The Linux kernel netfs subsystem had a vulnerability where netfs_release_folio() incorrectly updated the zero_point based on the local file size (i_size) rather than the remote server's file size (remote_i_size). This discrepancy occurs when there is uncommitted data in the pagecache beyond the folio being released but the server's EOF is within that folio (i_size > remote_i_size). The zero_point should be limited to remote_i_size because i_size reflects local cache updates not yet written to the server. This fix prevents erroneous assumptions about zero-filled data beyond the server EOF, which could affect read operations. The issue was discovered using the fsx test tool with specific operations on CIFS with default caching.

The vulnerability could cause the kernel to incorrectly assume zero-filled data beyond the server's EOF, potentially leading to incorrect read results or short reads. This may affect file system consistency or data integrity when reading cached data that is locally updated but not yet committed to the server. No known exploits in the wild have been reported.

A fix has been applied to the Linux kernel to correctly update zero_point using remote_i_size instead of i_size. Users should update to a kernel version that includes this fix. Since this is a kernel-level fix, upgrading to the patched kernel version is the recommended mitigation. Patch status is not explicitly confirmed here; check the official Linux kernel advisories or vendor updates for the exact fixed versions and apply updates accordingly.