On Thu, Sep 17, 2020 at 07:58:59PM -0700, Eric Biggers wrote:
On Thu, Sep 17, 2020 at 10:05:35PM -0400, Sasha Levin wrote:
From: Theodore Ts'o tytso@mit.edu
[ Upstream commit 244adf6426ee31a83f397b700d964cff12a247d3 ]
This fixes the direct I/O versus writeback race which can reveal stale data, and it improves the tail latency of commits on slow devices.
Link: https://lore.kernel.org/r/20200125022254.1101588-1-tytso@mit.edu Signed-off-by: Theodore Ts'o tytso@mit.edu Signed-off-by: Sasha Levin sashal@kernel.orgI
Any particular reason to be backporting this? I thought I saw some fixes for dioread_nolock go by, after it was made the default. Are you getting all of those fixes too?
Agreed, making dioread_nolock the default has enough issues that it's not something that I'd suggest backporting at this point. It's a fundamental behavioral change that it's not something we should change in a stable kernel.
- Ted