Thanks, # this is a thread of my older series, v7 (current one is v9), tho your comment is still valid so, thanks !
On Thu, 19 Jun 2025 20:43:06 +0900, Benjamin Berg wrote: > > Hi, > > I am not sure that fxsaveq/fxrstorq is guaranteed to be sufficient. I > don't think that catches some of the newer AVX registers for example. I've added fxsaveq/fxrstorq parts with your suggestion, test-signal-restore.c. I think I may not know all of userspace situations with register usage so, my current approach is to address each issue when we faced real issue with a concrete test case. > At the same time, dealing with all that properly could be rather > annoying. I don't really have a solution for that, normal UM does not > need to deal with many of those details … thanks for the input. I'm also/always trying to study a lot from existing code base. -- Hajime
