| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
|
|
|
| |
Due to API change, the patch also fixes affected BSPs and uses
value provided by MPU CTRL spec option there.
Sponsored-By: Precidata
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The comment above bintime2timespec() says:
When converting between timestamps on parallel timescales of differing
resolutions it is historical and scientific practice to round down.
However, the delta_nsec value is a time difference and not a timestamp. Also
the rounding errors accumulate in the frequency accumulator, see hardpps().
So, rounding to the closest integer is probably slightly better.
Reviewed by: imp
Pull Request: https://github.com/freebsd/freebsd-src/pull/604
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Let A be the current calculation of the frequency accumulator (pps_fcount)
update in pps_event()
scale = (uint64_t)1 << 63;
scale /= captc->tc_frequency;
scale *= 2;
bt.sec = 0;
bt.frac = 0;
bintime_addx(&bt, scale * tcount);
bintime2timespec(&bt, &ts);
hardpps(tsp, ts.tv_nsec + 1000000000 * ts.tv_sec);
and hardpps(..., delta_nsec):
u_nsec = delta_nsec;
if (u_nsec > (NANOSECOND >> 1))
u_nsec -= NANOSECOND;
else if (u_nsec < -(NANOSECOND >> 1))
u_nsec += NANOSECOND;
pps_fcount += u_nsec;
This change introduces a new calculation which is slightly simpler and more
straight forward. Name it B.
Consider the following sample values with a tcount of 2000000100 and a
tc_frequency of 2000000000 (2GHz).
For A, the scale is 9223372036. Then scale * tcount is 18446744994337203600
which is larger than UINT64_MAX (= 18446744073709551615). The result is
920627651984 == 18446744994337203600 % UINT64_MAX. Since all operands are
unsigned the result is well defined through modulo arithmetic. The result of
bintime2timespec(&bt, &ts) is 49. This is equal to the correct result
1000000049 % NANOSECOND.
In hardpps(), both conditional statements are not executed and pps_fcount is
incremented by 49.
For the new calculation B, we have 1000000000 * tcount is 2000000100000000000
which is less than UINT64_MAX. This yields after the division with tc_frequency
the correct result of 1000000050 for delta_nsec.
In hardpps(), the first conditional statement is executed and pps_fcount is
incremented by 50.
This shows that both methods yield roughly the same results. However, method B
is easier to understand and requires fewer conditional statements.
Reviewed by: imp
Pull Request: https://github.com/freebsd/freebsd-src/pull/604
|
|
|
|
|
| |
Reviewed by: imp
Pull Request: https://github.com/freebsd/freebsd-src/pull/604
|
|
|
|
|
|
|
| |
Move the pseq increment. This makes it possible to reuse registers earlier.
Reviewed by: imp
Pull Request: https://github.com/freebsd/freebsd-src/pull/604
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Use local variables for the captured timehand and timecounter in pps_event().
This fixes a potential issue in the nsec preparation for hardpps(). Here the
timecounter was accessed through the captured timehand after the generation was
checked.
Make a snapshot of the relevent timehand values early in pps_event(). Check
the timehand generation only once during the capture and event processing. Use
atomic_thread_fence_acq() similar to the other readers.
Reviewed by: imp
Pull Request: https://github.com/freebsd/freebsd-src/pull/604
|
|
|
|
|
|
|
|
| |
This ensures that the timecounter and the tc_get_timecount handler belong
together.
Reviewed by: imp
Pull Request: https://github.com/freebsd/freebsd-src/pull/604
|
|
|
|
| |
Sponsored by: Rubicon Communications, LLC ("Netgate")
|
|
|
|
|
|
|
|
|
|
|
|
| |
Since 32c203577a5e by phk in 1999 (Make even more of the PPSAPI
implementations generic), the "nsec" parameter of hardpps() is a time
difference and no longer a time point. Change the name to "delta_nsec"
and adjust the comment.
Remove comment about a clock tick adjustment which is no longer in the code.
Pull Request: https://github.com/freebsd/freebsd-src/pull/640
Reviewed by: imp
|
|
|
|
|
|
|
|
|
|
|
| |
The third argument to this function indicates whether the supplied
ticker is fixed or variable, i.e. requiring calibration. Give this
argument a type and name that better conveys this purpose.
Reviewed by: kib, markj
MFC after: 1 week
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D35459
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
(which is used to calculate cputime from cpu ticks) has some imprecision and,
worse, huge timestep (about 20 minutes on 4GHz CPU) near 53.4 days of elapsed
time.
kern_time.c/cputick2timespec() (it is used for clock_gettime() for
querying process or thread consumed cpu time) Uses cputick2usec()
and then needlessly converting usec to nsec, obviously losing
precision even with fixed cputick2usec().
kern_time.c/kern_clock_getres() uses some weird (anyway wrong)
formula for getting cputick resolution.
PR: 262215
Reviewed by: gnn
Differential Revision: https://reviews.freebsd.org/D34558
|
|
|
|
|
|
|
| |
The goto label was directly after the loop, so we can replace the goto
with a break.
Close #4847.
|
|
|
|
|
|
|
|
| |
Add an assert to _Thread_Initialize_scheduler_and_wait_nodes() which may
help a static analyzer. Use a do/while loop since we have at least one
scheduler.
Update #4832.
|
|
|
|
|
|
|
| |
Several SMP message processing functions returned a value. This value
was always unused.
Close #4822.
|
|
|
|
|
|
| |
Do not use a chained assignment for code clarity.
Close #4818.
|
|
|
|
| |
For example, the QorIQ T4240 has more than 16 TLB1 entries.
|
|
|
|
| |
Close #4759.
|
|
|
|
|
|
|
|
|
| |
This patch changes the license to BSD-2 for all source files where the
copyright is held by Aeroflex Gaisler, Cobham Gaisler, or Gaisler Research.
Some files also includes copyright right statements from OAR and/or
embedded Brains in addition to Gaisler.
Updates #3053.
|
|
|
|
|
|
|
| |
Set the thread ID register in the CPU context.
Update #3835.
Close #4753.
|
|
|
|
|
|
|
| |
Where CPU_ENABLE_ROBUST_THREAD_DISPATCH == TRUE, the only supported interrupt
level allowed to set is 0 (interrupts enabled). This constraint is enforced by
the API level functions which return an error status for other interrupt
levels.
|
| |
|
|
|
|
|
|
|
| |
The AArch64 TRM specifies that when FPCR is set to trap floating point
exceptions, the FPSR exception bits are not set. This ensures that FPSR
is updated as FENV expects even if floating point exception traps are
enabled.
|
|
|
|
|
| |
The _RISCV_Map_cpu_index_to_hardid() and _RISCV_Map_hardid_to_cpu_index()
functions must be available to all riscv BSPs.
|
|
|
|
| |
This avoids a pedantic warning about a zero size Context_Control_fp.
|
| |
|
|
|
|
|
|
|
| |
Ensure that the IDLE storage allocator did allocate a suffiently large area.
Update #3835.
Update #4524.
|
|
|
|
|
|
|
|
|
|
|
|
| |
By default, allocate the IDLE task storage areas from the RTEMS Workspace.
This avoids having to estimate the thread-local storage size in the default
configuration.
Add the application configuration option CONFIGURE_IDLE_TASK_STORAGE_SIZE to
request a static allocation of the task storage area for IDLE tasks.
Update #3835.
Update #4524.
|
|
|
|
|
|
| |
Add the INTERNAL_ERROR_IDLE_THREAD_CREATE_FAILED fatal error in case the
creation of an idle thread fails. This may happen due to a failing create
extension provided by the application.
|
|
|
|
| |
Update #3835.
|
|
|
|
|
|
|
|
| |
Allow the IDLE stack allocator to change the stack size. This can be
used by applications with a very dynamic thread-local storage size to
adjust the thread storage area of the IDLE tasks dynamically.
Update #4524.
|
|
|
|
|
| |
For most CPU ports this was already the case. This makes it possible to use
the size as an object alignment using RTEMS_ALIGNED().
|
|
|
|
|
|
|
|
| |
The Microchip PolarFire SoC support is implemented as a
riscv BSP variant to boot with any individual hart(cpu core)
or SMP based on the boot HARTID configurable and support
components are 4 CPU Cores (U54), Interrupt controller (PLIC),
Timer (CLINT), UART.
|
|
|
|
|
|
|
| |
Directly use "static inline" which is available in C99 and later. This brings
the RTEMS implementation closer to standard C.
Close #3935.
|
|
|
|
| |
Update #4702.
|
|
|
|
|
|
|
| |
All CPU ports used the same _CPU_Counter_difference() implementation. Remove
this CPU port interface and mandate a monotonically increasing CPU counter.
Close #3456.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The VRSAVE feature of the Altivec unit can be used to reduce the amount of
Altivec registers which need to be saved/restored during interrupt processing
and context switches.
In order to use the VRSAVE optimization a corresponding multilib (-mvrsave) is
required, see GCC configuration. The -mvrsave option must be added to the
ABI_FLAGS of the BSP.
Currently only the -mcpu=e6500 based QorIQ BSP support this optimization.
Update #4712.
|
|
|
|
|
|
|
|
| |
Added two pragmas to address, and changed the value of
AARCH64_EXCEPTION_MAKE_ENUM_64_BIT to INT_MAX because the old value was not
in range of an int.
Updates #4662
|
|
|
|
|
|
| |
rtl-mdreloc-aarch64.c and elf_machdep.h came from NetBSD.
Updates #4682
|
|
|
|
|
|
|
|
|
|
| |
The rtems_task_delete() directive is basically just a combined pthread_cancel()
and pthread_join(). In addition, it removes the PTHREAD_DETACHED state. The
exit value returned by pthread_join() of threads cancelled by
rtems_task_delete() should reflect this by getting a PTHREAD_CANCELED value
instead of NULL which could be a normal exit value.
Close #4680.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Threads may join the thread termination of another thread using the
pthread_join() or rtems_task_delete() directives. The thread cancel operation
used a special case priority boosting mechanism implemented by
_Thread_Raise_real_priority(). The problem was that this approach
* is not transitive,
* does not account for priority adjustments of the calling task
while waiting for the join,
* does not support clustered scheduling, and
* does not detect deadlocks.
All these problems are fixed by using a priority inheritance thread queue for
the join operation.
Close #4679.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The uniprocessor schedulers had some special case logic for the
PRIORITY_PSEUDO_ISR priority. Tasks with a priority of PRIORITY_PSEUDO_ISR
were allowed to preempt a not preemptible task. If other higher priority task
are made ready while a PRIORITY_PSEUDO_ISR task preempts a not preemptible
task, then the other tasks run before the not preemptible task. This made the
RTEMS_NO_PREEMPT mode ineffective.
Remove the PRIORITY_PSEUDO_ISR special case logic. This simplifies the
uniprocessor schedulers. Move the uniprocessor-specific scheduler support to
the new header file <rtems/score/scheduleruniimpl.h>.
Close #2365.
|
|
|
|
|
|
|
|
|
| |
This alters the AArch64 page table generation and mapping code and MMU
configuration to use page table level 0 in addition to levels 1, 2, and
3. This allows the mapping of up to 48 bits of memory space and is the
maximum that can be mapped without relying on additional processor
extensions. Mappings are restricted based on the number of physical
address bits that the CPU supports.
|
|
|
|
|
|
|
|
| |
Commit 21275b58a5a69c3c838082ffc8a7a3641f32ea9a ("score: Static
Objects_Information initialization") introduced an off-by-one error in the
maintenance of inactive objects.
Close #4677.
|
|
|
|
|
|
|
| |
With unlimited objects the object maximum may be larger than the sum of active
and inactive objects.
Update #4677.
|
|
|
|
|
|
| |
Dirty or zero also the part of the .noinit section used by RTEMS.
Close #4678.
|
|
|
|
|
|
| |
Conditional expressions with inline functions are not optimized away if
optimization is disabled. Avoid such expressions to prevent dead
branches.
|
|
|
|
| |
This function is only used in SMP configurations.
|
|
|
|
| |
This function was never actually used and is dead code.
|