On Mon, Aug 18, 2008 at 05:24:46PM -0700, Andrew Morton wrote: > On Mon, 18 Aug 2008 15:32:24 +0200 > Nick Piggin <npiggin@xxxxxxx> wrote: > > XEN and PAT and such do not like deferred TLB flushing because they can't > > always handle multiple aliasing virtual addresses to a physical address. They > > now call vm_unmap_aliases() in order to flush any deferred mappings. That call > > is very expensive (well, actually not a lot more expensive than a single vunmap > > under the old scheme), however it should be OK if not called too often. > > What are the prospects now for making vunmap safe from atomic (or > interrupt) contexts? That's something which people keep on trying to > do and all the other memory-freeing functions permit it. We've tried lazy unmap with dma_free_coherent() on ARM and had one report of success and another of filesystem corruption. Thankfully vmap isn't used for this, but is used for ARMs ioremap. > > +/*** Per cpu kva allocator ***/ > > + > > +/* > > + * vmap space is limited especially on 32 bit architectures. Ensure there is > > + * room for at least 16 percpu vmap blocks per CPU. > > + */ > > +#if 0 /* constant vmalloc space size */ > > +#define VMALLOC_SPACE (VMALLOC_END-VMALLOC_START) > > kill? > > > +#else > > +#if BITS_PER_LONG == 32 > > +#define VMALLOC_SPACE (128UL*1024*1024) > > +#else > > +#define VMALLOC_SPACE (128UL*1024*1024*1024) > > +#endif > > +#endif > > So VMALLOC_SPACE has type unsigned long, whereas it previously had type > <god-knows-what-usually-unsigned-long>. Fair enough. So the generic code knows enough about all the platforms Linux runs on to be able to dictate that there shall be 128MB of space available on all platforms? Second question - will ARMs separate module area still work with this code in place (which allocates regions in a different address space using __get_vm_area and __vmalloc_area)? -- Russell King Linux kernel 2.6 ARM Linux - http://www.arm.linux.org.uk/ maintainer of: -- To unsubscribe from this list: send the line "unsubscribe linux-arch" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
