2023-06-23 21:11:33

by Tavian Barnes

Subject: [PATCH] fs/nfsd: Fix creation time serialization order

From: Tavian Barnes <[email protected]>

In nfsd4_encode_fattr(), TIME_CREATE was being written out after all
other times. However, they should be written out in an order that
matches the bit flags in bmval1, which in this case are

#define FATTR4_WORD1_TIME_ACCESS (1UL << 15)
#define FATTR4_WORD1_TIME_CREATE (1UL << 18)
#define FATTR4_WORD1_TIME_DELTA (1UL << 19)
#define FATTR4_WORD1_TIME_METADATA (1UL << 20)
#define FATTR4_WORD1_TIME_MODIFY (1UL << 21)

so TIME_CREATE should come second.

I noticed this on a FreeBSD NFSv4.2 client, which supports creation
times. On this client, file times were weirdly permuted. With this
patch applied on the server, times looked normal on the client.

Fixes: e377a3e698fb ("nfsd: Add support for the birth time attribute")
Link: https://unix.stackexchange.com/q/749605/56202
Signed-off-by: Tavian Barnes <[email protected]>
---
fs/nfsd/nfs4xdr.c | 14 +++++++-------
1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
index 76db2fe29624..3037c5b0623e 100644
--- a/fs/nfsd/nfs4xdr.c
+++ b/fs/nfsd/nfs4xdr.c
@@ -3354,6 +3354,13 @@ nfsd4_encode_fattr(struct xdr_stream *xdr, struct svc_fh *fhp,
p = xdr_encode_hyper(p, (s64)stat.atime.tv_sec);
*p++ = cpu_to_be32(stat.atime.tv_nsec);
}
+ if (bmval1 & FATTR4_WORD1_TIME_CREATE) {
+ p = xdr_reserve_space(xdr, 12);
+ if (!p)
+ goto out_resource;
+ p = xdr_encode_hyper(p, (s64)stat.btime.tv_sec);
+ *p++ = cpu_to_be32(stat.btime.tv_nsec);
+ }
if (bmval1 & FATTR4_WORD1_TIME_DELTA) {
p = xdr_reserve_space(xdr, 12);
if (!p)
@@ -3374,13 +3381,6 @@ nfsd4_encode_fattr(struct xdr_stream *xdr, struct svc_fh *fhp,
p = xdr_encode_hyper(p, (s64)stat.mtime.tv_sec);
*p++ = cpu_to_be32(stat.mtime.tv_nsec);
}
- if (bmval1 & FATTR4_WORD1_TIME_CREATE) {
- p = xdr_reserve_space(xdr, 12);
- if (!p)
- goto out_resource;
- p = xdr_encode_hyper(p, (s64)stat.btime.tv_sec);
- *p++ = cpu_to_be32(stat.btime.tv_nsec);
- }
if (bmval1 & FATTR4_WORD1_MOUNTED_ON_FILEID) {
u64 ino = stat.ino;

--
2.41.0

2023-06-23 21:51:49

by Chuck Lever

Subject: Re: [PATCH] fs/nfsd: Fix creation time serialization order

On Fri, Jun 23, 2023 at 05:09:06PM -0400, [email protected] wrote:
> From: Tavian Barnes <[email protected]>
>
> In nfsd4_encode_fattr(), TIME_CREATE was being written out after all
> other times. However, they should be written out in an order that
> matches the bit flags in bmval1, which in this case are
>
> #define FATTR4_WORD1_TIME_ACCESS (1UL << 15)
> #define FATTR4_WORD1_TIME_CREATE (1UL << 18)
> #define FATTR4_WORD1_TIME_DELTA (1UL << 19)
> #define FATTR4_WORD1_TIME_METADATA (1UL << 20)
> #define FATTR4_WORD1_TIME_MODIFY (1UL << 21)
>
> so TIME_CREATE should come second.
>
> I noticed this on a FreeBSD NFSv4.2 client, which supports creation
> times. On this client, file times were weirdly permuted. With this
> patch applied on the server, times looked normal on the client.
>
> Fixes: e377a3e698fb ("nfsd: Add support for the birth time attribute")
> Link: https://unix.stackexchange.com/q/749605/56202
> Signed-off-by: Tavian Barnes <[email protected]>

I'm not especially familiar with this area of the protocol, but this
looks correct at first glance. I've applied this to nfsd-fixes for
v6.5.

Out of interest, what type of filesystem does your server export?


> ---
> fs/nfsd/nfs4xdr.c | 14 +++++++-------
> 1 file changed, 7 insertions(+), 7 deletions(-)
>
> diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
> index 76db2fe29624..3037c5b0623e 100644
> --- a/fs/nfsd/nfs4xdr.c
> +++ b/fs/nfsd/nfs4xdr.c
> @@ -3354,6 +3354,13 @@ nfsd4_encode_fattr(struct xdr_stream *xdr, struct svc_fh *fhp,
> p = xdr_encode_hyper(p, (s64)stat.atime.tv_sec);
> *p++ = cpu_to_be32(stat.atime.tv_nsec);
> }
> + if (bmval1 & FATTR4_WORD1_TIME_CREATE) {
> + p = xdr_reserve_space(xdr, 12);
> + if (!p)
> + goto out_resource;
> + p = xdr_encode_hyper(p, (s64)stat.btime.tv_sec);
> + *p++ = cpu_to_be32(stat.btime.tv_nsec);
> + }
> if (bmval1 & FATTR4_WORD1_TIME_DELTA) {
> p = xdr_reserve_space(xdr, 12);
> if (!p)
> @@ -3374,13 +3381,6 @@ nfsd4_encode_fattr(struct xdr_stream *xdr, struct svc_fh *fhp,
> p = xdr_encode_hyper(p, (s64)stat.mtime.tv_sec);
> *p++ = cpu_to_be32(stat.mtime.tv_nsec);
> }
> - if (bmval1 & FATTR4_WORD1_TIME_CREATE) {
> - p = xdr_reserve_space(xdr, 12);
> - if (!p)
> - goto out_resource;
> - p = xdr_encode_hyper(p, (s64)stat.btime.tv_sec);
> - *p++ = cpu_to_be32(stat.btime.tv_nsec);
> - }
> if (bmval1 & FATTR4_WORD1_MOUNTED_ON_FILEID) {
> u64 ino = stat.ino;
>
> --
> 2.41.0
>

2023-06-23 22:07:54

by Tavian Barnes

Subject: Re: [PATCH] fs/nfsd: Fix creation time serialization order

On Fri, Jun 23, 2023 at 5:48 PM Chuck Lever <[email protected]> wrote:
> On Fri, Jun 23, 2023 at 05:09:06PM -0400, [email protected] wrote:
> > From: Tavian Barnes <[email protected]>
> >
> > In nfsd4_encode_fattr(), TIME_CREATE was being written out after all
> > other times. However, they should be written out in an order that
> > matches the bit flags in bmval1, which in this case are
> >
> > #define FATTR4_WORD1_TIME_ACCESS (1UL << 15)
> > #define FATTR4_WORD1_TIME_CREATE (1UL << 18)
> > #define FATTR4_WORD1_TIME_DELTA (1UL << 19)
> > #define FATTR4_WORD1_TIME_METADATA (1UL << 20)
> > #define FATTR4_WORD1_TIME_MODIFY (1UL << 21)
> >
> > so TIME_CREATE should come second.
> >
> > I noticed this on a FreeBSD NFSv4.2 client, which supports creation
> > times. On this client, file times were weirdly permuted. With this
> > patch applied on the server, times looked normal on the client.
> >
> > Fixes: e377a3e698fb ("nfsd: Add support for the birth time attribute")
> > Link: https://unix.stackexchange.com/q/749605/56202
> > Signed-off-by: Tavian Barnes <[email protected]>
>
> I'm not especially familiar with this area of the protocol, but this
> looks correct at first glance. I've applied this to nfsd-fixes for
> v6.5.

Great, thanks!

> Out of interest, what type of filesystem does your server export?

It's a btrfs filesystem.

--
Tavian Barnes

2023-06-27 16:23:42

by Jeff Layton

Subject: Re: [PATCH] fs/nfsd: Fix creation time serialization order

On Fri, 2023-06-23 at 17:09 -0400, [email protected] wrote:
> From: Tavian Barnes <[email protected]>
>
> In nfsd4_encode_fattr(), TIME_CREATE was being written out after all
> other times. However, they should be written out in an order that
> matches the bit flags in bmval1, which in this case are
>
> #define FATTR4_WORD1_TIME_ACCESS (1UL << 15)
> #define FATTR4_WORD1_TIME_CREATE (1UL << 18)
> #define FATTR4_WORD1_TIME_DELTA (1UL << 19)
> #define FATTR4_WORD1_TIME_METADATA (1UL << 20)
> #define FATTR4_WORD1_TIME_MODIFY (1UL << 21)
>
> so TIME_CREATE should come second.
>
> I noticed this on a FreeBSD NFSv4.2 client, which supports creation
> times. On this client, file times were weirdly permuted. With this
> patch applied on the server, times looked normal on the client.
>
> Fixes: e377a3e698fb ("nfsd: Add support for the birth time attribute")
> Link: https://unix.stackexchange.com/q/749605/56202
> Signed-off-by: Tavian Barnes <[email protected]>
> ---
> fs/nfsd/nfs4xdr.c | 14 +++++++-------
> 1 file changed, 7 insertions(+), 7 deletions(-)
>
> diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
> index 76db2fe29624..3037c5b0623e 100644
> --- a/fs/nfsd/nfs4xdr.c
> +++ b/fs/nfsd/nfs4xdr.c
> @@ -3354,6 +3354,13 @@ nfsd4_encode_fattr(struct xdr_stream *xdr, struct svc_fh *fhp,
> p = xdr_encode_hyper(p, (s64)stat.atime.tv_sec);
> *p++ = cpu_to_be32(stat.atime.tv_nsec);
> }
> + if (bmval1 & FATTR4_WORD1_TIME_CREATE) {
> + p = xdr_reserve_space(xdr, 12);
> + if (!p)
> + goto out_resource;
> + p = xdr_encode_hyper(p, (s64)stat.btime.tv_sec);
> + *p++ = cpu_to_be32(stat.btime.tv_nsec);
> + }
> if (bmval1 & FATTR4_WORD1_TIME_DELTA) {
> p = xdr_reserve_space(xdr, 12);
> if (!p)
> @@ -3374,13 +3381,6 @@ nfsd4_encode_fattr(struct xdr_stream *xdr, struct svc_fh *fhp,
> p = xdr_encode_hyper(p, (s64)stat.mtime.tv_sec);
> *p++ = cpu_to_be32(stat.mtime.tv_nsec);
> }
> - if (bmval1 & FATTR4_WORD1_TIME_CREATE) {
> - p = xdr_reserve_space(xdr, 12);
> - if (!p)
> - goto out_resource;
> - p = xdr_encode_hyper(p, (s64)stat.btime.tv_sec);
> - *p++ = cpu_to_be32(stat.btime.tv_nsec);
> - }
> if (bmval1 & FATTR4_WORD1_MOUNTED_ON_FILEID) {
> u64 ino = stat.ino;
>

Reviewed-by: Jeff Layton <[email protected]>