New features and bugfix for microADB by JianyuWang0623 · Pull Request #42 · spiriou/microADB

JianyuWang0623 · 2024-10-11T09:15:42Z

Summary

New features for microADB, please see the commit message for details.

a34e817: The host report "error: closed" after adb reboot
861e8cb: Fix crash
19345e8: Skip environment variables when exec logcat

Details for `19345e8`

Log

$ adb -s emulator-5554 logcat
nsh: export: command not found

Packet

0000   00 00 00 00 00 00 00 00 00 00 00 00 08 00 45 00   ..............E.
0010   00 64 44 d7 40 00 40 06 f7 ba 7f 00 00 01 7f 00   .dD.@.@.........
0020   00 01 d6 05 15 b3 5f a7 fe 39 f6 e2 f8 80 80 18   ......_..9......
0030   02 00 fe 58 00 00 01 01 08 0a 29 24 9c b4 29 24   ...X......)$..)$
0040   9c b4 73 68 65 6c 6c 3a 65 78 70 6f 72 74 20 41   ..shell:export A
0050   4e 44 52 4f 49 44 5f 4c 4f 47 5f 54 41 47 53 3d   NDROID_LOG_TAGS=
0060   22 27 27 22 3b 20 65 78 65 63 20 6c 6f 67 63 61   "''"; exec logca
0070   74 00

Test

Selftest
CI

Signed-off-by: dengwenqi <dengwenqi@xiaomi.com>

Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>

uv__run_closing_handles libuv/src/unix/core.c:365 uv_run libuv/src/unix/core.c:464 adb_hal_run microADB/hal/hal_uv.c:69 adbd_main /home/djz/workspace/vela_rp/apps/system/adb/adb_main.c:157 nxtask_startup sched/task_startup.c:70 nxtask_start task/task_start.c:134 pre_start sim/sim_initialstate.c:52 Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>

host_memalign sim/posix/sim_hostmemory.c:168 host_realloc sim/posix/sim_hostmemory.c:193 mm_realloc sim/sim_heap.c:334 mm_malloc sim/sim_heap.c:272 mm_zalloc sim/sim_heap.c:402 mm_calloc sim/sim_heap.c:387 calloc umm_heap/umm_calloc.c:73 uv__calloc libuv/src/uv-common.c:94 uv_loop_init libuv/src/unix/loop.c:40 uv_default_loop libuv/src/uv-common.c:821 adb_hal_create_context microADB/hal/hal_uv.c:35 adbd_main /home/djz/workspace/vela_rp/apps/system/adb/adb_main.c:151 nxtask_startup sched/task_startup.c:70 nxtask_start task/task_start.c:134 pre_start sim/sim_initialstate.c:52 Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>

Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>

__assert nuttx/libs/libc/assert/lib_assert.c:36 usb_uv_kick apps/system/adb/microADB/hal/hal_uv_client_usb.c:197 adb_hal_apacket_release /media/liangchaozhong/ssd/x4b/apps/system/adb/microADB/hal/hal_uv_packet.c:55 adb_uv_close_client apps/system/adb/microADB/hal/hal_uv.c:100 usb_uv_on_close apps/system/adb/microADB/hal/hal_uv_client_usb.c:209 uv__finish_close apps/system/libuv/libuv/src/unix/core.c:352 adb_hal_run apps/system/adb/microADB/hal/hal_uv.c:69 adbd_main Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>

Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>

Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

and apply other minor style change Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

since ID_QUIT is always handled in state_wait_cmd Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

…e_wait_cmd_data since the same check is already done in read_from_packet and state_wait_cmd Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

instead of syncmsg::dent and syncmsg::data Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

since it's never used at all Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

This reverts commit 917f97deadd1d70e58347098086cf5ed8e8d0254. Reason for revert: pc need to recv header first

Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>

Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>

…eing overwritten When the windows/mac PC sends multiple files in a write packet, and the last DONE packet sent is incomplete, the remaining bytes of DONE are in the next wirte packet, which will lead to the following two situations. 1. When the sync message has not been completely received, we should maintain the state machine state current frame: 2f 63 6f 6d 2e 76 65 6c 61 2e 63 68 61 72 74 2e /com.vela.chart. 64 65 6d 6f 2f 69 31 38 6e 2f 65 6e 2e 6a 73 6f demo/i18n/en.jso 6e 2c 33 33 32 30 36 44 41 54 41 13 00 00 00 7b n,33206DATA....{ 22 61 22 3a 7b 22 62 22 3a 22 68 65 6c 6c 6f 22 "a":{"b":"hello" 7d 7d 44 4f 4e 45 88 e2 }}DONE.. next frame: 57 52 54 45 3a 00 00 00 01 00 00 00 00 04 00 00 WRTE:........... c6 ac 01 00 a8 ad ab ba 88 65 53 45 4e 44 43 00 .........eSENDC. 00 00 2f 64 61 74 61 2f 63 6f 6d 2e 76 65 6c 61 ../data/com.vela 2e 63 68 61 72 74 2e 64 65 6d 6f 2f 63 6f 6d 2e .chart.demo/com. 76 65 6c 61 2e 63 68 61 72 74 2e 64 65 6d 6f 2f vela.chart.demo/ The complete DONE package is: | 44 4f 4e 45 88 e2 | 88 65 | current next 2. When the device receives DONE, it needs to reply OKAY to the HOST.If the current apcket is reused, OKAY will overwrite the unprocessed content, so a local buffer needs to be used. before wirte: 57 52 54 45 3a 00 00 00 01 00 00 00 00 04 00 00 WRTE:........... c6 ac 01 00 a8 ad ab ba 88 65 53 45 4e 44 43 00 .........eSENDC. after write: 57 52 54 45 3a 00 00 00 01 00 00 00 00 04 00 00 WRTE:........... c6 ac 01 00 a8 ad ab ba 41 59 00 00 00 00 00 00 ........AY...... ^ ^ the buffer ".eSENDC" -> "AY......" Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>

Signed-off-by: guohao15 <guohao15@xiaomi.com>

Signed-off-by: getian <getian@xiaomi.com>

Log: $ adb -s emulator-5554 logcat nsh: export: command not found Packet: 0000 00 00 00 00 00 00 00 00 00 00 00 00 08 00 45 00 ..............E. 0010 00 64 44 d7 40 00 40 06 f7 ba 7f 00 00 01 7f 00 .dD.@.@......... 0020 00 01 d6 05 15 b3 5f a7 fe 39 f6 e2 f8 80 80 18 ......_..9...... 0030 02 00 fe 58 00 00 01 01 08 0a 29 24 9c b4 29 24 ...X......)$..)$ 0040 9c b4 73 68 65 6c 6c 3a 65 78 70 6f 72 74 20 41 ..shell:export A 0050 4e 44 52 4f 49 44 5f 4c 4f 47 5f 54 41 47 53 3d NDROID_LOG_TAGS= 0060 22 27 27 22 3b 20 65 78 65 63 20 6c 6f 67 63 61 "''"; exec logca 0070 74 00 t. Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>

There is a situation similar to "adb shell ps", after the device response data to the host, the host sends okay packet, and then the device should return close packet. In the original code, it is possible that the child process is slow to execute, resulting in the main process receiving okay packet, but the svc has not been released yet, resulting in no close packet being returned. Signed-off-by: yangsong8 <yangsong8@xiaomi.com>

This reverts commit 037d0f1f04518837b26888cc2d3c451be08ea7fd. Reason for revert: can not reserve problem well Signed-off-by: yangsong8 <yangsong8@xiaomi.com>

There is a situation similar to "adb shell ps", after the device response data to the host, the host sends okay packet, and then the device should return close packet. In the original code, it is possible that the child process is slow to execute, resulting in the main process receiving okay packet, but the svc has not been released yet, resulting in no close packet being returned. Signed-off-by: yangsong8 <yangsong8@xiaomi.com>

Signed-off-by: yangsong8 <yangsong8@xiaomi.com>

JianyuWang0623 · 2024-10-11T09:18:23Z

Most of commits by @xiaoxiang781216， @Donny9， @yangsong8-ai， @dengwenqi123 and @guohao15
CC: @GUIDINGLI

xiaoxiang781216 · 2024-10-11T09:41:51Z

@JianyuWang0623 maybe need split the huge pr to small pr.

JianyuWang0623 · 2024-10-11T10:25:33Z

@JianyuWang0623 maybe need split the huge pr to small pr.

OK, set to draft, and will split later.

spiriou · 2024-10-11T11:20:53Z

@@ -0,0 +1,2 @@
+*.o


you can use another build folder with CMake, it doesn't have to be in the source folder.

@spiriou I think this change is for Makefile only since NuttX Makefile based build system doesn't support the out of tree.

spiriou · 2024-10-11T11:24:54Z


 adb_context_t* adb_hal_create_context(void) {
-    adb_context_uv_t *adbd = &g_adbd_context;
+    adb_context_uv_t *adbd = malloc(sizeof(adb_context_uv_t));


what is the usecase of having multiple contexts?
can't we fit everything into the same event loop?

no, just make the code simple and avoid consuming memory before launch adbd.

xiaoxiang781216 · 2024-10-13T03:55:03Z

 static int usb_uv_write(adb_client_t *c, apacket *p) {
    int ret;
-    uv_buf_t buf;
+    uv_buf_t buf[2];


@JianyuWang0623 let's drop the patch and reverted patch

@JianyuWang0623 let's drop the patch and reverted patch

get, will update later

xiaoxiang781216 · 2024-10-13T03:58:55Z

    svc = adb_client_find_service(client, p->msg.arg1, 0);
    if (!svc) {
-        adb_hal_apacket_release(client, p);
+        send_close_frame(client, p, p->msg.arg1, p->msg.arg0);


let's drop this and previous patch

JianyuWang0623 · 2024-10-13T16:07:49Z

Sorry for the large PR, splited to below (doing, part):

QEMU PIPE - Implement ADB function through qemu pipe #43
Hotplug detect - Support USB hotplug detecting for microADB #44
Bug fixes for ADB - Bug fixes for ADB #45
Fix environment error for ADB - Skip environment variables when exec logcat #46
TODO: The remaining 10 commits depend on Bug fixes for ADB #45, will update this PR after it merged.

Keep Draft status until all other related PRs were merged.

dengwenqi123 and others added 28 commits October 11, 2024 16:49

fix: git status/git status --ignore normalization check problem

f3b9be1

Signed-off-by: dengwenqi <dengwenqi@xiaomi.com>

microADB: using timer to check usb hotplug event

d6b1a7d

Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>

microADB: setup timer when adbd run with usb plugout

d096f23

Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>

microADB: using inotify to check usb hotplug event

e2fcaa6

Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>

microADB: call service_close to send CLSE frame when child task exit

d021ceb

Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>

Send response before adb_reboot_impl()

a34e817

Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>

hal_uv_client_usb.c: Call uv_write with one buffer instead two

1f0a87c

Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

hal/hal_uv.c: Remove g_adbd_context global variable

9771f27

Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

adb_client.c: Merge the handle of A_CNXN and A_AUTH to switch block

6f95403

Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

Add on_ prefix to close_cb like other callback in adb_tcp_socket_s

35c683a

and apply other minor style change Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

Remove ID_QUIT handler from state_wait_cmd_data

bf816e1

since ID_QUIT is always handled in state_wait_cmd Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

Remove SYNC_TEMP_BUFF_SIZE check from state_process_send_sym and stat…

c38ec55

…e_wait_cmd_data since the same check is already done in read_from_packet and state_wait_cmd Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

Always report ID_DONE and ID_FAIL through syncmsg::status

b23add5

instead of syncmsg::dent and syncmsg::data Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

Remove the assignment of client->socket.data

a998092

since it's never used at all Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>

Revert "hal_uv_client_usb.c: Call uv_write with one buffer instead two"

1d28038

This reverts commit 917f97deadd1d70e58347098086cf5ed8e8d0254. Reason for revert: pc need to recv header first

Check whether client has services before close

861e8cb

Signed-off-by: wangjianyu3 <wangjianyu3@xiaomi.com>

microadb: using dents message to responde request "LIST"

bc1fad4

Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>

adb:do not send sigkill when shell close

86ec8aa

Signed-off-by: guohao15 <guohao15@xiaomi.com>

microADB: Implement ADB function through qemu pipe (3/3)

7f51157

Signed-off-by: getian <getian@xiaomi.com>

Revert "adb: send close packet in on_child_exit method"

6a319a0

This reverts commit 037d0f1f04518837b26888cc2d3c451be08ea7fd. Reason for revert: can not reserve problem well Signed-off-by: yangsong8 <yangsong8@xiaomi.com>

adb: check svc exiting flag in shell_write function

bb979b0

Signed-off-by: yangsong8 <yangsong8@xiaomi.com>

JianyuWang0623 marked this pull request as draft October 11, 2024 10:19

spiriou reviewed Oct 11, 2024

View reviewed changes

xiaoxiang781216 reviewed Oct 13, 2024

View reviewed changes

JianyuWang0623 mentioned this pull request Oct 13, 2024

Implement ADB function through qemu pipe #43

Merged

This was referenced Oct 13, 2024

Support USB hotplug detecting for microADB #44

Open

Bug fixes for ADB #45

Open

Skip environment variables when exec logcat #46

Merged

Navigation Menu

New features and bugfix for microADB - #42

New features and bugfix for microADB#42
JianyuWang0623 wants to merge 28 commits into
spiriou:masterfrom
JianyuWang0623:br_wjy_microadb_1011

JianyuWang0623 commented Oct 11, 2024

Uh oh!

JianyuWang0623 commented Oct 11, 2024 •

edited

Loading

Uh oh!

xiaoxiang781216 commented Oct 11, 2024

Uh oh!

JianyuWang0623 commented Oct 11, 2024

Uh oh!

spiriou Oct 11, 2024

Uh oh!

xiaoxiang781216 Oct 13, 2024 •

edited

Loading

Uh oh!

spiriou Oct 11, 2024

Uh oh!

xiaoxiang781216 Oct 13, 2024

Uh oh!

xiaoxiang781216 Oct 13, 2024

Uh oh!

JianyuWang0623 Oct 13, 2024

Uh oh!

xiaoxiang781216 Oct 13, 2024

Uh oh!

JianyuWang0623 commented Oct 13, 2024 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

Conversation

JianyuWang0623 commented Oct 11, 2024

Summary

Details for 19345e8

Test

Uh oh!

JianyuWang0623 commented Oct 11, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

xiaoxiang781216 commented Oct 11, 2024

Uh oh!

JianyuWang0623 commented Oct 11, 2024

Uh oh!

spiriou Oct 11, 2024

Choose a reason for hiding this comment

Uh oh!

xiaoxiang781216 Oct 13, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

spiriou Oct 11, 2024

Choose a reason for hiding this comment

Uh oh!

xiaoxiang781216 Oct 13, 2024

Choose a reason for hiding this comment

Uh oh!

xiaoxiang781216 Oct 13, 2024

Choose a reason for hiding this comment

Uh oh!

JianyuWang0623 Oct 13, 2024

Choose a reason for hiding this comment

Uh oh!

xiaoxiang781216 Oct 13, 2024

Choose a reason for hiding this comment

Uh oh!

JianyuWang0623 commented Oct 13, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

Details for `19345e8`

JianyuWang0623 commented Oct 11, 2024 •

edited

Loading

xiaoxiang781216 Oct 13, 2024 •

edited

Loading

JianyuWang0623 commented Oct 13, 2024 •

edited

Loading