summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Add a example skeleton pluginKevin Rocard2014-02-277-0/+369
| | | | | Change-Id: I16c4e9c90395648dc3d583687c97f4b72401be8b Signed-off-by: Kevin Rocard <kevinx.rocard@intel.com>
* Merge pull request #1 from 01org/cmake_cleanupsGuillaume Denneulin2014-02-205-81/+34
|\ | | | | Cmake cleanups
| * CMake: detect the XInclude feature in libxml2David Wagner2014-02-201-1/+9
| | | | | | | | | | | | | | | | | | xmlserializer needs the XInclude feature (its usage in structure files is optional but xmlserializer needs the symbols for compiling). Verbosely fail if libxml2 has not been configured with XInclude support. Signed-off-by: David Wagner <david.wagner@intel.com>
| * CMake: Force POSIX threads to be availableDavid Wagner2014-02-201-2/+6
| | | | | | | | | | | | | | The FindThreads module in CMake can detect several thread libraries. We only want the phtread library, so enforce it. Signed-off-by: David Wagner <david.wagner@intel.com>
| * CMake: remove WIN32 commentsDavid Wagner2014-02-202-2/+2
| | | | | | | | | | | | | | | | | | At the time of writing the makefiles, it wasn't clear what this argument was for. According to the documentation, it is only useful for GUI applications on windows. test-platform and remote-process are command-line tools and this does not apply. Signed-off-by: David Wagner <david.wagner@intel.com>
| * CMake: export client and plugin headers in separate directoriesDavid Wagner2014-02-201-76/+9
| | | | | | | | | | | | | | | | | | client headers are exporter in parameter/client; plugin headers in parameter/plugin. Also, reduce the list of exported plugin headers to the bare minimum. Signed-off-by: David Wagner <david.wagner@intel.com>
| * xmlserializer/CMake: fail verbosely if libxml2 is not foundDavid Wagner2014-02-191-0/+8
|/ | | | | | Also, append the libxml2 cflags if any. Signed-off-by: David Wagner <david.wagner@intel.com>
* Add a READMEDavid Wagner2014-02-181-0/+15
| | | | Signed-off-by: David Wagner <david.wagner@intel.com>
* CMake: add rules for installing headersDavid Wagner2014-02-182-0/+106
| | | | | | | | | | | libparameter abd libxmlserializer headers. Concerning libparameter headers: only core headers (needed for the plugins) are currently exported. We'll need to export the client headers too, but we must make a distiction between the two classes of headers. Also, some currently exported core headers might not be needed (some cleanup is needed). Signed-off-by: David Wagner <david.wagner@intel.com>
* CMake: beautification, fix some commentsDavid Wagner2014-02-186-13/+16
| | | | Signed-off-by: David Wagner <david.wagner@intel.com>
* CMake: clarify why libparameter doesn't link with libremote-processorDavid Wagner2014-02-171-0/+1
| | | | | | It is accessed through dlopen(). Signed-off-by: David Wagner <david.wagner@intel.com>
* CMake: add installation instructionDavid Wagner2014-02-175-0/+10
| | | | | | | | | Run "cmake -DCMAKE_INSTALL_PREFIX=/path/to/install ." for choosing a custom prefix (optional: if it is not specified, a default directory will be chosen). Run "make install" for installing all generated files. Signed-off-by: David Wagner <david.wagner@intel.com>
* Add CMake MakefilesDavid Wagner2014-02-177-0/+252
| | | | | | | 1) Generate the Makefiles with "cmake"; 2) compile all targets with "make". Signed-off-by: David Wagner <david.wagner@intel.com>
* Import convert util templatesDavid Wagner2014-02-173-3/+179
| | | | | | Import an external dependency: a template-based conversion toolkit. Signed-off-by: David Wagner <david.wagner@intel.com>
* Android.mk: Remove the libparameter_includes hackDavid Wagner2014-02-141-33/+0
| | | | | | The plugins' Android.mk will directly access libparameter's include dir. Signed-off-by: David Wagner <david.wagner@intel.com>
* Rename libutility to libpfw_utilityDavid Wagner2014-02-133-6/+6
| | | | | | That name was too generic and may have conflicted in the future. Signed-off-by: David Wagner <david.wagner@intel.com>
* Remove some useless LOCAL_C_INCLUDES directivesDavid Wagner2014-02-132-2/+0
| | | | Signed-off-by: David Wagner <david.wagner@intel.com>
* Reintroduce hardcoded include pathsDavid Wagner2014-02-136-10/+10
| | | | | | | | | | Pathmaps for libxml2, stlport and icu4c are not defined in vanilla AOSP. As temporary fix for compilation on vanilla AOSP, we need to hardcode them again. Commit 55f41bcc3edf282f236539bb26bd6dc8638f235e introduced the use of include-path-for. Signed-off-by: David Wagner <david.wagner@intel.com>
* Import the 'naive tokenizer' library in libutilityDavid Wagner2014-02-134-3/+114
| | | | | | | remote-process uses the 'naive tokenizer'. Let's make it part of the utility and make remote-process link against libutility. Signed-off-by: David Wagner <david.wagner@intel.com>
* Change the license to 3-clause BSDDavid Wagner2014-02-13246-5100/+6626
| | | | | | | Add license header in all source files and Makefiles, Add a "COPYING" file containing the license text. Signed-off-by: David Wagner <david.wagner@intel.com>
* In structure XML files, implement component library files inclusionGuillaume Denneulin2014-02-1310-55/+173
| | | | | | | | | | | | | | BZ: 168727 In the PFW structure file, it is not possible to include a component library from another XML file. Implement the possibility to import component from another XML file that would be included in a structure XML file and that would describe a component library. Change-Id: Id6125140de1c8e9882375d01199f695b929f45e2 Signed-off-by: Guillaume Denneulin <guillaume.denneulin@intel.com>
* Correct signed hexadecimal integer ranges conversionSylvain Pichon2014-02-131-1/+4
| | | | | | | | | | | BZ: 165527 PFW does not convert hexadecimal signed range values correctly This patch correct signed range values interpretation Change-Id: I5c060c637781821b70b769cabc31347b0e5e5be1 Signed-off-by: Sylvain Pichon <sylvainx.pichon@intel.com>
* Add integer conversion that can take care of sign extensionGuillaume Denneulin2014-02-128-5/+73
| | | | | | | | | | | | | | BZ: 163707 When using signed integers of size 8 or 16, the bit of sign needs to be propagated if the API called by the plugin wait for "int" type Add the virtual toInteger function in CTypeElement for each parameter type to take care of conversion to "int". Specialize this function in the class CIntegerParameterType to take care of sign extension. Change-Id: I41183dccbcc21212299d1dde86b3ad4ba8432ce4 Signed-off-by: Guillaume Denneulin <guillaume.denneulin@intel.com>
* Make hostDomainGenerator.sh verbose by defaultFrédéric Boisnard2014-02-121-4/+4
| | | | | | | | | | | | | BZ: 151780 The option --verbose must be provided in order to make hostDomainGenerator.sh verbose. However, this option is very usefull when the generation fails during a build, and it would be nice to have it enabled by default. Make this option the default one, and add a nonverbose option instead. Change-Id: I59195b7a28e25a5f18871d6baf2ae4f1310f49d7 Signed-off-by: Frédéric Boisnard <fredericx.boisnard@intel.com>
* Fix hostDomainGenerator.sh concurrency issuesFrédéric Boisnard2014-02-123-60/+114
| | | | | | | | | | | | | | | | | | | | | | | | | | BZ: 151780 When multiple process run hostDomainGenerator.sh, some race conditions occur, despite current locking mechanism. In consequence, the script has been modified in order to make sure that available socket ports can be chosen for test-platform and PFW processes: - portAllocator.py: python script which creates a new socket using a random available port, and displays the port chosen. This port number is then pushed to the end of the available ports list by the system. - hostDomainGenerator.sh: uses portAllocator.py to get free ports for test-platform and the PFW. Fixed multiple issues as well: - When an error occur, a cleanup function was called and sometimes would kill concurrent processes. Now, only the test-platform and PFW instances started by the current hostDomainGenerator.sh script will be stopped, - Test-platform startup was not checked correctly, - The creation of the symlink to libremote-processor.so sometimes failed, leading to an early cleanup. Change-Id: I33b4b2ad5a6c2bbc6f8283222eb2f16f05286734 Signed-off-by: Frédéric Boisnard <fredericx.boisnard@intel.com>
* Allow starting test-platform as a DaemonFrédéric Boisnard2014-02-123-16/+163
| | | | | | | | | | | | | | | | | | BZ: 151780 Currently, test-platform does not return when it is launched. In particular, one has to try to connect to it in order to know if it has correctly been initialized. Added a new '-d' option to test-platform to start it as a daemon: test-platform [-d] <file path> [port number, default 5001] When test-platform is started as a daemon, a child process is created and the main process immediatly returns its status. Change-Id: I70a33691909c958904cf50d156a563b998f92657 Signed-off-by: Frédéric Boisnard <fredericx.boisnard@intel.com>
* Improve verbosity during PFW socket creationFrédéric Boisnard2014-02-123-4/+16
| | | | | | | | | | | | | BZ: 151780 Debugging XML generation from .pfw files is uneasy because the log messages are lacking ports & sockets information. Added the port number to the logs of the PFW when a new socket is created. Change-Id: I688f954068a6819ecc9e65f7664ac7a98ca0dabf Signed-off-by: Frédéric Boisnard <fredericx.boisnard@intel.com>
* Add Exit command to test-platformFrédéric Boisnard2014-02-123-9/+41
| | | | | | | | | | | | | BZ: 151780 The only way to exit test-platform is to kill it. However during the build process, the hostDomainGenerator.sh script needs to start and exit test-platform, without knowing its PID. Allow exiting the test-plaform though a new 'exit' command. Change-Id: Ifb94ea1c2017a0b23e25b42a06e2ceeae69ace89 Signed-off-by: Frédéric Boisnard <fredericx.boisnard@intel.com>
* Add export include directiveFrancois Gaffie2014-02-121-0/+25
| | | | | | | | | | BZ: 156679 Due to split of plugins and core, and to avoid hard coded path, needs to add an include directive in utility component makefile. Change-Id: Ifbe6bc2b9ea692cd482f1c6934f9eb54d16eae68 Signed-off-by: Francois Gaffie <francois.gaffie@intel.com>
* [core] Optimize log handling.Guillaume Denneulin2014-02-122-12/+21
| | | | | | | | | | | BZ: 154623 When logging, the pfw was limited to 512 bytes. Remove this limitation and optimize the buffer allocated. Change-Id: I91537f30a519d93a4b1e55a23fe02f42dec3af8a Signed-off-by: Guillaume Denneulin <guillaume.denneulin@intel.com>
* Close all dynamic libraries when PFW is destroyedRenaud de Chivre2014-02-124-3/+23
| | | | | | | | | | | | | BZ: 159294 The PFW opens many dynamic libs but does not keep any track of them. This prevents to close them cleanly and have border effect when trying to do coverage measurement. The patch ensures that the PFW closes its libraries when destroyed. Change-Id: Idd1f69bfbc719e8e000914bef990c809ec5ed206 Signed-off-by: Renaud de Chivre <renaud.de.chivre@intel.com>
* remote-processor: configure the socket with the TCP_NODELAY optionDavid Wagner2014-02-121-0/+8
| | | | | | | | | | | | | | | | | | | | | | BZ: 154316 The "man 7 tcp" page speaks for itself: TCP_NODELAY If set, disable the Nagle algorithm. This means that segments are always sent as soon as possible, even if there is only a small amount of data. When not set, data is buffered until there is a sufficient amount to send out, thereby avoiding the frequent sending of small packets, which results in poor utilization of the network. The behaviour described above if this option is not set introduces a lot of overhead when several commands are sent with remote-processor because they are not sent in burst mode: instead, remote-processor wait for the answer to each command. See "man 7 tcp" for all the details. Change-Id: I6c8ede4428629a16e9a2927d73a2897873a57d37 Signed-off-by: David Wagner <david.wagner@intel.com>
* remote-process: allow reading commands from stdinDavid Wagner2014-02-123-44/+146
| | | | | | | | | | | | | | | | BZ: 154316 The remote-process utility was only able to send one command at a time. This patch adds a new usage: if the command is omitted, remote-process reads stdin and sends each line as commands. It stops whenever it encounters an error of any kind (internal or returned by the server). Example of the new usage: remote-process localhost 5000 < commands.txt Change-Id: I2478668807b5541544c5752e68e112c23c85af15 Signed-off-by: David Wagner <david.wagner@intel.com>
* Make "Path not found" error more verboseKevin Rocard2014-02-123-9/+11
| | | | | | | | | | | | | BZ: 156207 When an error occurs while using a path one does not know which path it refers to. Append the unfound path to the error : "Path not found: /the/path/that/was/not/found" Change-Id: If53c1a5c301906b464d38f604d6e97029f83ae3f Signed-off-by: Kevin Rocard <kevinx.rocard@intel.com>
* Modify include handling to comply with new plugins repoGuillaume Denneulin2014-02-122-0/+59
| | | | | | | | | | | BZ: 156675 Plugins are moving to new repositories. Modify include handling to comply inter-repo include dependency. Change-Id: I4f65ab4a7f1f524b59dffc567a679154e3394147 Signed-off-by: Guillaume Denneulin <guillaume.denneulin@intel.com>
* Update xml generation scriptsFrancois Gaffie2014-02-122-4/+8
| | | | | | | | | | | | | | | | | BZ: 99956 XML scripts does not support empty tuning file. This patches updates XML generation scripts to support empty tuning file. Change-Id: I2d3e11a950524c468c78e3b99cdebf3538b9e053 Signed-off-by: Francois Gaffie <francois.gaffie@intel.com> Reviewed-on: http://android.intel.com:8080/138519 Reviewed-by: Quintero, Jorge <jorge.quintero@intel.com> Reviewed-by: Centelles, Sylvain <sylvain.centelles@intel.com> Reviewed-by: Rocard, KevinX <kevinx.rocard@intel.com> Tested-by: Barthes, FabienX <fabienx.barthes@intel.com>
* Adds default value pair for inclusive criterionFrancois Gaffie2014-02-123-2/+57
| | | | | | | | | | | | | | | | | | | | BZ: 99956 Inclusive criterion type does not allow to check if the bitfield is empty, i.e., if zero is not available. This patch adds the default value pair "none, 0" to allow check either if the criterion "Is" 0, and establishes consistency with inclusion rule. It also allows to convert a concatenated chain of string value, where "|" is considered as AND rule into the corresponding numerical value. Change-Id: I4ee7c0f69228a727964ab80050ac12c6babd4f58 Signed-off-by: Francois Gaffie <francois.gaffie@intel.com> Signed-off-by: Jorge Quintero <jorge.quintero@intel.com> Reviewed-on: http://android.intel.com:8080/138518 Reviewed-by: Centelles, Sylvain <sylvain.centelles@intel.com> Tested-by: Barthes, FabienX <fabienx.barthes@intel.com>
* Add getbitposition API for bitfieldFrancois Gaffie2014-02-121-0/+8
| | | | | | | | | | | | | | | | | | | BZ: 99956 Bitfield parameter type does not allow to retrieve the position of the bit. This patch introduces a new API to retrieve the bit position in the bitfield Change-Id: I54bf8a446e7bd53bed6acda274d32966a737c877 Signed-off-by: Francois Gaffie <francois.gaffie@intel.com> Reviewed-on: http://android.intel.com:8080/138517 Reviewed-by: Quintero, Jorge <jorge.quintero@intel.com> Reviewed-by: Centelles, Sylvain <sylvain.centelles@intel.com> Tested-by: Barthes, FabienX <fabienx.barthes@intel.com> Reviewed-by: Balestriere, VianneyX <vianneyx.balestriere@intel.com> Tested-by: Balestriere, VianneyX <vianneyx.balestriere@intel.com>
* Allow component in parameter block in xsd schemasKevin Rocard2014-02-124-9/+8
| | | | | | | | | | | | | | | | | | BZ: 154054 Components are now allowed in parameter block since the patch 140555. Reflect that in xsd schemas. Change-Id: I43c03d6fd7754096b4e19d7fa5d10dce6bb2b10e Signed-off-by: Kevin Rocard <kevinx.rocard@intel.com> Reviewed-on: http://android.intel.com:8080/146241 Reviewed-by: cactus <cactus@intel.com> Reviewed-by: Denneulin, Guillaume <guillaume.denneulin@intel.com> Reviewed-by: De Chivre, Renaud <renaud.de.chivre@intel.com> Tested-by: Barthes, FabienX <fabienx.barthes@intel.com> Reviewed-by: Balestriere, VianneyX <vianneyx.balestriere@intel.com> Tested-by: Balestriere, VianneyX <vianneyx.balestriere@intel.com>
* fix build errorJin Wei2014-02-122-2/+2
| | | | | | | host version of lib libicuuc has changed to libicuuc-host Change-Id: I936a8e196d01d8cacf4ce0d64370b9cddccd8b3a Signed-off-by: Jin Wei <wei.a.jin@intel.com>
* fix the build errorXiaokang Qin2014-02-121-0/+1
| | | | | Change-Id: Ide3521a161a3e0903531b2093c87679a80b5870c Signed-off-by: Xiaokang Qin <xiaokang.qin@intel.com>
* Fix typo in PFW when loading pluginsFrédéric Boisnard2014-02-122-2/+2
| | | | | | | | | | | | BZ: 151782 There is a typo in the error message when a plugin is not found (folowings instead of following). This patch updates the error message. Change-Id: Ifdb749fd90d646d0c04522874e879a329520de0f Signed-off-by: Frédéric Boisnard <fredericx.boisnard@intel.com>
* [xml generation] Ignore errors in clean upKevin Rocard2014-02-121-4/+7
| | | | | | | | | | | | BZ: 152121 As the hostDomainGenerator clean up function can be called after any command, some clean commands were sometime irrelevant. Clean commands should not fail. Change-Id: I0e472919cd77abe393bca090a95d1c72cf77d58c Signed-off-by: Kevin Rocard <kevinx.rocard@intel.com>
* Avoid integer overflow while calculating max valuesDmitry Shkurko2014-02-123-4/+12
| | | | | | | | | | | | BZ: 134249 The behavior is undefined in the case of signed integer overflow for enum and fixed point parameter types. Modify the behavior to handle correctly the signed integers. Change-Id: Idbd0798a39f826853ae1afcd05cebd897675b9a8 Signed-off-by: Dmitry Shkurko <Dmitry.V.Shkurko@intel.com>
* Export lightRoutingUpdate for external buildKevin Rocard2014-02-121-0/+8
| | | | | | | | | | | | | | BZ: 145379 As the lightRoutingUpdate.sh script is in private, it is not exported for external release. As a result external clients can not use lightRoutingUpdate to generate xml from .pfw. Export lightRoutingUpdate to make it available in $ANDROID_HOST_OUT/system/bin after compilation. Change-Id: Id174ddb4dd0e31a7c5ede66e91b0db341745ff21 Signed-off-by: Kevin Rocard <kevinx.rocard@intel.com>
* hostDomainGenerator.sh: fix lock permissionsFrédéric Boisnard2014-02-121-1/+10
| | | | | | | | | | | | | | | | BZ: 149851 The hostDomainGenerator.sh script uses a lockfile to protect some operations from concurrent execution. This mechanism uses a file that is created if it does not already exist. In this case, the file permissions are too stricts, and other users are unable to run the script anymore. This patch changes the permission on the lockfile, allowing other users to access it. Change-Id: Iec250a4022cb8f97973fa92020b27dc0e83890c6 Signed-off-by: Frédéric Boisnard <fredericx.boisnard@intel.com>
* Fix hostDomainGenerator socket-checking mechanism for Ubuntu 13.04Frédéric Boisnard2014-02-121-1/+1
| | | | | | | | | | | | | | | | | | BZ: 148729 The script hostDomainGenerator.sh doesn't work fine on Ubuntu 13.04. The processing is based on the external tool "ss" to monitor opened sockets, in order to check if existing instances of test-platform are currently started. It appears that the behavior of this tool is different in the latest Ubuntu release: requests of opened sockets always return multiple false results. In consequence, the script cannot find any available port to start a new test-platform instance. This patch updates the way opened sockets are checked, using "grep" instead of the "ss" filter mechanism. Change-Id: I2e1245c524d021d99389a0ec028105ed1f706666 Signed-off-by: Frédéric Boisnard <fredericx.boisnard@intel.com>
* [PFW] Add possibility to put mapping info in component nodesKevin Rocard2014-02-121-2/+0
| | | | | | | | | | | | | BZ: 147490 Components did not support instantiation mapping (mapping that instantiate subsystem objects) which leaded to artificial parameter bloc use, ie a useless level. Treat the components as parameter block as for instantiation mappings. Change-Id: I64834dbdcbda9ab32a29d6c70dc12da5b5fe6d35 Signed-off-by: Georges-Henri Baron <georges-henrix.baron@intel.com>
* Allow concurent generation of XML routing filesFrédéric Boisnard2014-02-121-0/+19
| | | | | | | | | | | | | | | | | | BZ: 145938 XML generation is impossible when multiple instances of the script hostDomainGeneration are run (this is typically the case with the current Android Make process). This is because the way the ports used to communicate with test-platform and parameter-framework are chosen is not protected agains concurrent executions. This patch protects the initialization of the test-platform and parameter-framework agains concurrent executions using a flock. In particular, it ensures that the ports chosen are reserved to a specific process. Change-Id: I3d8fee2b20313814c79f5390851ad57dc696cc28 Signed-off-by: Frédéric Boisnard <fredericx.boisnard@intel.com>
* HostDomainGenerator: Do not try to use port if already in useKevin Rocard2014-02-121-0/+13
| | | | | | | | | | | | | | | | | | BZ: 144262 Currently, the hostDomainGenerator script doesn't check if the port used by the testPlatform is already used. This patch adds this new check. Change-Id: I323575f97f7437c01ce4e2d0b127aa658639af06 Signed-off-by: Kevin Rocard <kevinx.rocard@intel.com> Reviewed-on: http://android.intel.com:8080/136085 Reviewed-by: Boisnard, FredericX <fredericx.boisnard@intel.com> Reviewed-by: De Chivre, Renaud <renaud.de.chivre@intel.com> Tested-by: Barthes, FabienX <fabienx.barthes@intel.com> Reviewed-by: cactus <cactus@intel.com> Tested-by: cactus <cactus@intel.com>
generated by cgit v1.1 at 2025-01-10 16:58:53 +0000