On Fri, Mar 07, 2025 at 09:53:40AM -0800, Dave Hansen wrote: >On 3/7/25 08:41, Chao Gao wrote: >> Note that this issue does not cause any functional problems because the >> guest fpstate is allocated using vmalloc(), which aligns the size to a >> full page, providing enough space for all existing supervisor components. >> On Emerald Rapids CPUs, the guest fpstate after this correction is ~2880 >> bytes. > >How about we move up the fpstate pointers at allocation time so they >just scrape the end of the vmalloc buffer? Basically, move the >page-alignment padding to the beginning of the first page instead of the >end of the last page. That sounds like a good way to detect similar errors and might be helpful for all other vmalloc'ed buffers. I can try to implement this for the fpstate pointers. The patch will be put at the end of the series or even in a separate series.
