This commit turns cpu_init() into a SoC-agnostic function by
removing any SoC specific calls and isolating them into their
own SoC implementation.
We start this by isolating all IRQs initialization code from
the legacy-pc target, pic_init() and spurious IRQ7 registration,
into a new interface: irq_init() from irq.h. Future SoCs will have
to provide their own implementation of this interface.
This model is to be followed by future initialization code that
we may need to add and which is not common to all x86 SoCs.
The x86/init/common/ folder holds all cpu initialization
code - idt and gdt setup, interrupts and cpu initialization.
On this folder will also sit any SoC specific implementation of
the functions called from cpu_init().
Now the cpu/x86/ provides a Makefile.x86_common and a
Makefile.x86_pc. The former includes the common Makefile
and adds legacy pc specific implementations (currently,
drivers only) into the building context, while the latter
has everything that defines the bootstrap of a x86 CPU.
This commit also fixes platform/galileo/ so it includes the
correct makefile - Makefile.x86_quarkX1000. Galileo uses
a Quark X1000 SoC which is not an IBM Generic PC-like CPU,
but it does provide most of a PCs peripherals through
its "Legacy Bridge". Thus, it makes sense that QuarkX1000's
Makefile includes code from the legacy_pc x86 cpu.
All drivers implemented so far are for chips which are only available
on legacy x86 PCs. This commit moves them into a more appropriate folder,
also making the cpu/x86/drivers/ folder ready for other x86 based SoCs.
Currently, it is common to see Contiki's core/ interfaces implementations
spread in both cpu/ and platform/. We here take one step further starting
an effort to centralize all of these in platform's code instead.
This commit starts this by adding platform/galileo/core/ and its sys/
subfolder, adding a stubbed mtarch.h and moving clock and rtimer
implementations to this new folder. From now on we should concentrate
implementation from Contiki's core/ interfaces into the appropriate
subfolder in platform/galileo/core/.
Note that this is not the current fashion followed on other platforms
and cpus folders, as most of them add the core interface implementation
into its subfolder directly. For instance, on CC2538DK,
core/dev/button-sensor.h is implemented in platform/cc2538dk/dev/
directly, while on Galileo it would sit at platform/galileo/core/dev/.
We believe ours is a better approach to organize and escalate a
platform's code base.
We also remove previous x86 mtarch.h and mtarch.c since they weren't used
at all - both native and cooja platforms have their own mtarch
implementations.
The 8259a PIC has a well known problem of generating flaky
IRQ7 interrupts. The correct solution is to always check
if an IRQ7 interrupt is real or not by probing the PIC's ISR
register. This check is only mandatory if the IRQ7 is actually
being used by the system. More importantly, the handler should
NEVER send and EOI if the interrupt was spurious.
This patch addresses this issue by implementing a fake empty
handler for this IRQ and, as stated, NOT sending the EOI.
This commit implements pic_eoi(int irq) and a helper macro PIC_INT(irq).
This first checks which PICs should be 'acked' given an IRQ number, while
the macro returns the actual system interrupt number for the IRQ according
to the offset used on the PIC initialization.
This patch implements the pic_unmask_irq() helper and uses it where
applicable. This function zeros the corresponding bit from the IRQ
number in IMR register.
This patch doesn't implement the pic_mask_irq() helper since it is not
useful at this moment.
This patch adds a driver for the 8254 Programmable Interrupt Timer (PIT).
The driver introduced by this patch programs the PIT to generate interrupt
periodically. The interrupt frequency can be configured by the user.
On each PIT interrupt, a callback configured by the user is called. As
expected, that callback is executed in interrupt context so the user
should be aware of what it is not supposed to do (e.g. to call blocking
functions).
Issues marked as FIXME are all related to missing APIs on the PIC driver
so they will be addressed by a future commit.
This patch adds a driver for Real-Time Clock (RTC). The RTC timer is
suitable to implement some operating system features such as the
system clock. Actually, the RTC will be used to implement the system
clock in galileo platform.
The driver introduced by this patch programs the RTC to generate
interrupt periodically. The interrupt frequency can be configured by the
user. On each RTC interrupt, a callback configured by the user is called.
As expected, that callback is executed in interrupt context so the user
should be aware of what it is not supposed to do (e.g. to call blocking
functions).
This patch also adds the inb() helper function to helpers.h. The helpers
is a wrapper for assembly 'in' instruction.
The Programmable Interrupt Controller is a chip responsible for
translating hardware interrupts to system interrupts. When it
receives an Interrupt Request (IRQ), it triggers the appropriate
interrupt line reaching the appropriate IDT gate, following a
previously setup offset.
There are 2 daisy-chained PICs. PIC1 handles IRQs 0-7 and PIC2
handles IRQs 8-15. If no vector offset is set, an IRQ0, for instance,
would trigger the interrupt 0, clashing with the "Division by zero exception"
handler. Thus the IRQs must be remapped.
This patch implements the PICs initialization through their 4
Initialization Command Words (ICWs) in a very "canonical" way:
- ICW1: the initializing command;
- ICW2: the vector offset for the PIC1 and PIC2 (we add an offset of 32 positions);
- ICW3: the inter-PICs wiring setup (we connect PIC2 to PIC1's IRQ2);
- ICW4: extra systems information (we set PIC1 as Master and PIC2 as slave).
It then masks the Interrupt Mask Register, blocking all IRQs but #2 initially.
These must be unmasked on demand. The IMR is 8-bits long, so setting the n^th bit to 1
would DISABLE the IRQ n while setting it to 0 would ENABLE IRQ n.
As stated, this is an implementation of the legacy 8259 PIC. More
investigation is needed so we decide if it is enough or if we need
the (newer) APIC implementation instead.
This patch also adds the outb() helper function to helpers.h. The helpers
is a wrapper for assembly 'out' instruction.
Finally, since we now properly support hardware interrupts, this patch
also enables IRQs in platform main().
More information:
- Quark X1000 Datasheet, section 21.12, page 898.
- http://wiki.osdev.org/8259_PIC
- http://stanislavs.org/helppc/8259.html
This patch sets an interrupt handler for Double Fault exception during
CPU initialization. In case such exception is raised, we halt the system.
This way, we avoid the system to triple fault (due to an unhandled
interrupt for instance), leaving no trace about what cause the triple
fault.
This patch introduces the interrupt.h header file which provides some
helper macros to set a interrupt handler and disable/enable maskable
hardware interrupts.
Since there is no easy way to write an Interrupt Service Routines
(ISR) in C (for further information on this, see [1]), we introduce
the SET_INTERRUPT_HANDLER helper macro.
The macro does two things:
1) Defines an assembly trampolin to a C function that will, indeed,
handle the interrupt.
2) Sets the corresponding interrupt gate descriptor in IDT.
The macro usage is pretty straightforward. The macro is defined as
SET_INTERRUPT_HANDLER(num, has_error_code, handler) where:
@num: Interrupt number (0-255)
@has_error_code: 0 if processor doesn't push error code onto the
stack. Otherwise, set this argument to 1.
@handler: Pointer to function that should be called once the
interrupt is raised. In case has_error_code == 0
the function prototype should be the following:
void handler(void)
Otherwise, it should be:
void handler(struct interrupt_context context)
For instance, let's say we want to set a handler for a device interrupt
(for example, interrupt number 101). Remember, hardware interrupts don't
have error code. So we should have something like this:
void interrupt_handler(void)
{
/* Handling code here */
}
void my_device_init(void)
{
...
SET_INTERRUPT_HANDLER(101, 0, interrupt_handler);
...
}
Now, let's say we want to set an interrupt handler for Page Fault
(interrupt number 14). Some exceptions, such as Page Fault, pushes an
error code onto the stack and may require registers values in order
to be properly be handled. Thus, the code should look like this:
void pagefault_handler(struct interrupt_context context)
{
/* Handling code here */
}
void init_memory(void)
{
...
SET_INTERRUPT_HANDLER(14, 1, pagefault_handler);
...
}
For further information about exceptions and error code, refer to Intel
Combined Manual, Vol. 3, Sections 6.3 and 6.13.
Finally, we don't define any API to unregister interrupt handlers since
we believe that it wouldn't be useful at all, at least at this moment.
Considering Contiki's context, interrupt handler registration is pretty
"static" and defined at compile-time by platform code (or the device
drivers used by the platform).
[1] http://wiki.osdev.org/Interrupt_Service_Routines
This patch defines the cpu_init() function which should encapsulate
all code related to x86 CPU initialization. For now, this function
initializes GDT and IDT.
This patch adds code to handle Interrupt Descriptor Table (IDT)
initialization. The IDT is initialized with null descriptors
therefore any interrupt at this point will cause a triple fault.
The IDT initialization is part of x86 CPU initialization.
Strictly speaking, there is no need to use attribute packed in struct
intr_gate_desc however we use it for readability reasons.
This patch adds the helpers.h. This file should contain only x86-related
helper functions and macros. For now, we define the BIT macro and halt()
helpers which will be used in upcoming patches.
Additionally, this patch also changes loader.S to call the halt().
This patch adds code to initialize the Global Descriptor Table. For
simplicity, the memory is organized following the flat model. Thus,
memory appears to Contiki as a single, continuous address space. Code,
data, and stack are all contained in this address space (so called
linear address space).
The macros to manipulate bits from segment descriptor and the
set_descriptor() helper are based on the ones described in [1].
[1] http://wiki.osdev.org/GDT_Tutorial
This patch changes Galileo's buildsystem to print the elf sections
sizes after a new image is built. This way we can easily track how
these sections increase or decrease after any change.
To achieve that, we define a custom linking rule which is pretty much
the same as the default linking rule define in Makefile.include, but
we run 'size' command after the image is built.
Currently there are only one platform using CPU x86: Cooja. The
elfloader-x86.c is rather a POSIX implementation, so the Galileo port
won't use it for now. This patch fixes this by moving this source file to
be included by the platforms using it instead of the cpu's Makefile.
The peripheral core clocks of the PWM timers are gated in PM1+, so these
power modes must be disabled if a PWM timer is running. Use
lpm_register_peripheral() to handle this automatically and dynamically.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
The peripheral core clock of the general-purpose timers used by the PWM
driver is the system clock, not the I/O clock.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
According to the Procedure Call Standard for the ARM Architecture
(AAPCS) - ABI r2.09 [1], §5.2.1.2, the stack pointer must be
double-word-aligned at a public interface. The stack implementation
being full-descending, this requires that the top of stack be
double-word-aligned too.
[1] http://infocenter.arm.com/help/topic/com.arm.doc.ihi0042e/IHI0042E_aapcs.pdf
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
Coffee is placed by default at the beginning of the flash memory, right
before the firmware. This avoids the memory gaps that there could be
before and after Coffee if it were placed after the firmware, because it
is unlikely that the end of the firmware is aligned with a flash page
boundary, and the CCA is not flash-page-aligned. Thanks to that, Coffee
is also always in the same flash area if its size remains unchanged,
even if the firmware changes, which makes it possible to keep the Coffee
files when reprogramming the firmware after a partial flash erase
command.
The default configuration of Coffee is set to use sensible values for a
typical usage on this SoC, i.e. for sensor data logging.
The default size of Coffee is set to 0 in order not to waste flash if
Coffee is unused.
COFFEE_CONF_CUSTOM_PORT can be defined to a header file to be used with
"#include" in order to override the default CC2538 port of Coffee. This
makes it possible to use Coffee with an external memory device rather
than with the internal flash memory, without having to alter the Contiki
files.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
No entry point was defined, so it defaulted to the beginning of the
.text output section where the vector table is located by default in
Contiki. Actually, the vector table may be located elsewhere, and the
ROM-based boot loader first reads the CCA to find the vector table.
Consequently, this commit sets the entry point to the CCA, which fixes
both the entry point and the initial symbol reference, so this commit
also removes the now-unneeded "__used__" and "KEEP" keywords from the
CCA.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
The current CC2538 linker script in Contiki places the vector table at
the beginning of the flash memory / .text output section. However, this
location is arbitrary (the only requirement is that the vector table is
512-byte aligned), and custom linker scripts may be used with Contiki,
which means that Contiki may be used with a vector table placed
elsewhere. Thus, using the flash/.text start address in the CCA and as
the default NVIC VTABLE value was wrong.
This commit rather uses the address of the vectors[] array from
startup-gcc.c, which makes it possible to freely move around the vector
table without breaking anything or having to use a custom startup-gcc.c
and to configure the NVIC driver for that. Moreover, referencing the
vectors[] array naturally prevents it and its input section from being
garbage-collected by the linker, so this commit also removes the
now-unneeded "used" and "KEEP" keywords from the vector table.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
The only thing needed for VTABLE is the absolute address of the vector
table. Splitting it between code/SRAM base and offset complicates things
and brings nothing.
Consequently, this commit merges the NVIC VTABLE configurations into a
single one giving the vector table absolute address.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
Define the flash memory page and word sizes. These definitions are
grouped with the flash lock bit page and CCA definitions, so flash-cca.h
is renamed to flash.h.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
Define the available CC2538 devices and their features, and use them to
define the linker script memory regions. The .nrdata output section is
now always defined in order to trigger an error if it is used but no
memory is available for it. The CC2538 device used by Contiki is made a
configuration option, the CC2538SF53 device being the default.
This makes more sense than defining the flash memory address and size as
configuration options like previously, all the more not all values are
possible and all the features are linked by each device.
This change also makes it possible to:
- use the correct SRAM parameters for the CC2538NF11,
- know at build time if the AES, SHA, ECC and RSA hardware features are
available on the selected CC2538 device.
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
This commit fixes nearly all of the reported doxygen warnings.
I tried to not clutter the log with removed trailing spaces.
Removed whitespace and converted tab/spaces for all files affected by this commit
are in a separate branch.
This commit applies a number of improvements to the logic used when trying to drop to a CC13xx/CC26xx low-power mode:
* We identify whether there are any pending etimers by using `etimer_pending()` instead of `etimer_next_expiration_time()`. This subsequently allows us to also identify whether an etimer is set to fire at time 0.
* We run a larger portion of the code with the global interrupt disabled. This prevents a number of messy conditions that can occur if an interrupt fires after we have started the low-power sequence.
* We check whether there are pending events earlier in the sequence.
* We make sure to schedule a next wakeup event even when an LPM module prohibits deep sleep and forces sleep instead.
This fixes some of the issues discussed in #1236
The AON RTC CH1 event handler aims to schedule the next compare event on the next 512 RTC counter boundary. However, the current calculation of "now" takes place too early within the interrupt handler. In some cases, this results in the next event getting scheduled too soon in the future or on some extreme cases even in the past.
AON RTC compare events cannot happen within 2 SCLK_LF cycles after a clearance (4 RTC ticks in the 16.16 format). Thus, if the next 512 boundary is too soon (5 ticks for margin), we skip it altogether. When this happens, etimers that would have expired on the skipped tick will expire 1 tick later instead. Skipping a tick has no negative impact on our s/w clock counter, since this is always derived directly from the hardware counter.
The current logic attempts to send `CMD_SET_TX_POWER` before saving the new power setting. If `CMD_SET_TX_POWER` fails the power setting will not get saved. As a result, when the RFC is powered off, all attempts to change TX power will fail.
This commit changes this logic to always save the new TX setting as requested by the user. If the RFC is powered up, we apply it immediately. If it is powered down, the new setting will automatically be applied next time we send `CMD_RADIO_SETUP`.
Fixes#1340
CC_FASTCALL was introduced many years ago for the cc65 tool chain. It was never used for another tool chain. With a798b1d648 the cc65 tool chain doesn't need CC_FASTCALL anymore.
So far 80 column display was an attribute of a cc65 platform. Now each cc65 application can ask for 80 column display by defining WITH_80COL. Of course this is ignored by platforms incapable of 80 column display.
I see three types of application:
* Applications not benefitting from 80 column at all and in fact looking better with 40 column display. These are now using 40 column display. Examples: ethconfig, ipconfig
* Applications taking advantage of 80 column display if it is available without drawbacks. These stay as they were. Examples: Telnet server, web server, wget
* Applications needing 80 column display so urgently that it is likely desirable even if the display becomes harder to read. These come now in both flavors allowing the user to choose. Examples: IRC, web browser
Note: This change doesn't actually introduce any 80 column display with drawbacks. This if left to a subsequent change.
As discussed in #1294, every time a CC13xx/CC26xx RF interrupt fires, we clear all interrupt flags unconditionally. This may result in missed events. This patch fixes this bug by clearing only the flag that triggered the interrupt in the first place.
Fixes#1294
