The assembly directive ".rept ... .endr" allows us to write the
init_page_table much shorter. To make things further simpler,
set the text and stack area as Normal Memory, and the other sections
as Device attribute.
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
The DDR PHY training function, ddrphy_prepare_training() would not
work if compiled with GCC 4.9.
The struct ddrphy (arch/arm/include/asm/arch-uniphier/ddrphy-regs.h)
is specified with __packed because it represents a hardware register
mapping, but it turned out to cause a problem on GCC 4.9.
If -mno-unaligned-access is specified (yes, it is in
arch/arm/cpu/armv7/config.mk), GCC 4.9 is aware of the
__attribute__((packed)) and generates extra instructions to perform
the memory access in a way that does not cause unaligned access.
(Actually it is not need here because the register base, the first
argument of the ddrphy_prepare_training(), is always given with a
4-byte aligned address.)
Anyway, as a result, readl() / writel() is divided into byte-wise
accesses. The problem is that this hardware only accepts 4-byte
register access. Byte-wise accesses lead to unexpected behavior.
There are some options to avoid this problem.
[1] Remove -mno-unaligned-access
[2] Add __aligned(4) along with __packed to struct ddrphy
[3] Remove __packed from struct ddrphy
[1] solves the problem for ARMv7, but it does not for pre-ARMv6 and
ARMv6-M architectures where -mno-unaligned-access is default.
So, [1] does not seem reasonable in terms of code portability.
Both [2] and [3] work well, but [2] seems too much. All the members
of struct ddrphy have the u32 type. No padding would be inserted
even if __packed is dropped.
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Reviewed-by: Tom Rini <trini@ti.com>
2 recent sunxi changes have removed the usage of lowlevel_init by moving some
code around and then setting CONFIG_SKIP_LOWLEVEL_INIT.
This is problematic for 2 reasons:
1) It does not just stop s_init from being called, it also stops
cpu_init_cp15 from getting called, which is undesirable.
2) We want u-boot.bin to be usable standalone, without SPL, some people e.g.
use an upstream u-boot.bin together with Allwinner's boot0 loader. So
u-boot.bin must (re)initialize the gpios, timer, etc.
This commit restores the lowlevel_init / s_init usage, while keeping the
changes to no longer use the global-data (gd) struct in the SPL.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
The global_data pointer (gd) has already been set before board_init_f()
is called. We should not assign it again. We should also not use gdata since
it is going away.
Signed-off-by: Simon Glass <sjg@chromium.org>
This does nothing now, so drop it. We have SPL anyway to do our low-level
init.
Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
The current sunxi implementation uses gdata, which is going away. It also
sets up DRAM before board_init_f() in SPL.
There is really no reason to do much in s_init() since board_init_f() is
called immediately afterwards. The only change is that we need our own
implementation of board_init_f() which sets up DRAM before the BSS (which
is in DRAM) is cleared.
The s_init() code runs once for SPL and again for U-Boot proper. We
shouldn't need to init the clock/timer/gpio/i2c init twice, so just have it
in SPL.
Signed-off-by: Simon Glass <sjg@chromium.org>
We need to get rid of this SPL-specific setting of the global_data pointer.
It is already set up in start.S immediately before board_init_f() is called,
and there may be information there that is needed (e.g. pre-reloc malloc
info).
Signed-off-by: Simon Glass <sjg@chromium.org>
Prior to this change we set the gd pointer early so that we can store
data in it. This becomes problematic for DM changes as well as being
odd in general. Re-work the code paths so that we don't need to set the
gd pointer so early and instead can rely upon the normal setting of it.
In order to do this we do need to move certain calls from s_init into
spl_board_init(), mainly preloader_console_init and
save_omap_boot_params.
Tested on: Beaglebone Black, AM43xx GP EVM, Beagleboard, Beagleboard xM,
OMAP5 uEVM, DRA7xx EVM
Signed-off-by: Tom Rini <trini@ti.com>
Tested-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
All the 74xx_7xx boards are still non-generic boards:
P3G4, ZUMA, ppmc7xx, ELPPC, mpc7448hpc2
Acked-by: Marek Vasut <marex@denx.de>
Acked-by: Stefan Roese <sr@denx.de>
Acked-by: York Sun <yorksun@freescale.com>
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Nye Liu <nyet@zumanetworks.com>
Cc: Roy Zang <tie-fei.zang@freescale.com>
Now TQM8xx is the only remaining board family of mpc8xx.
It uses its own linker script, board/tqc/tqm8xx/u-boot.lds.
arch/powerpc/cpu/mpc8xx/u-boot.lds is not used by any boards.
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Wolfgang Denk <wd@denx.de>
Since commit 843125daeb (ppc4xx: remove HH405 board), CONFIG_HH405
is not defined.
Since commit d526330479 (ppc4xx: remove PMC405), CONFIG_PMC405
is not defined.
Acked-by: Stefan Roese <sr@denx.de>
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Matthias Fuchs <matthias.fuchs@esd.eu>
As a preparation to ARCv2 port submission we rename "arc700" folder to
"arcv1" which stands for ARCv1 ISA also known as ARCompact.
This will allow us to add more flavours of binary-compatible ARCv1 CPUs
like ARC600 if needed later on and all required ARCv2 CPUs (which are
binary incompatible with ARCv1) in "arcv2" folder in subsequent commits.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Igor Guryanov <guryanov@synopsys.com>
Both ARCangel4 and AXS10x are FPGA-based boards so they may have
different CPUs. For now we have only 1 option (ARC700) and we define
this as default in arch Kconfig.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
"reset.c" and "cpu.c" have no architecture-specific code at all.
Others are applicable to either ARC CPU.
This change is a preparation to submission of ARCv2 architecture port.
Even though ARCv1 and ARCv2 ISAs are not binary compatible most of
built-in modules still have the same programming model - AUX registers
are mapped in the same addresses and hold the same data (new featues
extend existing ones).
So only low-level assembly code (start-up, interrupt handlers) is left
as CPU(actually ISA)-specific. This significantyl simplifies maintenance
of multiple CPUs/ISAs.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Igor Guryanov <guryanov@synopsys.com>
* use better symbols for relocatable region boundaries
("__image_copy_start" instead of "CONFIG_SYS_TEXT_BASE")
* remove useless debug messages because they will only show up in case
of both problem (when normal "if" branch won't be taken) and DEBUG take
place which is pretty rare situation.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Even though existing implementation works fine in preparation to
submission of ARCv2 architecture we need this change.
In case of ARCv2 interrupt vector table consists of just addresses
of corresponding handlers. And if those addresses will be in .text
section then assembler will encode them as everything in .text section
as middle-endian and then on real execution CPU will read swapped
addresses and will jump into the wild.
Once introduced new section is situated so .text section remains the
first which allows us to use common linker option for linking everything
to a specified CONFIG_SYS_TEXT_BASE.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Igor Guryanov <guryanov@synopsys.com>
Depending on MMU presence in CPU there're differences in HW behavior.
For example address of instruction that caused exception is put in
ECR register if MMU exists and in ERET register otherwise.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Igor Guryanov <guryanov@synopsys.com>
To disable interrupts we need to reset corresponding flags in STATUS32
register. For this we need to OR flags for interrupts level1 and level2
and then AND with current value in STATUS32.
Before that implementation was incorrect.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Igor Guryanov <guryanov@synopsys.com>
Exception cause register (ECR) contains value that describes a reason
for exception that has happened. This helps a lot to figure-out what
went wrong.
Now we print this register contents when dumping registers.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Igor Guryanov <guryanov@synopsys.com>
Some cache operations ({i|d}cache_{enable|disable|status} or
flush_dcache_all) are built and used even if CONFIG_SYS_{I|D}CACHE_OFF
is set.
This is required for force disable of caches on early boot.
What if something was executed before U-boot and enabled caches
(low-level bootloaders, previously run kernel etc.)?
But if CPU doesn't really have caches any attempt to access
cache-related AUX registers triggers instruction error exception.
So for convenience we'll try to avoid exceptions by checking if CPU
actually has caches (we check separately data and instruction cache
existence) at all.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Igor Guryanov <guryanov@synopsys.com>
The axp221 / axp223's N_VBUSEN pin can be configured as an output rather
then an input, and this is used on some boards to control usb-vbus0, add
support for this.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Most of the usb-controller init code found in ehci-sunxi.c also is necessary
to init the otg usb controller, so move it to a common place.
While at it also update various #ifdefs / defines for sun8i support.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
"adr rX, text_end" only works if the label is close. Adding further code
to the other functions will prevent this. So move the containing
function close to label. No functional change.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Based on the original version by Marc Zyngier. It adds a psci_cpu_off
implementation for the A20 SoC. The mechanism works by first preparing
the calling CPU to go offline (disable and flush cache, disable SMP),
then requesting CPU 0 to pull the plug. The request is sent as FIQ on
SGI15.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Add support for VGA directly from the sunxi SoC / display engine.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Add lcd output support, see the new Kconfig entries and doc/README.video for
how to enable / configure this.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Modify sunxi_lcdc_pll_set to work with both tcon0 and tcon1, this is a
preparation patch for adding lcd support.
While at it also swap the divider search order, searching from low to
high, as the comment above the code says we should do. In cases where there
are multiple solutions this will result in picking a lower pll clock and
divider, which is more stable and saves power.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Some boards use GPIO-s on the pmic, one example of this is the A13-OLinuXino
board, which uses gpio0 of the axp209 for the lcd-power signal.
This commit adds support for gpio pins on the AXP209 pmic, the sunxi_gpio.c
changes are universal, adding gpio support for the other AXP pmics (when
necessary) should be a matter of adding the necessary axp_gpio_foo functions
to their resp. drivers, and add "#define AXP_GPIO" to their header file.
Note this commit only adds support for the non device-model version of the
gpio code, patches for adding support to the device-model version are very
welcome.
The string representation for these gpio-s is AXP0-#, the 0 in the AXP0 prefix
is there in case we need to support gpio-s on more then 1 pmic in the future.
At least A80 boards have 2 pmics, and we may end up needing to support gpio-s
on both.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Move a few mux defines around so that all the mux defines are properly sorted
by port number.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Add a write to the "unknown" (*) register to enable auto input sync, when
initially adding sunxi hdmi output support this magic write from the android
kernel code was missed, causing lcdc -> hdmi encoder sync problems.
With this write added, we can drop the modesetting retries and the extra
delays added to work around these sync problems.
With the retries dropped there also is no need to 0 all the enable flags at
the beginning of the modeset, as they are initialized to 0 already by
engines_init.
*) "unknown" is the actual name of this register in the android kernel sources
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Acked-by: Anatolij Gustschin <agust@denx.de>
So far we've been programming the hdmi-encoder to send out dvi data over the
hdmi connector. This works well for most devices, including hdmi devices, but
not all devices accept dvi data on a hdmi input.
Add support for sending proper hdmi data over the hdmi output found on most
sunxi boards. This can be turned on by adding monitor=hdmi as option to the
video-mode env. variable.
A follow up patch will determine whether to send dvi or hdmi automatically when
EDID is used.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Acked-by: Anatolij Gustschin <agust@denx.de>
Add DDC & EDID support and use it to automatically select the native mode of
the attached monitor. This can be disabled by adding edid=0 as option
to the video-mode env. variable.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Acked-by: Anatolij Gustschin <agust@denx.de>
PLL1 on sun6i / sun8i also has a p factor which divides the clock by
2^p (to the power p). On sun6i the p factor is ignored, but on sun8i it is
used and we were setting it to 1, resulting in the CPU running at 504 MHz
instead of 1008 MHz, this commit fixes this.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
After reboot, reset or even short power off, DRAM typically retains
the old stale data for some period of time (for this type of memory,
the bits of data are stored in slowly discharging capacitors).
The current sun6i/sun8i DRAM size detection logic, which is
inherited from the Allwinner code, relies on using a large magic
signature with the hope that it is unique enough and unlikely to
ever accidentally match this leftover garbage data in RAM. But
this approach is inherently unsafe, as can be demonstrated using
the following test program:
/***** A testcase for reproducing the problem ******/
void main(int argc, char *argv[])
{
size_t size, i;
uint32_t *buf;
/* Allocate the buffer */
if (argc < 2 || !(size = (size_t)atoi(argv[1]) * 1048576) ||
!(buf = malloc(size))) {
printf("Need buffer size in MiB as a cmdline argument\n");
exit(1);
}
/* Fill it with the Allwinner DRAM "magic" values */
for (i = 0; i < size / 4; i++)
buf[i] = 0xaa55aa55 + ((uintptr_t)&buf[i] / 4) % 64;
/* Try to reboot */
system("reboot");
/* And wait */
for (;;) {}
}
/***************************************************/
If this test program is run on the device (giving it a large
chunk of memory), then the DRAM size detection logic in u-boot
gets confused after reboot and fails to initialize DRAM properly.
A better approach is not to rely on luck and abstain from making
any assumptions about the properties of the leftover garbage
data in RAM. Instead just use a more reliable code for testing
whether two different addresses refer to the same memory location.
Signed-off-by: Siarhei Siamashka <siarhei.siamashka@gmail.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Based on the register / dram_para headers from the Allwinner u-boot / linux
sources + the init sequences from boot0.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
The sun8i boot0 code fills the DRAM with a "random" pattern before comparing
it at different offsets to do columns, etc. detection. The sun6i boot0 code
does not do it, instead relying on the memory contents being random enough
to begin with for the memcmp to properly detect the wrap-around address, iow
it is working purely by chance. Since our sun6i dram code was modelled after
the boot0 code it contained the same issue.
This commit fixes this by filling the memory with a unique, distinct pattern.
The new mctl_mem_fill function this introduces is added as an inline helper
in dram.h, so that it can be shared with the sun8i dram code.
While at it move mctl_mem_matches to dram.h for re-use in sun8i too.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
The await_completion helper is already copy pasted between the sun4i and sun6i
dram code, and we need it for sun8i too, so lets make it an inline helper in
dram.h, rather then adding yet another copy.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Our old hardcoded k and m values are based on PLL5 being configured in steps
of 48 MHz, which is correct for sun6i where the DRAM PLL runs at twice the
DRAM CLK, which is usually configured in 24 MHz step. But on the A23 (sun8i)
the PLL5 runs at half the DRAM CLK, so we require 12 MHz steps.
This commit adjusts clock_set_pll5 to automatically select the best k and m
depending on the requested clk rate.
Suggested-by: Siarhei Siamashka <siarhei.siamashka@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
The sun8i dram code sometimes wants to enable sigma delta mode,
add a parameter to allow this.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
