Rob Clark [Thu, 27 Jul 2017 12:04:19 +0000 (08:04 -0400)]
efi_loader: indent entry/exit prints to show nesting level
This should make it easier to see when a callback back to UEFI world
calls back in to the u-boot world, and generally match up EFI_ENTRY()
and EFI_EXIT() calls.
Signed-off-by: Rob Clark <robdclark@gmail.com>
[agraf: remove static from const var] Signed-off-by: Alexander Graf <agraf@suse.de>
Rob Clark [Thu, 27 Jul 2017 12:04:17 +0000 (08:04 -0400)]
efi_loader: Add an EFI_CALL() macro
Rather than open-coding EFI_EXIT() + callback + EFI_ENTRY(), introduce
an EFI_CALL() macro. This makes callbacks into UEFI world (of which
there will be more in the future) more concise and easier to locate in
the code.
Signed-off-by: Rob Clark <robdclark@gmail.com> Signed-off-by: Alexander Graf <agraf@suse.de>
Rob Clark [Thu, 27 Jul 2017 12:04:16 +0000 (08:04 -0400)]
efi_loader: only evaluate EFI_EXIT()'s ret once
There are a couple spots doing things like:
return EFI_EXIT(some_fxn(...));
which I handn't noticed before. With addition of printing return value
in the EFI_EXIT() macro, now the fxn call was getting evaluated twice.
Which we didn't really want.
Signed-off-by: Rob Clark <robdclark@gmail.com> Signed-off-by: Alexander Graf <agraf@suse.de>
Rob Clark [Wed, 26 Jul 2017 18:34:05 +0000 (14:34 -0400)]
efi_loader: fix bug in efi_get_memory_map
When booting shim -> fallback -> shim -> grub -> linux the memory map is
a bit larger than the size linux passes in on the first call. But in
the EFI_BUFFER_TOO_SMALL case we were not passing back the updated size
to linux so it would loop forever.
Signed-off-by: Rob Clark <robdclark@gmail.com> Signed-off-by: Alexander Graf <agraf@suse.de>
Alexander Graf [Wed, 26 Jul 2017 11:41:05 +0000 (13:41 +0200)]
efi_loader: Fix configuration table override
Before commit 7cbc12415d ("efi_loader: initalize EFI object list
only once") we recreated the world on every bootefi invocation.
That included the object tree as well as the configuration tables.
Now however we don't recreate them, which means we must not explicitly
override the configuration tables, as otherwise we may lose our SMBIOS
table from the configuration table list on second bootefi invocation.
This patch makes bootefi call our normal configuration table modification
APIs to add/remove the FDT instead of recreating all tables from scratch.
That way the SMBIOS table gets preserved across multiple invocations.
Rob Clark [Mon, 24 Jul 2017 14:39:01 +0000 (10:39 -0400)]
efi_loader: expose protocols via GUID
shim.efi (or rather gnu-efi's LibLocateProtocol() which shim.efi uses)
resolves protocols via efi_locate_handle() so the console protocols
need to be added to the efi object list.
Signed-off-by: Rob Clark <robdclark@gmail.com>
[agraf: whitespace fixes] Signed-off-by: Alexander Graf <agraf@suse.de>
Rob Clark [Mon, 24 Jul 2017 14:39:00 +0000 (10:39 -0400)]
efi_loader: add helper macro to construct protocol objects
There are a bunch of protocols which should be exposed by GUID but are
not. Add a helper macro to create an efi_object, to avoid much typing.
Note that using the pointer for efiobj->handle is semi-arbitrary. We
just need a unique value to match the efiobj supporting the protocol
with the handle that LocateHandle() returns..
See LibLocateProtocol() in gnu-efi. It does LocateHandle() to find all
the handles, and then loops over them calling HandleProtocol() with the
GUID of the protocol it is trying to find.
Signed-off-by: Rob Clark <robdclark@gmail.com> Signed-off-by: Alexander Graf <agraf@suse.de>
efi_loader: parameter types for CreateEvent, SetTimer
The first argument 'type' of CreateEvent is an 32bit unsigned
integer bitmap and not an enum.
The second argument 'type' of SetTimer take values of an
enum which is called EFI_TIMER_DELAY in the UEFI standard.
To avoid confusion rename efi_event_type to efi_timer_delay.
Reported-by: Alexander Graf <agraf@suse.de> Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Signed-off-by: Alexander Graf <agraf@suse.de>
Rob Clark [Thu, 20 Jul 2017 11:59:39 +0000 (07:59 -0400)]
efi_loader: workaround for grub lsefi bug
Patch has also been sent to fix grub to not ignore the error returned
and treat protocol_buffer_count as valid. But that that might take a
while to trickle into distro's, so this workaround might be useful.
Signed-off-by: Rob Clark <robdclark@gmail.com> Signed-off-by: Alexander Graf <agraf@suse.de>
Set up a timer event and the WaitForKey event.
In the notify function of the timer event check for console input
and signal the WaitForKey event accordingly.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Signed-off-by: Alexander Graf <agraf@suse.de>
efi_loader: disk: iterate only over valid block devices
The efi_loader currently stops iterating over the available
block devices stopping at the first device that fails.
This may imply that no block device is found.
With the patch efi_loader only iterates over valid devices.
It is based on patch 06d592bf52f6 (dm: core: Add uclass_first/next_device_check())
which is currently in u-boot-dm.git.
For testing I used an odroid-c2 with a dts including
&sd_emmc_a {
status = "okay";
};
This device does not exist on the board and cannot be initialized.
Without the patch:
=> bootefi hello
## Starting EFI application at 01000000 ...
WARNING: Invalid device tree, expect boot to fail
mmc_init: -95, time 1806
Found 0 disks
Hello, world!
## Application terminated, r = 0
With the patch:
=> bootefi hello
## Starting EFI application at 01000000 ...
WARNING: Invalid device tree, expect boot to fail
mmc_init: -95, time 1806
Scanning disk mmc@70000.blk...
Scanning disk mmc@72000.blk...
Card did not respond to voltage select!
mmc_init: -95, time 9
Scanning disk mmc@74000.blk...
Found 3 disks
Hello, world!
## Application terminated, r = 0
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Reviewed-by: Andreas Färber <afaerber@suse.de> Signed-off-by: Alexander Graf <agraf@suse.de>
bootefi: allow return without EFI_BOOT_SERVICES.Exit
The Unified Extensible Firmware Interface Specification, version 2.7,
defines in chapter 2.1.2 - UEFI Application that an EFI application may
either directly return or call EFI_BOOT_SERVICES.Exit().
Unfortunately U-Boot makes the incorrect assumption that
EFI_BOOT_SERVICES.Exit() is always called.
So the following application leads to a memory exception on the aarch64
architecture when returning:
For the implementation of UninstallMultipleProtocolInterfaces we
need to call efi_uninstall_protocol_interface. In internal calls
we should not pass through EFI_EXIT.
The patch introduces a wrapper function
efi_uninstall_protocol_interface_ext.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Signed-off-by: Alexander Graf <agraf@suse.de>
For the implementation of InstallMultipleProtocolInterfaces we
need to call efi_install_protocol_interface. In internal calls
we should not pass through EFI_EXIT.
The patch introduces a wrapper function
efi_install_protocol_interface_ext.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Signed-off-by: Alexander Graf <agraf@suse.de>
Masahiro Yamada [Thu, 22 Jun 2017 08:57:50 +0000 (17:57 +0900)]
dm: ofnode: use fdt32_t for DT property value to fix sparse warning
DTB is encoded in big endian. When we retrieve property values,
we need to use fdt32_to_cpu (aka be32_to_cpu) for endian conversion.
This is a bit error-prone, but sparse is useful to detect endian
mismatch.
We need to use (fdt32_t *) instead of (u32 *) for a pointer of a
property value. Otherwise sparse warns "cast to restricted __be32".
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Masahiro Yamada [Thu, 22 Jun 2017 08:10:11 +0000 (17:10 +0900)]
dm: include <dm/util.h> from driver/core/dump.c
Include <dm/util.h> to fix sparse warnings:
symbol 'dm_dump_all' was not declared. Should it be static?
symbol 'dm_dump_uclass' was not declared. Should it be static?
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com> Reviewed-by: Bin Meng <bmeng.cn@gmail.com> Acked-by: Simon Glass <sjg@chromium.org>
Masahiro Yamada [Thu, 22 Jun 2017 07:54:05 +0000 (16:54 +0900)]
dm: ofnode: rename ofnode_read_prop() to ofnode_get_property()
This function returns the pointer to the value of a node property.
The current name ofnode_read_prop() is confusing. Follow the naming
of_get_property() from Linux.
The return type (const u32 *) is wrong. DT property values can be
strings as well as integers. This is why of_get_property/fdt_getprop
returns an opaque pointer.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com> Acked-by: Simon Glass <sjg@chromium.org>
Tom Rini [Tue, 13 Jun 2017 11:10:35 +0000 (07:10 -0400)]
fdt: Check for NULL return from fdt_getprop in 'fdt set'
While the previous pass through fixed one place where we knew that
fdt_getprop would be given a positive len, in the case of 'fdt set' we
do not, so check that we did no get NULL from fdt_getprop().
Cc: Simon Glass <sjg@chromium.org> Reported-by: Coverity (CID: 163249)
Fixes 72c98ed1ab48 ("fdt: Add a check to do_fdt() for coverity") Signed-off-by: Tom Rini <trini@konsulko.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Simon Glass [Mon, 12 Jun 2017 12:22:01 +0000 (06:22 -0600)]
tegra: fdt: Ensure that the console UART is enabled
Many tegra boards have the console UART node disabled. With livetree this
prevents serial from working since it does not 'force' the console to be
bound. Updates the affected boards to fix this error.
The boards were checked with:
for b in $(grep tegra boards.cfg |grep -v integrator | \
awk '{print $7}' | sort); do
echo $b;
fdtgrep -c nvidia,tegra20-uart b/$b/u-boot.dtb |grep okay;
done
Simon Glass [Mon, 12 Jun 2017 12:21:31 +0000 (06:21 -0600)]
dm: core: Add functions to obtain node's address/size cells
The of_n_addr_cells() and of_n_size_cells() functions are useful for
getting the size of addresses in a node, but in a few places U-Boot needs
to obtain the actual property value for a node without walking up the
stack. Add functions for this and just the existing code to use it.
Add a comment to the existing ofnode functions which do not do the right
thing with a flat tree.
Simon Glass [Mon, 12 Jun 2017 12:21:29 +0000 (06:21 -0600)]
dm: core: Add dev_read_resource() to read device resources
Add a function which reads resources from a device, such as the device
hardware address. This uses the "reg" property in the device.
Unlike other functions there is little sense in inlining this when
livetree is not being used because it has some logic in it and this would
just bloat the code size.
Simon Glass [Wed, 21 Jun 2017 03:28:49 +0000 (21:28 -0600)]
binman: Put our local modules ahead of system modules
If a system module is named the same as one of those used by binman we
currently pick the system module. Adjust the ordering so that our modules
are chosen instead.
The module conflict reported was 'tools' from jira-python. I cannot access
that package to test it.
Signed-off-by: Simon Glass <sjg@chromium.org> Reported-by: Kevin Hilman <khilman@baylibre.com> Acked-by: Kevin Hilman <khilman@baylibre.com>
Simon Glass [Mon, 19 Jun 2017 04:09:05 +0000 (22:09 -0600)]
sandbox: Stop printing platdata at the start of SPL
Currently we have code which prints out platform data at the start of SPL.
Now that we have tests for dtoc this is probably not necessary. Drop it.
Update test_ofplatdata to check for empty output since it is useful to
check that sandbox_spl works as expected.
Simon Glass [Mon, 19 Jun 2017 04:09:04 +0000 (22:09 -0600)]
dtoc: Add a comment about string replace in conv_name_to_c()
This function uses several separate string replaces where a regular
expression might seem more reasonable. Add a comment justifying the way it
is currently done.
Simon Glass [Mon, 19 Jun 2017 04:09:03 +0000 (22:09 -0600)]
dtoc: Move the main logic into the dtb_platdata file
Collect the main logic of dtoc into a function and put it into
dtb_platdata. This will allow tests to use this function instead of
duplicating the code themselves.
Simon Glass [Mon, 19 Jun 2017 04:09:01 +0000 (22:09 -0600)]
dtoc: Pass include_disabled explicitly
This option is the only one actually used by the dtb_platdata class. Pass
it explicitly to avoid needing to pass the whole option object to the
constructor.
Simon Glass [Mon, 19 Jun 2017 04:08:58 +0000 (22:08 -0600)]
dtoc: Split out the main class into its own file
To simplify running tests we should move this class into its own file.
This allows the tests to import it without having to import dtoc.py, which
runs the tests.
Simon Glass [Fri, 16 Jun 2017 03:39:33 +0000 (21:39 -0600)]
moveconfig: Allow automatic location and adding of 'imply'
By using a Kconfig parser we can find the location of each option in the
Kconfig tree. Using the information from the database we can then
automatically add an 'imply' option into the right place if requested by
the user.
Add a -a option to support adding 'imply' options. Display the location of
any existing 'imply' option so that progress can be examined. Add a -A
option to hide any existing 'imply' options so that already-completed
additions need not be considered further.
Simon Glass [Fri, 16 Jun 2017 03:39:32 +0000 (21:39 -0600)]
moveconfig: Allow control of which implying configs are shown
Sometimes it is useful to display CONFIG_TARGET or CONFIG_CMD configs. Add
an option to control this.
Also we generally ignore implying configs which affect fewer than 5
boards. But sometimes it is useful to show those those, so add an option
that reduces the minimum to two.
ERRATUM configs are never useful for implying things, so ignore those.
Simon Glass [Fri, 16 Jun 2017 03:39:31 +0000 (21:39 -0600)]
moveconfig: Allow piping in 'git show --stat' output
It is useful to be able to process only a subset of boards to save time.
Often that subset is defined by the defconfig files in a git commit. This
change allows things like:
# Build the database
./tools.moveconfig.py -b
# Find some implying configs
./tools/moveconfig.py -i CONFIG_X
# Add some 'imply' statements to Kconfig files
./tools/moveconfig.py -i CONFIG_X -a CONFIG_A,CONFIG_B
# Reprocess the defconfig files to see if we can drop some changes
git show --stat | ./tools/moveconfig.py -s -d -
# Update the commit, with fewer defconfig changes
gii commit -au
Where the commit contains defconfig files, this will reprocess them to
take account of the imply statements that you added.
Simon Glass [Fri, 16 Jun 2017 03:37:53 +0000 (21:37 -0600)]
sandbox: Enable more console options
Enable the pre-console buffer, displaying the model and post-relocation
console announce on sandbox. Also add a model name to the device tree.
This allows testing of these features.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com> Tested-by: Stephen Warren <swarren@nvidia.com>
Simon Glass [Fri, 16 Jun 2017 03:37:52 +0000 (21:37 -0600)]
console: Use map_sysmem() for the pre-relocation console
At present this feature casts the address to a pointer. Use the
map_sysmem() function so that it will work correctly on sandbox.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com> Tested-by: Bin Meng <bmeng.cn@gmail.com> Tested-by: Stephen Warren <swarren@nvidia.com>
Simon Glass [Fri, 16 Jun 2017 03:37:50 +0000 (21:37 -0600)]
Allow displaying the U-Boot banner on a video display
At present the U-Boot banner is only displayed on the serial console. If
this is not visible to the user, the banner does not show. Some devices
have a video display which can usefully display this information.
Add a banner which is printed after relocation only on non-serial devices
if CONFIG_DISPLAY_BOARDINFO_LATE is defined.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com> Tested-by: Bin Meng <bmeng.cn@gmail.com> Tested-by: Stephen Warren <swarren@nvidia.com>
Simon Glass [Fri, 16 Jun 2017 18:51:42 +0000 (12:51 -0600)]
display_options: Refactor to allow obtaining the banner
Move the display options code into a separate function so that the U-Boot
banner can be obtained from other code. Adjust the 'version' command to
use it.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com> Tested-by: Bin Meng <bmeng.cn@gmail.com> Tested-by: Stephen Warren <swarren@nvidia.com>
Simon Glass [Thu, 15 Jun 2017 03:28:46 +0000 (21:28 -0600)]
dm: ahci: Add a driver for SCSI on AHCI
Some AHCI drivers use SCSI under the hood. Rather than making the AHCI
driver be in the SCSI uclass it makes sense to have the AHCI device create
a SCSI device as a child. That way we can handle any AHCI-specific
operations rather than trying to pretend tha the device is just SCSI.
To handle this we need to provide a way for AHCI drivers to bind a SCSI
device as its child, and probe it. Add functions for this.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
projects / u-boot / log