- Added support for creating and managing opaque IPC NT handles in the
WDDM layer.
- Introduced a new flag `shareableWithoutNTHandle` to indicate if memory
can be shared without an NT handle.
- Updated the `isShareableMemory` method to accommodate this new flag.
- Added debug variable EnableShareableWithoutNTHandle to control the
behavior of sharing memory without NT handles until requested.
- Updated Linux path to enable sharing DMA Buf FDs between processes
for use in pidfd_getfd
- Updated getfd and get IPC Handle helper functions to support opaque or
previous versions
Related-To: NEO-15345 , NEO-15346 , NEO-15347, NEO-10380
Signed-off-by: Neil R. Spruit <neil.r.spruit@intel.com>
in case of CPU address range wider than GPU address range, we reserve
CPU address range to avoid CPU side allocations in our non-SVM heaps
Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
- Added support for creating and managing opaque IPC NT handles in the
WDDM layer.
- Introduced a new flag `shareableWithoutNTHandle` to indicate if memory
can be shared without an NT handle.
- Updated the `isShareableMemory` method to accommodate this new flag.
- Added debug variable EnableShareableWithoutNTHandle to control the
behavior of sharing memory without NT handles until requested.
- Updated Linux path to enable sharing DMA Buf FDs between processes
for use in pidfd_getfd
Related-To: NEO-15345 , NEO-15346 , NEO-15347, NEO-10380
Signed-off-by: Neil R. Spruit <neil.r.spruit@intel.com>
Related-To: NEO-15156, GSD-9939
Support for start address hint in zeVirtualMemReserve.
If it fails to find pStart then it defaults to the base line
allocateWithCustomAlignment(...)
Signed-off-by: Chandio, Bibrak Qamar <bibrak.qamar.chandio@intel.com>
Related-To: NEO-15156, GSD-9939
Support for start address hint in zeVirtualMemReserve.
If it fails to find pStart then it defaults to the base line
allocateWithCustomAlignment(...)
Signed-off-by: Chandio, Bibrak Qamar <bibrak.qamar.chandio@intel.com>
When allocation is saved for reuse, remove it from allocations map.
On get from reuse, reinsert it with new allocId.
Allocations saved for reuse should not be made resident for indirect
access.
Only applicable on aub/tbx csr types.
Related-To: NEO-6893
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
On getting from reuse update alloc id and increment allocations counter.
This is to ensure that making indirect allocations resident picks up the
reuse allocation.
Only applicable on aub/tbx csr types.
Related-To: NEO-6893
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
- readOnly ISA allocations must be written through CPU pointer
- command buffer allocation in singleAddressSpaceSbaTracking mode cannot
be readonly - it is written by SBA tracking commands
- this change is fixing ZET_ENABLE_PROGRAM_DEBUGGING mode 2
Related-To: GSD-10359
Signed-off-by: Mateusz Hoppe <mateusz.hoppe@intel.com>
on blocking free wait for engines completion before putting into reuse
mark allocation as completed and do not check is in use status on
getting
Related-To: NEO-6893, HSD-18042686276
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
Do not reuse internal or imported allocations.
Mark as aub writable on getting from reuse.
Related-To: NEO-6893
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
adjust ULTs
disable usm reuse in multi device l0
don't initialize usm caches if max reuse size is 0
check requested alignment when returning from cache
guard against double free
Related-To: NEO-6893
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
- Add isaPaddingIncluded flag to AllocationProperties to control ISA
padding
- Modify MemoryManager to skip automatic padding when flag is set
- Update ISAPoolAllocator to manage padding explicitly and set the
flag
- Use actual buffer size for HeapAllocator instead of requested size
- Add test covering different padding scenarios
This prevents double padding addition which was wasting memory when
ISAPoolAllocator managed its own padding on top of MemoryManager's
automatic padding.
Related-To: NEO-12287, HSD-18042276431
Signed-off-by: Fabian Zwoliński <fabian.zwolinski@intel.com>
Move getting usm pool for ptr to its own method.
Move trying to free via pooling to its own method.
Use base ptr of pool for tracking in IPCHandleMap.
Track ipc handle refcount in freeMem.
Return error when trying to use a not allocated ptr within pool.
Related-To: NEO-6893
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
Allow the application to force storageInfo.localOnly and get the
out-of-memory returned if not possible.
This is a windows-only feature supported on discrete platforms.
Related-To: NEO-13428
Signed-off-by: Maciej Bielski <maciej.bielski@intel.com>
DeferredDeleter instance could get deleted before worker thread called
unlock() on queue mutex leading to an error
Related-To: NEO-14896
Signed-off-by: Oskar Hubert Weber <oskar.hubert.weber@intel.com>
Related-To: NEO-14360
Current gmm usage type of these resources is causing
them to be cached, which is incorrect.
Signed-off-by: Szymon Morek <szymon.morek@intel.com>
Use device compression defaults when allocating pool.
Ignore compression hints when allocating from pool.
Also remove unused late initialization methods.
Related-To: NEO-6893
Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
