From: J. Bruce Fields <[email protected]>
Thanks to Casey for reminding me that 5661 gives a special meaning to a
value of 0 in the stateid's seqid field, so we should instead be
starting out stateid's with si_generation 1.
Reported-by: Casey Bodley <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
---
fs/nfsd/nfs4state.c | 6 +++---
1 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index 49c3dd1..7caa812 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -251,7 +251,7 @@ alloc_init_deleg(struct nfs4_client *clp, struct nfs4_stateid *stp, struct svc_f
dp->dl_stateid.si_boot = boot_time;
dp->dl_stateid.si_stateownerid = current_delegid++;
dp->dl_stateid.si_fileid = 0;
- dp->dl_stateid.si_generation = 0;
+ dp->dl_stateid.si_generation = 1;
fh_copy_shallow(&dp->dl_fh, ¤t_fh->fh_handle);
dp->dl_time = 0;
atomic_set(&dp->dl_count, 1);
@@ -2303,7 +2303,7 @@ init_stateid(struct nfs4_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *
stp->st_stateid.si_boot = boot_time;
stp->st_stateid.si_stateownerid = oo->oo_owner.so_id;
stp->st_stateid.si_fileid = fp->fi_id;
- stp->st_stateid.si_generation = 0;
+ stp->st_stateid.si_generation = 1;
stp->st_access_bmap = 0;
stp->st_deny_bmap = 0;
__set_bit(open->op_share_access & ~NFS4_SHARE_WANT_MASK,
@@ -3898,7 +3898,7 @@ alloc_init_lock_stateid(struct nfs4_lockowner *lo, struct nfs4_file *fp, struct
stp->st_stateid.si_boot = boot_time;
stp->st_stateid.si_stateownerid = lo->lo_owner.so_id;
stp->st_stateid.si_fileid = fp->fi_id;
- stp->st_stateid.si_generation = 0;
+ stp->st_stateid.si_generation = 1;
stp->st_access_bmap = 0;
stp->st_deny_bmap = open_stp->st_deny_bmap;
stp->st_openstp = open_stp;
--
1.7.4.1
On 2011-08-31 14:19, J. Bruce Fields wrote:
> On Wed, Aug 31, 2011 at 01:27:58PM -0700, Benny Halevy wrote:
>> Don't we also need the following?
>>
>> As per RFC5661, skip zero seqid when wrapping around
>> while updating a stateid.
>
> Yep, done:
>
> http://marc.info/?l=linux-nfs&m=131439774426154&w=2
yup, thanks!
>
> I need to send out my patches one at a time rather than fifteen at once
> if I want people to read them....
My bad :-)
Benny
>
> --b.
>
>>
>> Signed-off-by: Benny Halevy <[email protected]>
>>
>> git diff --stat -p -M
>> fs/nfsd/state.h | 5 ++++-
>> 1 files changed, 4 insertions(+), 1 deletions(-)
>>
>> diff --git a/fs/nfsd/state.h b/fs/nfsd/state.h
>> index 77f98b8..7d18d3e 100644
>> --- a/fs/nfsd/state.h
>> +++ b/fs/nfsd/state.h
>> @@ -289,7 +289,10 @@ struct nfs4_client_reclaim {
>> static inline void
>> update_stateid(stateid_t *stateid)
>> {
>> - stateid->si_generation++;
>> + u32 new = stateid->si_generation + 1;
>> + if (new == 0)
>> + new = 1;
>> + stateid->si_generation = new;
>> }
>>
>> /* A reasonable value for REPLAY_ISIZE was estimated as follows:
>>
>> On 2011-08-31 12:53, J. Bruce Fields wrote:
>>> From: J. Bruce Fields <[email protected]>
>>>
>>> Thanks to Casey for reminding me that 5661 gives a special meaning to a
>>> value of 0 in the stateid's seqid field, so we should instead be
>>> starting out stateid's with si_generation 1.
>>>
>>> Reported-by: Casey Bodley <[email protected]>
>>> Signed-off-by: J. Bruce Fields <[email protected]>
>>> ---
>>> fs/nfsd/nfs4state.c | 6 +++---
>>> 1 files changed, 3 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
>>> index 49c3dd1..7caa812 100644
>>> --- a/fs/nfsd/nfs4state.c
>>> +++ b/fs/nfsd/nfs4state.c
>>> @@ -251,7 +251,7 @@ alloc_init_deleg(struct nfs4_client *clp, struct nfs4_stateid *stp, struct svc_f
>>> dp->dl_stateid.si_boot = boot_time;
>>> dp->dl_stateid.si_stateownerid = current_delegid++;
>>> dp->dl_stateid.si_fileid = 0;
>>> - dp->dl_stateid.si_generation = 0;
>>> + dp->dl_stateid.si_generation = 1;
>>> fh_copy_shallow(&dp->dl_fh, ¤t_fh->fh_handle);
>>> dp->dl_time = 0;
>>> atomic_set(&dp->dl_count, 1);
>>> @@ -2303,7 +2303,7 @@ init_stateid(struct nfs4_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *
>>> stp->st_stateid.si_boot = boot_time;
>>> stp->st_stateid.si_stateownerid = oo->oo_owner.so_id;
>>> stp->st_stateid.si_fileid = fp->fi_id;
>>> - stp->st_stateid.si_generation = 0;
>>> + stp->st_stateid.si_generation = 1;
>>> stp->st_access_bmap = 0;
>>> stp->st_deny_bmap = 0;
>>> __set_bit(open->op_share_access & ~NFS4_SHARE_WANT_MASK,
>>> @@ -3898,7 +3898,7 @@ alloc_init_lock_stateid(struct nfs4_lockowner *lo, struct nfs4_file *fp, struct
>>> stp->st_stateid.si_boot = boot_time;
>>> stp->st_stateid.si_stateownerid = lo->lo_owner.so_id;
>>> stp->st_stateid.si_fileid = fp->fi_id;
>>> - stp->st_stateid.si_generation = 0;
>>> + stp->st_stateid.si_generation = 1;
>>> stp->st_access_bmap = 0;
>>> stp->st_deny_bmap = open_stp->st_deny_bmap;
>>> stp->st_openstp = open_stp;
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
Don't we also need the following?
As per RFC5661, skip zero seqid when wrapping around
while updating a stateid.
Signed-off-by: Benny Halevy <[email protected]>
git diff --stat -p -M
fs/nfsd/state.h | 5 ++++-
1 files changed, 4 insertions(+), 1 deletions(-)
diff --git a/fs/nfsd/state.h b/fs/nfsd/state.h
index 77f98b8..7d18d3e 100644
--- a/fs/nfsd/state.h
+++ b/fs/nfsd/state.h
@@ -289,7 +289,10 @@ struct nfs4_client_reclaim {
static inline void
update_stateid(stateid_t *stateid)
{
- stateid->si_generation++;
+ u32 new = stateid->si_generation + 1;
+ if (new == 0)
+ new = 1;
+ stateid->si_generation = new;
}
/* A reasonable value for REPLAY_ISIZE was estimated as follows:
On 2011-08-31 12:53, J. Bruce Fields wrote:
> From: J. Bruce Fields <[email protected]>
>
> Thanks to Casey for reminding me that 5661 gives a special meaning to a
> value of 0 in the stateid's seqid field, so we should instead be
> starting out stateid's with si_generation 1.
>
> Reported-by: Casey Bodley <[email protected]>
> Signed-off-by: J. Bruce Fields <[email protected]>
> ---
> fs/nfsd/nfs4state.c | 6 +++---
> 1 files changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> index 49c3dd1..7caa812 100644
> --- a/fs/nfsd/nfs4state.c
> +++ b/fs/nfsd/nfs4state.c
> @@ -251,7 +251,7 @@ alloc_init_deleg(struct nfs4_client *clp, struct nfs4_stateid *stp, struct svc_f
> dp->dl_stateid.si_boot = boot_time;
> dp->dl_stateid.si_stateownerid = current_delegid++;
> dp->dl_stateid.si_fileid = 0;
> - dp->dl_stateid.si_generation = 0;
> + dp->dl_stateid.si_generation = 1;
> fh_copy_shallow(&dp->dl_fh, ¤t_fh->fh_handle);
> dp->dl_time = 0;
> atomic_set(&dp->dl_count, 1);
> @@ -2303,7 +2303,7 @@ init_stateid(struct nfs4_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *
> stp->st_stateid.si_boot = boot_time;
> stp->st_stateid.si_stateownerid = oo->oo_owner.so_id;
> stp->st_stateid.si_fileid = fp->fi_id;
> - stp->st_stateid.si_generation = 0;
> + stp->st_stateid.si_generation = 1;
> stp->st_access_bmap = 0;
> stp->st_deny_bmap = 0;
> __set_bit(open->op_share_access & ~NFS4_SHARE_WANT_MASK,
> @@ -3898,7 +3898,7 @@ alloc_init_lock_stateid(struct nfs4_lockowner *lo, struct nfs4_file *fp, struct
> stp->st_stateid.si_boot = boot_time;
> stp->st_stateid.si_stateownerid = lo->lo_owner.so_id;
> stp->st_stateid.si_fileid = fp->fi_id;
> - stp->st_stateid.si_generation = 0;
> + stp->st_stateid.si_generation = 1;
> stp->st_access_bmap = 0;
> stp->st_deny_bmap = open_stp->st_deny_bmap;
> stp->st_openstp = open_stp;
On Wed, Aug 31, 2011 at 02:29:22PM -0700, Benny Halevy wrote:
> On 2011-08-31 14:19, J. Bruce Fields wrote:
> > On Wed, Aug 31, 2011 at 01:27:58PM -0700, Benny Halevy wrote:
> >> Don't we also need the following?
> >>
> >> As per RFC5661, skip zero seqid when wrapping around
> >> while updating a stateid.
> >
> > Yep, done:
> >
> > http://marc.info/?l=linux-nfs&m=131439774426154&w=2
>
> yup, thanks!
>
> >
> > I need to send out my patches one at a time rather than fifteen at once
> > if I want people to read them....
>
> My bad :-)
I'm gonna start inserting little "respond here if you read this far"
comments.
And actually I didn't read Casey's report carefully: he was just
complaining about delegation stateid's. I don't think they matter much,
as their seqid field never changes, but may as well use the same as the
others. And while we're there, make this uniform for v4.0 as well.
--b.
commit 80809fa40f8cfb3cf0bf006e3218bbd00584b38d
Author: J. Bruce Fields <[email protected]>
Date: Wed Aug 31 15:47:21 2011 -0400
nfsd4: make delegation stateid's seqid start at 1
Thanks to Casey for reminding me that 5661 gives a special meaning to a
value of 0 in the stateid's seqid field, so all stateid's should start
out with si_generation 1. We were doing that in the open and lock
cases for minorversion 1, but not for the delegation stateid, and not
for openstateid's with v4.0.
It doesn't *really* matter much for v4.0 or for delegation stateid's
(which never get the seqid field incremented), but we may as well do the
same for all of them.
Reported-by: Casey Bodley <[email protected]>
Signed-off-by: J. Bruce Fields <[email protected]>
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index 49c3dd1..dc83ca1 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -251,7 +251,7 @@ alloc_init_deleg(struct nfs4_client *clp, struct nfs4_stateid *stp, struct svc_f
dp->dl_stateid.si_boot = boot_time;
dp->dl_stateid.si_stateownerid = current_delegid++;
dp->dl_stateid.si_fileid = 0;
- dp->dl_stateid.si_generation = 0;
+ dp->dl_stateid.si_generation = 1;
fh_copy_shallow(&dp->dl_fh, ¤t_fh->fh_handle);
dp->dl_time = 0;
atomic_set(&dp->dl_count, 1);
@@ -2303,6 +2303,7 @@ init_stateid(struct nfs4_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *
stp->st_stateid.si_boot = boot_time;
stp->st_stateid.si_stateownerid = oo->oo_owner.so_id;
stp->st_stateid.si_fileid = fp->fi_id;
+ /* note will be incremented before first return to client: */
stp->st_stateid.si_generation = 0;
stp->st_access_bmap = 0;
stp->st_deny_bmap = 0;
@@ -2893,7 +2894,6 @@ nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nf
status = nfs4_upgrade_open(rqstp, fp, current_fh, stp, open);
if (status)
goto out;
- update_stateid(&stp->st_stateid);
} else {
status = nfs4_new_open(rqstp, &stp, fp, current_fh, open);
if (status)
@@ -2904,9 +2904,8 @@ nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nf
release_open_stateid(stp);
goto out;
}
- if (nfsd4_has_session(&resp->cstate))
- update_stateid(&stp->st_stateid);
}
+ update_stateid(&stp->st_stateid);
memcpy(&open->op_stateid, &stp->st_stateid, sizeof(stateid_t));
if (nfsd4_has_session(&resp->cstate))
@@ -3898,6 +3897,7 @@ alloc_init_lock_stateid(struct nfs4_lockowner *lo, struct nfs4_file *fp, struct
stp->st_stateid.si_boot = boot_time;
stp->st_stateid.si_stateownerid = lo->lo_owner.so_id;
stp->st_stateid.si_fileid = fp->fi_id;
+ /* note will be incremented before first return to client: */
stp->st_stateid.si_generation = 0;
stp->st_access_bmap = 0;
stp->st_deny_bmap = open_stp->st_deny_bmap;
On Wed, Aug 31, 2011 at 01:27:58PM -0700, Benny Halevy wrote:
> Don't we also need the following?
>
> As per RFC5661, skip zero seqid when wrapping around
> while updating a stateid.
Yep, done:
http://marc.info/?l=linux-nfs&m=131439774426154&w=2
I need to send out my patches one at a time rather than fifteen at once
if I want people to read them....
--b.
>
> Signed-off-by: Benny Halevy <[email protected]>
>
> git diff --stat -p -M
> fs/nfsd/state.h | 5 ++++-
> 1 files changed, 4 insertions(+), 1 deletions(-)
>
> diff --git a/fs/nfsd/state.h b/fs/nfsd/state.h
> index 77f98b8..7d18d3e 100644
> --- a/fs/nfsd/state.h
> +++ b/fs/nfsd/state.h
> @@ -289,7 +289,10 @@ struct nfs4_client_reclaim {
> static inline void
> update_stateid(stateid_t *stateid)
> {
> - stateid->si_generation++;
> + u32 new = stateid->si_generation + 1;
> + if (new == 0)
> + new = 1;
> + stateid->si_generation = new;
> }
>
> /* A reasonable value for REPLAY_ISIZE was estimated as follows:
>
> On 2011-08-31 12:53, J. Bruce Fields wrote:
> > From: J. Bruce Fields <[email protected]>
> >
> > Thanks to Casey for reminding me that 5661 gives a special meaning to a
> > value of 0 in the stateid's seqid field, so we should instead be
> > starting out stateid's with si_generation 1.
> >
> > Reported-by: Casey Bodley <[email protected]>
> > Signed-off-by: J. Bruce Fields <[email protected]>
> > ---
> > fs/nfsd/nfs4state.c | 6 +++---
> > 1 files changed, 3 insertions(+), 3 deletions(-)
> >
> > diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> > index 49c3dd1..7caa812 100644
> > --- a/fs/nfsd/nfs4state.c
> > +++ b/fs/nfsd/nfs4state.c
> > @@ -251,7 +251,7 @@ alloc_init_deleg(struct nfs4_client *clp, struct nfs4_stateid *stp, struct svc_f
> > dp->dl_stateid.si_boot = boot_time;
> > dp->dl_stateid.si_stateownerid = current_delegid++;
> > dp->dl_stateid.si_fileid = 0;
> > - dp->dl_stateid.si_generation = 0;
> > + dp->dl_stateid.si_generation = 1;
> > fh_copy_shallow(&dp->dl_fh, ¤t_fh->fh_handle);
> > dp->dl_time = 0;
> > atomic_set(&dp->dl_count, 1);
> > @@ -2303,7 +2303,7 @@ init_stateid(struct nfs4_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *
> > stp->st_stateid.si_boot = boot_time;
> > stp->st_stateid.si_stateownerid = oo->oo_owner.so_id;
> > stp->st_stateid.si_fileid = fp->fi_id;
> > - stp->st_stateid.si_generation = 0;
> > + stp->st_stateid.si_generation = 1;
> > stp->st_access_bmap = 0;
> > stp->st_deny_bmap = 0;
> > __set_bit(open->op_share_access & ~NFS4_SHARE_WANT_MASK,
> > @@ -3898,7 +3898,7 @@ alloc_init_lock_stateid(struct nfs4_lockowner *lo, struct nfs4_file *fp, struct
> > stp->st_stateid.si_boot = boot_time;
> > stp->st_stateid.si_stateownerid = lo->lo_owner.so_id;
> > stp->st_stateid.si_fileid = fp->fi_id;
> > - stp->st_stateid.si_generation = 0;
> > + stp->st_stateid.si_generation = 1;
> > stp->st_access_bmap = 0;
> > stp->st_deny_bmap = open_stp->st_deny_bmap;
> > stp->st_openstp = open_stp;
On 2011-08-31 17:33, J. Bruce Fields wrote:
> On Wed, Aug 31, 2011 at 02:29:22PM -0700, Benny Halevy wrote:
>> On 2011-08-31 14:19, J. Bruce Fields wrote:
>>> On Wed, Aug 31, 2011 at 01:27:58PM -0700, Benny Halevy wrote:
>>>> Don't we also need the following?
>>>>
>>>> As per RFC5661, skip zero seqid when wrapping around
>>>> while updating a stateid.
>>>
>>> Yep, done:
>>>
>>> http://marc.info/?l=linux-nfs&m=131439774426154&w=2
>>
>> yup, thanks!
>>
>>>
>>> I need to send out my patches one at a time rather than fifteen at once
>>> if I want people to read them....
>>
>> My bad :-)
>
> I'm gonna start inserting little "respond here if you read this far"
> comments.
>
> And actually I didn't read Casey's report carefully: he was just
> complaining about delegation stateid's. I don't think they matter much,
> as their seqid field never changes, but may as well use the same as the
> others. And while we're there, make this uniform for v4.0 as well.
Ack :)
>
> --b.
>
> commit 80809fa40f8cfb3cf0bf006e3218bbd00584b38d
> Author: J. Bruce Fields <[email protected]>
> Date: Wed Aug 31 15:47:21 2011 -0400
>
> nfsd4: make delegation stateid's seqid start at 1
>
> Thanks to Casey for reminding me that 5661 gives a special meaning to a
> value of 0 in the stateid's seqid field, so all stateid's should start
> out with si_generation 1. We were doing that in the open and lock
> cases for minorversion 1, but not for the delegation stateid, and not
> for openstateid's with v4.0.
>
> It doesn't *really* matter much for v4.0 or for delegation stateid's
> (which never get the seqid field incremented), but we may as well do the
> same for all of them.
>
> Reported-by: Casey Bodley <[email protected]>
> Signed-off-by: J. Bruce Fields <[email protected]>
>
> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> index 49c3dd1..dc83ca1 100644
> --- a/fs/nfsd/nfs4state.c
> +++ b/fs/nfsd/nfs4state.c
> @@ -251,7 +251,7 @@ alloc_init_deleg(struct nfs4_client *clp, struct nfs4_stateid *stp, struct svc_f
> dp->dl_stateid.si_boot = boot_time;
> dp->dl_stateid.si_stateownerid = current_delegid++;
> dp->dl_stateid.si_fileid = 0;
> - dp->dl_stateid.si_generation = 0;
> + dp->dl_stateid.si_generation = 1;
> fh_copy_shallow(&dp->dl_fh, ¤t_fh->fh_handle);
> dp->dl_time = 0;
> atomic_set(&dp->dl_count, 1);
> @@ -2303,6 +2303,7 @@ init_stateid(struct nfs4_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *
> stp->st_stateid.si_boot = boot_time;
> stp->st_stateid.si_stateownerid = oo->oo_owner.so_id;
> stp->st_stateid.si_fileid = fp->fi_id;
> + /* note will be incremented before first return to client: */
> stp->st_stateid.si_generation = 0;
> stp->st_access_bmap = 0;
> stp->st_deny_bmap = 0;
> @@ -2893,7 +2894,6 @@ nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nf
> status = nfs4_upgrade_open(rqstp, fp, current_fh, stp, open);
> if (status)
> goto out;
> - update_stateid(&stp->st_stateid);
> } else {
> status = nfs4_new_open(rqstp, &stp, fp, current_fh, open);
> if (status)
> @@ -2904,9 +2904,8 @@ nfsd4_process_open2(struct svc_rqst *rqstp, struct svc_fh *current_fh, struct nf
> release_open_stateid(stp);
> goto out;
> }
> - if (nfsd4_has_session(&resp->cstate))
> - update_stateid(&stp->st_stateid);
> }
> + update_stateid(&stp->st_stateid);
> memcpy(&open->op_stateid, &stp->st_stateid, sizeof(stateid_t));
>
> if (nfsd4_has_session(&resp->cstate))
> @@ -3898,6 +3897,7 @@ alloc_init_lock_stateid(struct nfs4_lockowner *lo, struct nfs4_file *fp, struct
> stp->st_stateid.si_boot = boot_time;
> stp->st_stateid.si_stateownerid = lo->lo_owner.so_id;
> stp->st_stateid.si_fileid = fp->fi_id;
> + /* note will be incremented before first return to client: */
> stp->st_stateid.si_generation = 0;
> stp->st_access_bmap = 0;
> stp->st_deny_bmap = open_stp->st_deny_bmap;
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html