Bin Meng [Sun, 14 Oct 2018 03:52:02 +0000 (20:52 -0700)]
x86: Ensure no instruction sets of MMX/SSE are generated in 64-bit build
With the '-march=core2' fix, it seems that we have some luck that
the 64-bit U-Boot boots again. However if we examine the disassembly
codes there are still SSE instructions elsewhere which means passing
cpu type to GCC is not enough to prevent it from generating these
instructions. A simple test case is doing a 'bootefi selftest' from
the U-Boot shell and it leads to a reset too.
The 'bootefi selftest' reset is even seen with the image created by
the relative older GCC 5.4.0, the one shipped by Ubuntu 16.04.
The reset actually originates from undefined instruction exception
caused by these SSE instructions. To keep U-Boot as a bootloader as
simple as possible, we don't want to handle such advanced SIMD stuff.
To make sure no MMX/SSE instruction sets are generated, tell GCC not
to do this. Note AVX is out of the question as CORE2 is old enough
to support AVX yet.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com> Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Reviewed-by: Simon Glass <sjg@chromium.org>
Bin Meng [Sun, 14 Oct 2018 03:52:01 +0000 (20:52 -0700)]
x86: Specify -march=core2 to build 64-bit U-Boot proper
With newer kernel.org GCC (7.3.0 or 8.1.0), the u-boot.rom image
built for qemu-x86_64 target does not boot. It keeps resetting
soon after the 32-bit SPL jumps to 64-bit proper. Debugging shows
that the reset happens inside env_callback_init().
Executing "xorps %xmm0,%xmm0" causes CPU to immediately reset.
However older GCC like 5.4.0 (the one shipped by Ubuntu 16.04)
does not generate such instructions that utilizes SSE for this
function - env_callback_init() and U-Boot boots without any issue.
Explicitly specifying -march=core2 for newer GCC allows U-Boot
proper to boot again. Examine assembly codes of env_callback_init
and there is no SSE instruction in that function hence U-Boot
continues to boot.
core2 seems to be the oldest arch in GCC that supports 64-bit.
Like 32-bit U-Boot build we use -march=i386 which is the most
conservative cpu type so that the image can run on any x86
processor, let's do the same for the 64-bit U-Boot build.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com> Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Reviewed-by: Simon Glass <sjg@chromium.org>
Simon Glass [Sun, 2 Sep 2018 23:02:24 +0000 (17:02 -0600)]
Enable CONFIG_TIMER_EARLY with bootstage
In initr_bootstage() we call bootstage_mark_name() which ends up calling
timer_get_us(). This call happens before initr_dm(), which inits driver
model.
On x86 we set gd->timer to NULL in the transition from board_init_f()
to board_init_r(). See board_init_f_r() for this assignment. So U-Boot
knows there is no timer available in the period immediately after
relocation.
On x86 the timer_get_us() call is implemented as calls to get_ticks() and
get_tbclk(). Both of these call dm_timer_init() to set up the timer, if
gd->timer is NULL and the early timer is not available.
However dm_timer_init() cannot succeed before initr_dm() is called.
So it seems that on x86 if we want to use CONFIG_BOOTSTAGE we must enable
CONFIG_TIMER_EARLY. Update the Kconfig to handle this.
Note: On most architectures we can rely on the pre-relocation memory still
being available, so that gd->timer pointers to a valid timer device and
everything works correctly. Admittedly this is not strictly correct since
the timer device is set up by pre-relocation U-Boot, but normally this is
fine. On x86 the 'CAR' (cache-as-RAM) memory used by pre-relocation U-Boot
disappears in board_init_f_r() and any attempt to access it will hang.
This is the reason why we must mark the timer as invalid when we get to
board_init_f_r().
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Simon Glass [Sun, 2 Sep 2018 23:02:26 +0000 (17:02 -0600)]
binman: Add support for Intel reference code
Some platforms use this instead of FSP to set up the platform, including
memory. Add support for this in binman. This is needed for
chromebook_samus, for example.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Marek Vasut [Wed, 10 Oct 2018 19:27:09 +0000 (21:27 +0200)]
test: Add test for PCI device without compat string and with DT node
Add test which checks if a PCI device described in DT with an
entry and reg = <...> property, but without compatible string
results in a valid U-Boot PCI udevice with the udevice.node
populated with reference to this DT node. Also check if the
other PCI device without a DT node does not contain any bogus
udevice.node.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com> Cc: Simon Glass <sjg@chromium.org> Cc: Tom Rini <trini@konsulko.com> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Marek Vasut [Wed, 10 Oct 2018 19:27:08 +0000 (21:27 +0200)]
test: Add PCI device entry without compat string and with DT node
Add PCI entry without compatible string and with a DT node only with
reg = <...> property into the DT. This is needed for the tests to
verify whether such a setup creates an U-Boot PCI device with the
DT node associated with it in udevice.node.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com> Cc: Simon Glass <sjg@chromium.org> Cc: Tom Rini <trini@konsulko.com> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Marek Vasut [Wed, 10 Oct 2018 19:27:07 +0000 (21:27 +0200)]
pci: Update documentation to make 'compatible' string optional
Reword the documentation to make it clear the compatible string is now
optional, yet still matching on it takes precedence over PCI IDs and
PCI classes.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com> Cc: Simon Glass <sjg@chromium.org> Cc: Tom Rini <trini@konsulko.com> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Marek Vasut [Wed, 10 Oct 2018 19:27:06 +0000 (21:27 +0200)]
pci: Support parsing PCI controller DT subnodes
The PCI controller can have DT subnodes describing extra properties
of particular PCI devices, ie. a PHY attached to an EHCI controller
on a PCI bus. This patch parses those DT subnodes and assigns a node
to the PCI device instance, so that the driver can extract details
from that node and ie. configure the PHY using the PHY subsystem.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com> Cc: Simon Glass <sjg@chromium.org> Cc: Tom Rini <trini@konsulko.com> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Adam Ford [Sun, 14 Oct 2018 20:10:50 +0000 (15:10 -0500)]
Convert CONFIG_FLASH_CFI_DRIVER et al to Kconfig
This converts the following to Kconfig:
CONFIG_FLASH_CFI_DRIVER
CONFIG_SYS_FLASH_USE_BUFFER_WRITE
CONFIG_FLASH_CFI_MTD
CONFIG_SYS_FLASH_PROTECTION
CONFIG_SYS_FLASH_CFI
Signed-off-by: Adam Ford <aford173@gmail.com>
[trini: Re-migrate] Signed-off-by: Tom Rini <trini@konsulko.com>
Tom Rini [Sun, 7 Oct 2018 18:20:30 +0000 (14:20 -0400)]
travis: Add sandbox/clang-7 support
To make testing with clang support easier, add sandbox/clang-7
combination to our testing matrix. To facilitate this, switch to using
the "sources" method that the travis.yml file supports to list
additional repositories and add the official one for llvm-7. Due to
buildman not supporting using clang at this time add logic to manually
build a single sandbox configuration in the expected output directory so
that we can still invoke all of our tests.
Cc: Simon Glass <sjg@chromium.org> Signed-off-by: Tom Rini <trini@konsulko.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Marek Vasut [Wed, 10 Oct 2018 22:13:54 +0000 (00:13 +0200)]
bootcount: Make bootcount magic configurable
Add new Kconfig option, SYS_BOOTCOUNT_MAGIC, to select the boot
counter magic word. This can be useful ie. in case the entire
boot counter register is not usable.
Signed-off-by: Marek Vasut <marex@denx.de> Cc: Tom Rini <trini@konsulko.com>
Bin Meng [Mon, 8 Oct 2018 23:53:54 +0000 (16:53 -0700)]
travis: Switch to i386 version toolchain for x86
Currently this uses x86_64 version toolchain for x86 build in
travis-ci. Change it to i386 version to avoid updating the
buildman toolchain path every time when the toolchain version
number is changed, eg: from 7.3.0 to 8.1.0.
Adam Ford [Sun, 7 Oct 2018 14:58:25 +0000 (09:58 -0500)]
ARM: mach-omap2: Kconfig: Make SYS_MPUCLK dependent on AM33XX
This value is unly used in arch/arm/mach-omap2/am33xx/
clock_am33xx.c, so let's make it dependent on AM33XX since
that is the only way this file gets compiled into the code
according to the Makefile.
Signed-off-by: Adam Ford <aford173@gmail.com> Reviewed-by: Tom Rini <trini@konsulko.com>
[trini: Fix symbol name] Signed-off-by: Tom Rini <trini@konsulko.com>
Makefile: add LPC32xx precondition for building platform images
To prevent accidental build failures the change converts a number of
NXP LPC32xx specific image targets to be conditionally dependent on
target build configuration. The wrapped image targets always contain
a U-Boot SPL binary and the images are supposed to be directly flashed
on a NAND flash device for read access by LPC32xx NAND MLC controller.
The explicit arch specific build symbol allows to group supported
boards, generalize common config options and it will serve as
a dependency for platform only drivers.
Two related board defconfigs are resynced after the change.
Marek Vasut [Thu, 18 Oct 2018 18:28:19 +0000 (20:28 +0200)]
ARM: rmobile: Drop PRR syscon driver
The PRR syscon driver is available too late for Multi DTB build
of U-Boot. Replace it with simple check whether a platform is
Gen3 or not and produce an address of the PRR.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Marek Vasut [Thu, 18 Oct 2018 18:27:16 +0000 (20:27 +0200)]
ARM: dts: rmobile: Build -u-boot variants of DTs
Build the -u-boot variants of the device trees so they can be included
in Multi-DTB fitImage, which in turn allows us to build single U-Boot
image for multiple boards.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
ARM: rmobile: Fix module clock controls refer status on Gen3
When referring to the MSTPSR register, it contains the clock
status of SYS, RT, SECURE, and controlling SMSTPCR using this
value has the problem of being affected by the RT and SECURE
status.This patch changes the reference register to SMSTPCR.
Tom Rini [Wed, 17 Oct 2018 11:20:52 +0000 (07:20 -0400)]
Merge tag 'signed-efi-2018.11' of git://github.com/agraf/u-boot
Patch queue for efi - 2018-10-17
A few bug fixes for the 2018.11 release:
- Fix block seeking on 32bit
- Fix execution with DEBUG set
- Fix a few Coverity found bugs
- Fix warnings
Heinrich Schuchardt (13):
efi_loader: fix relocation on x86_64
efi_loader: correct signature of GetPosition, SetPosition
efi_loader: execute efi_save_gd() first
efi_loader: efi_allocate_pool(EFI_ALLOCATE_ANY_PAGES, ...)
efi_loader: error handling in read_console()
efi_loader: return type efi_console_register()
efi_loader: superfluous statement in is_dir()
efi_loader: memory leak in efi_set_variable()
efi_loader: remove lcd.h from efi_net.c
arm: do not include efi_loader.h twice
efi_loader: fix typo in efi_boottime.c
efi_selftest: creating new handle in controller test
efi_loader: efi_dp_get_next_instance() superfluous statement
Tom Rini (2):
efi_loader: Fix warning in efi_load_image()
fs: fat: Fix warning in normalize_longname()
arm64: zynqmp: Add new command for TCM initialization
This patch adds new zynqmp command "zynqmp tcminit mode" to
initialize TCM. TCM needs to be initialized before accessing
to avoid ECC errors. This new command helps to perform
the same. It also makes tcm_init() as global and uses it for
doing the TCM initialization.
Signed-off-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
arm64: zynqmp: Move TCM initialization to a separate routine
This patch moves TCM initialization to a separate routine to
make it modular and can be reused if required. It also prints
warning message now as it writes to TCM.
Signed-off-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Virtual QEMU board is generating DTB self and putting it to
VERSAL_QEMU_DTB_ADDR address.
Board is using CONFIG_OF_BOARD which ensures that u-boot is aligned with
board created by QEMU.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Michal Simek [Wed, 22 Aug 2018 12:55:27 +0000 (14:55 +0200)]
arm64: versal: Add support for new Xilinx Versal ACAPs
Xilinx is introducing Versal, an adaptive compute acceleration platform
(ACAP), built on 7nm FinFET process technology. Versal ACAPs combine
Scalar Processing Engines, Adaptable Hardware Engines, and Intelligent
Engines with leading-edge memory and interfacing technologies to deliver
powerful heterogeneous acceleration for any application. The Versal AI
Core series has five devices, offering 128 to 400 AI Engines. The series
includes dual-core Arm Cortex™-A72 application processors, dual-core Arm
Cortex-R5 real-time processors, 256KB of on-chip memory with ECC, more
than 1,900 DSP engines optimized for high-precision floating point with
low latency.
The patch is adding necessary infrastructure in place without enabling
platform which is done in separate patch.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Tom Rini [Sun, 30 Sep 2018 14:33:42 +0000 (10:33 -0400)]
fs: fat: Fix warning in normalize_longname()
As observed with clang:
fs/fat/fat_write.c:1024:13: warning: comparison of constant 128
with expression of type 'char' is always false
[-Wtautological-constant-out-of-range-compare]
if ((0x80 <= c) && (c <= 0xff))
~~~~ ^ ~
fs/fat/fat_write.c:1024:25: warning: comparison of constant 255
with expression of type 'char' is always true
[-Wtautological-constant-out-of-range-compare]
if ((0x80 <= c) && (c <= 0xff))
~ ^ ~~~~
Fixes: 25bb9dab14f4 ("fs: fat: check and normalize file name") Signed-off-by: Tom Rini <trini@konsulko.com> Signed-off-by: Alexander Graf <agraf@suse.de>
Tom Rini [Sun, 30 Sep 2018 14:38:15 +0000 (10:38 -0400)]
efi_loader: Fix warning in efi_load_image()
As observed with clang:
lib/efi_loader/efi_boottime.c:1624:7: warning: variable 'info'
is used uninitialized whenever 'if' condition is true [-Wsometimes-uninitialized]
if (ret != EFI_SUCCESS)
^~~~~~~~~~~~~~~~~~
lib/efi_loader/efi_boottime.c:1653:7: note: uninitialized use
occurs here
free(info);
^~~~
lib/efi_loader/efi_boottime.c:1624:3: note: remove the 'if' if
its condition is always false
if (ret != EFI_SUCCESS)
^~~~~~~~~~~~~~~~~~~~~~~
lib/efi_loader/efi_boottime.c:1602:31: note: initialize the
variable 'info' to silence this warning
struct efi_loaded_image *info;
^
= NULL
Rather than change how we unwind the function it makes the most sense to
initialize info to NULL so that we can continue to pass it to free().
Fixes: c982874e930d ("efi_loader: refactor efi_setup_loaded_image()") Signed-off-by: Tom Rini <trini@konsulko.com> Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Signed-off-by: Alexander Graf <agraf@suse.de>
Fixes: 99b8db7291ce ("arm: print information about loaded UEFI images") Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Signed-off-by: Alexander Graf <agraf@suse.de>
If DEBUG is defined we may be calling EFI_CALL already during the
initialization of the EFI subsystem. We must make sure efi_save_gd() has
already been called at that moment.
Anyway it is better to have this call in one location instead of three.
This fixes an illegal memory access occurring since 4e6b5d6503ce
("efi_loader: create root node") with DEBUG = 1.
Fixes: 4e6b5d6503ce ("efi_loader: create root node") Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Signed-off-by: Alexander Graf <agraf@suse.de>
Fixes: b6dd57773719 ("efi_loader: use correct types in EFI_FILE_PROTOCOL") Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Signed-off-by: Alexander Graf <agraf@suse.de>
Currently the relocation of the EFI runtime on x86_64 fails. This renders
the EFI subsystem unusable. The ELF relocation records for x86_64 contain
an addend field.
Always write the function name into error messages related to the EFI
runtime relocation.
Break an excessively long line.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Reviewed-by: Bin Meng <bmeng.cn@gmail.com> Tested-by: Bin Meng <bmeng.cn@gmail.com> Signed-off-by: Bin Meng <bmeng.cn@gmail.com> Signed-off-by: Alexander Graf <agraf@suse.de>
Michal Simek [Wed, 27 Jun 2018 13:53:59 +0000 (15:53 +0200)]
arm64: zynqmp: Enable CDC ethernet gadget for zcu100/Ultra96
Ethernet is not present on this board that's why there are two other
options how to wire the board to ethernet. The first is asix_eth usb
host converter which is already enabled by default. The second option is
to use USB CDC/RNDIS ethernet gadget.
This patch is enabling CDC which is working with Linux.
With new bind/unbind command there is no need to call usb_ether_init()
from platform code and use for example these commands:
There shouldn't be a need to call private spl_load_fpga_image function
because the whole sequence should be already handled by fpga framework.
If there is missing loading bistream by chunks it should be done via
fpga framework instead of having private hooks.
Also spl_load_fpga_image() weak function is not used anywhere and
opening a way for not reviewed hacks out of mainline U-Boot is not the
right way to go.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Tom Rini [Mon, 15 Oct 2018 11:20:07 +0000 (07:20 -0400)]
Merge tag 'arc-more-updates-for-2018.11-rc2-2' of git://git.denx.de/u-boot-arc
More fixes and improvements for ARC here:
Fixes (this time included for real):
* Take care of global uninitialized variables
They used to be put right after .bss section and were never
zeroed as they should be. Now merged with normal .bss
Improvements:
* Print more verbose CPU info for boards built on real silicon
* Add support for SD-card detection on all ARC boards
* Quite a few fixes for IoT DK
- Support reset by command
- Print of CPU freq on boot
- Link for eFlash etc
Alexey Brodkin [Thu, 11 Oct 2018 20:12:05 +0000 (23:12 +0300)]
ARC: Don't use COMMON section for global not-initialized variables
By default GCC puts global non-initialized variables in COMMON section.
And we used to ignore existence of COMMON section in our linker
scripts though smart LD silently appended it right after .bss.
And the problem here is variables from COMMON section even though
require zeroing in run-time were not zeroed as they were placed
right after __bss_end symbol.
It was a pure luck we never faced serious problem due to this,
but now it is fixed.
Now as for some other architectures we'll just force GCC to put
those global variables in normal .bss section.
This solution is much nicer than adding COMMON section to each and
every linker script.
Alexey Brodkin [Thu, 11 Oct 2018 11:14:13 +0000 (14:14 +0300)]
iot_dk: Link for eFlash
That's what we'll have in production.
But note it won't work for loading via JTAG as
eFlash is not directly writable, one needs to use
prebootloader to flash uboot.bin from SD-card into eFlash.
So now we may detect MMC/SD-card existence and
instead of completely misleading message on missing card:
------------------------>8-----------------------
Loading Environment from FAT... Card did not respond to voltage select!
------------------------>8-----------------------
we now get very clear one:
------------------------>8-----------------------
Loading Environment from FAT... MMC: no card present
------------------------>8-----------------------
Alexey Brodkin [Wed, 10 Oct 2018 11:03:47 +0000 (14:03 +0300)]
iot_dk/hsdk: Implement its own print_cpuinfo()
ARC IDENTITY register only encodes major architecture
type and version while for a particular board/silicon we
may know better which template was used and so we may identify
CPU more precise, which exactly we do here.
Alexey Brodkin [Wed, 10 Oct 2018 10:59:33 +0000 (13:59 +0300)]
ARC: make generic print_cpuinfo() weak
This allows board to override print_cpuinfo() because
they might know better which ARChitect template was used.
This way we may not only derive base architecture type and
version but more meaningful things like "ARC EM7D" instead of
simple "ARC EM", "ARC HS36" instead of "ARC HS".
NOR flash name MT35X_QLKA and MT25Q_** used on NXP board has
manufacturer id as 0x2C, which are rather for newer flashes
after the split of Micron from ST.
So macro for this micron manufacturer id.
Signed-off-by: Suresh Gupta <suresh.gupta@nxp.com> Signed-off-by: Yogesh Gaur <yogeshnarayan.gaur@nxp.com> Signed-off-by: Ashish Kumar <Ashish.Kumar@nxp.com>
[jagan: updated commit message] Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Pankaj Bansal [Wed, 10 Oct 2018 08:38:34 +0000 (14:08 +0530)]
driver: net: fsl-mc: Add support of multiple phys for dpmac
Till now we have had cases where we had one phy device per dpmac.
Now, with the upcoming products (LX2160AQDS), we have cases, where there
are sometimes two phy devices for one dpmac. One phy for TX lanes and
one phy for RX lanes. to handle such cases, add the support for multiple
phys in ethernet driver. The ethernet link is up if all the phy devices
connected to one dpmac report link up. also the link capabilities are
limited by the weakest phy device.
i.e. say if there are two phys for one dpmac. one operates at 10G without
autoneg and other operate at 1G with autoneg. Then the ethernet interface
will operate at 1G without autoneg.
Signed-off-by: Pankaj Bansal <pankaj.bansal@nxp.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Pankaj Bansal [Wed, 10 Oct 2018 08:38:33 +0000 (14:08 +0530)]
driver: net: fsl-mc: initialize dpmac irrespective of phy
The dpmac initalization should not depend on phy.
As the phy is not necessary to be present for dpmac to function.
Therefore, remove dpmac initialization dependency from phy.
Signed-off-by: Pankaj Bansal <pankaj.bansal@nxp.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Pankaj Bansal [Wed, 10 Oct 2018 08:38:32 +0000 (14:08 +0530)]
driver: net: fsl-mc: Modify the dpmac link detection method
when there is no phy present for a dpmac, a dummy phy device is created.
when we move to multiple phy method, we need to create as many dummy phy
devices.
Change this method so that we don't need to create dummy phy devices.
We always report linkup if no phy is present.
Signed-off-by: Pankaj Bansal <pankaj.bansal@nxp.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
The interruption support had be removed for ARM architecture and
the function udelay_masked() is no more used except in some timer.c
files and have the same content than udelay() or __udelay().
This patch update each timer.c implementing this function and
remove the associated prototype in u-boot-arm.h.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
projects / u-boot.git / log