This problem is also reported as bug #715238.
The page above makes adb completely unusable. While the patch changes the
server side of adb, the client side still connects to inet socket (look at
_adb_connect() in adb_client.c) w/o luck, certainly.
So either drop the patch or change the client side too. In the latter case
an additional logic is
* Hilko Bengen:
> If I am not mistaken, the dangerous default behavior is encoded into the
> build_local_name() function and the attached patch should fix things.
I have just uploaded an NMU (android-tools_4.2.2+git20130529-2.1.dsc)
that addresses this issue to DELAYED/2.
Cheers,
-Hilko
--
To
Control: tags -1 +patch
If I am not mistaken, the dangerous default behavior is encoded into the
build_local_name() function and the attached patch should fix things.
Cheers,
-Hilko
diff --git a/core/adb/adb.c b/core/adb/adb.c
index 399d43a..d3e87c7 100644
--- a/core/adb/adb.c
+++ b/core/adb/adb
Package: android-tools-adb
Version: 4.1.1+git20120801-1
Severity: grave
Tags: security
adb starts a daemon on first call without asking the user. This daemon
listens on a TCP port on localhost:
| $ id -u
| 1000
| $ netstat -tlpen | …
| Proto Local Address Foreign Address State User PID/Program
5 matches
Mail list logo
