oweals/u-boot.git
4 years agotest: Add the beginnings of some string tests
Simon Glass [Wed, 8 Apr 2020 14:32:55 +0000 (08:32 -0600)]
test: Add the beginnings of some string tests

There are quite a few string functions in U-Boot with no tests. Make a
start by adding a test for strtoul().

Signed-off-by: Simon Glass <sjg@chromium.org>
4 years agolib: strto: Stop detection when invalid char is used
Michal Simek [Wed, 8 Apr 2020 08:09:16 +0000 (10:09 +0200)]
lib: strto: Stop detection when invalid char is used

This issue has been found when mtd partition are specified. Autodetection
code should stop when the first invalid char is found.

Here is the example of commands:
setenv mtdids nand0=memory-controller@e000e000
setenv mtdparts "mtdparts=nand0:4m(boot),4m(env),64m(kernel),96m(rootfs)"
mtd list

Before:
Zynq> mtd list
List of MTD devices:
* nand0
  - type: NAND flash
  - block size: 0x20000 bytes
  - min I/O: 0x800 bytes
  - OOB size: 64 bytes
  - OOB available: 16 bytes
  - ECC strength: 1 bits
  - ECC step size: 2048 bytes
  - bitflip threshold: 1 bits
  - 0x000000000000-0x000010000000 : "nand0"
  - 0x000000000000-0x000000400000 : "boot"
  - 0x000000400000-0x000000800000 : "env"
  - 0x000000800000-0x000006c00000 : "kernel"
  - 0x000006c00000-0x000010000000 : "rootfs"

Where it is visible that kernel partition has 100m instead of 64m

After:
Zynq> mtd list
* nand0
  - type: NAND flash
  - block size: 0x20000 bytes
  - min I/O: 0x800 bytes
  - OOB size: 64 bytes
  - OOB available: 16 bytes
  - ECC strength: 1 bits
  - ECC step size: 2048 bytes
  - bitflip threshold: 1 bits
  - 0x000000000000-0x000010000000 : "nand0"
  - 0x000000000000-0x000000400000 : "boot"
  - 0x000000400000-0x000000800000 : "env"
  - 0x000000800000-0x000004800000 : "kernel"
  - 0x000004800000-0x00000a800000 : "rootfs"

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Fixes: 0486497e2b5f ("lib: Improve _parse_integer_fixup_radix base 16 detection")
Tested-by: Heiko Schocher <hs@denx.de>
Tested-by: Pali Rohár <pali@kernel.org>
4 years agomtd: nand: pxa3xx: fix raw read when last_chunk_size == 0
Baruch Siach [Sun, 5 Apr 2020 16:19:31 +0000 (19:19 +0300)]
mtd: nand: pxa3xx: fix raw read when last_chunk_size == 0

Commit 6293b0361d9 ("mtd: nand: pxa3xx: add raw read support") added the
local data_len variable in handle_data_pio() to track read size, but
forgot to update the condition of drain_fifo() call. That happens to
work when the layout last_chunk_size != 0. But when last_chunk_size ==
0, drain_fifo() is not called to read the last chunk, which leads to
"Wait timeout!!!" error. Fix this.

Fixes: 6293b0361d9 ("mtd: nand: pxa3xx: add raw read support")
Cc: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
4 years agocommon/board_f: Make reserve_mmu generic
Ovidiu Panait [Sun, 29 Mar 2020 17:57:41 +0000 (20:57 +0300)]
common/board_f: Make reserve_mmu generic

Introduce arch_reserve_mmu to allow for architecture-specific reserve_mmu
routines. Also, define a weak nop stub for it.

Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
4 years agoarm: asm/cache.c: Introduce arm_reserve_mmu
Ovidiu Panait [Sun, 29 Mar 2020 17:57:40 +0000 (20:57 +0300)]
arm: asm/cache.c: Introduce arm_reserve_mmu

As a preparation for turning reserve_mmu into an arch-specific variant,
introduce arm_reserve_mmu on ARM. It implements the default routine for
reserving memory for MMU TLB and needs to be weakly defined in order to allow
for machines to override it.

Without this decoupling, after introducing arch_reserve_mmu, there would be two
weak definitions for it, one in common/board_f.c and one in
arch/arm/lib/cache.c.

Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
4 years agocommon/board_f: Move arm-specific reserve_mmu to arch/arm/lib/cache.c
Ovidiu Panait [Sun, 29 Mar 2020 17:57:39 +0000 (20:57 +0300)]
common/board_f: Move arm-specific reserve_mmu to arch/arm/lib/cache.c

Move the ARM-specific reserve_mmu definition from common/board_f.c
to arch/arm/lib/cache.c.

Signed-off-by: Ovidiu Panait <ovpanait@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
4 years agoMerge tag 'u-boot-stm32-20200424' of https://gitlab.denx.de/u-boot/custodians/u-boot-stm
Tom Rini [Fri, 24 Apr 2020 18:07:59 +0000 (14:07 -0400)]
Merge tag 'u-boot-stm32-20200424' of https://gitlab.denx.de/u-boot/custodians/u-boot-stm

- Solve stm32mp15 pinctrl dts issue (patch conflict in branches master and next)
- Split device tree for DHCOR Som and AV 96 board
- Update PLL4 setting in AV96 board
- Enable bootd, iminfo, imxtract on DHCOM

4 years agoRevert "sunxi: Fix PHY regression on A20-OLinuXino-Lime2 and A20-Olimex-SOM-EVB"
Tom Rini [Fri, 24 Apr 2020 17:13:33 +0000 (13:13 -0400)]
Revert "sunxi: Fix PHY regression on A20-OLinuXino-Lime2 and A20-Olimex-SOM-EVB"

While the change is correct, generally, it was not intended to be pushed
just yet.

This reverts commit b897306341024695d17296efc1f9d83d06368209.

Signed-off-by: Tom Rini <trini@konsulko.com>
4 years agoMerge branch '2020-04-24-master-imports'
Tom Rini [Fri, 24 Apr 2020 17:04:38 +0000 (13:04 -0400)]
Merge branch '2020-04-24-master-imports'

- Assorted minor bugfixes.
- Resync fixdep with Linux v5.7-rc1
- Numerous changes to reduce SPL in various cases including when we have
  read-only env support.
- Allow mkimage to align the header on FIT images to a specific size.

4 years agosunxi: Fix PHY regression on A20-OLinuXino-Lime2 and A20-Olimex-SOM-EVB
Tom Rini [Fri, 24 Apr 2020 16:30:45 +0000 (12:30 -0400)]
sunxi: Fix PHY regression on A20-OLinuXino-Lime2 and A20-Olimex-SOM-EVB

When moving the PHYLIB PHY drivers around in Kconfig we did not at the
same time perform a careful migration of the related drivers and
sub-options.  This lead to the case where previously Kconfig-enabled
driver choices were now disabled on some platforms.  Correct this by
enabling both the PHY driver and sub-option on the above referenced
platforms.

Fixes: af2cbfd6b982 ("drivers: net: Provide Kconfig menu for PHYLIB")
Fixes: 8728c97eff5b ("configs: Re-sync")
Reported-by: Dario <dario86@tutamail.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
---
After checking back on the original commit I can see that these are the
only two platforms to have been broken in the change.

4 years agoARM: stm32: Enable bootd, iminfo, imxtract on DHCOM
Marek Vasut [Wed, 1 Apr 2020 15:13:15 +0000 (17:13 +0200)]
ARM: stm32: Enable bootd, iminfo, imxtract on DHCOM

Enable these standard U-Boot commands for image manipulation and for
starting the default boot command using 'boot' command in U-Boot shell.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Patrick Delaunay <patrick.delaunay@st.com>
Cc: Patrice Chotard <patrice.chotard@st.com>
Reviewed-by: Patrick Delaunay <patrick.delaunay@st.com>
4 years agotravis: Correct error checking when building boards
Simon Glass [Sun, 19 Apr 2020 23:02:32 +0000 (17:02 -0600)]
travis: Correct error checking when building boards

At present if buildman reports an error, the travis build still succeeds.

This is because the travis script does not stop when it sees errors; nor
does it automatically return the exit code. Also the current error
checking never triggers since 'ret' is not set.

Fix this by setting 'ret' correctly.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reported-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Michal Simek <michal.simek@xilinx.com>
4 years agomkimage: fit: Free buf directly in fit_extract_data()
Bin Meng [Sat, 18 Apr 2020 08:59:11 +0000 (01:59 -0700)]
mkimage: fit: Free buf directly in fit_extract_data()

If given ptr to free() is NULL, no operation is performed.
Hence we can just free buf directly in fit_extract_data().

Signed-off-by: Bin Meng <bin.meng@windriver.com>
4 years agomkimage: fit: Unmmap the memory before closing fd in fit_import_data()
Lihua Zhao [Sat, 18 Apr 2020 08:59:10 +0000 (01:59 -0700)]
mkimage: fit: Unmmap the memory before closing fd in fit_import_data()

Without calling munmap(), the follow-up call to open() the same file
with a flag O_TRUNC seems not to cause any issue on Linux, but it fails
on Windows with error like below:

    Can't open kernel_fdt.itb.tmp: Permission denied

Fix this by unmapping the memory before closing fd in fit_import_data().

Signed-off-by: Lihua Zhao <lihua.zhao@windriver.com>
Signed-off-by: Bin Meng <bin.meng@windriver.com>
4 years agotools: Remove the out-of-date MinGW support codes
Bin Meng [Sat, 18 Apr 2020 08:59:09 +0000 (01:59 -0700)]
tools: Remove the out-of-date MinGW support codes

MinGW build for U-Boot tools has been broken for years. The official
support of Windows build is now MSYS2. Remove the MinGW support codes.

Signed-off-by: Bin Meng <bin.meng@windriver.com>
4 years agodoc: invalid doc link in rstFlatTable.py
Heinrich Schuchardt [Sat, 18 Apr 2020 00:16:34 +0000 (02:16 +0200)]
doc: invalid doc link in rstFlatTable.py

Remove an invalid documentation link in rstFlatTable.py. This synchronizes
the file with Linux next-20200413.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
4 years agospl: Kconfig: de-dup SPL_DM_GPIO definition
Joel Johnson [Fri, 17 Apr 2020 15:42:29 +0000 (09:42 -0600)]
spl: Kconfig: de-dup SPL_DM_GPIO definition

Two nearly concurrent commits (d4d65e112 and bcee8d676) added a
SPL_DM_GPIO symbol. Resolve the duplication in favor of the version
in drivers/gpio/Kconfig.

Signed-off-by: Joel Johnson <mrjoel@lixil.net>
Reviewed-by: Frieder Schrempf <frieder.schrempf@kontron.de>
4 years agofixdep: Re-sync with Linux 5.7-rc1
Masahiro Yamada [Thu, 16 Apr 2020 05:01:45 +0000 (14:01 +0900)]
fixdep: Re-sync with Linux 5.7-rc1

fixdep is a standalone host program, so we can just re-sync it with
the latest Linux in one commit.

I kept the U-Boot own code block surrounded by
/* hack for U-Boot */ ... /* U-boot hack end */.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
4 years agofixdep: handle CONFIG_IS_ENABLE() and friends for TPL
Masahiro Yamada [Thu, 16 Apr 2020 05:01:44 +0000 (14:01 +0900)]
fixdep: handle CONFIG_IS_ENABLE() and friends for TPL

Since commit f1c6e1922eb5 ("spl: dm: use CONFIG_IS_ENABLED to test for
the DM option"), CONFIG_IS_ENABLED() handles CONFIG_TPL_* options, but
fixdep still cannot because it hard-codes the "SPL_" prefix as follows:

    char tmp_buf[256] = "SPL_"; /* hack for U-Boot */

Take care of the "TPL_" prefix too.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
4 years ago.mailmap: map Ruchika Gupta's mail address
Heinrich Schuchardt [Wed, 15 Apr 2020 18:47:38 +0000 (18:47 +0000)]
.mailmap: map Ruchika Gupta's mail address

Freescale mail addresses are not valid anymore.

Reported-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
4 years agodlmalloc: remove unit test support in SPL
Heinrich Schuchardt [Wed, 15 Apr 2020 16:46:23 +0000 (18:46 +0200)]
dlmalloc: remove unit test support in SPL

We cannot run unit tests in SPL. So remove the unit test support.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
4 years agolib: do not provide hexdump in SPL
Heinrich Schuchardt [Wed, 15 Apr 2020 16:46:22 +0000 (18:46 +0200)]
lib: do not provide hexdump in SPL

SPL should not be enlarged by building with CONFIG_HEXDUMP=y.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Tom Rini <trini@konsulko.com>
4 years agocommon: image_sign_info helper functions in SPL
Heinrich Schuchardt [Wed, 15 Apr 2020 16:46:21 +0000 (18:46 +0200)]
common: image_sign_info helper functions in SPL

Do not build image_sign_info helper functions in SPL if not needed.

Fixes: b983cc2da0ba ("lib: rsa: decouple rsa from FIT image verification")
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Tom Rini <trini@konsulko.com>
4 years agolib: do not build OID registry in SPL
Heinrich Schuchardt [Wed, 15 Apr 2020 16:46:19 +0000 (18:46 +0200)]
lib: do not build OID registry in SPL

The OID registry is only used by crypto functions that are not built in
SPL. So we should not build it in SPL.

Fixes: a9b45e6e8382 ("lib: add oid registry utility")
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Tom Rini <trini@konsulko.com>
4 years agotools: image-host.c: use correct output format
Heinrich Schuchardt [Sat, 11 Apr 2020 16:36:04 +0000 (18:36 +0200)]
tools: image-host.c: use correct output format

When building on a 32bit host the following warning occurs:

tools/image-host.c: In function ‘fit_image_read_data’:
tools/image-host.c:310:42: warning: format ‘%ld’ expects argument of type
‘long int’, but argument 3 has type ‘ssize_t’ {aka ‘int’} [-Wformat=]
   printf("Can't read all file %s (read %ld bytes, expexted %ld)\n",
                                        ~~^
                                        %d
          filename, n, sbuf.st_size);
                    ~

n is of type ssize_t so we should use %zd for printing.

Fixes: 7298e422504e ("mkimage: fit: add support to encrypt image with aes")
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
4 years agoMakefile: ARMv7-M has no CPSR register
Heinrich Schuchardt [Fri, 10 Apr 2020 09:10:32 +0000 (11:10 +0200)]
Makefile: ARMv7-M has no CPSR register

Compiling on ARMv7-M fails when trying to address the CPSR register which
is not available on this architecture.

Atomic functions refer to the CPSR register if compiled with
arch/arm/include/asm/proc-armv/system.h. On ARMv7-M we should hence
use arch/arm/thumb1/include/asm/proc-armv/system.h instead.

Cf. https://stackoverflow.com/questions/61097841/error-selected-processor-does-not-support-requested-special-purpose-register

Reported-by: Sicris Rey Embay <sicris.embay@gmail.com>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
4 years agoqemu: don't allow to select 32- and 64-bit
Heinrich Schuchardt [Fri, 10 Apr 2020 07:37:21 +0000 (09:37 +0200)]
qemu: don't allow to select 32- and 64-bit

TARGET_QEMU_ARM_64BIT and TARGET_QEMU_ARM_32BIT should be mutually
exclusive.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
4 years agofvp: Add support for loading Android boot images via semihosting
Peter Collingbourne [Sat, 4 Apr 2020 02:58:24 +0000 (19:58 -0700)]
fvp: Add support for loading Android boot images via semihosting

FVP now loads an Android boot image named boot.img if available,
otherwise it falls back to the existing code path.

Signed-off-by: Peter Collingbourne <pcc@google.com>
Reviewed-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
4 years agomkimage: fit_image: Add option to make fit header align
Kever Yang [Mon, 30 Mar 2020 03:56:24 +0000 (11:56 +0800)]
mkimage: fit_image: Add option to make fit header align

The image is usually stored in block device like emmc, SD card, make the
offset of image data aligned to block(512 byte) can avoid data copy
during boot process.
eg. SPL boot from FIT image with external data:
- SPL read the first block of FIT image, and then parse the header;
- SPL read image data separately;
- The first image offset is the base_offset which is the header size;
- The second image offset is just after the first image;
- If the offset of imge does not aligned, SPL will do memcpy;
The header size is a ramdon number, which is very possible not aligned, so
add '-B size'to specify the align size in hex for better performance.

example usage:
  ./tools/mkimage -E -f u-boot.its -B 0x200 u-boot.itb

Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Punit Agrawal <punit1.agrawal@toshiba.co.jp>
Reviewed-by: Tom Rini <trini@konsulko.com>
4 years agolibfdt: Make fdtdec_get_child_count() available for HOST
Kever Yang [Mon, 30 Mar 2020 03:56:23 +0000 (11:56 +0800)]
libfdt: Make fdtdec_get_child_count() available for HOST

The tool need to use fdtdec_get_child_count(), make it available for
HOST_CC.

Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Punit Agrawal <punit1.agrawal@toshiba.co.jp>
Reviewed-by: Simon Glass <sjg@chromium.org>
4 years agotool: use ALIGN() to align the size
Kever Yang [Mon, 30 Mar 2020 03:56:22 +0000 (11:56 +0800)]
tool: use ALIGN() to align the size

Use the ALIGN() for size align so that the code is more readable.

Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Punit Agrawal <punit1.agrawal@toshiba.co.jp>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
4 years agotools: imx8mimage: remove redundant code
Kever Yang [Mon, 30 Mar 2020 03:56:21 +0000 (11:56 +0800)]
tools: imx8mimage: remove redundant code

The align for fit_size has been done twice, remove the first one for it
does not make any sense.

Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Punit Agrawal <punit1.agrawal@toshiba.co.jp>
Reviewed-by: Tom Rini <trini@konsulko.com>
4 years agotools: kwbimage: use common ALIGN to do the size align
Kever Yang [Mon, 30 Mar 2020 03:56:20 +0000 (11:56 +0800)]
tools: kwbimage: use common ALIGN to do the size align

The ALIGN() is now available at imagetool.h, migrate to use it.

Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Punit Agrawal <punit1.agrawal@toshiba.co.jp>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Stefan Roese <sr@denx.de>
4 years agotools: mkimage: use common ALIGN to do the size align
Kever Yang [Mon, 30 Mar 2020 03:56:19 +0000 (11:56 +0800)]
tools: mkimage: use common ALIGN to do the size align

The ALIGN() is now available at imagetool.h, migrate to use it.

Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Punit Agrawal <punit1.agrawal@toshiba.co.jp>
Reviewed-by: Tom Rini <trini@konsulko.com>
4 years agotool: aisimage: use ALIGN instead of self defiend macro
Kever Yang [Mon, 30 Mar 2020 03:56:18 +0000 (11:56 +0800)]
tool: aisimage: use ALIGN instead of self defiend macro

The ALIGN() is available at imagetool.h, no need to self define one.

Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Punit Agrawal <punit1.agrawal@toshiba.co.jp>
Reviewed-by: Tom Rini <trini@konsulko.com>
4 years agotool: Move ALIGN_MASK to header as common MACRO
Kever Yang [Mon, 30 Mar 2020 03:56:17 +0000 (11:56 +0800)]
tool: Move ALIGN_MASK to header as common MACRO

The ALIGN code is need by many files who need handle structure or image
align, so move the macro to imagetool.h file.

Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Punit Agrawal <punit1.agrawal@toshiba.co.jp>
Reviewed-by: Tom Rini <trini@konsulko.com>
4 years agodocs: Fix conf.py for Sphinx 2.0
Jonathan Corbet [Fri, 20 Mar 2020 06:18:30 +0000 (02:18 -0400)]
docs: Fix conf.py for Sphinx 2.0

commit 3bc8088464712fdcb078eefb68837ccfcc413c88 upstream.

Our version check in Documentation/conf.py never envisioned a world where
Sphinx moved beyond 1.x.  Now that the unthinkable has happened, fix our
version check to handle higher version numbers correctly.

Signed-off-by: Jonathan Corbet <corbet@lwn.net>
[rebase for u-boot]
Signed-off-by: Sean Anderson <seanga2@gmail.com>
4 years agoenv/sf.c: drop private CMD_SAVEENV logic
Rasmus Villemoes [Wed, 19 Feb 2020 09:47:43 +0000 (09:47 +0000)]
env/sf.c: drop private CMD_SAVEENV logic

Deciding whether to compile the env_sf_save() function based solely on
CONFIG_SPL_BUILD is wrong: For U-Boot proper, it leads to a build
warning in case CONFIG_CMD_SAVEENV=n (because the env_save_ptr() macro
causes the function to indeed not be referenced anywhere). And for
SPL, when one selects CONFIG_SPL_SAVEENV, one obviously expects to
actually be able to save the environment.

Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
4 years agoenv/ext4.c: remove CONFIG_CMD_SAVEENV ifdef
Rasmus Villemoes [Wed, 19 Feb 2020 09:47:42 +0000 (09:47 +0000)]
env/ext4.c: remove CONFIG_CMD_SAVEENV ifdef

Removing this ifdef/endif pair yields a "defined but unused warning"
for CONFIG_CMD_SAVEENV=n, but that vanishes if we use the ENV_SAVE_PTR
macro instead. This gives slightly better compile testing, and
moreover, it's possible to have

  CONFIG_CMD_SAVEENV=n
  CONFIG_SPL_SAVEENV=y
  SPL_ENV_IS_IN_EXT4=y

in which case env_ext4_save would erroneously not be compiled in.

Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
4 years agoenv/fat.c: remove private CMD_SAVEENV logic
Rasmus Villemoes [Wed, 19 Feb 2020 09:47:41 +0000 (09:47 +0000)]
env/fat.c: remove private CMD_SAVEENV logic

Always compile the env_fat_save() function, and let
CONFIG_IS_ENABLED(SAVEENV) (via the ENV_SAVE_PTR macro) decide whether
it actually ends up being compiled in.

Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
4 years agoenv_internal.h: add alternative ENV_SAVE_PTR macro
Rasmus Villemoes [Wed, 19 Feb 2020 09:47:40 +0000 (09:47 +0000)]
env_internal.h: add alternative ENV_SAVE_PTR macro

The current definition of the env_save_ptr does not take SPL_SAVEENV
into account. Moreover, the way it is implemented means that drivers
need to guard the definitions of their _save methods with ifdefs to
avoid "defined but unused" warnings in case CMD_SAVEENV=n.

The ifdeffery can be avoided by using a "something ? x : NULL"
construction instead and still have the compiler elide the _save
method when it is not referenced. Unfortunately we can't just switch
the existing env_save_ptr macro, since that would give a lot of build
errors unless all the ifdeffery is removed at the same time.
Conversely, removing that ifdeffery first would merely lead to the
"defined but unused" warnings temporarily, but for some storage
drivers it requires a bit more work than just removing their private
CMD_SAVEENV logic.

So introduce an alternative to env_save_ptr, which for lack of a
better name is simply uppercased, allowing one to update storage
drivers piecemeal to both reduce their ifdeffery and honour
CONFIG_SPL_SAVEENV.

Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
4 years agoenv: add SAVEENV as an alias of the CMD_SAVEENV symbol
Rasmus Villemoes [Wed, 19 Feb 2020 09:47:39 +0000 (09:47 +0000)]
env: add SAVEENV as an alias of the CMD_SAVEENV symbol

Currently, testing whether to compile in support for saving the
environment is a bit awkward when one needs to take SPL_SAVEENV into
account, and quite a few storage drivers currently do not honour
SPL_SAVEENV.

To make it a bit easier to decide whether environment saving should be
enabled, introduce SAVEENV as an alias for the CMD_SAVEENV
symbol. Then one can simply use

  CONFIG_IS_ENABLED(SAVEENV)

Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
4 years agobootcount_ext: Add flag to enable/disable bootcount
Frédéric Danis [Tue, 17 Mar 2020 16:59:09 +0000 (17:59 +0100)]
bootcount_ext: Add flag to enable/disable bootcount

After a successful upgrade, multiple problem during boot sequence may
trigger the altbootcmd process.
This patch adds a version and an upgrade_available entries to the
bootcount file to enable/disable the bootcount check.
When failing to read the bootcount file it will consider that bootcount is
enabled, acting as previously, and update the file accordingly.

The bootcount file is only saved when `upgrade_available` is true, this
allows to save writes to the filesystem.

Signed-off-by: Frédéric Danis <frederic.danis@collabora.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
4 years agomake env_entry::callback conditional on !CONFIG_SPL_BUILD
Rasmus Villemoes [Thu, 27 Feb 2020 13:56:12 +0000 (13:56 +0000)]
make env_entry::callback conditional on !CONFIG_SPL_BUILD

The callback member of struct env_entry is always NULL for an SPL
build. Removing it thus saves a bit of run-time memory in the
SPL (when CONFIG_SPL_ENV_SUPPORT=y) since struct env_entry is embedded
in struct env_entry_node - i.e. about 2KB for the normal case of
512+change hash table entries.

Two small fixups are needed for this, all other references to the
callback member are already under !CONFIG_SPL_BUILD: Don't initialize
.callback in set_flags() - hsearch_r doesn't use that value
anyway. And make env_callback_init() initialize ->callback to NULL for
a new entry instead of relying on an unused or deleted entry having
NULL in ->callback.

Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Reviewed-by: Simon Glass <sjg@chromium.org>
4 years agolib/hashtable.c: don't test ->callback in SPL
Rasmus Villemoes [Thu, 27 Feb 2020 13:56:11 +0000 (13:56 +0000)]
lib/hashtable.c: don't test ->callback in SPL

In SPL, environment callbacks are not supported, so e->callback is
always NULL. Removing this makes the SPL a little smaller (about 400
bytes in my ppc build) with no functional change.

Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Reviewed-by: Simon Glass <sjg@chromium.org>
4 years agolib/hashtable.c: create helper for calling env_entry::callback
Rasmus Villemoes [Thu, 27 Feb 2020 13:56:11 +0000 (13:56 +0000)]
lib/hashtable.c: create helper for calling env_entry::callback

This is preparation for compiling out the "call the callback" code and
associated error handling for SPL, where ->callback is always NULL.

Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Reviewed-by: Simon Glass <sjg@chromium.org>
4 years agoenv: remove callback.o for an SPL build
Rasmus Villemoes [Thu, 27 Feb 2020 13:56:10 +0000 (13:56 +0000)]
env: remove callback.o for an SPL build

env.h says this about about callback declarations (U_BOOT_ENV_CALLBACK):

 * For SPL these are silently dropped to reduce code size, since environment
 * callbacks are not supported with SPL.

So env_callback_init() does a lot of work to not find anything in the
guaranteed empty env_clbk list. Drop callback.o entirely from the link
and stub out the only public function defined in callback.o. This cuts
about 600 bytes from the SPL on my ppc build.

Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Reviewed-by: Simon Glass <sjg@chromium.org>
4 years agoeth: mtk-eth: add mt7531 switch support in mediatek eth driver
Landen Chao [Tue, 18 Feb 2020 08:49:37 +0000 (16:49 +0800)]
eth: mtk-eth: add mt7531 switch support in mediatek eth driver

mt7531 is a 7-ports switch with 5 embedded giga phys, and uses the same
MAC design of mt7530. The cpu port6 supports SGMII only. The cpu port5
supports RGMII or SGMII in different model.

mt7531 is connected to mt7622 via both RGMII and SGMII interfaces.
In this patch, mt7531 cpu port5 or port6 is configured to maximum
capability to align CPU MAC setting.

The dts has been committed in the commit 6efa450565cdc ("arm: dts:
mediatek: add ethernet and sgmii dts node for mt7622")

Signed-off-by: Landen Chao <landen.chao@mediatek.com>
Tested-by: Frank Wunderlich <frank-w@public-files.de>
4 years agowatchdog: mpc8xx_wdt: Allow selection of watchdog mode through environment
Charles Frey [Wed, 19 Feb 2020 16:50:15 +0000 (16:50 +0000)]
watchdog: mpc8xx_wdt: Allow selection of watchdog mode through environment

The mpc8xx watchdog can work either in 'reset mode' or 'NMI mode'.
The selection can be done at startup only.
It is desirable to select the mode without rebuilding U-boot.
It is also desirable to disable the watchdog without rebuilding.

At watchdog startup, check environment variable 'watchdog_mode'.
If it is 'off', the watchdog is not started. If it is 'nmi',
the watchdog is started in NMI mode. Otherwise, it is started
in reset mode which is the default mode.

Signed-off-by: Charles Frey <charles.frey@c-s.fr>
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
4 years agoARM: dts: stm32: Split AV96 into DHCOR SoM and AV96 board
Marek Vasut [Tue, 31 Mar 2020 17:51:36 +0000 (19:51 +0200)]
ARM: dts: stm32: Split AV96 into DHCOR SoM and AV96 board

The AV96 is in fact an assembly of DH Electronics DHCOR SoM on top
of an AV96 reference board. Split the DTs to reflect that and make
sure to DHCOR SoM can be reused on other boards easily.

It is also highly recommended to configure the board for the DHCOM
make stm32mp15_dhcom_basic_defconfig
make DEVICE_TREE=stm32mp15xx-dhcor-avenger96
as that permits reusing the board code for the DH components, like
accessing and reading out the ethernet MAC from EEPROM.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Cc: Patrick Delaunay <patrick.delaunay@st.com>
Cc: Patrice Chotard <patrice.chotard@st.com>
Change-Id: I7db47280d4eb0d668eb4e006355240271154f97f

4 years agoARM: dts: stm32: Adjust PLL4 settings on AV96
Marek Vasut [Tue, 31 Mar 2020 17:51:35 +0000 (19:51 +0200)]
ARM: dts: stm32: Adjust PLL4 settings on AV96

The PLL4 is supplying SDMMC12, SDMMC3 and SPDIF with 120 MHz and
FDCAN with 96 MHz. This isn't good for the SDMMC interfaces, which
can not easily divide the clock down to e.g. 50 MHz for high speed
SD and eMMC devices, so those devices end up running at 30 MHz as
that is 120 MHz / 4. Adjust the PLL4 settings such that both PLL4P
and PLL4R run at 100 MHz instead, which is easy to divide to 50MHz
for optimal operation of both SD and eMMC, SPDIF clock are not that
much slower and FDCAN is also unaffected.

Reviewed-by: Patrice Chotard <patrice.chotard@st.com>
Reviewed-by: Patrick Delaunay <patrick.delaunay@st.com>
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Cc: Patrick Delaunay <patrick.delaunay@st.com>
Cc: Patrice Chotard <patrice.chotard@st.com>
Reviewed-by: Patrick Delaunay <patrick.delaunay@st.com>
4 years agoARM: dts: stm32mp1: remove file stm32mp157-pinctrl.dtsi
Patrick Delaunay [Tue, 21 Apr 2020 10:27:35 +0000 (12:27 +0200)]
ARM: dts: stm32mp1: remove file stm32mp157-pinctrl.dtsi

Remove the unnecessary file stm32mp157-pinctrl.dtsi and
solve all issues introduced by the commit 891483186052b2598 ("Merge branch
 'next'") after a conflict on the patch applied in the next branch in
commit 1a4f57c895cc ("ARM: dts: stm32mp1: DT alignment with Linux 5.6-rc1")

Need to reapplied on stm32mp15-pinctrl.dtsi the the 3 patches
applied previously on file "stm32mp157-pinctrl.dtsi" in v2020.04
- commit 4fdbe6487daa ("ARM: dts: stm32: Add alternate pinmux for SDMMC1
 direction pins")'
- commit 5fdcba64027f ("ARM: dts: stm32: Add alternate pinmux for SDMMC2
 pins 4-7")'
- commit 955de5111112 ("ARM: dts: stm32: Add alternate pinmux for ethernet
 RGMII")'

Cc: Marek Vasut <marex@denx.de>
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Reviewed-by: Patrice Chotard <patrice.chotard@st.com>
4 years agodm: pinctrl: Use right device pointer for configuring pinctrl
Lokesh Vutla [Wed, 22 Apr 2020 17:25:31 +0000 (22:55 +0530)]
dm: pinctrl: Use right device pointer for configuring pinctrl

commit 719cab6d2e2bf ("dm: pinctrl: convert pinctrl-single to livetree")
converted pinctrl driver to livetree. In this conversion, the call to
read pinctrl-single,pins/bits property is provided with pinctrl device
pointer instead of pinctrl config pointer. Because of this none of the
pins gets configured. Fix it by passing the right udevice pointer.

Fixes: 719cab6d2e2bf ("dm: pinctrl: convert pinctrl-single to livetree")
Reported-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Patrick Delaunay <patrick.delaunay@st.com>
Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com>
4 years agoMerge https://gitlab.denx.de/u-boot/custodians/u-boot-riscv
Tom Rini [Thu, 23 Apr 2020 12:24:47 +0000 (08:24 -0400)]
Merge https://gitlab.denx.de/u-boot/custodians/u-boot-riscv

- Adds few DT related fixes required for Linux EFI stub to work on
  RISC-V.
- Makes SBI v0.2 the default SBI version to work with OpenSBI v0.7.
- Revert "riscv: qemu: clear kernel-start/-end in device tree as
  workaround for BBL"
- Remove unnecessary CONFIG_IS_ENABLED().

4 years agoriscv: Move all fdt fixups together
Atish Patra [Tue, 21 Apr 2020 18:15:04 +0000 (11:15 -0700)]
riscv: Move all fdt fixups together

Keep all the fdt fixups together for better code management.

Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
4 years agoriscv: Copy the reserved-memory nodes to final DT
Atish Patra [Tue, 21 Apr 2020 18:15:03 +0000 (11:15 -0700)]
riscv: Copy the reserved-memory nodes to final DT

The DT used by U-Boot may be different from the DT being passed to
the OS if the DT is loaded from external media such as network or
mmc. In that case, the reserved-memory node needs to be copied to
the DT passed to the OS.

Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
4 years agoriscv: Setup reserved-memory node for FU540
Atish Patra [Tue, 21 Apr 2020 18:15:02 +0000 (11:15 -0700)]
riscv: Setup reserved-memory node for FU540

FU540 uses OF_SEPARATE instead of OF_PRIOR_STAGE.

Enable OF_BOARD_FIXUP to update the DT with reserved-memory node.

Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
4 years agoriscv: Provide a mechanism to fix DT for reserved memory
Atish Patra [Tue, 21 Apr 2020 18:15:01 +0000 (11:15 -0700)]
riscv: Provide a mechanism to fix DT for reserved memory

In RISC-V, M-mode software can reserve physical memory regions
by setting appropriate physical memory protection (PMP) csr. As the
PMP csr are accessible only in M-mode, S-mode U-Boot can not read
this configuration directly. However, M-mode software can pass this
information via reserved-memory node in device tree so that S-mode
software can access this information.

This patch provides a framework to copy to the reserved-memory node
from one DT to another. This will be used to update the DT used by
U-Boot and the DT passed to the next stage OS.

Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
4 years agofdtdec: Fix boundary check
Atish Patra [Tue, 21 Apr 2020 18:15:00 +0000 (11:15 -0700)]
fdtdec: Fix boundary check

In U-Boot, the reserved memory end address is considered as a inclusive
address. This notion is followed while adding a reserved memory node to
the DT.

For example:
end_address = start_address + size - 1

Follow the same notion and fix the end address computation while checking
for existing nodes.

Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
4 years agoriscv: Add boot hartid to device tree
Atish Patra [Tue, 21 Apr 2020 18:14:59 +0000 (11:14 -0700)]
riscv: Add boot hartid to device tree

Linux booting protocol mandates that register "a0" contains the hartid.
However, U-Boot can not pass the hartid via a0 during standard UEFI
protocol. DT nodes are commonly used to pass such information to the OS.

Add a DT node under chosen node to indicate the boot hartid. EFI stub
in Linux kernel will parse this node and pass it to the real kernel
in "a0" before jumping to it.

Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Rick Chen <rick@andestech.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
4 years agoriscv: Make SBI v0.2 the default SBI version
Bin Meng [Thu, 16 Apr 2020 15:09:33 +0000 (08:09 -0700)]
riscv: Make SBI v0.2 the default SBI version

To work with latest OpenSBI release (v0.7 or above) that has the HSM
extension support, select the SBI v0.2 support by default.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Atish Patra <atish.patra@wdc.com>
4 years agoriscv: Add Kconfig option for SBI v0.2
Bin Meng [Thu, 16 Apr 2020 15:09:32 +0000 (08:09 -0700)]
riscv: Add Kconfig option for SBI v0.2

SBI v0.2 is more scalable and extendable to handle future needs
for RISC-V supervisor interfaces. For example, with SBI v0.2 HSM
extension, only a single hart need to boot and enter operating
system. The booting hart can bring up secondary harts one by one
afterwards.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Atish Patra <atish.patra@wdc.com>
4 years agoriscv: Add SMP Kconfig option dependency for U-Boot proper
Bin Meng [Thu, 16 Apr 2020 15:09:31 +0000 (08:09 -0700)]
riscv: Add SMP Kconfig option dependency for U-Boot proper

U-Boot proper running in S-mode only need SMP support when using
SBI v0.1. With SBI v0.2 HSM extension, it does not need implement
multicore boot in U-Boot proper.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Atish Patra <atish.patra@wdc.com>
4 years agoriscv: Introduce SPL_SMP Kconfig option for U-Boot SPL
Bin Meng [Thu, 16 Apr 2020 15:09:30 +0000 (08:09 -0700)]
riscv: Introduce SPL_SMP Kconfig option for U-Boot SPL

With SBI v0.2 HSM extension, only a single hart need to boot and
enter operating system. The booting hart can bring up secondary
harts one by one afterwards.

For U-Boot running in SPL, SMP can be turned on, while in U-Boot
proper, SMP can be optionally turned off if using SBI v0.2 HSM.

Introduce a new SPL_SMP Kconfig option to support this.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Atish Patra <atish.patra@wdc.com>
4 years agoriscv: Merge unnecessary SMP ifdefs in start.S
Bin Meng [Thu, 16 Apr 2020 15:09:29 +0000 (08:09 -0700)]
riscv: Merge unnecessary SMP ifdefs in start.S

Two consecutive SMP ifdefs blocks can be combined into one.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Atish Patra <atish.patra@wdc.com>
4 years agoriscv: qemu: Remove the simple-bus driver for the SoC node
Bin Meng [Thu, 16 Apr 2020 15:09:28 +0000 (08:09 -0700)]
riscv: qemu: Remove the simple-bus driver for the SoC node

Prior to QEMU v3.1.0, QEMU generated the 'virt' SoC node with a
"riscv-virtio-soc" compatible string, and a "simple-bus" driver
was created to accommodate that special case in U-Boot.

Starting from QEMU v3.1.0, the SoC node was set as a "simple-bus",
hence the special simple-bus driver is no longer needed.

Update the doc to mention the latest tested QEMU version 4.2.0.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
4 years agoRevert "riscv: qemu: clear kernel-start/-end in device tree as workaround for BBL"
Lukas Auer [Tue, 14 Apr 2020 21:24:46 +0000 (23:24 +0200)]
Revert "riscv: qemu: clear kernel-start/-end in device tree as workaround for BBL"

The commit was added as a workaround required in QEMU when using BBL as
the supervisor binary interface (SBI) for Linux. We are now using
OpenSBI to provide the SBI, the workaround is therefore not required
anymore and can be removed.

This reverts commit 897206c5cc5c6ac0dc2ab851044e42baada3785b.

Signed-off-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
4 years agoriscv: ax25: cache: Remove SPL_RISCV_MMODE config check
Pragnesh Patel [Sat, 14 Mar 2020 13:42:28 +0000 (19:12 +0530)]
riscv: ax25: cache: Remove SPL_RISCV_MMODE config check

CONFIG_IS_ENABLED(FOO) will check FOO config option for U-Boot,
SPL and TPL, so remove unnecessary CONFIG_IS_ENABLED()

Signed-off-by: Pragnesh Patel <pragnesh.patel@sifive.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
4 years agoMerge https://gitlab.denx.de/u-boot/custodians/u-boot-marvell
Tom Rini [Wed, 22 Apr 2020 17:00:21 +0000 (13:00 -0400)]
Merge https://gitlab.denx.de/u-boot/custodians/u-boot-marvell

- mvebu bubt cmd: Add A38x support (Joel)
- Clearfog: Fix SCSI boot duplication (Joel)
- Armada-37xx: Fix DDR PHY clock divider values (Marek)

4 years agoMerge tag 'mmc-2020-4-22' of https://gitlab.denx.de/u-boot/custodians/u-boot-mmc
Tom Rini [Wed, 22 Apr 2020 12:58:41 +0000 (08:58 -0400)]
Merge tag 'mmc-2020-4-22' of https://gitlab.denx.de/u-boot/custodians/u-boot-mmc

- iproc_sdhci memory leak fix and enable R1B resp quirk
- more mmc cmds and several mmc updates from Heinirich
- Use bounce buffer for tmio sdhci
- Alignment check for tmio sdhci

4 years agodrivers: mmc: rpmb: do not build for SPL
Heinrich Schuchardt [Wed, 15 Apr 2020 16:28:09 +0000 (18:28 +0200)]
drivers: mmc: rpmb: do not build for SPL

RPMB support is used by the 'mmc rpmb' command and by the OP-TEE support.
We do not need it in SPL.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
4 years agospl: mmc: Rename spl_boot_partition() to spl_mmc_boot_partition()
Harald Seiler [Wed, 15 Apr 2020 09:33:31 +0000 (11:33 +0200)]
spl: mmc: Rename spl_boot_partition() to spl_mmc_boot_partition()

This function is only relevant to the MMC driver so calling it
spl_boot_partition() might be confusing.  Rename it to
spl_mmc_boot_partition() to make its purpose more clear (and bring
it in line with spl_mmc_boot_mode()).

Signed-off-by: Harald Seiler <hws@denx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
4 years agospl: mmc: Rename spl_boot_mode() to spl_mmc_boot_mode()
Harald Seiler [Wed, 15 Apr 2020 09:33:30 +0000 (11:33 +0200)]
spl: mmc: Rename spl_boot_mode() to spl_mmc_boot_mode()

The function's name is misleading as one might think it is used
generally to select the boot-mode when in reality it is only used by the
MMC driver to find out in what way it should try reading U-Boot Proper
from a device (either using a filesystem, a raw sector/partition, or an
eMMC boot partition).

Rename it to spl_mmc_boot_mode() to make it more obvious what this
function is about.

Link: https://lists.denx.de/pipermail/u-boot/2020-April/405979.html
Signed-off-by: Harald Seiler <hws@denx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
4 years agommc: tmio: sdhi: Implement get_b_max function
Marek Vasut [Sat, 4 Apr 2020 10:45:06 +0000 (12:45 +0200)]
mmc: tmio: sdhi: Implement get_b_max function

Implement get_b_max() for the Renesas R-Car SDHI controller driver, limit
the b_max per hardware capabilities such that select Gen2 controllers have
16bit block transfer limit, the rest has 32bit block transfer limit and on
Gen3, the block transfer limit on addresses above the 32bit boundary is set
to 1/4 of the malloc area.

Originally, on Gen3, the block transfers above the 32bit area were limited
to PIO only, which resulted in (R8A7795 Salvator-X , HS200 eMMC):
  => time mmc read 0x0000000700000000 0 0x10000
  time: 0.151 seconds
  => time mmc read 0x0000000700000000 0 0x100000
  time: 11.090 seconds
with bounce buffer in place and b_max adjustment in place:
  => time mmc read 0x0000000700000000 0 0x10000
  time: 0.156 seconds
  => time mmc read 0x0000000700000000 0 0x100000
  time: 2.349 seconds

Note that the bounce buffer does mallocate and free the bounce buffer
for every transfer. Experiment which removes this results in further
increase of read speed, from 2.349s to 2.156s per 512 MiB of data,
which is not such a significant improvement anymore. It might however
be interesting to have bounce buffer directly in the MMC core or even
block core.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Cc: Peng Fan <peng.fan@nxp.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Tom Rini <trini@konsulko.com>
4 years agommc: Add option to adjust b_max before long read
Marek Vasut [Sat, 4 Apr 2020 10:45:05 +0000 (12:45 +0200)]
mmc: Add option to adjust b_max before long read

Add getter function which permits adjusting the maximum number of
blocks that could be read in a single sustained read transfer based
on the location of the source/target buffer and length, before such
transfer starts.

This is mainly useful on systems which have various DMA restrictions
for different memory locations, e.g. DMA limited to 32bit addresses,
and where a bounce buffer is used to work around such restrictions.
Since the U-Boot bounce buffer is mallocated, it's size is limited
by the malloc area size, and the read transfer to such a buffer must
also be limited. However, as not all areas are limited equally, the
b_max should be adjusted accordinly as needed to avoid degrading
performance unnecessarily.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Cc: Peng Fan <peng.fan@nxp.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Tom Rini <trini@konsulko.com>
4 years agommc: tmio: sdhi: Use bounce buffer to avoid DMA limitations
Marek Vasut [Sat, 4 Apr 2020 10:45:04 +0000 (12:45 +0200)]
mmc: tmio: sdhi: Use bounce buffer to avoid DMA limitations

The R-Car SDHI DMA controller has various restrictions. To work around
those restrictions without falling back to PIO, implement bounce buffer
with custom alignment check function which tests for those limitations.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Cc: Peng Fan <peng.fan@nxp.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Tom Rini <trini@konsulko.com>
4 years agoARM: rmobile: Increase malloc area size
Marek Vasut [Sat, 4 Apr 2020 10:45:03 +0000 (12:45 +0200)]
ARM: rmobile: Increase malloc area size

Increase the malloc area size significantly to cater for bounce buffer
used by the SDHI driver.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Cc: Peng Fan <peng.fan@nxp.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Tom Rini <trini@konsulko.com>
4 years agocommon: bouncebuf: Permit passing custom alignment check function
Marek Vasut [Sat, 4 Apr 2020 10:45:02 +0000 (12:45 +0200)]
common: bouncebuf: Permit passing custom alignment check function

Add extended version of the bounce_buffer_start(), which permits passing in
a custom alignment checker function for the buffer. This is useful e.g. on
systems with various DMA restrictions and where the checker function might
be more complex than a simple CPU cache alignment check.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Cc: Peng Fan <peng.fan@nxp.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Tom Rini <trini@konsulko.com>
4 years agommc: adjust Kconfig for mmc sub-commands
Heinrich Schuchardt [Tue, 31 Mar 2020 17:39:28 +0000 (17:39 +0000)]
mmc: adjust Kconfig for mmc sub-commands

All sub-commands of the mmc command should be shown in the Kconfig menu
next to the mmc command. This includes:

* mmc bkops
* mmc rpmb
* mmc swrite

The mmc rpmb sub-command is not usable without CONFIG_SUPPORT_EMMC_RPMB.
Add the missing dependency.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
4 years agodrivers: mmc: iproc_sdhci: move host.mmc init before sdhci_setup_cfg
Rayagonda Kokatanur [Tue, 31 Mar 2020 05:34:06 +0000 (11:04 +0530)]
drivers: mmc: iproc_sdhci: move host.mmc init before sdhci_setup_cfg

move host.mmc before sdhci_setup_cfg

Signed-off-by: Rayagonda Kokatanur <rayagonda.kokatanur@broadcom.com>
Signed-off-by: Bharat Kumar Reddy Gooty <bharat.gooty@broadcom.com>
4 years agodrivers: mmc: iproc_sdhci: fix compilation warning
Rayagonda Kokatanur [Tue, 31 Mar 2020 05:34:05 +0000 (11:04 +0530)]
drivers: mmc: iproc_sdhci: fix compilation warning

set_ios_post return type changed from void to int, correcting
the same to fix compilation warning.

Signed-off-by: Rayagonda Kokatanur <rayagonda.kokatanur@broadcom.com>
Signed-off-by: Bharat Kumar Reddy Gooty <bharat.gooty@broadcom.com>
4 years agodrivers: mmc: iproc_sdhci: enable broken R1B response quirk
Bharat Kumar Reddy Gooty [Tue, 31 Mar 2020 05:34:04 +0000 (11:04 +0530)]
drivers: mmc: iproc_sdhci: enable broken R1B response quirk

Enable SDHCI_QUIRK_BROKEN_R1B quirk.

Signed-off-by: Bharat Kumar Reddy Gooty <bharat.gooty@broadcom.com>
Signed-off-by: Rayagonda Kokatanur <rayagonda.kokatanur@broadcom.com>
4 years agodrivers: mmc: iproc_sdhci: fix possible memory leak
Bharat Kumar Reddy Gooty [Tue, 31 Mar 2020 05:34:03 +0000 (11:04 +0530)]
drivers: mmc: iproc_sdhci: fix possible memory leak

Free the pointer variable 'iproc_sdhci' upon failure to fix
possible memory leak.

Signed-off-by: Bharat Kumar Reddy Gooty <bharat.gooty@broadcom.com>
Signed-off-by: Rayagonda Kokatanur <rayagonda.kokatanur@broadcom.com>
4 years agocmd: mmc: provide boot area protection command
Heinrich Schuchardt [Mon, 30 Mar 2020 05:24:19 +0000 (07:24 +0200)]
cmd: mmc: provide boot area protection command

Provide command 'mmc wp' to power on write protect boot areas on eMMC
devices.

The B_PWR_WP_EN bit in the extended CSD register BOOT_WP is set. The boot
area are write protected until the next power cycle occurs.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
4 years agocmd: mmc: display write protect state of boot partition
Heinrich Schuchardt [Mon, 30 Mar 2020 05:24:18 +0000 (07:24 +0200)]
cmd: mmc: display write protect state of boot partition

Boot partitions of eMMC devices can be power on or permanently write
protected. Let the 'mmc info' command display the protection state.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
4 years agommc: export mmc_send_ext_csd()
Heinrich Schuchardt [Mon, 30 Mar 2020 05:24:17 +0000 (07:24 +0200)]
mmc: export mmc_send_ext_csd()

Export function mmc_send_ext_csd() for reading the extended CSD register.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
4 years agommc: EXT_CSD registers for write protection
Heinrich Schuchardt [Mon, 30 Mar 2020 05:24:16 +0000 (07:24 +0200)]
mmc: EXT_CSD registers for write protection

Add the EXT_CSD register definition related to write protection.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
4 years agocmd: mvebu: bubt: show image boot device
Joel Johnson [Fri, 17 Apr 2020 15:38:08 +0000 (09:38 -0600)]
cmd: mvebu: bubt: show image boot device

When a mismatch is found trying to write an image for one boot method
to a different boot device, print an error message including the image
header marked target boot device type.

Signed-off-by: Joel Johnson <mrjoel@lixil.net>
Reviewed-by: Stefan Roese <sr@denx.de>
4 years agoarm: mvebu: clearfog: enable bubt command
Joel Johnson [Fri, 17 Apr 2020 15:38:07 +0000 (09:38 -0600)]
arm: mvebu: clearfog: enable bubt command

With support added for Armada 38x, include the bubt command in
ClearFog defconfig.

Signed-off-by: Joel Johnson <mrjoel@lixil.net>
Reviewed-by: Stefan Roese <sr@denx.de>
4 years agocmd: mvebu: bubt: verify A38x target device type
Joel Johnson [Fri, 17 Apr 2020 15:38:06 +0000 (09:38 -0600)]
cmd: mvebu: bubt: verify A38x target device type

Ensure that the device to which an image is being written includes
header information indicating boot support for the destination
device.

This is derived from the support in the SolidRun master-a38x vendor
fork.

Signed-off-by: Joel Johnson <mrjoel@lixil.net>
Reviewed-by: Stefan Roese <sr@denx.de>
4 years agocmd: mvebu: bubt: correct U-Boot spelling
Joel Johnson [Fri, 17 Apr 2020 15:38:05 +0000 (09:38 -0600)]
cmd: mvebu: bubt: correct U-Boot spelling

Replace "U-BOOT" text with correct spelling

Signed-off-by: Joel Johnson <mrjoel@lixil.net>
Reviewed-by: Stefan Roese <sr@denx.de>
4 years agocmd: mvebu: bubt: add A38x support
Joel Johnson [Fri, 17 Apr 2020 15:38:04 +0000 (09:38 -0600)]
cmd: mvebu: bubt: add A38x support

Add support for Armada 38x devices in bubt flashing utility.
This is based on (and streamlined from) the support in the SolidRun
master-a38x vendor fork.

Signed-off-by: Joel Johnson <mrjoel@lixil.net>
Reviewed-by: Stefan Roese <sr@denx.de>
4 years agoarm: mvebu: correct SPL boot configs for SPI/MMC
Joel Johnson [Fri, 17 Apr 2020 07:19:05 +0000 (01:19 -0600)]
arm: mvebu: correct SPL boot configs for SPI/MMC

Update mvebu SPL boot selection mechanism for the move to driver model
usage by ensuring that the required driver support for SPI and MMC
booting is available in SPL when the respective boot method is
selected.

Previously, all mvebu boards selected a boot method (implicitly
MVEBU_SPL_BOOT_DEVICE_SPI for many) even if SPL booting wasn't used.
This changes mvebu boot method selection to depend on SPL usage which
resolves the issue with aarch64 boards which don't use SPL getting an
implicit boot device selection resulting in unmet dependencies. The
32-bit arm boards do use SPL, but I'm led to conclude that most aren't
intentionally using the MVEBU_SPL_BOOT_DEVICE selection since none have
SPL_DM_SPI enabled in their defconfig even though they still implicitly
select the SPI boot method.

This also results in the new addition of SPL_GPIO_SUPPORT to helios4.
The mainline dts for helios4 includes the cd-gpios entry for sdhci with
identical addresses as the clearfog dts. I don't have a helios4 board
to confirm, but based on the current source conclude that the board
itself is either wired to pull the signal low for eMMC, or the default
MMC boot isn't fully functional in mainline. In either case, as far as
I can tell, including the GPIO support will at least cause no
regression.

Tested on SolidRun ClearFog devices.

Signed-off-by: Joel Johnson <mrjoel@lixil.net>
Reviewed-by: Stefan Roese <sr@denx.de>
4 years agoarm: mvebu: clearfog: adjust SCSI boot duplication
Joel Johnson [Fri, 17 Apr 2020 07:06:53 +0000 (01:06 -0600)]
arm: mvebu: clearfog: adjust SCSI boot duplication

Fix duplication resulting from merging of multiple related series.
Commits cecf38a75bd02fd29f, and 201a500de added or adjusted SCSI
boot support for ClearFog, but in slightly different locations which
didn't result in a merge conflict.

Signed-off-by: Joel Johnson <mrjoel@lixil.net>
Reviewed-by: Stefan Roese <sr@denx.de>
4 years agoclk: armada-37xx-periph: fix DDR PHY clock divider values
Marek Behún [Tue, 14 Apr 2020 22:59:18 +0000 (00:59 +0200)]
clk: armada-37xx-periph: fix DDR PHY clock divider values

Register value table for DDR PHY clock divider are wrong. They should be
0 or 1 for divide-by-2 or divide-by-4, respectively. Not 1 or 2. Current
values do not make sense, since 2 cannot be achieved, because the
register is only 1 bit long (mask is set to 1).

This fixes clk dump reporting DDR PHY clock rate differently from Linux.

Signed-off-by: Marek Behún <marek.behun@nic.cz>
Reviewed-by: Stefan Roese <sr@denx.de>
4 years agoMerge tag 'dm-pull-21apr20' of git://git.denx.de/u-boot-dm
Tom Rini [Tue, 21 Apr 2020 21:53:23 +0000 (17:53 -0400)]
Merge tag 'dm-pull-21apr20' of git://git.denx.de/u-boot-dm

Various improvements to buildman summary output

4 years agoMerge https://gitlab.denx.de/u-boot/custodians/u-boot-fsl-qoriq
Tom Rini [Tue, 21 Apr 2020 19:20:42 +0000 (15:20 -0400)]
Merge https://gitlab.denx.de/u-boot/custodians/u-boot-fsl-qoriq

- Backplane support and bug fixes

4 years agobuildman: Change the exit codes
Simon Glass [Thu, 9 Apr 2020 16:49:45 +0000 (10:49 -0600)]
buildman: Change the exit codes

The current exit codes of 128 and 129 are useful in that they do not
conflict with those returned by tools, but they are not actually valid.
It seems better to pick some codes which work with 'bit bisect run'.

Update them to 100 (for errors) and 101 (for warnings).

Signed-off-by: Simon Glass <sjg@chromium.org>
4 years agobuildman: Update the TODO items
Simon Glass [Thu, 9 Apr 2020 21:08:54 +0000 (15:08 -0600)]
buildman: Update the TODO items

A few of these have been done. Drop those and add some new ideas.

Signed-off-by: Simon Glass <sjg@chromium.org>
4 years agobuildman: Add an option to ignore migration warnings
Simon Glass [Thu, 9 Apr 2020 21:08:53 +0000 (15:08 -0600)]
buildman: Add an option to ignore migration warnings

These are becoming more common now. They cause boards to show warnings
which can be mistaking for compiler warnings.

Add a buildman option to ignore them. This option works only with the
summary option (-s). It does not affect the build process.

Signed-off-by: Simon Glass <sjg@chromium.org>