summaryrefslogtreecommitdiff
path: root/drivers/net
Commit message (Collapse)AuthorAgeLines
* drivers/fsl-mc: flib changes for mc 8.0.0Prabhakar Kushwaha2015-08-03-129/+261
| | | | | | | | | MC firware version 8.0.0 contains new command flags. This patch contains modifications in FLIB files to support the new command flags. Signed-off-by: Itai Katz <itai.katz@freescale.com> Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Reviewed-by: York Sun <yorksun@freescale.com>
* Merge branch 'master' of git://www.denx.de/git/u-boot-imxTom Rini2015-08-02-15/+5
|\
| * net: fec_mxc: remove useless struct nbufAlbert ARIBAUD \(3ADEV\)2015-07-10-15/+5
| | | | | | | | | | | | | | | | This locally defined struct is actually only used once and as an opaque type. Remove it for clarity. Signed-off-by: Albert ARIBAUD (3ADEV) <albert.aribaud@3adev.fr> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
* | Merge branch 'zynq' of git://www.denx.de/git/u-boot-microblazeTom Rini2015-07-28-7/+19
|\ \
| * | net: gem: Extend timeout valueMichal Simek2015-07-28-1/+1
| | | | | | | | | | | | | | | | | | Extend time for MDIO. (Because of zed board) Signed-off-by: Michal Simek <michal.simek@xilinx.com>
| * | zynq: gem: Setting up WRAP bit for one TX bdMichal Simek2015-07-28-1/+2
| | | | | | | | | | | | | | | | | | | | | Setting up WRAP bit to indicate that this is the last TX BD in the chain. Signed-off-by: Michal Simek <michal.simek@xilinx.com>
| * | zynq: gem: Increase the Rx buffer descriptors to 32Siva Durga Prasad Paladugu2015-07-28-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Increase the Rx Buffer descriptors to 32. This will avoid Rx buffer descriptors overflow if more packets were received at one shot before we process the received ones. This fixes the issue of intermittent timeouts during tftp on a 1Gb connection with tftp server running on windows. Signed-off-by: Siva Durga Prasad Paladugu <sivadur@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
| * | zynqmp: gem: Flush the rx buffers while transmittingSiva Durga Prasad Paladugu2015-07-28-3/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | Flush and invalidate the rx buffers while sending the tx packet it self as armv8 does flush also while doing invalidation. Signed-off-by: Siva Durga Prasad Paladugu <sivadur@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
| * | zynqmp: gem: Set data bus width to 64bit for arm64Siva Durga Prasad Paladugu2015-07-28-1/+8
| | | | | | | | | | | | | | | | | | | | | | | | Set the data bus width to 64-bit AMBA Databus width in config register. Signed-off-by: Siva Durga Prasad Paladugu <sivadur@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
* | | keystone2: net: add mcast function to keyston2 Ethernet driverVitaly Andrianov2015-07-27-0/+10
|/ / | | | | | | | | | | | | | | | | | | | | The MCAST_TFTP support requires that network drivers has mcast functon implemented. This commit adds dummy keystone2_eth_bcast_addr() to meet the requirement. As far as the driver doesn't use ALE and doesn't filter any incoming packets, the function is empty. Signed-off-by: Vitaly Andrianov <vitalya@ti.com> Reviewed-by: Tom Rini <trini@konsulko.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
* | dm: eth: Avoid blocking on packet receptionSimon Glass2015-07-21-4/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Some devices can take a long time to work out whether they have a new packet or now. For example the ASIX USB Ethernet dongle can take 5 seconds to do this, since it waits until it gets a new packet on the wire before allowing the USB bulk read packet to be submitted. At present with driver mode the Ethernet receive code reads 32 packets. This can take a very long time if we must wait for all 32 packets. The old code (before driver model) worked by reading a single set of packets from the USB device, then processing all the packets with in. It would be nice to use the same behaviour with driver model. Add a flag to the receive method which indicates that the driver should try to find a packet if available, by consulting the hardware. When the flag is not set, it should just return any packet data it has already received. If there is none, it should return -EAGAIN so that the loop will terminate. Signed-off-by: Simon Glass <sjg@chromium.org>
* | dm: eth: Add driver-model support to the rtl8169 driverSimon Glass2015-07-21-49/+187
| | | | | | | | | | | | | | This driver is used by the Intel Minnowmax board. Convert it to driver model so it can use the new Ethernet implementation. Signed-off-by: Simon Glass <sjg@chromium.org>
* | driver/ldpaa_eth:Avoid infinite loop in ldpaa_eth_rxPrabhakar Kushwaha2015-07-20-1/+8
| | | | | | | | | | | | | | | | Change infinite loop mechanism to timer based polling for QBMAN release in ldpaa_eth_rx. Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Reviewed-by: York Sun <yorksun@freescale.com>
* | driver/ldpaa_eth: Avoid TX conf framesPrabhakar Kushwaha2015-07-20-111/+4
| | | | | | | | | | | | | | | | | | | | | | Polling of TX conf frames is not a mandatory option. Packets can be transferred via WRIOP without TX conf frame. Configure ldpaa_eth driver to use TX path without confirmation frame Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: York Sun <yorksun@freescale.com>
* | driver/ldpaa_eth: Add timeout handling DQRR entry readPrabhakar Kushwaha2015-07-20-14/+22
| | | | | | | | | | | | | | | | | | | | | | Volatile command does not return frame immidiately, need to wait till a frame is available in DQRR. Ideally it should be a blocking call. Add timeout handling for DQRR frame instead of retry counter. Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: York Sun <yorksun@freescale.com>
* | driver/ldpaa_eth: Retry enqueue if portal was busyPrabhakar Kushwaha2015-07-20-2/+30
| | | | | | | | | | | | | | | | | | | | | | Do not immediately return if the enqueue function returns -EBUSY; re-try mulitple times. if timeout occures, release the buffer. Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: York Sun <yorksun@freescale.com>
* | drivers/fsl-mc: dynamically create ICID pool in DPCStuart Yoder2015-07-20-0/+36
| | | | | | | | | | | | | | | | | | | | delete any existing ICID pools in the DPC and create a new one based on the stream ID partitioning for the SoC Signed-off-by: Stuart Yoder <stuart.yoder@freescale.com> Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Reviewed-by: York Sun <yorksun@freescale.com>
* | armv8/fsl-lsch3: partition stream IDsStuart Yoder2015-07-20-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | Stream IDs on ls2085a devices are not hardwired and are programmed by sw. There are a limited number of stream IDs available, and the partitioning of them is scenario dependent. This header defines the partitioning between legacy, PCI, and DPAA2 devices. Signed-off-by: Stuart Yoder <stuart.yoder@freescale.com> Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Reviewed-by: York Sun <yorksun@freescale.com>
* | drivers: fsl-mc: Return error for major version mismatchPrabhakar Kushwaha2015-07-20-2/+7
| | | | | | | | | | | | | | | | | | | | Management complex major version should match to the firmware present in flash. Return error during mismatch of major version. Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: York Sun <yorksun@freescale.com>
* | drivers: fsl-mc: Update qbman driverPrabhakar Kushwaha2015-07-20-37/+53
| | | | | | | | | | | | | | | | | | | | | | | | Update qbman driver - As per latest available qbman driver - Use of atomic APIs Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> CC: Geoff Thorpe <Geoff.Thorpe@freescale.com> CC: Haiying Wang <Haiying.Wang@freescale.com> CC: Roy Pledge <Roy.Pledge@freescale.com> Reviewed-by: York Sun <yorksun@freescale.com>
* | drivers: fsl-mc: Update flibs to mc-0.6.0.1Prabhakar Kushwaha2015-07-20-6/+8
| | | | | | | | | | | | | | | | | | | | Update flibs changes to mc-0.6.0.1 for dpmang, dprc, dpni and dpio objects Also rename qbman_portal_ce/ci_paddr to qbman_portal_ce/ci_offset in dpio_attr. These are now offsets from the SoC QBMan portals base. Signed-off-by: J. German Rivera <German.Rivera@freescale.com> Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Reviewed-by: York Sun <yorksun@freescale.com>
* | drivers/fsl-mc: Autoload AOIP image from NOR flashJ. German Rivera2015-07-20-0/+23
| | | | | | | | | | | | | | | | | | Load AIOP image from NOR flash into DDR so that the MC firmware the MC fw can start it at boot time Signed-off-by: J. German Rivera <German.Rivera@freescale.com> Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Reviewed-by: York Sun <yorksun@freescale.com>
* | driver/ldpaa_eth:Flush buffer before seeding BMAN after TX_confPrabhakar Kushwaha2015-07-20-0/+1
| | | | | | | | | | | | | | | | Flush buffer before releasing to BMan after TX_conf to ensure, the core does not have any cachelines that the WRIOP will DMA to. Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Reviewed-by: York Sun <yorksun@freescale.com>
* | drivers/fsl-mc: Make MC boot error messages more readableJ. German Rivera2015-07-20-18/+13
| | | | | | | | | | | | | | | | | | Make it easier for the user to notice when the MC firmware had problems booting. Signed-off-by: J. German Rivera <German.Rivera@freescale.com> Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Reviewed-by: York Sun <yorksun@freescale.com>
* | driver/fsl_debug_server: Fix the DDR hide logic for LS2085aBhupesh Sharma2015-07-20-8/+0
|/ | | | | | | | | | This patch fixes the DDR hide logic for LS2085a, correcting the way the Debug Server FW and MC FW images are placed on the top of system DDR and how the rest of the system DDR space is made visibile to Linux. Signed-off-by: Bhupesh Sharma <bhupesh.sharma at freescale.com> Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com> Reviewed-by: York Sun <yorksun@freescale.com>
* net: designware: Program MAC address to hardware after soft resetBin Meng2015-07-08-0/+6
| | | | | | | | | | commit f566c99 "net: Update hardware MAC address if it changes in env" removes writing MAC address to designware controller after soft reset. This makes designware ethernet port fail to work. Actually the MAC address should always be programmed after soft reset. Signed-off-by: Bin Meng <bmeng.cn@gmail.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
* Move defaults from config_cmd_default.h to KconfigJoe Hershberger2015-06-25-0/+1
| | | | | | | | | | This sets the default commands Kconfig to match include/config_cmd_default.h commands in the common/Kconfig and removes them from include/configs. Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> [trini: rastaban, am43xx_evm_usbhost_boot, am43xx_evm_ethboot updates] Signed-off-by: Tom Rini <trini@konsulko.com>
* net: Remove all calls to net_random_ethaddr()Joe Hershberger2015-05-20-9/+0
| | | | | | | | | | | | | Remove the calls to net_random_ethaddr() that some boards and some drivers are calling. This is now implemented inside of net/eth.c Enable the feature for all boards that previously enabled it. Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Acked-By: Michael Walle <michael@walle.cc> (for the lsxl board part) Series-changes: 2 -Fixed bfin build errors
* net: Remove all references to CONFIG_ETHADDR and friendsJoe Hershberger2015-05-19-14/+3
| | | | | | | | | | We really don't want boards defining fixed MAC addresses in their config so we just remove the option to set it in a fixed way. If you must have a MAC address that was not provisioned, then use the random MAC address functionality. Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: Simon Glass <sjg@chromium.org>
* net: phy: Add support for all targets which requires MANUAL_RELOCMichal Simek2015-05-19-0/+16
| | | | | | | | | | | Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers passed to phy_register function. This patch was tested on Microblaze architecture. Signed-off-by: Michal Simek <michal.simek@xilinx.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
* net/phy: refactor RTL8211F initializationShengzhou Liu2015-05-19-8/+25
| | | | | | | | | | RTL8211F needs to enalbe TXDLY for RGMII during phy initialization, so move it to rtl8211f_config for early initialization. Signed-off-by: Shengzhou Liu <Shengzhou.Liu@freescale.com> cc: Joe Hershberger <joe.hershberger@gmail.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
* net: Update hardware MAC address if it changes in envJoe Hershberger2015-05-19-19/+0
| | | | | | | | | | | | When the ethaddr changes in the env, the hardware should also be updated so that MAC filtering will work properly without resetting U-Boot. Also remove the manual calls to set the hwaddr that was included in a few drivers as a result of the framework not doing it. Reported-by: Michal Simek <michal.simek@xilinx.com> Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Tested-by: Michal Simek <michal.simek@xilinx.com>
* net/phy: Add support for realtek RTL8211FShengzhou Liu2015-05-19-0/+85
| | | | | | | | | RTL8211F has different registers from RTL8211E. This patch adds support for RTL8211F PHY which can be found on Freescale's T1023 RDB board. Signed-off-by: Shengzhou Liu <Shengzhou.Liu@freescale.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
* sandbox: eth: Add a function to skip ping timeoutsJoe Hershberger2015-05-05-0/+17
| | | | | | | | | When called, the next call to receive will trigger a 10-second leap forward in time to avoid waiting for time to pass when tests are evaluating timeout behavior. Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Acked-by: Simon Glass <sjg@chromium.org>
* sunxi: emac: Remove non driver-model codeHans de Goede2015-05-04-69/+0
| | | | | | | | | All sunxi boards now use the driver-model, so remove the non driver-model code. Signed-off-by: Hans de Goede <hdegoede@redhat.com> Reviewed-by: Simon Glass <sjg@chromium.org> Acked-by: Ian Campbell <ijc@hellion.org.uk>
* sunxi: emac: Add driver model supportHans de Goede2015-05-04-0/+81
| | | | | | | | Modify the sunxi-emac eth driver to support driver model. Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Ian Campbell <ijc@hellion.org.uk> Reviewed-by: Stefan Roese <sr@denx.de>
* sunxi: emac: Rename DMA_CPU_TRRESHOLD to EMAC_RX_BUFSIZEHans de Goede2015-05-04-2/+2
| | | | | | | | | | | Besides being spelled wrong, the DMA_CPU_TRRESHOLD define actually has nothing to do with DMA as we only use mmio fifo access. Rename it to EMAC_RX_BUFSIZE to properly reflect what it does. Signed-off-by: Hans de Goede <hdegoede@redhat.com> Reviewed-by: Simon Glass <sjg@chromium.org> Acked-by: Ian Campbell <ijc@hellion.org.uk> Reviewed-by: Stefan Roese <sr@denx.de>
* sunxi: emac: Prepare for driver-model supportHans de Goede2015-05-04-44/+71
| | | | | | | | | | Split all the core functionality out into functions taking a struct emac_eth_dev *priv argument as preparation for adding driver-model support. Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Ian Campbell <ijc@hellion.org.uk> Reviewed-by: Stefan Roese <sr@denx.de>
* sunxi: emac: port to phylibHans de Goede2015-05-04-42/+69
| | | | | | | | | This is a preparation-patch for adding device-model support to the emac driver. Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Ian Campbell <ijc@hellion.org.uk> Reviewed-by: Stefan Roese <sr@denx.de>
* board/ls2085qds: Add support ethernetPrabhakar Kushwaha2015-04-23-0/+84
| | | | | | | | | Add support of ethernet: - eth.c: mapping lane to slot for (0x2A, 0x07) - ls2085a.c: To enable/disable dpmac and get link type Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Signed-off-by: York Sun <yorksun@freescale.com>
* driver/ldpaa: Add support of WRIOP static data structurePrabhakar Kushwaha2015-04-23-5/+147
| | | | | | | | | | | | Wire rate IO Processor (WRIOP) provide support of receive and transmit ethernet frames from the ethernet MAC. Here Each WRIOP block supports upto 64 DPMACs. Create a house keeping data structure to support upto 16 DPMACs and store external phy related information. Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Signed-off-by: York Sun <yorksun@freescale.com>
* net/memac_phy: reuse driver for little endian SoCsShaohui Xie2015-04-23-24/+45
| | | | | | | | | | | | | The memac for PHY management on little endian SoCs is similar on big endian SoCs, so we modify the driver by using I/O accessor function to handle the endianness, so the driver can be reused on little endian SoCs, we introduce CONFIG_SYS_MEMAC_LITTLE_ENDIAN for little endian SoCs, if the CONFIG_SYS_MEMAC_LITTLE_ENDIAN is defined, the I/O access is little endian, if not, the I/O access is big endian. Move fsl_memac.h out of powerpc include. Signed-off-by: Shaohui Xie <Shaohui.Xie@freescale.com> Signed-off-by: York Sun <yorksun@freescale.com>
* drivers/fsl-mc: Changed MC firmware loading for new boot architectureJ. German Rivera2015-04-23-124/+370
| | | | | | | | | | | Changed MC firmware loading to comply with the new MC boot architecture. Flush D-cache hierarchy after loading MC images. Add environment variables "mcboottimeout" for MC boot timeout in milliseconds, "mcmemsize" for MC DRAM block size. Check MC boot status before calling flib functions. Signed-off-by: J. German Rivera <German.Rivera@freescale.com> Signed-off-by: York Sun <yorksun@freescale.com>
* net/phy/cortina: Fix compilation warningpankaj chauhan2015-04-23-2/+2
| | | | | | | | Fix comilation warning which is emitted when firmware address is more than 32 bit. Signed-off-by: pankaj chauhan <pankaj.chauhan@freescale.com> Signed-off-by: York Sun <yorksun@freescale.com>
* driver/ldpaa_eth: Update ldpaa ethernet driverPrabhakar Kushwaha2015-04-23-24/+38
| | | | | | | | | | | | | | | | | | | | | | | | | | Fix flush_dcache_range() input parameter to use start and end addresses. Change ethernet interface name to DPNI. Update entry criteria for ldpaa_eth_stop. Ethernet stack first stop the device before performing next operation. At the time of Ethernet driver registration, net_dev->state is set as ETH_STATE_INIT So take care net_dev->state as ETH_STATE_INIT in ldpaa_eth_stop. Undef CONFIG_PHYLIB temorarily because ldpaa_eth driver currently does not support PHYLIB. Instead of clearing pull descriptor one time, clear it before issuing any volatile dequeue command. Volatile command does not return frame immidiately, wait till a frame is available in DQRR. This frame can be valid or expired. Flush buffer before releasing to BMan ensure the core does not have any cachelines that the WRIOP will DMA to. Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Signed-off-by: pankaj chauhan <pankaj.chauhan@freescale.com> Signed-off-by: Roy Pledge <Roy.Pledge@freescale.com> Signed-off-by: York Sun <yorksun@freescale.com>
* drivers/net/e1000.c: Cleanup whitespaceMinghuan Lian2015-04-23-4/+4
| | | | | | | | | The patch removes unnecessary whitespace to fix checkpatch's warning: unnecessary whitespace before a quoted newline Signed-off-by: Minghuan Lian <Minghuan.Lian@freescale.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: York Sun <yorksun@freescale.com>
* driver/ldpaa_eth: Add LDPAA Ethernet driverPrabhakar Kushwaha2015-04-23-3/+871
| | | | | | | | | | | | | | | LDPAA Ethernet driver is a freescale's new ethernet driver based on Layerscape architecture. Every ethernet driver controls on DPNI object. Where all DPNIs share one common DPBP and DPIO object to support Rx and Tx flows. Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> CC: Cristian Sovaiala <cristian.sovaiala@freescale.com> CC: Bogdan Hamciuc <bogdan.hamciuc@freescale.com> CC: J. German Rivera <German.Rivera@freescale.com> [York Sun: s/NetReceive/net_process_received_packet] Reviewed-by: York Sun <yorksun@freescale.com>
* driver/fsl-mc: Add support of MC FlibsPrabhakar Kushwaha2015-04-21-106/+2444
| | | | | | | | | | | | | | | | | | | | | | Freescale's Layerscape Management Complex (MC) provide support various objects like DPRC, DPNI, DPBP and DPIO. Where: DPRC: Place holdes for other MC objectes like DPNI, DPBP, DPIO DPBP: Management of buffer pool DPIO: Used for used to QBMan portal DPNI: Represents standard network interface These objects are used for DPAA ethernet drivers. Signed-off-by: J. German Rivera <German.Rivera@freescale.com> Signed-off-by: Lijun Pan <Lijun.Pan@freescale.com> Signed-off-by: Stuart Yoder <stuart.yoder@freescale.com> Signed-off-by: Geoff Thorpe <Geoff.Thorpe@freescale.com> Signed-off-by: Haiying Wang <Haiying.Wang@freescale.com> Signed-off-by: Cristian Sovaiala <cristian.sovaiala@freescale.com> Signed-off-by: pankaj chauhan <pankaj.chauhan@freescale.com> Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com> Reviewed-by: York Sun <yorksun@freescale.com>
* armv8/fsl-lsch3: Add Freescale Debug Server driverBhupesh Sharma2015-04-21-0/+4
| | | | | | | | | | | | | The Debug Server driver is responsible for loading the Debug server FW on the Service Processor (Cortex-A5 core) on LS2085A like SoCs and then polling for the successful initialization of the same. TOP MEM HIDE is adjusted to ensure the space required by Debug Server FW is accounted for. MC uses the DDR area which is calculated as: MC DDR region start = Top of DDR - area reserved by Debug Server FW Signed-off-by: Bhupesh Sharma <bhupesh.sharma@freescale.com> Reviewed-by: York Sun <yorksun@freescale.com>
* net: pch_gbe: Fix pch_gbe device nameBin Meng2015-04-20-1/+1
| | | | | | | The name "pch_gbe.%x" exceeds the limit of the name in the 'struct eth_device'. Rename it as just "pch_gbe". Signed-off-by: Bin Meng <bmeng.cn@gmail.com>