In a multi-subDevice device:
- If creating the queue with the parent device handle:
- Use subDevice0 engines.
- If creating the queue with one of the subDevice handles:
- Use that subDevice engines
In a device without subDevices:
- No changes. All engines of device are used.
Change-Id: I11cf08cbab08d168deb135111f884c0c023ba384
Signed-off: Jaime Arteaga <jaime.a.arteaga.molina@intel.com>
- Enables truncation on windows for kernel name > 256 to
account for the null terminator.
Change-Id: I97ba4d5a65d9675d6c0da70b8eab5bdff28b7dc0
Signed-off-by: Spruit, Neil R <neil.r.spruit@intel.com>
Add support for Platform Monitoring Technology,
and read temperature and energy counters.
Change-Id: Ic2c79e902400148cc538267c498fc1d0185c81f8
Signed-off-by: Matias Cabral <matias.a.cabral@intel.com>
- When peforming Image Copy from/to memory, the size of the image and thus the
memory should be expected to be at maximum the size of the
image width, height, & depth in the user's descriptor.
- When a user passes in a nullptr for the destination or source region the
width, height, & depth of the image region should match the image descriptor
that was passed in by the user at image create.
- The Physical image allocated may be larger than the user requested.
This new size cannot be used in the Copy to/From memory otherwise device data
corruption will occur.
Change-Id: I950f6fd1d5ddc9911dd235f84e1cf00618e4e2a3
Signed-off-by: Spruit, Neil R <neil.r.spruit@intel.com>
In this Change:
- Initialize local variables used in pci source.
- Move setting of temperature handle's type to constructor of
TemperatureImp class as temperature handle's type is used in
init() of TemperatureImp class.
- Allow only ENOENT errno at the end of FsAccess::listDirectory function.
Because in this function all directory entries will be read in a
loop. And loop will exit only after readdir() finished reading
all entries in directory and thus readdir() will exit, when it fails
reading any further entry. Thus system will return ENOENT for readdir.
Change-Id: Ibd3ac3f841b114fd87edea1410750b33f3a14e5b
Signed-off-by: Jitendra Sharma <jitendra.sharma@intel.com>
- added dual handle support for RAS Correctable and Uncorrectable Errors.
- added reset counter for RAS.
- added Os Specific ULT for RAS
Change-Id: Ia10115bf6720ab211f549571e810ec0d6c0801ec
- Changed Global Driver to be a void * to avoid auto add of Global
Driver Destructor to run before destruction of other L0 data structures
that might be enqueued to destory in static object destructors.
- Added register of library unload driverdestructor to cleanup
driver/device as the last destructor run.
Change-Id: I8ba6c5c27424b942a86a2613edd52fc682ab1c64
Signed-off-by: Spruit, Neil R <neil.r.spruit@intel.com>
Use hardware limit, instead of the one used for device capabilities,
so applications can fully use the available memory in the device.
Change-Id: I910c610d7a3af254724a810c3c60b9da8d5d64a7
Signed-off: Jaime Arteaga <jaime.a.arteaga.molina@intel.com>
If after reading clients directory, while actually trying to
access the individual clients file, we receive a error
ZET_RESULT_ERROR_NOT_AVAILABLE, then it means that this process/client
might be killed/removed. Hence dont return error to user.
Instead simply skip that client's traversal further and move on
to next available client.
Change-Id: If86610a89d7b2421fd85560e5a223391b76d78a0
Signed-off-by: Jitendra Sharma <jitendra.sharma@intel.com>
Create two handles for Global and Compute temperature
sensors.
Change-Id: I5da90135a803df71372c9c4be9dc891ca64fc214
Signed-off-by: Jitendra Sharma <jitendra.sharma@intel.com>
