reviewers noticed. Clearly
we'd all lost track of far too many options.
(I've also cleaned up a few random instances of fprintf(3) + exit(2).)
Bug: N/A
Test: manual
Change-Id: I3e8440848a24e30d928de9eded505916bc324786
The standard (RFC 1122 - 4.2.2.13) says that if we call close on a
socket while we have pending data, a TCP RST should be sent to the
other end to notify it that we didn't read all of its data. However,
this can result in data that we've succesfully written out to be dropped
on the other end. To avoid this, instead of immediately closing a
socket, call shutdown on it instead, and then read from the file
descriptor until we hit EOF or an error before closing.
Bug: http://b/74616284
Test: ./test_adb.py
Test: ./test_device.py
Change-Id: I36f72bd14965821dc23de82774b0806b2db24f13
Previously, we were returning the result of SendOkay/SendFail in a few
places after handling a host request, which is incorrect for two
reasons. First, the return type of SendOkay/SendFail is bool, and
handle_host_request was expected to return 0 on success. Second, we
don't care if the SendOkay fails; if we got to that point, we're done
with the request, regardless of whether we succeeded to report our
result. The result of this was a spurious failure result reported after
the initial result, which was ignored by the adb client.
Test: manually straced adb server
Test: python test_adb.py
Test: python test_device.py
Change-Id: I7d45ba527e1faccbbae5b15e7a0d1557b0a84858
An IOVector is a collection of immutable reference counted blocks which
can have its head detached at an arbitrary index. This is extremely
useful for implementing packet-framed protocols like adb on top of a
stream protocol like TCP: a stream reader can read blocks, append them
to the end of the IOVector, and then pull packets off of the front.
This also lends itself naturally towards scatter/gather I/O, which will
enable us to read data from disk and send it across the wire with a
theoretical minimum number of copies in USB, and one extra copy over
TCP.
Since this is basically a generalization of std::deque<Range>, delete
Range and replace its uses with IOVector.
Test: adb_test
Test: wine adb_test.exe
Change-Id: I06561ad0bb25a3a51b378b61d257b5b04b41d9c4
Remove fdevent_install and fdevent_remove in favor of using
fdevent_create and fdevent_destroy, so that we can put RAII types (i.e.
unique_fd) into fdevent without worrying about -Wexit-time-destructors
or structs that are freed instead of deleted.
Bug: http://b/79786774
Test: python test_device.py
Change-Id: I8471cc00574ed492fe1b196944976cdaae8b7cff
Add two states: connecting and authorizing, to disambiguate the offline
and unauthorized states, respectively.
Previously, devices would transition as follows:
offline -> unauthorized -> offline -> online
offline -> unauthorized (when actually unauthorized)
With this patch:
connecting -> authorizing -> online
connecting -> authorizing -> unauthorized (when actually unauthorized)
This allows test automation and the like to distinguish between offline
devices, unauthorized devices, and working devices without having to
do retry loops with arbitrary sleeps on their end.
Bug: http://b/79257434
Test: adb_test
Test: adbd_test
Test: manually plugging in a device with `while true; do adb shell echo foo; done`
Change-Id: I036d9b593b51a27a59ac3fc57da966fd52658567
Plumb the transport that we received the adb reverse request on through
to reverse_service, instead of trying to get a unique transport on
devices that have multiple active transports (e.g. a device with USB
(even unplugged) connected via TCP).
Bug: http://b/37066218
Bug: http://b/71898863
Test: `echo foo | nc -l 12345 & adb reverse tcp:12345 tcp:12345; adb shell nc localhost 12345` on a device connected via TCP
Change-Id: Iae199ae787f2e344126bbcacca8544cfc9844a4c
Switch from using std::string as the type we use to hold our payload in
apacket to a custom reimplementation that doesn't zero initialize. This
improves bulk transfer throughput in the adb_benchmark microbenchmark
on walleye by ~20%.
Test: adb shell taskset f0 /data/benchmarktest64/adb_benchmark/adb_benchmark
Change-Id: Ibad797701eb1460c9321b0400c5b167b89b2b4d0
When we fail to write to a local socket peer, we might still have data
queued up to send to the other side. Defer closing the socket until
we've failed to both read and write.
Bug: http://b/74616284
Test: python test_device.py
Change-Id: Ifc4b8fe95369b4872e475c2ae4ee611dd2d8b9d7
asocket has a destructor now, so we have to delete it, or leak the
data pointed to by its std::string.
Bug: http://b/73257049
Test: manual testing with asan
Change-Id: Ia88199292cc74e10032a9a16226d3afc61c3e0be
Switch asocket over to taking a std::string instead of apacket* for
data. This allows us to remove asocket specific fields from apacket*.
Test: python test_device.py with x86_64 emulator, walleye
Test: adb_test on host
Change-Id: I9d157ff331a75ba49a54fdd4194e3f6cdff722f4
The comment that was previously here says that local_socket_list_lock
must be taken, but this function is exposed to external callers that
can't possibly take the lock.
Bug: http://b/65419665
Test: python test_device.py
Change-Id: I12d464933936b2a210a827ccf19ea201020d8d78
Extend device selection to allow selecting a specific transport via
monotonically increasing identifier (visible in devices -l).
This is useful when using multiple devices (like hikey960...) that
have identical bogus serial numbers like 0123456789ABCDEF.
Bug: http://b/37043226
Test: adb -t {1, 2, 9999999} {get-serialno, shell, features}
Change-Id: I55e5dc5a406a4eeee0012e39b52e8cd232e608a6
Relax the shell command length limits when talking to an adbd with the
shell protocol.
shell is pretty much the only service that takes an arbitrarily long
string, so this is somewhat safe.
Bug: http://b/37716055
Test: `adb shell $(python -c 'print "echo " + "f" * (32*1024)') | wc` on L and master
Change-Id: I0737fd2244530ef8080f300cd3a3549a1ab93465
When device goes offline, user usually has to manually replug the
usb device. This patch tries to solve two offline situations, all
because when adb on host is killed, the adbd on device is not notified.
1. When adb server is killed while pushing a large file to device,
the device is still reading the unfinished large message. So the
device thinks of the CNXN message as part of the previous unfinished
message, so it doesn't reply and the device is in offline state.
The solution is to add a write_msg_lock in atransport struct. And it
kicks the transport only after sending a whole message. By kicking
all transports before exit, we ensure that we don't write part of
a message to any device. So next time we start adb server, the device
should be waiting for a new message.
2. When adb server is killed while pulling a large file from device,
the device is still trying to send the unfinished large message. So
adb on host usually reads data with EOVERFLOW error. This is because
adb on host is reading less than one packet sent from device.
The solution is to use buffered read on host. The max packet size
of bulk transactions in USB 3.0 is 1024 bytes. By preparing an at least
1024 bytes buffer when reading, EOVERFLOW no longer occurs. And teach
adb host to ignore wrong messages.
To be safe, this patch doesn't change any logic on device.
Bug: http://b/32952319
Test: run python -m unittest -q test_device.DeviceOfflineTest
Test: on linux/mac/windows with bullhead, ryu.
Change-Id: Ib149d30028a62a6f03857b8a95ab5a1d6e9b9c4e
- moved __android_log_is_debuggable to a new public header
(log_properties.h)
- vendor version of sched_policy uses ALOG* instead SLOG*
Test: (sanity) liblog-unit-tests
Test: (sanity) libcutils_test (noting b/b/32972117, two tests continue
to fail)
Test: system/core as a whole makes with BOARD_VNDK_VERSION := current
now with no problems.
Test: boots/works on internal marlin
Bug: 33241851
(cherry picked from commit 1f83aa424f)
Merged-In: I5bc1f348dc0f0c8814bec5b5c3d2c52c825ab640
Change-Id: I5bc1f348dc0f0c8814bec5b5c3d2c52c825ab640
Add a command to reconnect offline/unauthorized devices, mainly for use
with the inotify-monitoring of vendor key directories added by 2e671202.
Bug: http://b/29273531
Test: manually tested with a sailfish + copying vendor keys
Change-Id: If34cccee4ae553ada65d128b57d03cba8c0d7c46
Use fixed length types for structs going over the wire, constify
arguments where possible, use char* instead of unsigned char* for
apacket data, and assorted other refactoring.
Bug: http://b/29273531
Test: python test_device.py with every combination of old/new adb and adbd
Change-Id: I0b6f818a32be5386985aa4519f542003cf427f9d
When the adb client sends a command to the adb server targeting a
particular device serial, it looks something like this:
host-serial:<serial>:<command>
But if <serial> happens to be an IPv6 address (e.g. when `adb connect`
targets IPv6), the current parsing code doesn't handle the additional
colons properly. This CL fixes the host-serial parsing to handle this
case.
This only affects commands that explicitly name a device serial, e.g.:
adb -s <IPv6> shell
adb -s <IPv6> forward <port> <port>
Implicitly using a single attached device was unaffected by this bug.
Bug: http://b/30891386
Test: `adb -s [fe80::ba27:ebff:feb1:934%eth2]:5555 shell` works now,
and new unittests pass.
Change-Id: Iffe784e61432ae94eb96ed3c8477900a3e807329
New MinGW prebuilts update includes pthreads and C++11 threads support.
Use mutex.h and condition_variable provided by MinGW.
Test: Build AOSP with new MinGW prebuilts
Change-Id: Ia8f890f86652612df3fc2618c2bfbb450a5a2f52
close_all_sockets was assuming that all registered local sockets used
local_socket_close as their close function. However, this is not true
for JDWP sockets.
Bug: http://b/28347842
Change-Id: I40a1174845cd33f15f30ce70828a7081cd5a087e
sockets.cpp was branching on whether a socket close function was
local_socket_close in order to avoid a potential deadlock if the socket
list lock was held while closing a peer socket.
Bug: http://b/28347842
Change-Id: I5e56f17fa54275284787f0f1dc150d1960256ab3
Currently targeting a device by serial requires matching the serial
number exactly. This CL relaxes the matching rules for local transports
to ignore protocol prefixes and make the port optional:
[tcp:|udp:]<hostname>[:port]
The purpose of this is to allow a user to set ANDROID_SERIAL to
something like "tcp:100.100.100.100" and have it work for both fastboot
and adb (assuming the device comes up at 100.100.100.100 in both
modes).
This CL also adds some unit tests for the modified functions to make
sure they work as expected.
Bug: 27340240
Change-Id: I006e0c70c84331ab44d05d0a0f462d06592eb879
The actual maximum length will depend on the version of the shell
protocol being used, and any additional parameters being passed through
(e.g. TERM=xterm-256color). This should be able to be raised to 64K for
devices with commit 3d2904c (L-MR1 and above), but that'll require some
plumbing.
Bug: http://b/20467103
Change-Id: Idf0c46af5b18b854110aba58df13a53297d2475f
I've been using these changes to compile with Visual Studio.
- GetFileBasename(): __FILE__ uses \ with Visual Studio.
- adb_trace.cpp: Apparently VS needs an ampersand before the function name.
- "expr1 ? : expr2" is a GCC extension.
- <algorithm> contains std::min().
- seekdir can't always be #define'd because some headers have members
named seekdir.
- adb_utils.cpp: Not really a compiler issue, just a random fix:
0x7F/DEL is not printable.
Change-Id: I0dfb634f1ba4ccbc0d1b9f71b00e838fbebb3b41
Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
If -d/-e fail, get-serialno and friends will now report an error
and return a failure status code on exit.
Also fix the behavior of -d/-e with $ANDROID_SERIAL --- -d/-e
should override $ANDROID_SERIAL, not the other way round.
I'm deleting my own comment here about always returning "unknown"
for scripts. I can't find any evidence that there are scripts
relying on that, so I think my comment meant "I fear that there
are scripts doing so".
Bug: http://b/24403699
Change-Id: Ie13a751f1137abcfe0cc6c46a0630ba5e02db676
Add has_write_error flag in asocket, so it will not wait on local_socket_closing_list
to write pending packets in local_socket_close(). Although it doesn't fix any problem,
it helps to make the code more stable.
Add a missing put_apacket() in error handling.
Add a check when adding local socket in local_socket_closing_list.
Bug: 23314034
Change-Id: I75b07ba8ee59b7f277fba2fb919db63065b291be
