The vulnerability in the Linux kernel drm/gpusvm component arises because VMAs marked with VM_IO or VM_PFNMAP are not backed by struct page objects, which are necessary for GPUSVM operations. Specifically, get_pages() depends on hmm_range_fault() to resolve struct pages for the target memory range. When an SVM range is created on these VMAs, get_pages() repeatedly fails, potentially causing an infinite loop within the GPU driver's page-fault handler. The resolution involves rejecting VMAs with VM_IO or VM_PFNMAP flags when creating SVM ranges and returning an -EIO error code to prevent the infinite loop scenario.

If exploited, this vulnerability could cause a GPU driver's page-fault handler to enter an infinite loop, potentially leading to denial of service or system instability. There is no indication of privilege escalation or code execution. No known exploits are reported in the wild.

A fix has been implemented in the Linux kernel to reject VMAs with VM_IO or VM_PFNMAP flags when creating SVM ranges, returning an error to prevent infinite loops. Users should apply the official kernel updates that include this fix. Since no patch links or vendor advisories are provided here, users should consult their Linux distribution's security advisories for the relevant kernel update.