| Age | Commit message (Collapse) | Author |
|---|
|
This commit introduces initial support for mandatory access control. As
one may recall, L5 follows "everything is memory". In order to interact
with a resource, a process must request it from the kernel in the form
of a (sometimes) syncable memory buffer.
Each resource as well as processes have an access level, if a process
attempts to request a resource with a higher access level than it, the
request is rejected by the kernel. However, if a process has a greater
than or equal access level as a resource, the request can be granted.
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
These sometimes cause issues so we are better off just stripping them
from the final binary.
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
As we rely on resetting the controller via GHC.HR, we'll need to
re-establish a link with each implemented port to actually bring them
online.
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
This commit introduces this initial bring-up logic for ports on the HBA.
We have allocated the command headers and each FIS RX area for each
respective command slot. More work to be done but this lays the
groundwork
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Typically filesystems are tree-like, however certain filesystems in the
form of images have fixed paths baked into them. This optimizes handling
so there isn't a need to unpack it into a rootfs/tmpfs
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Abort if the host controller does not support 64-bit addressing
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Here we ensure that the port is in a stopped state, we have also added
port descriptor lists to keep track of each port.
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
As of now, the bootloader handles setting up the MMIO regions and we'll
need huge page support (2 megs) for reassigning MMIO space
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
The global host control register (described by section 3.1.2 in the AHCI
spec) has an HBA reset bit which resets *everything* to a known state.
This means that after the reset, we'll have to indicate to the HBA that
we want to speak to it over the AHCI interface. We do it once before the
reset if needed, then we do it again after.
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
This commit introduces this initial ahci_init() logic by resetting HBA
state on startup.
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Add new clock device callback to get the elapsed time from the timer
being initialized, in microseconds.
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
This commit introduces a new framework we call "clkdev". It is used to
keep track of the clock sources on the system and be able to fetch them
later based on specific attributes.
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Here we add a write(2) stub and add it to the UNIX syscall interface. We
also move the UNIX syscall numbers into compat/unix/syscall.h
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Add routine to check if an address is valid within a process's address
space. More work will need to be done with this but this introduces a
great starting point
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
If we do not do this, we'll get stale x/y values the the characters will
never advance position when using multichar strings
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
The PCI_BAR_IO() checks if the BAR is an I/O type BAR. The value 1 is
returned if true, otherwise 0 if it is a memory type BAR.
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
This commit adds support for page-level cacheability attributes. We have
added the pmap_set_cache() as well as some MMU_CACHE_* bits
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
When an attached device is detected, the driver will need information
about the PCI device for the bring up process.
Signed-off-by: Ian Moffett <ian@osmora.org>
|
|
Signed-off-by: Ian Moffett <ian@osmora.org>
|
