The vulnerability in the Linux kernel's nfsd component involves the function _nfsd_copy_file_range() which samples the destination file's writeback error cursor after the copy loop to detect errors. Because the nfsd_file cache reuses a single struct file across requests for the same inode, concurrent COMMIT or stable WRITE operations can advance the writeback error cursor between the error appearing and the sampling by the copy worker. This causes errseq_check() to incorrectly return zero, setting NFSD4_COPY_F_COMMITTED and encoding wr_stable_how = FILE_SYNC4, leading the client to treat the copied data as durable even though writeback failed, resulting in silent data loss. The resolution involves sampling the writeback error cursor once at the start of the function to correctly detect errors occurring during the copy regardless of thread timing, aligning with patterns used in other nfsd write operations.

The vulnerability can cause silent data loss where the NFS client believes data has been durably written when in fact writeback errors occurred and the data was not properly persisted. This undermines data integrity guarantees in NFS file copy operations, potentially leading to corrupted or lost data without immediate detection.

A fix has been implemented in the Linux kernel to address this issue by adjusting when the writeback error cursor is sampled to ensure accurate error detection. Users should apply the official kernel update that includes this fix. Patch status is not explicitly stated in the provided data; therefore, verify the vendor advisory or Linux kernel release notes for the specific patch version and apply it accordingly.