The x86_64 ABI requires the stack pointer to be 16-byte aligned before the
call of a function and decreased by 8 at the function entrypoint (after
the return address has been pushed to the stack).
Currently, when a new Genode thread gets created, the initial stack
pointer is aligned to 16 byte. On Genode/Linux, the thread entry function
is entered by a 'call' instruction, so the stack pointer alignment at the
function entrypoint is correct. On Fiasco.OC and NOVA, however, the thread
entry function gets executed without a return address being pushed to the
stack, so at the function entrypoint the stack pointer is still aligned to
16 byte, which can cause problems with compiler-generated SSE
instructions.
With this patch, the stack pointer given to a new thread gets aligned to
16 bytes and decreased by 8 by default, since most of the currently
supported base platforms execute the thread entry function without pushing
a return address to the stack. For base-linux, the stack pointer gets
realigned to 16 bytes before the thread entry function gets called.
Fixes#1043.
Use a bit allocator for the allocation management of thread contexts,
instead of holding allocation information within the Thread_base objects,
which lead to race conditions in the past.
Moreover, extend the Thread_base class interface with the ability to
to add additional stacks to a thread, and associate the context they're
located in with the corresponding Thread_base object. Additional stacks
can be used to do user-level scheduling with stack switching, without breaking
Genode's API.
Fixes#1024Fixes#1036
Since RM sessions can be used as dataspaces and dataspace sizes are
supposed to have page granularity, RM session sizes should have page
granularity, too.
Fixes#972.
In hybrid mode, all programs are dynamic executables and ld.lib.so is
not supported. Therefore, only the "static" variant of the test can be
build and executed.
Revised region management detects region conflicts by using _soft_
mappings per default. Overmapping is activated for population of managed
dataspaces only. For more information see header documentation of
base-linux/src/base/env/rm_session_mmap.cc.
Fixes#883.
To prevent multiple execution of main-bootstrap, I moved the code to a
statically initialized object. The reason for this change is that
_main() is exeuted twice when starting dynamic binaries. Now, the object
is part of the base-common library which is linked with ld.lib.so.
For ARM support on N900, commit 4a9b1c6 changed the process library to
start dynamic binaries directly depending on the Linux kernel to comply
to the interp section info ("ld.lib.so"). This seems not required on
more recent platforms or kernel versions and also introduced challenging
corner cases in region handling on Linux. Therefore, this commit
restores the original behavior.
With this patch, if an ELF dataspace to be executed does not have a path
in the host file system, the dataspace content gets copied into a
temporary file whose path can be given to 'execve()'.
Fixes#879.
- if no affinity was set for a new thread before calling
Cpu_session::start(), the CPU session's affinity gets set for this
thread
- documentation fix: <affinity_space> -> <affinity-space>
Fixes#873.
This patch extends the 'Parent::session()' and 'Root::session()'
functions with an additional 'affinity' parameter, which is inteded to
express the preferred affinity of the new session. For CPU sessions
provided by core, the values will be used to select the set of CPUs
assigned to the CPU session. For other services, the session affinity
information can be utilized to optimize the locality of the server
thread with the client. For example, to enable the IRQ session to route
an IRQ to the CPU core on which the corresponding device driver (the IRQ
client) is running.
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.
- use the generic 'crt0.s' for Linux
- move the read-only '__dso_handle' definition into the '.text' section
- move the '__initial_sp' definition into the '.bss' section
- remove the '_main_utcb' definition
Part of #766.
This patch introduces a new platform 'linux_arm' for building and running
Genode/Linux on an ARM device.
Known limitations:
- libc 'setjmp()'/'longjmp()' doesn't currently save/restore floating
point registers
Fixes#746.
genode_until_run can be called now with a spawn id to able to reattach to a
spawned process (amt, serial output). Run scripts can now call genode_until_run
multiple times.
Try to free the metadata of a thread only if created by Genode. The code
did not cause any actual problems. I just stumbled over the inconsistency
while reviewing the code.
With this patch, the 'futex' syscall gets used for blocking and unblocking
of threads in the Linux-specific lock implementation.
The 'Native_thread_id' type, which was previously used in the
lock-internal 'Applicant' class to identify a thread to be woken up,
was not suitable anymore for implementing this change. With this patch,
the 'Thread_base*' type gets used instead, which also has the positive
effect of making the public 'cancelable_lock.h' header file
platform-independent.
Fixes#646.
This happened when I killed one of the genode clients which was
tracked via an expception_handler. In this case the wait4 syscall
returned 0 and the for(;;) was looped eternally. This caused an
100% CPU utilization for the core binary.
This patch simplifies the way of how Genode's base libraries are
organized. Originally, the base API was implemented in the form of many
small libraries such as 'thread', 'env', 'server', etc. Most of them
used to consist of only a small number of files. Because those libraries
are incorporated in any build, the checking of their inter-dependencies
made the build process more verbose than desired. Also, the number of
libraries and their roles (core only, non-core only, shared by both core
and non-core) were not easy to capture.
Hereby, the base libraries have been reduced to the following few
libraries:
- startup.mk contains the startup code for normal Genode processes.
On some platform, core is able to use the library as well.
- base-common.mk contains the parts of the base library that are
identical by core and non-core processes.
- base.mk contains the complete base API implementation for non-core
processes
Consequently, the 'LIBS' declaration in 'target.mk' files becomes
simpler as well. In the most simple case, only the 'base' library must
be mentioned.
Fixes#18
The distinction between 'ipc.h' and 'ipc_generic.h' is no more. The only
use case for platform-specific extensions of the IPC support was the
marshalling of capabilities. However, this case is accommodated by a
function interface ('_marshal_capability', '_unmarshal_capability'). By
moving the implementation of these functions from the headers into the
respective ipc libraries, we can abandon the platform-specific 'ipc.h'
headers.
core does not use POSIX threads when built for the 'lx_hybrid_x86'
platform, so we need to reserve the thread-context area via a segment in
the program to prevent clashes with vdso and shared libraries.
Fixes#639.
Some shared libraries of the host system contain search paths for finding
other needed shared libraries. These paths get evaluated only by a native
linker. To find all needed shared libraries, with this patch, the host
linker is used to link hybrid applications.
Fixes#645.
If page faults are handled concurrently (as for base-nova) the traverse lookup
call in rm_session_component must be thread safe, which it isn't.
If the faulting area is backed by nested dataspaces which are managed by
various rm_sessions then a race happens under following circumstances
(triggered occasionally by the bomb test).
The traverse lookup may return a pointer to a rm_session of a nested dataspace.
If the rm_session is in parallel subject to destruction it happened that faults
got enqueued to the faulters list of the deleted rm_session and internally to
a list of the current rm_session of the Rm_client.
During destruction of the faulting Rm_client the associated rm_session will
be dissolved from the Rm_client, which leads to dereferencing the
dangling pointer of the already destructed rm_session.
On base-nova the memory of the rm_session object get unmapped eventually, so
that the de-referencing of the dangling pointer caused page faults in core.
The memory on other kernels inside core never get unmapped so that the
bug doesn't trigger visible faults.
The patch replace the keeping of a rm_session pointer by keeping a
capability instead. The rm_session object must be looked up now explicitly in
the Object_pool implementation, which implements proper reference counting on
the rm_session object.
Issue #549
