c...@cubidou.net said: > > keep out paddr_t from the kernel ABI relevant to modules. > You won't keep bus_addr_t out of the ABI any time soon, so this is not > a fix.
It is not correct anyway to tie bus_addr_t to paddr_t -- the semantics of the buses don't depend on the CPU but on chipsets, or are defined MI. So leaving bus_addr_t at 32bit would be a good first step to get things untangled. > Anyway, this is why I ask about measurements. That would at least > allow a discussion based on facts instead of suppositions. Measurements are good, but then they need also to be done on a low-end board, like the somewhat popular Soekris. (I've got an ElanSC520 board with 16M RAM myself.) There is also increased kernel memory usage. Largest part due to growth of vm_page probably, and cache footprint caused by it. And if the box is memory starved anyway, the more the page lists are walked the more it matters. Just want to say one can prove or disprove anything with the right benchmark. m...@eterna.com.au said: > > "i386pae" and "i386"? > FWIW, having a different module load path and choosing the right set > of modules is an incoming feature but i have no time line for it. The appeal of such a solution is that it can be taken back easily if it is not needed anymore. If paddr_t gets extended it will stay, and there will be little motication to clean up the APIs. best regards Matthias ------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------ Forschungszentrum Juelich GmbH 52425 Juelich Sitz der Gesellschaft: Juelich Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498 Vorsitzender des Aufsichtsrats: MinDirig Dr. Karl Eugen Huthmacher Geschaeftsfuehrung: Prof. Dr. Achim Bachem (Vorsitzender), Dr. Ulrich Krafft (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt, Prof. Dr. Sebastian M. Schmidt ------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------
