This patch introduces new types for expressing CPU affinities. Instead
of dealing with physical CPU numbers, affinities are expressed as
rectangles in a grid of virtual CPU nodes. This clears the way to
conveniently assign sets of adjacent CPUs to subsystems, each of them
managing their respective viewport of the coordinate space.
By using 2D Cartesian coordinates, the locality of CPU nodes can be
modeled for different topologies such as SMP (simple Nx1 grid), grids of
NUMA nodes, or ring topologies.
This commit simplifies the creation of additional threads and VCPUs in L4linux.
By now, some Genode::Thread_base methods where overridden to use a Fiasco.OC
specific Cpu_session when creating threads.
Recent commit: 297538678e moved the actual creation
of the platform thread into the constructor of the generic Thread_base class.
Thereby the Vcpu class, which extended the Thread_base class, now unnecessarily
created two platform threads for each thread created via Vcpu. Nowadays, the
cpu_session capability is available via the Genode::env() environment. So we can
use the Thread_base parent class for the setup of the platform thread, and
afterwards create a Fiasco.OC specific cpu session client with the same cpu
session capability, Thread_base used for creation, to make use of the L4Linux
specific features of this interface (VCPU enabling, irq object creation etc.).
This commit comprises the following changes to enable L4Linux to use several
CPUs:
* change default configuration for x86 and ARM
* add atomic cmpxchg operation to l4re library
* implement l4_sleep (per thread)
* enable setting affinity for VCPUs and IRQs
* move "per CPU" section within linker-script (x86 only)
* introduce SMP run-script for pandaboard
* deactivate direct scheduler (Fiasco.OC syscall) access by L4Linux
This commit fixes several issues that were triggered e.g. by the
'noux_tool_chain' run-script (fix#208 in part). The following problems
are tackled:
* Don't reference count capability selectors within a task that are actually
controlled by core (all beneath 0x200000), because it's undecideable which
"version" of a capability selector we currently use, e.g. a thread gets
destroyed and a new one gets created immediately some other thread might
have a Native_capability pointing to the already destroyed thread's gate
capability-slot, that is now a new valid one (the one of the new thread)
* In core we cannot invalidate and remove a capability from the so called
Cap_map before each reference to it is destroyed, so don't do this in
Cap_session_component::free, but only reference-decrement within there,
the actual removal can only be done in Cap_map::remove. Because core also
has to invalidate a capability to be removed in all protection-domains
we have to implement a core specific Cap_map::remove method
* When a capability gets inserted into the Cap_map, and we detect an old
invalid entry with the dame id in the tree, don't just overmap that
invalid entry (as there exist remaining references to it), but just remove
it from the tree and allocate an new entry.
* Use the Cap_session_component interface to free a Pager_object when it
gets dissolved, as its also used for allocation
Due to recently introduces smart-pointers to Cap_index objects it's
necessary to always keep at least one reference as long as a corresponding
slot in the capability-space of a process is in use. This is especially
important for L4Linux that uses cap-slots directly without the given
abstractions of Genode.
This commit introduces a Cap_index class for Fiasco.OC's capabilities.
A Cap_index is a combination of the global capability id, that is used by Genode
to correctly identify a kernel-object, and a corresponding entry in a
protection-domain's (kernel-)capability-space. The cap-indices are non-copyable,
unique objects, that are held in a Cap_map. The Cap_map is used to re-find
capabilities already present in the protection-domain, when a capability is
received via IPC. The retrieval of capabilities effectively fixes issue #112,
meaning the waste of capability-space entries.
Because Cap_index objects are non-copyable (their address indicates the position
in the capability-space of the pd), they are inappropriate to use as
Native_capability. Therefore, Native_capability is implemented as a reference
to Cap_index objects. This design seems to be a good pre-condition to implement
smart-pointers for entries in the capability-space, and thereby closing existing
leaks (please refer to issue #32).
Cap_index, Cap_map, and the allocator for Cap_index objects are designed in a way,
that it should be relatively easy to apply the same concept to NOVA also. By now,
these classes are located in the `base-foc` repository, but they intentionally
contain no Fiasco.OC specific elements.
The previously explained changes had extensive impact on the whole Fiasco.OC
platform implementation, due to various dependencies. The following things had to
be changed:
* The Thread object's startup and destruction routine is re-arranged, to
enable another thread (that calls the Thread destructor) gaining the
capability id of the thread's gate to remove it from the Cap_map, the
thread's UTCB had to be made available to the caller, because there
is the current location of that id. After having the UTCB available
in the Thread object for that reason, the whole thread bootstrapping
could be simplified.
* In the course of changing the Native_capability's semantic, a new Cap_mapping
class was introduced in core, that facilitates the establishment and
destruction of capability mappings between core and it's client's, especially
mappings related to Platform_thread and Platform_task, that are relevant to
task and thread creation and destruction. Thereby, the destruction of
threads had to be reworked, which effectively removed a bug (issue #149)
where some threads weren't destroyed properly.
* In the quick fix for issue #112, something similar to the Cap_map was
introduced available in all processes. Moreover, some kind of a capability
map already existed in core, to handle cap-session request properly. The
introduction of the Cap_map unified both structures, so that the
cap-session component code in core had to be reworked too.
* The platform initialization code had to be changed sligthly due to the
changes in Native_capability
* The vcpu initialization in the L4Linux support library had to be adapted
according to the already mentioned changes in the Thread object's bootstrap
code.
Norman Feske
Stefan Kalkowski
Alexander Boettcher
Martin Stein
Genode Labs