On Mon, Nov 14, 2022 at 12:45:16PM -0800, Song Liu wrote: > On Sun, Nov 13, 2022 at 2:43 AM Mike Rapoport <rppt@xxxxxxxxxx> wrote: > > > > > > If RO data causes problems with direct map fragmentation, we can use > > > similar logic. I think we will need another tree in vmalloc for this case. > > > Since the logic will be mostly identical, I personally don't think adding > > > another tree is a big overhead. > > > > Actually, it would be interesting to quantify memory savings/waste as the > > result of using execmem_alloc() > > From a random system in our fleet, execmem_alloc() saves: > > 139 iTLB entries (1x 2MB entry vs, 140x 4kB entries), which is more than > 100% of L1 iTLB and about 10% of L2 TLB. Using 2M pages saves page table entries. They might be cached in iTLB and might be not because on a loaded system large part of iTBL will cache userspace mappings. > It wastes 1.5MB memory, which is 0.0023% of system memory (64GB). > > I believe this is clearly a good trade-off. The actual trade-off would be for 1.5MB of memory for yet unknown, or at least unpublished, performance improvement on a loaded system. > Thanks, > Song -- Sincerely yours, Mike.
