On Mon, 2020-11-16 at 17:41 +0000, Christoph Hellwig wrote: > On Mon, Nov 16, 2020 at 09:37:32AM -0800, Linus Torvalds wrote: > > > This discussion seems to be going down the path of requiring an IMA > > > filesystem hook for reading the file, again. That solution was > > > rejected, not by me. What is new this time? > > > > You can't read a non-read-opened file. Not even IMA can. > > > > So don't do that then. > > > > IMA is doing something wrong. Why would you ever read a file that can't be read? > > > > Fix whatever "open" function instead of trying to work around the fact > > that you opened it wrong. > > The "issue" with IMA is that it uses security hooks to hook into the > VFS and then wants to read every file that gets opened on a real file > system to "measure" the contents vs a hash stashed away somewhere. > Which has always been rather sketchy. There are security hooks, where IMA is co-located, but there are also IMA hooks where there isn't an IMA hook (e.g. ima_file_check). In all cases, the file needs to be read in order to calculate the file hash, which is then used for verifying file signatures (equivalent of secure boot) and extending the TPM (equivalent of trusted boot). Only after measuring and verifying the file integrity, should access be granted to the file. Whether filesystems can and should be trusted to provide the real file hashes is a separate issue. The decision as to which files should be measured or the signature verified is based on policy. thanks, Mimi
