Re: [PATCH v3] fs: Invalidate the cache for a parent block-device if fsync() is called for a partition
From: Andrew Morton <akpm@linux-foundation.org>
Date: 2012-01-31 18:58:26
Also in:
linux-fsdevel
On Tue, 31 Jan 2012 16:00:44 +0000 Niels de Vos [off-list ref] wrote:
On 01/26/2012 09:45 PM, Christoph Hellwig wrote:quoted
On Thu, Jan 26, 2012 at 01:40:51PM -0800, Andrew Morton wrote:quoted
The Right Thing To Do here is to make the kernel behave logically and predictably, then modify the userspace tools. But if we're modifying the userspace tools then we would just change userspace to issue a BLKFLSBUF to /dev/sda and leave the kernel alone.The right fix is to make partition and whole disk access coherent, which is fairly simply: - create the block device inode/mapping per gendisk, and only reference count it per block_device - make sure blkdev_get_block(s) applies the correct offset if used on partitionsThis surely looks like a better way to fix this issue. I am not sure yet how much work that would involve and if I am the right person to fix this. If nobody beats me to it, I might send a patch for review some (undefined) time later.
One concern I have with the proposal is that it would forever rule out support of >16T devices on 32-bit machines. At present with 64-bit sector_t and 32-bit pgoff_t, I think we'd have a reasonable chance of supporting, say, four 8T partitions on a 32T device. But if we were to switch the kernel from using four 4T address_spaces (sda1-4) over to using a single 32T address_space (sda) then we can rule it all out.