This commit implements lazy FP save/restore mechanism, required for user threads
when they use the FPU. The following caveats take place:
* Only support synchronous exceptions. No support for asynchronous ones.
* VFP opcodes are used instead of normal instructions to discard compiler warnings/errors.
* Support is limited to specific ARM subarchitectures we support and tested this commit on.
* Disable the FPU by default if users are not using it to avoid channels.
* FPU support is not verified yet.
* Will not work properly if using a VM that's running besides other VMs/threads that are using the FPU.
Currently we assume CPU IDs to be linear and use HW CPU IDs as indices
to SMP-related arrays/state. This may change in the future (when
cluster-based SoC are supported).
cpuIndexToID converts a CPU index to a bit position in the case of ARM/GIC,
in order to send IPIs.
- don't print header every time
- drop address, it's not very useful and clutters the output
- add pretty header
- use new leading spaces feature in kprintf
When compiling the kernel as a whole program it is possible that these functions may
be inlined and not emitted in the resulting binary. However at the same time the
compiler may itself emit calls to these functions. Marking these functions as
externally visible tells the compiler that there may be more usages of them than
it sees immediately in the source code, in this cases usages that the compiler
itself is going to generate
In file included from ./kernel/src/api/faults.c:16:
In file included from ./kernel/include/api/syscall.h:19:
./kernel/include/api/debug.h:51:38: error: comparison of constant 'SysSend' (-3) with expression of type
'seL4_Word' (aka 'unsigned long') is always false [-Werror,-Wtautological-constant-out-of-range-compare]
if (ksKernelEntry.syscall_no == SysSend ||
~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~
./kernel/include/api/debug.h:52:42: error: comparison of constant 'SysNBSend' (-4) with expression of type
'seL4_Word' (aka 'unsigned long') is always false [-Werror,-Wtautological-constant-out-of-range-compare]
ksKernelEntry.syscall_no == SysNBSend ||
~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
./kernel/include/api/debug.h:53:42: error: comparison of constant 'SysCall' (-1) with expression of type
'seL4_Word' (aka 'unsigned long') is always false [-Werror,-Wtautological-constant-out-of-range-compare]
ksKernelEntry.syscall_no == SysCall) {
~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~
- when CONFIG_DEBUG is enabled, track all threads
- when CONFIG_PRINTING is enabled, provide seL4_DebugDumpScheduler which
allows the user to dump the state of the kernel scheduler.
The Arch_hasVCPU function instead of cheaking for whether a thread has a VCPU was
specifically checking for whether a thread on aarch32 had a VCPU, as that was the
condition needed to be passed into sanitiseRegister. This made the implementation
on x86 extremely confusing as Arch_hasVCPU was implemented to return false due to
there being no need for sanitiseRegister on x86 to know whether there was a vcpu.
This commit changes hasVCPU variables to be an abstract archInfo boolean that can
be used arbitrarily by an architecture. Arch_hasVCPU function was also changed to
become Arch_getSanitiseRegisterInfo whose result only passed to sanitiseRegister.
This version of compile_assert is not expanded during verification builds and
can be used in cases where you would like an assertion that contains syntax
that is not supported by the C parser
In the original slow path, armv_contextSwitch_HWASID shoots first.
Verification can't reorder machine operations on hardware state we don't
model. This puts the fastpath into the same order as the slow path.
invokeVCPUReadReg should not be setting message registers for the return message unless
the user performed a call. In doing so we must refactor the call to readVCPUReg to
outside the introduced `if` condition since, as it performs machine operations, it
should always happen
