still nfs problems [Was: Linux 2.6.37-rc8]

[Trond Myklebust]

On Wed, 2011-01-05 at 15:52 +0000, Russell King - ARM Linux wrote: 
> On Wed, Jan 05, 2011 at 10:14:17AM -0500, Trond Myklebust wrote:
> > OK. So,the new behaviour in 2.6.37 is that we're writing to a series of
> > pages via the usual kmap_atomic()/kunmap_atomic() and kmap()/kunmap()
> > interfaces, but we can end up reading them via a virtual address range
> > that gets set up via vm_map_ram() (that range gets set up before the
> > write occurs).
> 
> kmap of lowmem pages will always reuses the existing kernel direct
> mapping, so there won't be a problem there.
> 
> > Do we perhaps need an invalidate_kernel_vmap_range() before we can read
> > the data on ARM in this kind of scenario?
> 
> Firstly, vm_map_ram() does no cache maintainence of any sort, nor does
> it take care of page colouring - so any architecture where cache aliasing
> can occur will see this problem.  It should not limited to ARM.
> 
> Secondly, no, invalidate_kernel_vmap_range() probably isn't sufficient.
> There's two problems here:
> 
> 	addr = kmap(lowmem_page);
> 	*addr = stuff;
> 	kunmap(lowmem_page);
> 
> Such lowmem pages are accessed through their kernel direct mapping.
> 
> 	ptr = vm_map_ram(lowmem_page);
> 	read = *ptr;
> 
> This creates a new mapping which can alias with the kernel direct mapping.
> Now, as this is a new mapping, there should be no cache lines associated
> with it.  (Looking at vm_unmap_ram(), it calls free_unmap_vmap_area_addr(),
> free_unmap_vmap_area(), which then calls flush_cache_vunmap() on the
> region.  vb_free() also calls flush_cache_vunmap() too.)
> 
> If the write after kmap() hits an already present cache line, the cache
> line will be updated, but it won't be written back to memory.  So, on
> a subsequent vm_map_ram(), with any kind of aliasing cache, there's
> no guarantee that you'll hit that cache line and read the data just
> written there.
> 
> The kernel direct mapping would need to be flushed.

We should already be flushing the kernel direct mapping after writing by
means of the calls to flush_dcache_page() in xdr_partial_copy_from_skb()
and all the helpers in net/sunrpc/xdr.c.

The only new thing is the read access through the virtual address
mapping. That mapping is created outside the loop in
nfs_readdir_xdr_to_array(), which is why I'm thinking we do need the
invalidate_kernel_vmap_range(): we're essentially doing a series of
writes through the kernel direct mapping (i.e. readdir RPC calls), then
reading the results through the virtual mapping.

i.e. we're doing

ptr = vm_map_ram(lowmem_pages);
while (need_more_data) {

for (i = 0; i < npages; i++) {
addr = kmap_atomic(lowmem_page[i]);
*addr = rpc_stuff;
flush_dcache_page(lowmem_page[i]);
kunmap_atomic(lowmem_page[i]);
}

invalidate_kernel_vmap_range(ptr); // Needed here?

read = *ptr;
}
vm_unmap_ram(lowmem_pages)

> I'm really getting to the point of hating the poliferation of RAM
> remapping interfaces - it's going to (and is) causing nothing but lots
> of pain on virtual cache architectures, needing more and more cache
> flushing interfaces to be created.
> 
> Is there any other solution to this?

Arbitrary sized pages. :-)

The problem here is that we want to read variable sized records (i.e.
readdir() records) from a multi-page buffer. We could do that by copying
those particular records that overlap with page boundaries, but that
would make for a fairly intrusive rewrite too.

-- 
Trond Myklebust
Linux NFS client maintainer

NetApp
Trond.Myklebust at netapp.com
www.netapp.com