At this point, the remaining places where we have a symbol that is
defined as CONFIG_... are in fairly odd locations. While as much dead
code has been removed as possible, some of these locations are simply
less obvious at first. In other cases, this code is used, but was
defined in such a way as to have been missed by earlier checks. Perform
a rename of all such remaining symbols to be CFG_... rather than
CONFIG_...
Signed-off-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
This reverts commit 65ba7add0d.
A weak extern is a nasty sight to behold: If the symbol is never
defined, on ARM, the linker will replace the function call with a NOP.
This behavior isn't well documented but there are at least some hints
to it [1].
When timer_read_counter() is not defined, this obviously does the wrong
thing here and it does so silently. The consequence is that a board
without timer_read_counter() will sleep for random amounts and generally
have erratic get_ticks() values.
Drop the __weak annotation of the extern so a linker error is raised
when timer_read_counter() is not defined. This is okay, the original
reason for the reverted change - breaking the sandbox build - no longer
applies.
Final sidenote: This was the only weak extern in the entire tree at
this time as far as I can tell. I guess we should avoid introduction of
them again as they are obviously a very big footgun.
[1]: https://stackoverflow.com/questions/31203402/gcc-behavior-for-unresolved-weak-functions
Fixes: 65ba7add0d ("time: add weak annotation to timer_read_counter declaration")
Reported-by: Serge Bazanski <q3k@q3k.org>
Signed-off-by: Harald Seiler <hws@denx.de>
The rest of the unmigrated CONFIG symbols in the CONFIG_SYS namespace do
not easily transition to Kconfig. In many cases they likely should come
from the device tree instead. Move these out of CONFIG namespace and in
to CFG namespace.
Signed-off-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Globally replace all occurances of WATCHDOG_RESET() with schedule(),
which handles the HW_WATCHDOG functionality and the cyclic
infrastructure.
Signed-off-by: Stefan Roese <sr@denx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Tom Rini <trini@konsulko.com> [am335x_evm, mx6cuboxi, rpi_3,dra7xx_evm, pine64_plus, am65x_evm, j721e_evm]
CONFIG_SYS_TIMER_RATE might be a dynamic value, i.e. a function call
instead of a static value, thus it has to be evaluated at runtime. If it
is a static value, the compiler should be able to optimize the unused
branches out.
This will be needed for kirkwoods dynamic CONFIG_SYS_TCLK setting.
Cc: Pali Rohár <pali@kernel.org>
Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
In order to finish moving this symbol to Kconfig for all platforms, we
need to do a few more things. First, for all platforms that define this
to a function, introduce CONFIG_DYNAMIC_SYS_CLK_FREQ, similar to
CONFIG_DYNAMIC_DDR_CLK_FREQ and populate clock_legacy.h. This entails
also switching all users from CONFIG_SYS_CLK_FREQ to get_board_sys_clk()
and updating a few preprocessor tests.
With that done, all platforms that define a value here can be converted
to Kconfig, and a fall-back of zero is sufficiently safe to use (and
what is used today in cases where code may or may not have this
available). Make sure that code which calls this function includes
<clock_legacy.h> to get the prototype.
Signed-off-by: Tom Rini <trini@konsulko.com>
Move this out of the common header and include it only where needed. In
a number of cases this requires adding "struct udevice;" to avoid adding
another large header or in other cases replacing / adding missing header
files that had been pulled in, very indirectly. Finally, we have a few
cases where we did not need to include <asm/global_data.h> at all, so
remove that include.
Signed-off-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Rini <trini@konsulko.com>
This error should not happen in normal use. Reduce the length of it to
save space in the image.
Add an empty spl.h file to sh since it appears to lack this.
Signed-off-by: Simon Glass <sjg@chromium.org>
get_ticks does not always succeed. Sometimes it can be called before the
timer has been initialized. If it does, it returns a negative errno.
This causes the timer to appear non-monotonic, because the value will
become much smaller after the timer is initialized.
No users of get_ticks which I checked handle errors of this kind. Further,
functions like tick_to_time mangle the result of get_ticks, making it very
unlikely that one could check for an error without suggesting a patch such
as this one.
This patch panics if we ever get an error. There are two cases in which
this can occur. The first is if we couldn't find/probe the timer for some
reason. One reason for this is if the timer is not available so early. This
likely indicates misconfiguration. Another reason is that the timer has an
invalid/missing device tree binding. In this case, panicing is also
correct. The second case covers errors calling get_count. This can only
occur if the timer is missing a get_count function (or on RISC-V, but that
should be fixed soon).
Fixes: c8a7ba9e6a
Signed-off-by: Sean Anderson <seanga2@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The current get_timer_us() uses 64-bit arithmetic on 32-bit machines.
When implementing microsecond-level timeouts, 32-bits is plenty. Add a
new function that uses an unsigned long. On 64-bit machines this is
still 64-bit, but this doesn't introduce a penalty. On 32-bit machines
it is more efficient.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This function belongs in time.h so move it over and add a comment.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
Add get_timer_us(), which is useful e.g. when we need higher
precision timestamps.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Tom Rini <trini@konsulko.com>
Cc: Simon Glass <sjg@chromium.org>
[trini: Fixup arch/arm/mach-bcm283x/include/mach/timer.h]
Signed-off-by: Tom Rini <trini@konsulko.com>
In the UEFI Stall() boottime service we need access to usec_to_tick().
Export the function.
Remove redundant implementation in arch/arm/mach-rockchip/rk_timer.c.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
We'd better use correct way to check if module has enabled.
for we have 3 timer MACRO:
- CONFIG_TIMER
- CONFIG_SPL_TIMER
- CONFIG_TPL_TIMER
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
When U-Boot started using SPDX tags we were among the early adopters and
there weren't a lot of other examples to borrow from. So we picked the
area of the file that usually had a full license text and replaced it
with an appropriate SPDX-License-Identifier: entry. Since then, the
Linux Kernel has adopted SPDX tags and they place it as the very first
line in a file (except where shebangs are used, then it's second line)
and with slightly different comment styles than us.
In part due to community overlap, in part due to better tag visibility
and in part for other minor reasons, switch over to that style.
This commit changes all instances where we have a single declared
license in the tag as both the before and after are identical in tag
contents. There's also a few places where I found we did not have a tag
and have introduced one.
Signed-off-by: Tom Rini <trini@konsulko.com>
Currently, mdelay() and udelay() are declared in include/common.h,
while ndelay() in include/linux/compat.h. It would be nice to
collect them into include/linux/delay.h like Linux.
While we are here, fix the ndelay() implementation; I used the
DIV_ROUND_UP() instead of (x)/1000 because it must wait *longer*
than the given period of time.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
In some cases the timer must be accessible before driver model is active.
Examples include when using CONFIG_TRACE to trace U-Boot's execution before
driver model is set up. Enable this option to use an early timer. These
functions must be supported by your timer driver: timer_early_get_count()
and timer_early_get_rate().
Signed-off-by: Simon Glass <sjg@chromium.org>
Adding timer init function in timer-uclass driver to create and
initialize the timer device on platforms where u-boot,dm-pre-reloc
is not used. Since there will be multiple timer devices in the
system, adding a tick-timer node in chosen node to know which
timer device to be used as tick timer in u-boot.
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
There are timers with a 64-bit counter value but current timer
uclass driver assumes a 32-bit one. Modify timer_get_count()
to ask timer driver to always return a 64-bit counter value,
and provide an inline helper function timer_conv_64() to handle
the 32-bit/64-bit conversion automatically.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Simon Glass <sjg@chromium.org>
Unfortunately 'unsigned long long' and 'uint64_t' are not necessarily
compatible on 64-bit machines. Use the correct typedef instead of
writing the supposed type out in full.
Signed-off-by: Simon Glass <sjg@chromium.org>
CONFIG_SYS_HZ is always defined as 1000 in config_fallbacks.h
(but some boards still have redundant definitions).
This commit moves the definition and the document in README to
Kconfig. Since lib/Kconfig can assure that CONFIG_SYS_HZ is 1000,
the sanity check in lib/time.c should be removed.
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Reviewed-by: Marek Vasut <marex@denx.de>
As I initially suspected overflow in time handling, I took a detailed
look at lib/time.c. This adds comments about units being used, reduces
amount of type casting being done, and makes __udelay() always wait at
least one tick. (Current code could do no delaying at all for short
delays).
Signed-off-by: Pavel Machek <pavel@denx.de>
If timer_init() is made a weak stub function, then it allows us to
remove several empty timer_init functions for those boards that
already have a timer initialized when u-boot starts. Architectures
that use the timer framework may also remove the need for timer.c.
Signed-off-by: Darwin Rambo <drambo@broadcom.com>
Reviewed-by: Tim Kryger <tim.kryger@linaro.org>
Commit 8dfafdde88 ("Introduce common timer functions") created a
common definition of usec_to_tick() which had a couple problems:
static unsigned long long usec_to_tick(unsigned long usec)
{
uint64_t tick = usec * get_tbclk();
That likely overflows.
usec *= get_tbclk();
That was an attempt to fix it by performing the multiply after the
promotion of usec to 64-bit, but was applied to the wrong variable,
which was never used.
This patch fixes these issues. A user-visible symptom of the problem was
the e.g. "dhcp zImage" using an ASIX USB Ethernet dongle would print:
Waiting for Ethernet connection... unable to connect.
... with no delay before "unable to connect". There are likely other
symptoms.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Rob Herring <rob.herring@calxeda.com>
Commit 8dfafdde88 introduced
new gcc warnings on MIPS64:
time.c: In function 'tick_to_time':
time.c:59:2: warning: comparison of distinct pointer types lacks a cast [enabled by default]
time.c:59:2: warning: passing argument 1 of '__div64_32' from incompatible pointer type [enabled by default]
In file included from time.c:10:0:
./u-boot-mips/include/div64.h:22:17: note: expected 'uint64_t *' but argument is of type 'long long unsigned int *'
time.c: In function 'usec_to_tick':
time.c:76:2: warning: comparison of distinct pointer types lacks a cast [enabled by default]
time.c:76:2: warning: passing argument 1 of '__div64_32' from incompatible pointer type [enabled by default]
In file included from time.c:10:0:
./u-boot-mips/include/div64.h:22:17: note: expected 'uint64_t *' but argument is of type 'long long unsigned int *'
Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
A weak annotation is needed in order to prevent link errors when
get_ticks is overridden. This fixes sandbox build.
Signed-off-by: Rob Herring <rob.herring@calxeda.com>
Many platforms duplicate pretty much the same timer code yet they all have
a 32-bit freerunning counter register. Create a common implementation that
minimally requires 2 or 3 defines to add timer support:
CONFIG_SYS_TIMER_RATE - Clock rate of the timer counter
CONFIG_SYS_TIMER_COUNTER - Address of 32-bit counter
CONFIG_SYS_TIMER_COUNTS_DOWN - Define if counter counts down
All functions are weak or ifdef'ed so they can still be overriden by any
platform.
Signed-off-by: Rob Herring <rob.herring@calxeda.com>
The standalone example does not have get_timer() defined, so we cannot
rely on it being available.
Move the timer function into boootstage.c to avoid this problem.
This corrects a build breakage for the standalone example on some boards.
Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Matthias Fuchs <matthias.fuchs@esd.eu>
Define timer_get_boot_us() which returns the number of microseconds
since boot. If undefined then we use get_timer() * 1000.
We can fit this in a 32-bit register which keeps everyone happy on
the efficiency side. It will wrap around after about an hour. If we
are still looking at it after an hour then we had better not be
timing the boot.
Signed-off-by: Simon Glass <sjg@chromium.org>
There are several mdelay() definitions in the driver and
board code. Remove them all and provide a common mdelay()
in lib/time.c.
Signed-off-by: Anatolij Gustschin <agust@denx.de>
Acked-by: Mike Frysinger <vapier@gentoo.org>
Now that the other architecture-specific lib directories have been
moved out of the top-level directory there's not much reason to have the
'_generic' suffix on the common lib directory.
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
