| Message ID | 20200623223904.31643-1-fllinden@amazon.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | client side user xattr (RFC8276) support | expand |
Hi Trond, Anna
Any comments on this series? Do you think it can be included for 5.9?
The server side patches are looking pretty good for 5.9.
Let me know if there's anything you want me to do/test.
Thanks,
- Frank
On 6/23/20, 3:39 PM, "Frank van der Linden" <fllinden@amazon.com> wrote:
v3:
* Rebase to v5.8-rc2
v2:
* Move nfs4 specific definitions to nfs_fs4.h
* Squash some patches together to avoid unused function warnings
when bisecting.
* Made determining server support a two-step process. First,
the extended attribute FATTR is verified to be supported, then
the value of the attributed is queried in fsinfo to determine
support.
* Fixed up Makefile to remove an unneeded extra line.
v1:
* Client side caching is now included in this patch set.
* As per the discussion, user extended attributes are enabled if
the client and server support them (e.g. they support 4.2 and
advertise the user extended attribute FATTR). There are no longer
options to switch them off on either the client or the server.
* The code is no longer conditioned on a config option.
* The number of patches has been reduced somewhat by merging
smaller, related ones.
Original combined client/server RFC:
https://patchwork.kernel.org/cover/11143565/
In general, these patches were, both server and client, tested as
follows:
* stress-ng-xattr with 1000 workers
* Test all corner cases (XATTR_SIZE_*)
* Test all failure cases (no xattr, setxattr with different or
invalid flags, etc).
* Verify the content of xattrs across several operations.
* Use KASAN and KMEMLEAK for a longer mix of testruns to verify
that there were no leaks (after unmounting the filesystem).
* Interop run against FreeBSD server/client implementation.
* Ran xfstests-dev, with no unexpected/new failures as compared
to an unpatched kernel. To fully use xfstests-dev, it needed
some modifications, as it expects to either use all xattr
namespaces, or none. Whereas NFS only suppors the "user."
namespace (+ optional ACLs). I will send the changes in
seperately.
* Test the memory shrinker for the client side cache by running
the client inside a 1G KVM VM, filling up the cache by running
an agressive multi-threaded xattr workload, and tracing the
shrinker when it kicked in. No memory allocation problems were
seen.
Frank van der Linden (13):
nfs,nfsd: NFSv4.2 extended attribute protocol definitions
nfs: add client side only definitions for user xattrs
NFSv4.2: define limits and sizes for user xattr handling
NFSv4.2: query the server for extended attribute support
NFSv4.2: add client side XDR handling for extended attributes
nfs: define nfs_access_get_cached function
NFSv4.2: query the extended attribute access bits
nfs: modify update_changeattr to deal with regular files
nfs: define and use the NFS_INO_INVALID_XATTR flag
nfs: make the buf_to_pages_noslab function available to the nfs code
NFSv4.2: add the extended attribute proc functions.
NFSv4.2: hook in the user extended attribute handlers
NFSv4.2: add client side xattr caching.
fs/nfs/Makefile | 2 +-
fs/nfs/client.c | 22 +-
fs/nfs/dir.c | 24 +-
fs/nfs/inode.c | 16 +-
fs/nfs/nfs42.h | 24 +
fs/nfs/nfs42proc.c | 248 ++++++++
fs/nfs/nfs42xattr.c | 1083 +++++++++++++++++++++++++++++++++++
fs/nfs/nfs42xdr.c | 438 ++++++++++++++
fs/nfs/nfs4_fs.h | 35 ++
fs/nfs/nfs4client.c | 31 +
fs/nfs/nfs4proc.c | 237 +++++++-
fs/nfs/nfs4super.c | 10 +
fs/nfs/nfs4xdr.c | 31 +
fs/nfs/nfstrace.h | 3 +-
include/linux/nfs4.h | 25 +
include/linux/nfs_fs.h | 12 +
include/linux/nfs_fs_sb.h | 6 +
include/linux/nfs_xdr.h | 60 +-
include/uapi/linux/nfs4.h | 3 +
include/uapi/linux/nfs_fs.h | 1 +
20 files changed, 2269 insertions(+), 42 deletions(-)
create mode 100644 fs/nfs/nfs42xattr.c
v3: * Rebase to v5.8-rc2 v2: * Move nfs4 specific definitions to nfs_fs4.h * Squash some patches together to avoid unused function warnings when bisecting. * Made determining server support a two-step process. First, the extended attribute FATTR is verified to be supported, then the value of the attributed is queried in fsinfo to determine support. * Fixed up Makefile to remove an unneeded extra line. v1: * Client side caching is now included in this patch set. * As per the discussion, user extended attributes are enabled if the client and server support them (e.g. they support 4.2 and advertise the user extended attribute FATTR). There are no longer options to switch them off on either the client or the server. * The code is no longer conditioned on a config option. * The number of patches has been reduced somewhat by merging smaller, related ones. Original combined client/server RFC: https://patchwork.kernel.org/cover/11143565/ In general, these patches were, both server and client, tested as follows: * stress-ng-xattr with 1000 workers * Test all corner cases (XATTR_SIZE_*) * Test all failure cases (no xattr, setxattr with different or invalid flags, etc). * Verify the content of xattrs across several operations. * Use KASAN and KMEMLEAK for a longer mix of testruns to verify that there were no leaks (after unmounting the filesystem). * Interop run against FreeBSD server/client implementation. * Ran xfstests-dev, with no unexpected/new failures as compared to an unpatched kernel. To fully use xfstests-dev, it needed some modifications, as it expects to either use all xattr namespaces, or none. Whereas NFS only suppors the "user." namespace (+ optional ACLs). I will send the changes in seperately. * Test the memory shrinker for the client side cache by running the client inside a 1G KVM VM, filling up the cache by running an agressive multi-threaded xattr workload, and tracing the shrinker when it kicked in. No memory allocation problems were seen. Frank van der Linden (13): nfs,nfsd: NFSv4.2 extended attribute protocol definitions nfs: add client side only definitions for user xattrs NFSv4.2: define limits and sizes for user xattr handling NFSv4.2: query the server for extended attribute support NFSv4.2: add client side XDR handling for extended attributes nfs: define nfs_access_get_cached function NFSv4.2: query the extended attribute access bits nfs: modify update_changeattr to deal with regular files nfs: define and use the NFS_INO_INVALID_XATTR flag nfs: make the buf_to_pages_noslab function available to the nfs code NFSv4.2: add the extended attribute proc functions. NFSv4.2: hook in the user extended attribute handlers NFSv4.2: add client side xattr caching. fs/nfs/Makefile | 2 +- fs/nfs/client.c | 22 +- fs/nfs/dir.c | 24 +- fs/nfs/inode.c | 16 +- fs/nfs/nfs42.h | 24 + fs/nfs/nfs42proc.c | 248 ++++++++ fs/nfs/nfs42xattr.c | 1083 +++++++++++++++++++++++++++++++++++ fs/nfs/nfs42xdr.c | 438 ++++++++++++++ fs/nfs/nfs4_fs.h | 35 ++ fs/nfs/nfs4client.c | 31 + fs/nfs/nfs4proc.c | 237 +++++++- fs/nfs/nfs4super.c | 10 + fs/nfs/nfs4xdr.c | 31 + fs/nfs/nfstrace.h | 3 +- include/linux/nfs4.h | 25 + include/linux/nfs_fs.h | 12 + include/linux/nfs_fs_sb.h | 6 + include/linux/nfs_xdr.h | 60 +- include/uapi/linux/nfs4.h | 3 + include/uapi/linux/nfs_fs.h | 1 + 20 files changed, 2269 insertions(+), 42 deletions(-) create mode 100644 fs/nfs/nfs42xattr.c