fix: Fix unrecoverable abort during device destruction

An inappropriate ioctl call for the device is observed when both zesInit
and zeInit are called in a process and Sysman device reset is called.
A condition has been added in the UNRECOVERABLE_IF to prevent crash
during cleanup.

Related-To: NEO-16439

Signed-off-by: Pratik Bari <[email protected]>

Author: pratikbariintel

shared/source/os_interface/linux/drm_neo.cpp

@@ -425,14 +425,14 @@ void Drm::destroyDrmContext(uint32_t drmContextId) {
     GemContextDestroy destroy{};
     destroy.contextId = drmContextId;
     auto retVal = ioctlHelper->ioctl(DrmIoctl::gemContextDestroy, &destroy);
-    UNRECOVERABLE_IF((retVal != 0) && (errno != ENODEV));
+    UNRECOVERABLE_IF((retVal != 0) && (errno != ENODEV) && (errno != ENOTTY));
 }
 
 void Drm::destroyDrmVirtualMemory(uint32_t drmVmId) {
     GemVmControl ctl = {};
     ctl.vmId = drmVmId;
     auto ret = ioctlHelper->ioctl(DrmIoctl::gemVmDestroy, &ctl);
-    UNRECOVERABLE_IF((ret != 0) && (errno != ENODEV));
+    UNRECOVERABLE_IF((ret != 0) && (errno != ENODEV) && (errno != ENOTTY));
 }
 
 int Drm::queryVmId(uint32_t drmContextId, uint32_t &vmId) {