]> git.dujemihanovic.xyz Git - u-boot.git/log
u-boot.git
9 years agoarm/dts/ls2085a: Add dts files for LS2085AQDS and LS2085ARDB
Haikun Wang [Fri, 26 Jun 2015 11:48:58 +0000 (19:48 +0800)]
arm/dts/ls2085a: Add dts files for LS2085AQDS and LS2085ARDB

Add dts source files for LS2085AQDS and LS2085ARDB boards.

Signed-off-by: Haikun Wang <haikun.wang@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarm/dts/ls2085a: Add DSPI dts node
Haikun Wang [Fri, 26 Jun 2015 11:48:45 +0000 (19:48 +0800)]
arm/dts/ls2085a: Add DSPI dts node

Add DSPI controller dts node in fsl-ls2085a.dtsi

Signed-off-by: Haikun Wang <haikun.wang@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarm/dts/ls2085a: Bring in ls2085a dts files from linux kernel
Haikun Wang [Fri, 26 Jun 2015 11:48:36 +0000 (19:48 +0800)]
arm/dts/ls2085a: Bring in ls2085a dts files from linux kernel

Bring in required device tree files for ls2085a from Linux. These are
initially unchanged and have a number of pieces not needed by U-Boot.

Signed-off-by: Haikun Wang <Haikun.Wang@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarm/ls102xa: Add little-endian mode support for audio IPs
Alison Wang [Tue, 9 Jun 2015 08:07:49 +0000 (16:07 +0800)]
arm/ls102xa: Add little-endian mode support for audio IPs

As SCFG_ENDIANCR register is added to choose little-endian or big-endian
for audio IPs on Rev2.0 silion, little-endian mode is selected.

Signed-off-by: Alison Wang <alison.wang@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarm/ls102xa: Add PSCI support for ls102xa
Wang Dongsheng [Thu, 4 Jun 2015 04:01:09 +0000 (12:01 +0800)]
arm/ls102xa: Add PSCI support for ls102xa

Base on PSCI services, implement CPU_ON/CPU_OFF for ls102xa platform.

Tested on LS1021AQDS, LS1021ATWR.
Test CPU hotplug times: 60K
Test kernel boot times: 1.2K

Signed-off-by: Wang Dongsheng <dongsheng.wang@freescale.com>
Acked-by: Alison Wang <alison.wang@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoARMv7: Factor out reusable timer_wait from sunxi/psci_sun7i.S
Wang Dongsheng [Thu, 4 Jun 2015 04:01:08 +0000 (12:01 +0800)]
ARMv7: Factor out reusable timer_wait from sunxi/psci_sun7i.S

timer_wait is moved from sunxi/psci_sun7i.S, and it can be converted
completely into a reusable armv7 generic timer. LS1021A will use it
as well.

Signed-off-by: Wang Dongsheng <dongsheng.wang@freescale.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarm: ls1021a: Remove the inappropriate use of the function 'sprintf'
Alison Wang [Mon, 11 May 2015 07:39:47 +0000 (15:39 +0800)]
arm: ls1021a: Remove the inappropriate use of the function 'sprintf'

As the function 'sprintf' does not check buffer boundaries but outputs
to the buffer 'enet' of fixed size (16), this patch removes the function
'sprintf', and uses 'strcpy' instead. It will assign the character
arrays 'enet' and 'phy' the corresponding character strings.

Signed-off-by: Alison Wang <alison.wang@freescale.com>
Reviewed-by: Joe Hershberger <joe.hershberger@ni.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoboard/ls2085a: Increase kernel_size value in env variable
Prabhakar Kushwaha [Wed, 1 Jul 2015 10:58:22 +0000 (16:28 +0530)]
board/ls2085a: Increase kernel_size value in env variable

Linux itb image size has been increased from 30MB.

So updating kernel_size to 40MB in env variable.

Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoboard/ls2085rdb: Export functions for standalone AQ FW load apps
Prabhakar Kushwaha [Sun, 28 Jun 2015 05:33:59 +0000 (11:03 +0530)]
board/ls2085rdb: Export functions for standalone AQ FW load apps

Export functions required by Aquntia PHY firmware load application.
functions are memset, strcpy, mdelay, mdio_get_current_dev,
phy_find_by_mask, mdio_phydev_for_ethname and miiphy_set_current_dev

Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agodriver/ldpaa_eth:Avoid infinite loop in ldpaa_eth_rx
Prabhakar Kushwaha [Thu, 2 Jul 2015 05:59:08 +0000 (11:29 +0530)]
driver/ldpaa_eth:Avoid infinite loop in ldpaa_eth_rx

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>
9 years agodriver/ldpaa_eth: Avoid TX conf frames
Prabhakar Kushwaha [Thu, 2 Jul 2015 05:59:07 +0000 (11:29 +0530)]
driver/ldpaa_eth: Avoid TX conf frames

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>
9 years agodriver/ldpaa_eth: Add timeout handling DQRR entry read
Prabhakar Kushwaha [Thu, 2 Jul 2015 05:59:06 +0000 (11:29 +0530)]
driver/ldpaa_eth: Add timeout handling DQRR entry read

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>
9 years agodriver/ldpaa_eth: Retry enqueue if portal was busy
Prabhakar Kushwaha [Thu, 2 Jul 2015 05:59:05 +0000 (11:29 +0530)]
driver/ldpaa_eth: Retry enqueue if portal was busy

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>
9 years agoarmv8/fsl-lsch3: device tree fixups for PCI stream IDs
Stuart Yoder [Thu, 2 Jul 2015 05:59:04 +0000 (11:29 +0530)]
armv8/fsl-lsch3: device tree fixups for PCI stream IDs

This patch adds the infrastructure to update device
tree nodes to convey SMMU stream IDs in the device
tree.  Fixups are implemented for PCI controllers
initially.

Signed-off-by: Stuart Yoder <stuart.yoder@freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agodrivers/fsl-mc: dynamically create ICID pool in DPC
Stuart Yoder [Thu, 2 Jul 2015 05:59:03 +0000 (11:29 +0530)]
drivers/fsl-mc: dynamically create ICID pool in DPC

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>
9 years agoarmv8/fsl-lsch3: partition stream IDs
Stuart Yoder [Thu, 2 Jul 2015 05:59:02 +0000 (11:29 +0530)]
armv8/fsl-lsch3: partition stream IDs

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>
9 years agodrivers: fsl-mc: Return error for major version mismatch
Prabhakar Kushwaha [Thu, 2 Jul 2015 05:59:01 +0000 (11:29 +0530)]
drivers: fsl-mc: Return error for major version mismatch

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>
9 years agodrivers: fsl-mc: Update qbman driver
Prabhakar Kushwaha [Thu, 2 Jul 2015 05:59:00 +0000 (11:29 +0530)]
drivers: fsl-mc: Update qbman driver

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>
9 years agodrivers: fsl-mc: Update flibs to mc-0.6.0.1
Prabhakar Kushwaha [Thu, 2 Jul 2015 05:58:59 +0000 (11:28 +0530)]
drivers: fsl-mc: Update flibs to mc-0.6.0.1

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>
9 years agodrivers/fsl-mc: Autoload AOIP image from NOR flash
J. German Rivera [Thu, 2 Jul 2015 05:58:58 +0000 (11:28 +0530)]
drivers/fsl-mc: Autoload AOIP image from NOR flash

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>
9 years agodriver/ldpaa_eth:Flush buffer before seeding BMAN after TX_conf
Prabhakar Kushwaha [Thu, 2 Jul 2015 05:58:57 +0000 (11:28 +0530)]
driver/ldpaa_eth:Flush buffer before seeding BMAN after TX_conf

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>
9 years agodrivers/fsl-mc: Make MC boot error messages more readable
J. German Rivera [Thu, 2 Jul 2015 05:58:56 +0000 (11:28 +0530)]
drivers/fsl-mc: Make MC boot error messages more readable

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>
9 years agoarmv8/ls2085a: enable debug server
Stuart Yoder [Thu, 28 May 2015 09:24:15 +0000 (14:54 +0530)]
armv8/ls2085a: enable debug server

Signed-off-by: Stuart Yoder <stuart.yoder at freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/fsl-lsch3: Support 256M mem split for MC & dbg-srvr
Prabhakar Kushwaha [Tue, 2 Jun 2015 05:25:52 +0000 (10:55 +0530)]
armv8/fsl-lsch3: Support 256M mem split for MC & dbg-srvr

The agreed split of the top of memory is 256M for debug server and 256M
 for MC. This patch implements the split.

 In addition, the MC mem must be 512MB aligned, so the amount of memory
 to hide must be 512MB to achieve that alignment.

Signed-off-by: Stuart Yoder <stuart.yoder@freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarm/errata: Update required bits for A57 cores erratas
Bhupesh Sharma [Thu, 28 May 2015 09:24:13 +0000 (14:54 +0530)]
arm/errata: Update required bits for A57 cores erratas

This patch updates the setting of required bits for A57 cores erratas
- 828024 and 826974

Signed-off-by: Bhupesh Sharma <bhupesh.sharma at freescale.com>
Signed-off-by: Dai Haruki <dai.haruki at freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agodriver/fsl_debug_server: Fix the DDR hide logic for LS2085a
Bhupesh Sharma [Thu, 28 May 2015 09:24:12 +0000 (14:54 +0530)]
driver/fsl_debug_server: Fix the DDR hide logic for LS2085a

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>
9 years agoarmv8/ls2085a: Update SoC README for DDR layout
Prabhakar Kushwaha [Thu, 28 May 2015 09:24:11 +0000 (14:54 +0530)]
armv8/ls2085a: Update SoC README for DDR layout

Update SoC README to provide details of
 - Memory regions
 - Memory used by MC and Debug server

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>
9 years agoarmv8/ls2085a: Expose all DDR region(s) to Linux
Bhupesh Sharma [Thu, 28 May 2015 09:24:10 +0000 (14:54 +0530)]
armv8/ls2085a: Expose all DDR region(s) to Linux

This patch allows u-boot to expose the complete DDR region(s) to Linux
(after subtracting the memory hidden via MEM_TOP_HIDE mechanism).

This allows the u-boot to support the 48-bit VA support provided by
ARM64 Linux in flavors 3.18 and above, by passing the appropriate
'memory' DTS nodes.

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>
9 years agoarmv8/ls2085ardb: Fix SPD address error on early boards
York Sun [Thu, 28 May 2015 09:24:09 +0000 (14:54 +0530)]
armv8/ls2085ardb: Fix SPD address error on early boards

Board rev C and earlier has duplicated SPD address on 2nd DDR
controller slots. It is fixed on rev D and later. SPD addresses
need to be updated accordingly.

Signed-off-by: York Sun <yorksun at freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
9 years agodriver/ddr/fsl: Add a hook to update SPD address
York Sun [Thu, 28 May 2015 09:24:08 +0000 (14:54 +0530)]
driver/ddr/fsl: Add a hook to update SPD address

In case SPD address changes between board revisions, updating SPD
address can be called from board file.

Signed-off-by: York Sun <yorksun at freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
9 years agoarmv8/ls2085a: Avoid hard-coding for board name print
Prabhakar Kushwaha [Thu, 28 May 2015 09:24:07 +0000 (14:54 +0530)]
armv8/ls2085a: Avoid hard-coding for board name print

LS2085A supports 6 personalities i.e. LS2045AE, LS2045A, LS2080AE,
LS2080A, LS2085AE and LS2085A personlities.

Instead of hard-coding, board name should change as per selected
personality.

Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/fsl-ch3: Add support to print SoC personality
Prabhakar Kushwaha [Thu, 28 May 2015 09:24:06 +0000 (14:54 +0530)]
armv8/fsl-ch3: Add support to print SoC personality

This patch adds support to print out the SoC personality.
Freescale LS20xx SoCs (compliant to Chassis-3 specifications) can
have 6 personalities: LS2045AE, LS2045A, LS2080AE, LS2080A,
LS2085AE and LS2085A

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>
9 years agoarmv8/fsl-lsch3: Fix DDR speed message
York Sun [Thu, 28 May 2015 09:24:05 +0000 (14:54 +0530)]
armv8/fsl-lsch3: Fix DDR speed message

DDR speed should be in MT/s, not MHz.

Signed-off-by: York Sun <yorksun at freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
9 years agoarmv8/ls2085RDB: Update board version print logic
Prabhakar Kushwaha [Thu, 28 May 2015 09:24:04 +0000 (14:54 +0530)]
armv8/ls2085RDB: Update board version print logic

As per updated board document, no need to substract 1 from arch[BRD]
bit field. Default value + 'A' represents the board revision.

So update board version print logic to reflect the same.

Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoboard/fsl/common: Fix eeprom system version endianness
Jaiprakash Singh [Thu, 28 May 2015 09:24:03 +0000 (14:54 +0530)]
board/fsl/common: Fix eeprom system version endianness

SYSTEM ID EPPROM always store SYSTEM version info in big endian format.
SoC with ARM or PowerPC core should read/write version info from eeprom
in BIG endian format.

So use cpu-specific APIs to read SYSTEM version.

Signed-off-by: Jaiprakash Singh <b44839 at freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/ls2085a: Increase the supported kernel size
Bhupesh Sharma [Thu, 28 May 2015 09:24:02 +0000 (14:54 +0530)]
armv8/ls2085a: Increase the supported kernel size

Increases the kernel size supported for LS2085A platforms:-
 - Update environment variables
 - Add ramdisk_size in bootargs env variable
 - Define  CONFIG_SYS_BOOTM_LEN to 64MB

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>
9 years agoarmv8/ls2085rdb: Update PCA9547PW slave address
Prabhakar Kushwaha [Thu, 28 May 2015 09:24:01 +0000 (14:54 +0530)]
armv8/ls2085rdb: Update PCA9547PW slave address

Primary Mux on I2C1 controller has slave address as 0x75.
So update its address.

Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/ls2085qds: Update SFP TX bit as "0" to enable XFI
Prabhakar Kushwaha [Thu, 28 May 2015 09:24:00 +0000 (14:54 +0530)]
armv8/ls2085qds: Update SFP TX bit as "0" to enable XFI

FPGA BRDCFG9[SFP_TX] should be clear in order to enable XFI ports.

Signed-off-by: Dai Haruki <Dai.Haruki at freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/ls2085a: call ft_pcie_setup() to change dts status
Prabhakar Kushwaha [Thu, 28 May 2015 09:23:59 +0000 (14:53 +0530)]
armv8/ls2085a: call ft_pcie_setup() to change dts status

call ft_pci_setup() to disable PCIe dts node if corresponding
PCIe controller is disabled according to RCW

Signed-off-by: Minghuan Lian <Minghuan.Lian at freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/ls2085a: Update LS2085a PCIe compatible
Prabhakar Kushwaha [Thu, 28 May 2015 09:23:58 +0000 (14:53 +0530)]
armv8/ls2085a: Update LS2085a PCIe compatible

Compatible field "fsl,20851a-pcie" is not correct.
So update it to "fsl,ls2085a-pcie"

Signed-off-by: Minghuan Lian <Minghuan.Lian at freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/ls2085aqds: Add support of SerDes protocol 0x49
Prabhakar Kushwaha [Thu, 28 May 2015 09:23:57 +0000 (14:53 +0530)]
armv8/ls2085aqds: Add support of SerDes protocol 0x49

SerDes Protocol 0x49 enables 4 SGMII, PEX4, SATA1 and SATA2.

Add support of 0x49 SerDes protocol to enable 4SGMII on slot4 of
ls2085aqds platform.

Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/ls2085a: Enable "date" command for QDS and RDB
Priyanka Jain [Thu, 28 May 2015 09:23:56 +0000 (14:53 +0530)]
armv8/ls2085a: Enable "date" command for QDS and RDB

Enable "date" command for QDS and RDB boards

Signed-off-by: Priyanka Jain <Priyanka.Jain at freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/ls2085ardb: add hwconfig setting for eSDHC
Yangbo Lu [Thu, 28 May 2015 09:23:55 +0000 (14:53 +0530)]
armv8/ls2085ardb: add hwconfig setting for eSDHC

Add hwconfig setting for eSDHC since it shares some pins with other
IP block.

Signed-off-by: Yangbo Lu <yangbo.lu at freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/ls2085ardb: Add eth & phy firmware loading support
Prabhakar Kushwaha [Thu, 28 May 2015 09:23:54 +0000 (14:53 +0530)]
armv8/ls2085ardb: Add eth & phy firmware loading support

Add support for board eth initialization and support for loading phy
firmware. PHY firmware needs to be loaded from board_eth_init() because
all the MACs are not initialized by ldpaa_eth driver.

Signed-off-by: pankaj chauhan <pankaj.chauhan at freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar at freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoMerge git://git.denx.de/u-boot-x86
Tom Rini [Wed, 15 Jul 2015 14:41:20 +0000 (10:41 -0400)]
Merge git://git.denx.de/u-boot-x86

9 years agopci: Disable expansion ROM address decoding when signature check fails
Bin Meng [Wed, 8 Jul 2015 05:06:41 +0000 (13:06 +0800)]
pci: Disable expansion ROM address decoding when signature check fails

We should not leave the expansion ROM address window open when there
is not a valid ROM.

Suggested-by: Matt Porter <mporter@konsulko.com>
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Simon Glass <sjg@chromium.org>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agopci: Configure expansion ROM during auto config process
Bin Meng [Wed, 8 Jul 2015 05:06:40 +0000 (13:06 +0800)]
pci: Configure expansion ROM during auto config process

Currently PCI expansion ROM address is assigned by a call to
pciauto_setup_rom() outside of the pci auto config process.
This does not work when expansion ROM is on a device behind
PCI bridge where bridge's memory limit register was already
programmed to a value that does not cover the newly assigned
expansion ROM address. To fix this, we should configure the
ROM address during the auto config process.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agodrivers: block: Remove the ata_piix driver
Bin Meng [Sat, 16 May 2015 01:33:16 +0000 (09:33 +0800)]
drivers: block: Remove the ata_piix driver

This driver was originally added to support the native IDE mode for
Intel chipset, however it has some bugs like not supporting ATAPI
devices, endianness issue, or even broken build when CONFIG_LAB48.
Given no board is using this driver as of today, rather than fixing
all these issues we just remove it from the source tree.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Adjust config option order in defconfig for Crown Bay and Minnowmax
Bin Meng [Thu, 9 Jul 2015 10:37:40 +0000 (18:37 +0800)]
x86: Adjust config option order in defconfig for Crown Bay and Minnowmax

Update crownbay_defconfig and minnowmax_defconfig with 'savedefconfig'
result so that the config option order matches Kconfig.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agotools: ifdtool: Write correct offset on 32-bit machine
Bin Meng [Mon, 6 Jul 2015 07:57:06 +0000 (15:57 +0800)]
tools: ifdtool: Write correct offset on 32-bit machine

On 32-bit machine strtol() returns LONG_MAX which is 0x7fffffff,
which is wrong for u-boot.rom components like u-boot-x86-16bit.bin.
Change to use strtoll() so that it works on both 32-bit and 64-bit
machines.

Reported-by: Fei Wang <wangfei.jimei@gmail.com>
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Add binary blob checksums for Minnowboard MAX
Simon Glass [Sat, 4 Jul 2015 00:28:28 +0000 (18:28 -0600)]
x86: Add binary blob checksums for Minnowboard MAX

To try to reduce the pain of confusion of binary blobs, add MD5 checksums
for the current versions. This may worsen the situation as new versions
appear, but it should still be possible to obtain these versions, and thus
get a working setup.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
9 years agodm: x86: baytrail: Correct PCI region 3 when driver model is used
Simon Glass [Sat, 4 Jul 2015 00:28:27 +0000 (18:28 -0600)]
dm: x86: baytrail: Correct PCI region 3 when driver model is used

Commit afbbd413a fixed this for non-driver-model. Make sure that the driver
model code handles this also.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
9 years agodm: x86: minnowmax: Move PCI to use driver model
Simon Glass [Sat, 4 Jul 2015 00:28:26 +0000 (18:28 -0600)]
dm: x86: minnowmax: Move PCI to use driver model

Adjust minnowmax to use driver model for PCI. This requires adding a device
tree node to specify the ranges, removing the board-specific PCI code and
ensuring that the host bridge is configured.

Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
9 years agox86: pci: Tidy up the generic x86 PCI driver
Simon Glass [Sat, 4 Jul 2015 00:28:25 +0000 (18:28 -0600)]
x86: pci: Tidy up the generic x86 PCI driver

This driver should use the x86 PCI configuration functions. Also adjust its
compatible string to something generic (i.e. without a vendor name).

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
9 years agox86: Add ROM image description for minnowmax
Simon Glass [Sat, 4 Jul 2015 00:28:24 +0000 (18:28 -0600)]
x86: Add ROM image description for minnowmax

The layout of the ROM is a bit hard to discover by reading the code. Add
a table to make it easier.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
9 years agodm: spi: Enable environment for minnowmax
Simon Glass [Sat, 4 Jul 2015 00:28:23 +0000 (18:28 -0600)]
dm: spi: Enable environment for minnowmax

Enable a SPI environment and store it in a suitable place.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Jagan Teki <jteki@openedev.com>
9 years agodm: spi: Correct BIOS protection logic for ICH9
Simon Glass [Sat, 4 Jul 2015 00:28:22 +0000 (18:28 -0600)]
dm: spi: Correct BIOS protection logic for ICH9

The logic is incorrect and currently has no effect. Fix it so that we can
write to SPI flash, since by default it is write-protected.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Andrew Bradford <andrew.bradford@kodakalaris.com>
9 years agodm: spi: Correct status register access width
Simon Glass [Sat, 4 Jul 2015 00:28:21 +0000 (18:28 -0600)]
dm: spi: Correct status register access width

The status register on ICH9 is a single byte, so use byte access when
writing to it, to avoid updating the control register also.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Jagan Teki <jteki@openedev.com>
9 years agox86: Configure VESA parameters before loading Linux kernel
Bin Meng [Mon, 6 Jul 2015 08:31:36 +0000 (16:31 +0800)]
x86: Configure VESA parameters before loading Linux kernel

Store VESA parameters to Linux setup header so that vesafb driver
in the kernel could work.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
Tested-by: Jian Luo <jian.luo4@boschrexroth.de>
9 years agox86: crownbay: Enable graphics support
Bin Meng [Mon, 6 Jul 2015 08:31:35 +0000 (16:31 +0800)]
x86: crownbay: Enable graphics support

Enable graphics support on Intel Crown Bay board With the help of
vgabios for Intel TunnelCreek IGD. Tested with an external LVDS
panel connected to X4 connector and SDVO adapter connected to X9
connector on the board.

Signed-off-by: Jian Luo <jian.luo4@boschrexroth.de>
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Remove MARK_GRAPHICS_MEM_WRCOMB
Bin Meng [Mon, 6 Jul 2015 08:31:34 +0000 (16:31 +0800)]
x86: Remove MARK_GRAPHICS_MEM_WRCOMB

MARK_GRAPHICS_MEM_WRCOMB is not referenced anywhere in the code,
hence remove it.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Move VGA option rom macros to Kconfig
Bin Meng [Mon, 6 Jul 2015 08:31:33 +0000 (16:31 +0800)]
x86: Move VGA option rom macros to Kconfig

Move X86_OPTION_ROM_FILE & X86_OPTION_ROM_ADDR to arch/x86/Kconfig
and rename them to VGA_BIOS_FILE & VGA_BIOS_ADDR which depend on
HAVE_VGA_BIOS. The new names are consistent with other x86 binary
blob options like HAVE_FSP/FSP_FILE/FSP_ADDR.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: cmd_mtrr: Improve MTRR list information
Bin Meng [Mon, 6 Jul 2015 08:31:32 +0000 (16:31 +0800)]
x86: cmd_mtrr: Improve MTRR list information

Print the meaningful base address and mask of an MTRR range without showing
the memory type encoding or valid bit.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: queensbay: Change CPU_ADDR_BITS to 32
Bin Meng [Mon, 6 Jul 2015 08:31:31 +0000 (16:31 +0800)]
x86: queensbay: Change CPU_ADDR_BITS to 32

Per CPUID:80000008h result, the maximum physical address bits of
TunnelCreek processor is 32 instead of default 36. This will fix
the incorrect decoding of MTRR range mask.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Setup fixed range MTRRs for legacy regions
Bin Meng [Mon, 6 Jul 2015 08:31:30 +0000 (16:31 +0800)]
x86: Setup fixed range MTRRs for legacy regions

We should setup fixed range MTRRs for some legacy regions like VGA
RAM and PCI ROM areas as uncacheable. Note FSP may setup these to
other cache settings, but we can override this in x86_cpu_init_f().

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agovideo: Add 32-bit color depth support for VBE
Jian Luo [Mon, 6 Jul 2015 08:31:29 +0000 (16:31 +0800)]
video: Add 32-bit color depth support for VBE

The TunnelCreek IGD VBE reports 32-bit color depth regardless 24-bit
color depth is configured. Since 24-bit mode already uses 4 bytes
internally, it should be OK to just add this option in switch case.

Signed-off-by: Jian Luo <jian.luo4@boschrexroth.de>
Acked-by: Simon Glass <sjg@chromium.org>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
9 years agox86: bios: Allow pci config read/write to host bridge in int1a_handler
Jian Luo [Mon, 6 Jul 2015 08:31:28 +0000 (16:31 +0800)]
x86: bios: Allow pci config read/write to host bridge in int1a_handler

We should allow pci config read/write to host bridge (b.d.f = 0.0.0)
in the int1a_handler() which is a valid pci device.

Signed-off-by: Jian Luo <jian.luo4@boschrexroth.de>
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: bios: Synchronize stack between real and protected mode
Jian Luo [Mon, 6 Jul 2015 08:42:06 +0000 (16:42 +0800)]
x86: bios: Synchronize stack between real and protected mode

PCI option rom may use different SS during its execution, so it is not
safe to assume esp pointed to the same location in the protected mode.

Signed-off-by: Jian Luo <jian.luo4@boschrexroth.de>
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agovideo: vesa_fb: Look up VGA device by class instead of id
Bin Meng [Mon, 6 Jul 2015 08:31:26 +0000 (16:31 +0800)]
video: vesa_fb: Look up VGA device by class instead of id

Per PCI spec, VGA device reports its class as standard 030000h in
its configuration space, so we can use it to determine if we need
run option rom instead of testing the supported vendor/device ids.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agodm: pci: Correct bus number when scanning sub-buses
Simon Glass [Sun, 7 Jun 2015 14:50:41 +0000 (08:50 -0600)]
dm: pci: Correct bus number when scanning sub-buses

The sub-bus passed to pciauto_prescan_setup_bridge() is incorrect. Fix it
so that sub-buses are numbered correctly.

Signed-off-by: Simon Glass <sjg@chromium.org>
9 years agodm: pci: Use the correct hose when configuring devices
Simon Glass [Sun, 7 Jun 2015 14:50:40 +0000 (08:50 -0600)]
dm: pci: Use the correct hose when configuring devices

Only the PCI controller has access to the PCI region information. Make sure
to use the controller (rather than any attached bridges) when configuring
devices.

This corrects a failure to scan and configure devices when driver model is
enabled for PCI.

Also add a comment to explain the problem.

Signed-off-by: Simon Glass <sjg@chromium.org>
9 years agox86: queensbay: Change PCIe root ports' interrupt routing
Bin Meng [Tue, 23 Jun 2015 04:18:55 +0000 (12:18 +0800)]
x86: queensbay: Change PCIe root ports' interrupt routing

So far interrupt routing works pretty well for any on-chip devices
on Intel Crown Bay. When inserting any PCIe card to any PCIe slot,
Linux kernel is smart enough to do interrupt swizzling and figure
out device's irq using its parent bridge's interrupt routing info
all the way up to its root port. In U-Boot all PCIe root ports'
interrupts were routed to PIRQ E/F/G/H before, while actually all
PCIe downstream ports received INTx are routed to PIRQ A/B/C/D
directly and not configurable. Now we change this mapping so that
any external PCIe device can work correctly.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: crownbay: Enable writing MP table
Bin Meng [Tue, 23 Jun 2015 04:18:54 +0000 (12:18 +0800)]
x86: crownbay: Enable writing MP table

Enable writing MP table for Intel Crown Bay board.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Update README.x86 for SMP support
Bin Meng [Tue, 23 Jun 2015 04:18:53 +0000 (12:18 +0800)]
x86: Update README.x86 for SMP support

Document U-Boot multi-processor support as well as configuration
tables like SFI and MP tables for SMP OS kernel.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Generate a valid MultiProcessor (MP) table
Bin Meng [Tue, 23 Jun 2015 04:18:52 +0000 (12:18 +0800)]
x86: Generate a valid MultiProcessor (MP) table

Implement write_mp_table() to create a minimal working MP table.
This includes an MP floating table, a configuration table header
and all of the 5 base configuration table entries. The I/O interrupt
assignment table entry is created based on the same information used
in the creation of PIRQ routing table from device tree. A check
duplicated entry logic is applied to prevent writing multiple I/O
interrupt entries with the same information.

Use a Kconfig option GENERATE_MP_TABLE to tell U-Boot whether we
need actually write the MP table at the F seg, just like we did for
PIRQ routing and SFI tables. With MP table existence, linux kernel
will switch to I/O APIC and local APIC to process all the peripheral
interrupts instead of 8259 PICs. This takes full advantage of the
multicore hardware and the SMP kernel.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Add MultiProcessor (MP) table APIs
Bin Meng [Tue, 23 Jun 2015 04:18:51 +0000 (12:18 +0800)]
x86: Add MultiProcessor (MP) table APIs

The MP table provides a way for the operating system to support
for symmetric multiprocessing as well as symmetric I/O interrupt
handling with the local APIC and I/O APIC. We provide a bunch of
APIs for U-Boot to write the floating table, configuration table
header as well as base and extended table entries.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Remove inline for lapic access routines
Bin Meng [Tue, 23 Jun 2015 04:18:50 +0000 (12:18 +0800)]
x86: Remove inline for lapic access routines

Remove inline for lapic access routines and expose lapic_read()
& lapic_write() as APIs to read/write lapic registers. Also move
stop_this_cpu() to mp_init.c as it has nothing to do with lapic.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Add I/O APIC register access routines
Bin Meng [Tue, 23 Jun 2015 04:18:49 +0000 (12:18 +0800)]
x86: Add I/O APIC register access routines

I/O APIC registers are addressed indirectly. Add io_apic_read() and
io_apic_write() routines to help register access. Two macros for I/O
APIC ID and version register offset are also added.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Clean up ioapic header file
Bin Meng [Tue, 23 Jun 2015 04:18:48 +0000 (12:18 +0800)]
x86: Clean up ioapic header file

Remove all the dead/unused macros from asm/ioapic.h.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Reduce PIRQ routing table size
Bin Meng [Tue, 23 Jun 2015 04:18:47 +0000 (12:18 +0800)]
x86: Reduce PIRQ routing table size

There is no need to populate multiple irq info entries with the same
bus number and device number, but with different interrupt pin. We
can use the same entry to store all the 4 interrupt pin (INT A/B/C/D)
routing information to reduce the whole PIRQ routing table size.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Ignore function number when writing PIRQ routing table
Bin Meng [Tue, 23 Jun 2015 04:18:46 +0000 (12:18 +0800)]
x86: Ignore function number when writing PIRQ routing table

In fill_irq_info() pci device's function number is written into
the table, however this is not really necessary. The function
number can be anything as OS doesn't care about this field,
neither does the PIRQ routing specification. Change to always
writing 0 as the function number.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Write correct bus number for the irq router
Bin Meng [Tue, 23 Jun 2015 04:18:45 +0000 (12:18 +0800)]
x86: Write correct bus number for the irq router

We should write correct bus number to the PIRQ routing table for the
irq router from device tree, instead of hard-coded zero.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: queensbay: Correct Topcliff device irqs
Bin Meng [Tue, 23 Jun 2015 04:18:44 +0000 (12:18 +0800)]
x86: queensbay: Correct Topcliff device irqs

There are 4 usb ports on the Intel Crown Bay board, 2 of which are
connected to Topcliff usb host 0 and the other 2 connected to usb
host 1. USB devices inserted in the ports connected to usb host 1
cannot get detected due to wrong IRQ assigned to the controller.
Actually we need apply the PCI interrupt pin swizzling logic to all
devices on the Topcliff chipset when configuring the PIRQ routing.

This was observed on usb ports, but device 6 and 10 irqs are also
wrong. Correct them all together.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: crownbay: Enable DM RTC support
Bin Meng [Tue, 23 Jun 2015 04:18:43 +0000 (12:18 +0800)]
x86: crownbay: Enable DM RTC support

Add a RTC node in the device tree to enable DM RTC support.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agocmd: date: Change to use CONFIG_DM_RTC instead of CONFIG_DM_I2C
Bin Meng [Tue, 23 Jun 2015 04:18:42 +0000 (12:18 +0800)]
cmd: date: Change to use CONFIG_DM_RTC instead of CONFIG_DM_I2C

Currently CONFIG_DM_I2C is used in cmd_date.c for driver model,
but it should be actually CONFIG_DM_RTC.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agodm: rtc: Support mc146818 driver in driver model
Bin Meng [Tue, 23 Jun 2015 04:18:41 +0000 (12:18 +0800)]
dm: rtc: Support mc146818 driver in driver model

Add driver model support to the mc146818 rtc driver. Also clean up
the driver a little bit for coding convention issues.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: crownbay: Add MP initialization
Bin Meng [Wed, 17 Jun 2015 03:15:39 +0000 (11:15 +0800)]
x86: crownbay: Add MP initialization

Intel Crown Bay board has a TunnelCreek processor which supports
hyper-threading. Add /cpus node in the crownbay.dts and enable
the MP initialization.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Simon Glass <sjg@chromium.org>
(modified to remove error:
   overriding the value of OF_CONTROL. Old value: "y", new value: "y")

9 years agox86: Clean up lapic codes
Bin Meng [Wed, 17 Jun 2015 03:15:38 +0000 (11:15 +0800)]
x86: Clean up lapic codes

This commit cleans up the lapic codes:
- Delete arch/x86/include/asm/lapic_def.h, and move register and bit
  defines into arch/x86/include/asm/lapic.h
- Use MSR defines from msr-index.h in enable_lapic() and disable_lapic()
- Remove unnecessary stuff like NEED_LAPIC, X86_GOOD_APIC and
  CONFIG_AP_IN_SIPI_WAIT
- Move struct x86_cpu_priv defines to asm/arch-ivybridge/bd82x6x.h, as
  it is not apic related and only used by ivybridge
- Fix coding convention issues

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Move lapic_setup() call into init_bsp()
Bin Meng [Wed, 17 Jun 2015 03:15:37 +0000 (11:15 +0800)]
x86: Move lapic_setup() call into init_bsp()

Currently lapic_setup() is called before calling mp_init(), which
then calls init_bsp() where it calls enable_lapic(), which was
already enabled in lapic_setup(). Hence move lapic_setup() call
into init_bsp() to avoid the duplication.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: Move MP initialization codes into a common place
Bin Meng [Wed, 17 Jun 2015 03:15:36 +0000 (11:15 +0800)]
x86: Move MP initialization codes into a common place

Most of the MP initialization codes in arch/x86/cpu/baytrail/cpu.c is
common to all x86 processors, except detect_num_cpus() which varies
from cpu to cpu. Move these to arch/x86/cpu/cpu.c and implement the
new 'get_count' method for baytrail and cpu_x86 drivers. Now we call
cpu_get_count() in mp_init() to get the number of CPUs.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: ivybridge: Remove SMP from CPU_SPECIFIC_OPTIONS
Bin Meng [Wed, 17 Jun 2015 03:15:35 +0000 (11:15 +0800)]
x86: ivybridge: Remove SMP from CPU_SPECIFIC_OPTIONS

Ivybridge is not ready for U-Boot MP initialization yet.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agodm: cpu: Add a new get_count method to cpu uclass
Bin Meng [Wed, 17 Jun 2015 03:15:34 +0000 (11:15 +0800)]
dm: cpu: Add a new get_count method to cpu uclass

Introduce a new method 'get_count' in the UCLASS_CPU ops to get
the number of CPUs in the system.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: kconfig: Fix minor nits in MAX_CPUS
Bin Meng [Fri, 12 Jun 2015 06:52:23 +0000 (14:52 +0800)]
x86: kconfig: Fix minor nits in MAX_CPUS

Move MAX_CPUS definition after SMP so that it shows below SMP in the
menuconfig. Also replace the leading spaces in the MAX_CPUS section
with tabs to conform coding standard.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: kconfig: Make MAX_CPUS and AP_STACK_SIZE depend on SMP
Bin Meng [Fri, 12 Jun 2015 06:52:22 +0000 (14:52 +0800)]
x86: kconfig: Make MAX_CPUS and AP_STACK_SIZE depend on SMP

MAX_CPUS and AP_STACK_SIZE are only meaningful when SMP is on.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agox86: dm: Clean up cpu drivers
Bin Meng [Fri, 12 Jun 2015 06:52:20 +0000 (14:52 +0800)]
x86: dm: Clean up cpu drivers

This commit does the following to clean up x86 cpu dm drivers:
- Move cpu_x86 driver codes from arch/x86/cpu/cpu.c to a dedicated
  file arch/x86/cpu/cpu_x86.c
- Rename x86_cpu_get_desc() to cpu_x86_get_desc() to keep consistent
  naming with other dm drivers
- Add a new cpu_x86_bind() in the cpu_x86 driver which does exactly
  the same as the one in the intel baytrail cpu driver
- Update intel baytrail cpu driver to use cpu_x86_get_desc() and
  cpu_x86_bind()

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agodm: cpu: Test against cpu_ops->get_info in cpu_get_info()
Bin Meng [Fri, 12 Jun 2015 06:52:19 +0000 (14:52 +0800)]
dm: cpu: Test against cpu_ops->get_info in cpu_get_info()

In cpu_get_info() it wrongly tests against cpu_ops->get_desc to see
if it is NULL. It should test against cpu_ops->get_info.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
9 years agodm: cpu: Fix undefined ENOSYS build error
Bin Meng [Fri, 12 Jun 2015 06:52:18 +0000 (14:52 +0800)]
dm: cpu: Fix undefined ENOSYS build error

Include <errno.h> otherwise ENOSYS is undefined.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agodm: spi: Correct minor nits in ICH driver
Simon Glass [Sun, 7 Jun 2015 14:50:33 +0000 (08:50 -0600)]
dm: spi: Correct minor nits in ICH driver

Tidy up three minor problems in this file.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Jagan Teki <jteki@openedev.com>
9 years agospi: sf: Print the error code on failure
Simon Glass [Sun, 7 Jun 2015 14:50:32 +0000 (08:50 -0600)]
spi: sf: Print the error code on failure

Rather than just 'ERROR', display the error code, which may be useful, at
least with driver model.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Marek Vasut <marex@denx.de>
Reviewed-by: Jagan Teki <jteki@openedev.com>
9 years agox86: fsp: Move FspInitEntry call to board_init_f()
Bin Meng [Sun, 7 Jun 2015 03:33:14 +0000 (11:33 +0800)]
x86: fsp: Move FspInitEntry call to board_init_f()

The call to FspInitEntry is done in arch/x86/lib/fsp/fsp_car.S so far.
It worked pretty well but looks not that good. Apart from doing too
much work than just enabling CAR, it cannot read the configuration
data from device tree at that time. Now we want to move it a little
bit later as part of init_sequence_f[] being called by board_init_f().
This way it looks and works better in the U-Boot initialization path.

Due to FSP's design, after calling FspInitEntry it will not return to
its caller, instead it jumps to a continuation function which is given
by bootloader with a new stack in system memory. The original stack in
the CAR is gone, but its content is perserved by FSP and described by
a bootloader temporary memory HOB. Technically we can recover anything
we had before in the previous stack, but that is way too complicated.
To make life much easier, in the FSP continuation routine we just
simply call fsp_init_done() and jump back to car_init_ret() to redo
the whole board_init_f() initialization, but this time with a non-zero
HOB list pointer saved in U-Boot's global data so that we can bypass
the FspInitEntry for the second time.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
Tested-by: Andrew Bradford <andrew.bradford@kodakalaris.com>
Tested-by: Simon Glass <sjg@chromium.org>