This method is supposed to extract platform data from the device tree. It
should be done before the device itself is probed. Move it earlier in the
device_probe() function.
Signed-off-by: Simon Glass <sjg@chromium.org>
Early in boot it is necessary to decode the PCI device/function values for
particular peripherals in the device tree or of-platdata. This is needed
in TPL where CONFIG_PCI is not defined.
To handle this, move pci_get_devfn() into a file that is built even when
CONFIG_PCI is not defined.
Also add a function for use by of-platdata, to convert a reg property to
a pci_dev_t.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
These functions are CPU-related and do not use driver model. Move them to
cpu_func.h
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Bootstage improvements for TPL, SPL
Various sandbox and dm improvements and fixes
-----BEGIN PGP SIGNATURE-----
iQFFBAABCgAvFiEEslwAIq+Gp8wWVbYnfxc6PpAIreYFAl24zDoRHHNqZ0BjaHJv
bWl1bS5vcmcACgkQfxc6PpAIreZRiwf+Mf7oG//zjiZcHaPP6KOFb1CjwqLSuiaI
eCYB6GEZXuSpc8MjAw+baAfIw8MMFuaCfPsUQ5dDo391neK0sW9kkEsdcXnQBpB3
GdJYcewLN1UuJZovriGobIisGc0GQHh8gmRcVGWSKoEt+gAQauRtVYBIUPuS3JG+
ihZgYmVgICa+3tqavIbP2oZdXLuLAxR65mfiQHtiJwdCN3OOzyls2v0T0slQX9GV
ln6EvTk8OMIudBmkB7YiTmJF7AB+3PH/uBaiYyTKO9gtjLhnmCFvG5HxJTRFOKvU
F4oVWZJDEcqfZvYgWg8i5vgokMP41nzLfBr7j5ifUX1f3xwCHy/Tpw==
=S4XI
-----END PGP SIGNATURE-----
Merge tag 'dm-pull-29oct19' of git://git.denx.de/u-boot-dm
- Fix for patman with email addresses containing commas
- Bootstage improvements for TPL, SPL
- Various sandbox and dm improvements and fixes
Use log() insted of debug() for uclass_find_device_by_seq function,
since this print is very much and we can filter it out with log()
interface.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Move #define to top of file as per docs:
Signed-off-by: Simon Glass <sjg@chromium.org>
This function assumes that the 'val' parameter has no masked bits set.
This is not defined by the function prototype though. Fix the function to
mask the value and update the documentation.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
This fixes the case where assigned-clocks is used to define a clock
defaults inside this same clock's node. This is used sometimes to setup a
default parents and/or rate for a clock.
example:
muxed_clock: muxed_clock {
clocks = <&clk_provider 0>, <&clk_provider 1>;
#clock-cells = <0>;
assigned-clocks = <&muxed_clock>;
assigned-clock-parents = <&clk_provider 1>;
};
It doesn't work in u-boot because the assigned-clocks are setup *before*
the clock is probed. (clk_set_parent() will likely crash or fail if called
before the device probe function)
Making it work by handling "assigned-clocks" in 2 steps: first before the
clk device is probed, and then after the clk device is probed.
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
At present the 'Index' column of 'dm tree' assumes there is
two digits, this patch increase it to 3 digits.
It also aligns output of 'dm uclass', assuming the same 3 digits index.
The boards with CONFIG_PINCTRL_FULL activated have one pinconfig
by pin configuration, so they can have more than 100 devices
pinconfig (for example with stm32mp157c-ev1 board we have
106 pinconfig node).
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The power domain associated with a device is enabled when probing,
but currently the domain remains enabled when the device is removed.
Some boards started to disable power domains for selected devices
via custom board_quiesce_devices(), but it doesn't work in many
cases, i. e. because devices still can be accessed later in
.remove() callback on behalf of dm_remove_devices_flags().
Utilize the DM core to power off the device power domain, but add a
device flag to be able to selectively let the power domain enabled
after device removal. This might be required for devices that must
remain enabled when booting OS, i. e. serial console for debug
output, etc.
Signed-off-by: Anatolij Gustschin <agust@denx.de>
Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
In some remoteproc cases, enabling the power domain of the core will
start running the core. In such cases image should be loaded before
enabling the power domain. But the current DM framework enables the
power-domain by default during probe. This is causing the remotecore
to start and crash as there is no valid image loaded.
In order to avoid this introduce a DM flag that doesn't allow for
enabling/disabling the power-domain by DM framework.
Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
When multiple power domains attached to a device, need power on
them all, so use dev_power_domain_on to do that.
Signed-off-by: Peng Fan <peng.fan@nxp.com>
For CONFIG_OF_PRIOR_STAGE, in the absence of a device tree alias for a
given device, use the next request number for that type of device.
This allows aliases to be used when they're available, while still
allowing unaliased devices to be probed.
Signed-off-by: Thomas Fitzsimmons <fitzsim@fitzsim.org>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Simon Glass <sjg@chromium.org>
This function ise effectively replaced by ofnode_read_pci_addr() which
works with flat tree. Delete it to avoid code duplication.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
At present PCI address transaction is not supported so drivers must
manually read the correct BAR after reading the device tree info. The
ns16550 has a suitable implementation, so move this code into the core
DM support.
Note that there is no live-tree equivalent at present.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
[bmeng: correct the unclear comments in test.dts]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
This function returns -ENODEV when there is no device. This is
inconsistent with other functions, such as uclass_find_next_device(),
which returns 0.
Update it and tidy up the incorrect '-1' values in the comments.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
At present this function is never called when of-platdata is enabled since
we never have a device tree. However, this function is responsible for
copying over the of-platdata, so we must call it. Otherwise the probe()
method would have to be used.
Correct this and fix the sandbox serial driver to not read from the device
tree and try to write to what is read-only platdata on some platforms.
Fixes: 396e343b3d (dm: core: Allow binding a device from a live tree)
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Two functions in this file return early for no good reason. Adjust the
code to match the standard DM style of returning 0 at the end of the
function on success.
Oddly enough this save 12 bytes of code size on ARM.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
The pr_debug() functions do not response to setting the log level and in
fact have their own separate log level. Use U-Boot logging instead.
Perhaps we should make these options redirect to log_debug(), etc.?
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
At present this code passes an fdt_addr_t pointer as a u64 pointer which
is not safe, since sizeof(fdt_addr_t) may be 4, e.g. with sandbox. Correct
this to avoid a stack corruption problem.
Fixes: e679d03b08 (core: ofnode: Add ofnode_get_addr_size_index)
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
[bmeng: correct one typo in the commit message]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
This reads the low cell of the PCI address from the wrong cell. Fix it.
Also fix the function that this code came from.
Fixes: 9e51204527 (dm: core: Add operations on device tree references)
Fixes: 4ea5243a3a (fdt: fix fdtdec_get_pci_addr() for CONFIG_PHYS_64BIT)
Signed-off-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This patch adds functions dev_read_u64_default & dev_read_u64
to read unsigned 64-bit values from devicetree.
Signed-off-by: T Karthik Reddy <t.karthik.reddy@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Some of the DM functions depend on OF_CONTROL, which is incorrect.
DM and DT are orthogonal. Add macro guards around such functions to
avoid compiling them in when DM is enabled, while OF_CONTROL is not.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Simon Glass <sjg@chromium.org>
Several options are presenting themselves on a various boards
where the options are clearly not used. (ie, SPL/TPL options
when SPL or TPL are not defined)
This patch is not attempting to be a complete list of items, but
more like low hanging fruit. In some instances, I wasn't sure
of DM was required, so I simply made them SPL or TPL.
This patch attempts to reduce some of the menuconfig noise
by defining dependencies so they don't appear when not used.
Signed-off-by: Adam Ford <aford173@gmail.com>
In some case with LIVE DT, some node always not valid, or not have
a valid name, eg. blk driver add by mmc.
Return fail instead of Assert for this kind of ofnode, and this
help with assert happen from time to time when of_live is enabled
and DEBUG is enabled.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Current dev_read_*() API lacks support to get address and size
of a "reg" property by name or index. Add support for the same.
Livetree support has been added but not tested on real hardware.
The existing unit tests testing reading address from device-tree
have been updated to test address as well as size.
Reviewed-by: Lokesh Vutla <lokeshvutla@ti.com>
Signed-off-by: Sekhar Nori <nsekhar@ti.com>
Reduce power domain calls when CONFIG_POWER_DOMAIN is disabled.
With gcc v8.2, this change saves 104 bytes.
Signed-off-by: Anatolij Gustschin <agust@denx.de>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Per device tree spec, "status" property can have a value of "okay",
or "disabled", but not "disable".
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Without a valid ofnode, it's meaningless to call clk_set_defaults()
to process various properties.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Add the following functions to translate DMA address to CPU address:
- dev_translate_dma_address()
- ofnode_translate_dma_address()
- of_translate_dma_address()
- fdt_translate_dma_address()
These functions work the same way as xxx_translate_address(), with the
difference that the translation relies on the "dma-ranges" property
instead of the "ranges" property.
Add related test. Test report:
=> ut dm fdt_translation
Test: dm_test_fdt_translation: test-fdt.c
Test: dm_test_fdt_translation: test-fdt.c (flat tree)
Failures: 0
Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
In device nodes with more than one entry in the reg property,
it is sometimes useful to regmap only of the entries. Add an
API regmap_init_mem_index() to facilitate this.
Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
`if (!strncmp(dev->name, name, strlen(name)))` might find out
the wrong device, it might find out `dram_pll_ref_sel`, when name is
`dram_pll`. So use strcmp to avoid such issue.
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This add missing parts for previous commit 06f94461a9
("fdt: Allow indicating a node is for U-Boot proper only")
At present it is not possible to specify that a node should be used before
relocation (in U-Boot proper) without it also ending up in SPL and TPL
device trees. Add a new "u-boot,dm-pre-proper" boolean property for this.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
This uclass is intended for devices that do not need any features from the
uclass, including binding children.
This will typically be used by devices that are used to bind child devices
but do not use dm_scan_fdt_dev() to do it. That is for example the case of
several USB wrappers that have 2 child devices (1 for device and 1 for
host) but bind only one at a any given time.
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
It was returning an int, which doesn't work if the u32 it is reading,
or the default value, will overflow a signed int.
While it could be made to work, when using a C standard/compiler where
casting negative signed values to unsigned has a defined behavior,
combined with careful casting, it seems obvious one is meant to use
ofnode_read_s32_default() with signed values.
Cc: Simon Glass <sjg@chromium.org>
Signed-off-by: Trent Piepho <tpiepho@impinj.com>
Add ofnode_get_addr_size_index function to fetch the address
and size of the reg space based on index.
Signed-off-by: Keerthy <j-keerthy@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Update the condition to translate only if this is enabled for SPL.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Testing has shown that the current DM implementation of a platform /
board specific translation offset, as its needed for the SPL on MVEBU
platforms is buggy. The translation offset is confingured too late,
after the driver bind functions are run. This may result in incorrect
address translations. With the current implementation its not possible
to configure the offset earlier, as the DM code has not run at all.
This patch now removed the set_/get_translation_offset() calls and
moves the translation offset into the GD variable translation_offset.
This variable will get used when CONFIG_TRANSLATION_OFFSET is enabled.
This option is enabled only for MVEBU on ARM32 platforms, where its
currenty needed and configured in the SPL.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Pierre Bourdon <delroth@gmail.com>
Cc: Baruch Siach <baruch@tkos.co.il>
Cc: Simon Glass <sjg@chromium.org>
Cc: Heiko Schocher <hs@denx.de>
Cc: Tom Rini <trini@konsulko.com>
Tested-by: Pierre Bourdon <delroth@gmail.com>
Tested-by: Baruch Siach <baruch@tkos.co.il>
Boards such as qemu-riscv, which receive their device tree at runtime,
for example from QEMU or firmware, are unable to add the appropriate
device tree properties to make devices available pre relocation.
Instead, they must rely on the DM_FLAG_PRE_RELOC flag to be set for the
required drivers.
Add the DM_FLAG_PRE_RELOC flag to the simple-bus driver to make devices
under it with drivers that have set the flag as well available pre
relocation for these boards.
Signed-off-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
The function dm_ofnode_pre_reloc should be used instead
of the function dm_fdt_pre_reloc and avoid duplicated code.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
v2019.04-rc3 sandbox U-Boot fails to boot when compiled with
-fsanitize=address and linked against -lasan, reporting [1].
Git bisecting shows that the issue is contributed by v2019.01 commit
1678754f5e ("core: ofnode: Fix ofnode_get_addr_index function").
The root cause seems to be the mismatch between sizeof(u64) and
sizeof(fdt_size_t) on sandbox. Luckily, thanks to the fact that the
size argument of both of_get_address() and fdtdec_get_addr_size_fixed()
is optional, we can pass NULL in its place, avoiding the problem.
[1] Backtrace reported by ASAN (gcc 8.1.0):
$> ./u-boot -d arch/sandbox/dts/sandbox.dtb
[..]
Reviewed-by: Simon Glass <sjg@chromium.org>
=================================================================
==10998==ERROR: AddressSanitizer: stack-buffer-overflow on address 0x7ffcc2331140 at pc 0x0000004eeeb0 bp 0x7ffcc2330f80 sp 0x7ffcc2330f70
WRITE of size 8 at 0x7ffcc2331140 thread T0
#0 0x4eeeaf in of_get_address drivers/core/of_addr.c:154
#1 0x4f7441 in ofnode_get_addr_index drivers/core/ofnode.c:263
#2 0x5b2a78 in sb_eth_ofdata_to_platdata drivers/net/sandbox.c:422
#3 0x4dccd8 in device_probe drivers/core/device.c:407
#4 0x753170 in eth_initialize net/eth-uclass.c:428
#5 0x47d9bf in initr_net common/board_r.c:557
#6 0x6bcfa7 in initcall_run_list lib/initcall.c:30
#7 0x47e1fe in board_init_r common/board_r.c:859
#8 0x4060e5 in main arch/sandbox/cpu/start.c:356
#9 0x7fb8d135482f in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x2082f)
#10 0x40a3a8 in _start (/path/to/u-boot/u-boot+0x40a3a8)
Address 0x7ffcc2331140 is located in stack of thread T0 at offset 32 in frame
#0 0x4f72b8 in ofnode_get_addr_index drivers/core/ofnode.c:255
This frame has 3 object(s):
[32, 36) 'size' <== Memory access at offset 32 partially overflows this variable
[96, 100) 'flags'
[160, 168) 'node'
HINT: this may be a false positive if your program uses some custom stack unwind mechanism or swapcontext
(longjmp and C++ exceptions *are* supported)
SUMMARY: AddressSanitizer: stack-buffer-overflow drivers/core/of_addr.c:154 in of_get_address
Shadow bytes around the buggy address:
0x10001845e1d0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x10001845e1e0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x10001845e1f0: 00 00 00 00 00 00 00 00 00 00 00 00 f1 f1 f1 f1
0x10001845e200: 04 f2 f2 f2 f2 f2 f2 f2 04 f2 f2 f2 f2 f2 f2 f2
0x10001845e210: 04 f2 f2 f2 f3 f3 f3 f3 00 00 00 00 00 00 00 00
=>0x10001845e220: 00 00 00 00 f1 f1 f1 f1[04]f2 f2 f2 f2 f2 f2 f2
0x10001845e230: 04 f2 f2 f2 f2 f2 f2 f2 00 f2 f2 f2 f3 f3 f3 f3
0x10001845e240: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x10001845e250: 00 00 00 00 f1 f1 f1 f1 00 00 f2 f2 f3 f3 f3 f3
0x10001845e260: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 f1 f1
0x10001845e270: f1 f1 00 f2 f2 f2 f3 f3 f3 f3 00 00 00 00 00 00
Shadow byte legend (one shadow byte represents 8 application bytes):
Addressable: 00
Partially addressable: 01 02 03 04 05 06 07
Heap left redzone: fa
Freed heap region: fd
Stack left redzone: f1
Stack mid redzone: f2
Stack right redzone: f3
Stack after return: f5
Stack use after scope: f8
Global redzone: f9
Global init order: f6
Poisoned by user: f7
Container overflow: fc
Array cookie: ac
Intra object redzone: bb
ASan internal: fe
Left alloca redzone: ca
Right alloca redzone: cb
==10998==ABORTING
'To' list:
git log --since=1year drivers/core/ofnode.c | grep "\-by: .*@" | \
sed 's/.*-by: //' | sort | uniq -c | sort -rn
10 Simon Glass <sjg@chromium.org>
3 Mario Six <mario.six@gdsys.cc>
2 Martin Fuzzey <mfuzzey@parkeon.com>
2 Marek Vasut <marek.vasut+renesas@gmail.com>
1 Tom Rini <trini@konsulko.com>
1 Masahiro Yamada <yamada.masahiro@socionext.com>
1 Keerthy <j-keerthy@ti.com>
1 Jens Wiklander <jens.wiklander@linaro.org>
1 Bin Meng <bmeng.cn@gmail.com>
Fixes: 1678754f5e ("core: ofnode: Fix ofnode_get_addr_index function")
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Change the function syscon_regmap_lookup_by_phandle()
introduced by commit 6c3af1f24e ("syscon: dm: Add a
new method to get a regmap from DTS") to have
Linux-compatible syscon API.
Same modification than commit e151a1c288 ("syscon: add
Linux-compatible syscon API") solves issue when the node
identified by the phandle has several compatibles and is
already bound to a dedicated driver.
See Linux commit bdb0066df96e ("mfd: syscon: Decouple syscon
interface from platform devices").
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
We can remove the pre reloc property in SPL and TPL device-tree:
- u-boot,dm-pre-reloc
- u-boot,dm-spl
- u-boot,dm-tpl
As only the needed node are kept by fdtgrep (1st pass).
The associated function (XXX_pre_reloc) are simple for SPL/TPL:
return always true.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
At present it is not possible to use the syscon devices for PCI devices
since a regmap is required. Since PCI uses a 3-cell address the conversion
of the 'reg' property to an address always fails. In any case, the regmap
is not useful with PCI since devices are accessed through the PCI bus
which regmap does not support.
Add a special case for PCI syscon devices, so that they don't set up a
regmap.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
