Skip to contentSkip to conversation

nfsd41: fix fore channel attribute initialization

PATCH In review [email protected] 3 participants 11 messages Sep 11, 2009 – Sep 21, 2009 2 messages not in the archive

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover nfsd41: fix fore channel attribute initialization

Fix forechannel attribute initialization

These two patches fix ca_maxresponsesize_cached, ca_maxresponsesize, and ca_maxrequestsize calculations for the forechannel.

Tested:

Kerberos NFSv4.1 mount:
Passes Connectathon tests. Passes the expected pynfs tests when tests are run individually.
When the pynfs tests are run all at once, there are too many sessions created, and the server code (init_forechannel_attrs) works as planned and fails the CREATE SESSION with NFS4ERR_SERVERFAULT.

The pynfs 4.1 tests need to be fixed to clean up after themselves!

-->Andy

0001-nfsd41-fix-NFSD_MIN_HDR_SEQ_SZ.patch 0002-nfsd41-add-RPC-header-size-to-fore-channel-negotiat.patch

#1 lore raw

PATCH 1/2 nfsd41: fix NFSD_MIN_HDR_SEQ_SZ

From: Andy Adamson <[email protected]>

The reply RPC header is 12 bytes, the NULL verifier is 8 bytes giving 20 not 24 bytes in the NFSD_MIN_HDR_SEQ_SZ calculation,

---
 fs/nfsd/nfs4state.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)
fs/nfsd/nfs4state.c +2 -2
index 46e9ac5..5ecb42c 100644
@@ -420,12 +420,12 @@ gen_sessionid(struct nfsd4_session *ses)
  * each time.  Therefore we can advertise a ca_maxresponssize_cached
  * value that is the number of bytes in our cache plus a few additional
  * bytes.  In order to stay on the safe side, and not promise more than
- * we can cache, those additional bytes must be the minimum possible: 24
+ * we can cache, those additional bytes must be the minimum possible: 20
  * bytes of rpc header (xid through accept state, with AUTH_NULL
  * verifier), 12 for the compound header (with zero-length tag), and 44
  * for the SEQUENCE op response:
  */
-#define NFSD_MIN_HDR_SEQ_SZ  (24 + 12 + 44)
+#define NFSD_MIN_HDR_SEQ_SZ  (20 + 12 + 44)
 
 /*
  * Give the client the number of ca_maxresponsesize_cached slots it
Signature · 1 line
1.6.2.5
#2 lore raw

PATCH 2/2 nfsd41: add RPC header size to fore channel negotiation

From: Andy Adamson <[email protected]>

Both the max request and the max response size include the RPC header with credential (request only) and verifier as well as the payload.

The RPCSEC_GSS credential and verifier are the largest. Kerberos is the only supported GSS security mechansim, so the Kerberos GSS credential and verifier sizes are used.

---
 fs/nfsd/nfs4state.c |   29 +++++++++++++++++++++++++----
 1 files changed, 25 insertions(+), 4 deletions(-)
fs/nfsd/nfs4state.c +25 -4
index 5ecb42c..e4d8f94 100644
@@ -428,6 +428,27 @@ gen_sessionid(struct nfsd4_session *ses)
 #define NFSD_MIN_HDR_SEQ_SZ  (20 + 12 + 44)
 
 /*
+ * The protocol defines ca_maxrequestsize as the XDR encoded size of the
+ * request including the RPC headers (including security flavor credentials
+ * and verifiers) but excludes any RPC transport framing headers.
+ * So we need to advertize a ca_maxrequestsize value that is the number of
+ * bytes of the maximum payload we support, plus some additional bytes to cover
+ * the maximum RPC header size. The RPCSEC_GSS security flavor has the
+ * largest credential and verifier, so we add 24 bytes for the RPC call header
+ * (xid through proceedure), 32 bytes of GSS credential, and 48 bytes of
+ * Kerberos GSS verifier.
+ */
+#define NFSD_MAX_CALL_HDR_SZ (24 + 32 + 48)
+
+/* The protocol defines ca_maxresponsesize as also including the RPC headers
+ * just as in ca_maxrequestsize. Once again, we use the maximum supported
+ * payload plus the largest RPC reply header which uses the RPCSEC_GSS
+ * security flavor. We add 12 bytes of RPC reply header (xid through 
+ * reply state) and 48 bytes of GSS Kerberos verifier.
+ */
+#define NFSD_MAX_REPLY_HDR_SZ (12 + 48)
+
+/*
  * Give the client the number of ca_maxresponsesize_cached slots it
  * requests, of size bounded by NFSD_SLOT_CACHE_SIZE,
  * NFSD_MAX_MEM_PER_SESSION, and nfsd_drc_max_mem. Do not allow more
@@ -488,12 +509,12 @@ static int init_forechannel_attrs(struct svc_rqst *rqstp,
 	/* headerpadsz set to zero in encode routine */
 
 	/* Use the client's max request and max response size if possible */
-	if (fchan->maxreq_sz > maxcount)
-		fchan->maxreq_sz = maxcount;
+	if (fchan->maxreq_sz > maxcount + NFSD_MAX_CALL_HDR_SZ)
+		fchan->maxreq_sz = maxcount + NFSD_MAX_CALL_HDR_SZ;
 	session_fchan->maxreq_sz = fchan->maxreq_sz;
 
-	if (fchan->maxresp_sz > maxcount)
-		fchan->maxresp_sz = maxcount;
+	if (fchan->maxresp_sz > maxcount + NFSD_MAX_REPLY_HDR_SZ)
+		fchan->maxresp_sz = maxcount + NFSD_MAX_REPLY_HDR_SZ;
 	session_fchan->maxresp_sz = fchan->maxresp_sz;
 
 	/* Use the client's maxops if possible */
Signature · 1 line
1.6.2.5
#3 lore raw

Re: [PATCH 2/2] nfsd41: add RPC header size to fore channel negotiation

Quoted 8 lines · On Fri, Sep 11, 2009 at 06:52:55PM -0400, [email protected] wrote:

From: Andy Adamson <[email protected]>

Both the max request and the max response size include the RPC header with credential (request only) and verifier as well as the payload.

The RPCSEC_GSS credential and verifier are the largest. Kerberos is the only supported GSS security mechansim, so the Kerberos GSS credential and verifier sizes are used.

Rather than trying to estimate this is might be simplest just to use what the server's using to allocate memory: RPCSVC_MAXPAGES. No, that also takes into account space for the reply. You could do

	PAGE_SIZE * (1 + (RPCSVC_MAXPAYLOAD+PAGE_SIZE-1)/PAGE_SIZE)

Actually, by design the server's real limit is actually on the sum of the request and the reply sizes.

What happens if we get a request such that both the request and reply are under our advertised limits, but the sum is too much? Can we just declare that no client will be that weird and that we shouldn't have to worry about it?

--b.

#4 lore raw

Re: [pnfs] [PATCH 2/2] nfsd41: add RPC header size to fore channel negotiation

Quoted 4 lines · On Mon, Sep 14, 2009 at 3:03 PM, J. Bruce Fields <[email protected]>= wrote:
Quoted 3 lines · On Fri, Sep 11, 2009 at 06:52:55PM -0400, [email protected] wrote:

From: Andy Adamson <[email protected]>

Both the max request and the max response size include the RPC heade=

r with

Quoted 3 lines
Quoted 3 lines

credential (request only) =A0and verifier as well as the payload.

The RPCSEC_GSS credential and verifier are the largest. Kerberos is =

the only

Quoted 1 line
Quoted 1 line

supported GSS security mechansim, so the Kerberos GSS credential and=

verifier

Quoted 4 lines
Quoted 1 line

sizes are used.

Rather than trying to estimate this is might be simplest just to use what the server's using to allocate memory: RPCSVC_MAXPAGES. =A0No, t=

hat

Quoted 3 lines

also takes into account space for the reply. =A0You could do

=A0 =A0 =A0 =A0PAGE_SIZE * (1 + (RPCSVC_MAXPAYLOAD+PAGE_SIZE-1)/PAGE_=

SIZE)

Quoted 3 lines

Actually, by design the server's real limit is actually on the sum of the request and the reply sizes.

I think the actual limit is svc_max_payload rounded up to a multiple of PAGE_SIZE plus PAGE_SIZE. which is a lot smaller than the sum of the request and reply sizes. See below.

Note that svc_max_payload is what is returned in nfs4_encode_fattr for MAXREAD and for MAXWRITE. These attributes use svc_max_payload in the same way this patch does - the maximum data size not including rpc headers.

I don't think the server wants is to advertise a MAXREAD/WRITE that it can't supply because the fore channel maxrequest/maxresponse is too small, so some additional space needs to be added to svc_max_payload for the fore channel.

Quoted 3 lines

What happens if we get a request such that both the request and reply are under our advertised limits, but the sum is too much? =A0Can we j=

ust

Quoted 1 line

declare that no client will be that weird and that we shouldn't have =

to

Quoted 1 line

worry about it?

I think the server already has this problem. In svc_init_buffer which sets up the pages for a server thread request/response handling, it uses sv_max_mesg / PAGE_SIZE + 1 with the comment

"extra page as we hold both request and reply. We assume one is at most one page"

where
sv_max_mesg =3D roundup(serv->sv_max_payload + PAGE_SIZE, PAGE_SIZE).

-->Andy

Quoted 7 lines

--b.
_______________________________________________
pNFS mailing list
[email protected]
http://linux-nfs.org/cgi-bin/mailman/listinfo/pnfs

A message that is not in the archive (sent off-list or not yet received): 89c397150909180947w69b0139dx732100109e2793b0-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org

#6 lore raw

Re: [pnfs] [PATCH 2/2] nfsd41: add RPC header size to fore channel negotiation

On Fri, Sep 18, 2009 at 12:47:57PM -0400, William A. (Andy) Adamson wro= te:

Quoted 1 line

On Mon, Sep 14, 2009 at 3:03 PM, J. Bruce Fields <bfields-uC3wQj2KruMpug/[email protected]=

Quoted 4 lines · g> wrote:
Quoted 4 lines · On Fri, Sep 11, 2009 at 06:52:55PM -0400, [email protected] wrote:
Quoted 3 lines

From: Andy Adamson <[email protected]>

Both the max request and the max response size include the RPC hea=

der with

Quoted 1 line
Quoted 1 line
Quoted 1 line

credential (request only) =C2=A0and verifier as well as the payloa=

d.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

The RPCSEC_GSS credential and verifier are the largest. Kerberos i=

s the only

Quoted 1 line
Quoted 1 line
Quoted 1 line

supported GSS security mechansim, so the Kerberos GSS credential a=

nd verifier

Quoted 3 lines
Quoted 3 lines
Quoted 1 line

sizes are used.

Rather than trying to estimate this is might be simplest just to us=

e

Quoted 1 line
Quoted 1 line

what the server's using to allocate memory: RPCSVC_MAXPAGES. =C2=A0=

No, that

Quoted 3 lines
Quoted 3 lines

also takes into account space for the reply. =C2=A0You could do

=C2=A0 =C2=A0 =C2=A0 =C2=A0PAGE_SIZE * (1 + (RPCSVC_MAXPAYLOAD+PAGE=

_SIZE-1)/PAGE_SIZE)

Quoted 2 lines
Quoted 2 lines

Actually, by design the server's real limit is actually on the sum =

of

Quoted 5 lines
Quoted 1 line

the request and the reply sizes.

=20
I think the actual limit is svc_max_payload rounded up to a multiple of PAGE_SIZE plus PAGE_SIZE. which is a lot smaller than the sum of the request and reply sizes. See below.

Right. I think you're agreeing with me?

Quoted 1 line

Note that svc_max_payload is what is returned in nfs4_encode_fattr fo=

r

Quoted 5 lines

MAXREAD and for MAXWRITE. These attributes use svc_max_payload in the same way this patch does - the maximum data size not including rpc headers.
=20
I don't think the server wants is to advertise a MAXREAD/WRITE that i=

t

Quoted 3 lines

can't supply because the fore channel maxrequest/maxresponse is too small, so some additional space needs to be added to svc_max_payload for the fore channel.

Yes.

Quoted 1 line
Quoted 1 line

What happens if we get a request such that both the request and rep=

ly

Quoted 1 line
Quoted 1 line

are under our advertised limits, but the sum is too much? =C2=A0Can=

we just

Quoted 1 line
Quoted 1 line

declare that no client will be that weird and that we shouldn't hav=

e to

Quoted 11 lines
Quoted 1 line

worry about it?

=20
I think the server already has this problem. In svc_init_buffer which sets up the pages for a server thread request/response handling, it uses sv_max_mesg / PAGE_SIZE + 1 with the comment
=20
"extra page as we hold both request and reply. We assume one is at most one page"
=20
where
sv_max_mesg =3D roundup(serv->sv_max_payload + PAGE_SIZE, PAGE_SIZE).

Right. The difference is that now it looks to me like we're actually going to start promising that we support the large request + large response case, when actually we don't.

I guess the problem's unlikely to arise, so maybe it's not worth fixing= =2E
But it's annoying to have yet another undocumented restriction on the compounds we'll accept.

--b.

#7 lore raw

Re: [pnfs] [PATCH 2/2] nfsd41: add RPC header size to fore channel negotiation

Quoted 1 line · On Fri, Sep 18, 2009 at 5:12 PM, J. Bruce Fields <[email protected]>= wrote:

On Fri, Sep 18, 2009 at 12:47:57PM -0400, William A. (Andy) Adamson w=

rote:

Quoted 1 line
Quoted 1 line

On Mon, Sep 14, 2009 at 3:03 PM, J. Bruce Fields <[email protected]=

Quoted 4 lines · rg> wrote:
Quoted 4 lines
Quoted 4 lines · On Fri, Sep 11, 2009 at 06:52:55PM -0400, [email protected] wrote:
Quoted 3 lines

From: Andy Adamson <[email protected]>

Both the max request and the max response size include the RPC he=

ader with

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

credential (request only) =A0and verifier as well as the payload.

The RPCSEC_GSS credential and verifier are the largest. Kerberos =

is the only

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

supported GSS security mechansim, so the Kerberos GSS credential =

and verifier

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

sizes are used.

Rather than trying to estimate this is might be simplest just to u=

se

Quoted 1 line
Quoted 1 line
Quoted 1 line

what the server's using to allocate memory: RPCSVC_MAXPAGES. =A0No=

, that

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

also takes into account space for the reply. =A0You could do

=A0 =A0 =A0 =A0PAGE_SIZE * (1 + (RPCSVC_MAXPAYLOAD+PAGE_SIZE-1)/PA=

GE_SIZE)

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Actually, by design the server's real limit is actually on the sum=

of

Quoted 7 lines
Quoted 5 lines
Quoted 1 line

the request and the reply sizes.

I think the actual limit is svc_max_payload rounded up to a multiple of PAGE_SIZE plus PAGE_SIZE. which is a lot smaller than the sum of the request and reply sizes. See below.

Right. =A0 I think you're agreeing with me?

yes!

Quoted 2 lines
Quoted 2 lines

Note that svc_max_payload is what is returned in nfs4_encode_fattr f=

or

Quoted 1 line
Quoted 1 line

MAXREAD and for MAXWRITE. These attributes use svc_max_payload in th=

e

Quoted 4 lines
Quoted 4 lines

same way this patch does - the maximum data size not including rpc headers.

I don't think the server wants is to advertise a MAXREAD/WRITE that =

it

Quoted 5 lines
Quoted 3 lines

can't supply because the fore channel maxrequest/maxresponse is too small, so some additional space needs to be added to svc_max_payload for the fore channel.

Yes.

=46or the additional space, shall we use what this patch calculates or some other metric?

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

What happens if we get a request such that both the request and re=

ply

Quoted 1 line
Quoted 1 line
Quoted 1 line

are under our advertised limits, but the sum is too much? =A0Can w=

e just

Quoted 1 line
Quoted 1 line
Quoted 1 line

declare that no client will be that weird and that we shouldn't ha=

ve to

Quoted 3 lines
Quoted 3 lines
Quoted 1 line

worry about it?

I think the server already has this problem. In svc_init_buffer whic=

h

Quoted 8 lines
Quoted 8 lines

sets up the pages for a server thread request/response handling, it uses sv_max_mesg / PAGE_SIZE + 1 with the comment

"extra page as we hold both request and reply. We assume one is at most one page"

where
sv_max_mesg =3D roundup(serv->sv_max_payload + PAGE_SIZE, PAGE_SIZE)=

=2E

Quoted 2 lines

Right. =A0The difference is that now it looks to me like we're actual=

ly

Quoted 2 lines

going to start promising that we support the large request + large response case, when actually we don't.

OK - I see your point. With MAXREAD or MAXWRITE we only promise either a large request or a large response per compound, not a large request and a large response in a single compound, e.g. a read and a write in the same compound.

Quoted 2 lines

I guess the problem's unlikely to arise, so maybe it's not worth fixi=

ng.

Quoted 2 lines

But it's annoying to have yet another undocumented restriction on the compounds we'll accept.

I wonder what other servers are doing.

-->Andy

Quoted 3 lines

--b.

A message that is not in the archive (sent off-list or not yet received): 89c397150909210529q5b119d9dmff958dfe2d859635-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org

#9 lore raw

Re: [pnfs] [PATCH 2/2] nfsd41: add RPC header size to fore channel negotiation

On Mon, Sep 21, 2009 at 08:29:30AM -0400, William A. (Andy) Adamson wro= te:

Quoted 1 line

On Fri, Sep 18, 2009 at 5:12 PM, J. Bruce Fields <bfields-uC3wQj2KruMpug/[email protected]=

Quoted 1 line · g> wrote:
Quoted 1 line

On Fri, Sep 18, 2009 at 12:47:57PM -0400, William A. (Andy) Adamson=

Quoted 1 line · wrote:
Quoted 1 line
Quoted 1 line

On Mon, Sep 14, 2009 at 3:03 PM, J. Bruce Fields <bfields@fieldses=

Quoted 1 line · =2Eorg> wrote:
Quoted 1 line
Quoted 1 line
Quoted 1 line

On Fri, Sep 11, 2009 at 06:52:55PM -0400, [email protected] wrot=

e:

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

From: Andy Adamson <[email protected]>

Both the max request and the max response size include the RPC =

header with

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

credential (request only) =C2=A0and verifier as well as the pay=

load.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

The RPCSEC_GSS credential and verifier are the largest. Kerbero=

s is the only

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

supported GSS security mechansim, so the Kerberos GSS credentia=

l and verifier

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

sizes are used.

Rather than trying to estimate this is might be simplest just to=

use

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

what the server's using to allocate memory: RPCSVC_MAXPAGES. =C2=

=A0No, that

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

also takes into account space for the reply. =C2=A0You could do

=C2=A0 =C2=A0 =C2=A0 =C2=A0PAGE_SIZE * (1 + (RPCSVC_MAXPAYLOAD+P=

AGE_SIZE-1)/PAGE_SIZE)

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Actually, by design the server's real limit is actually on the s=

um of

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

the request and the reply sizes.

I think the actual limit is svc_max_payload rounded up to a multip=

le

Quoted 1 line
Quoted 1 line
Quoted 1 line

of PAGE_SIZE plus PAGE_SIZE. which is a lot smaller than the sum o=

f

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

the request and reply sizes. See below.

Right. =C2=A0 I think you're agreeing with me?

=20
yes!
=20

Quoted 2 lines
Quoted 2 lines

Note that svc_max_payload is what is returned in nfs4_encode_fattr=

for

Quoted 1 line
Quoted 1 line
Quoted 1 line

MAXREAD and for MAXWRITE. These attributes use svc_max_payload in =

the

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

same way this patch does - the maximum data size not including rpc headers.

I don't think the server wants is to advertise a MAXREAD/WRITE tha=

t it

Quoted 1 line
Quoted 1 line
Quoted 1 line

can't supply because the fore channel maxrequest/maxresponse is to=

o

Quoted 1 line
Quoted 1 line
Quoted 1 line

small, so some additional space needs to be added to svc_max_paylo=

ad

Quoted 6 lines
Quoted 3 lines
Quoted 1 line

for the fore channel.

Yes.

=20
For the additional space, shall we use what this patch calculates or some other metric?

I guess something like roundup(serv->sv_max_payload + PAGE_SIZE, PAGE_SIZE) is the best we can do for now.

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

What happens if we get a request such that both the request and =

reply

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

are under our advertised limits, but the sum is too much? =C2=A0=

Can we just

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

declare that no client will be that weird and that we shouldn't =

have to

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

worry about it?

I think the server already has this problem. In svc_init_buffer wh=

ich

Quoted 1 line
Quoted 1 line
Quoted 1 line

sets up the pages for a server thread request/response handling, i=

t

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

uses sv_max_mesg / PAGE_SIZE + 1 with the comment

"extra page as we hold both request and reply. We assume one is at most one page"

where
sv_max_mesg =3D roundup(serv->sv_max_payload + PAGE_SIZE, PAGE_SIZ=

E).

Quoted 2 lines
Quoted 2 lines

Right. =C2=A0The difference is that now it looks to me like we're a=

ctually

Quoted 4 lines
Quoted 2 lines

going to start promising that we support the large request + large response case, when actually we don't.

=20
OK - I see your point. With MAXREAD or MAXWRITE we only promise eithe=

r

Quoted 6 lines

a large request or a large response per compound, not a large request and a large response in a single compound, e.g. a read and a write in the same compound.
=20

Quoted 2 lines

I guess the problem's unlikely to arise, so maybe it's not worth fi=

xing.

Quoted 1 line
Quoted 1 line

But it's annoying to have yet another undocumented restriction on t=

he

Quoted 3 lines
Quoted 1 line

compounds we'll accept.

=20
I wonder what other servers are doing.

Yeah. For now I guess we should ignore this. (If you wanted to, in th= e
same patch, add a note about this problem to the end of Documentation/filesystems/nfs41-server.txt, that would be good.)

--b.

#10 lore raw

Re: [PATCH 1/2] nfsd41: fix NFSD_MIN_HDR_SEQ_SZ

In reply to [email protected]
Quoted 4 lines · On Fri, Sep 11, 2009 at 06:52:54PM -0400, [email protected] wrote:

From: Andy Adamson <[email protected]>

The reply RPC header is 12 bytes, the NULL verifier is 8 bytes giving 20 not 24 bytes in the NFSD_MIN_HDR_SEQ_SZ calculation,

No, 24 is right. Maybe you forgot the accept stat at the end?:

xid
msg type
reply stat
verf flavor
verf length
accept stat

--b.

Quoted 28 lines
---
 fs/nfsd/nfs4state.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)
fs/nfsd/nfs4state.c +2 -2
index 46e9ac5..5ecb42c 100644
@@ -420,12 +420,12 @@ gen_sessionid(struct nfsd4_session *ses)
  * each time.  Therefore we can advertise a ca_maxresponssize_cached
  * value that is the number of bytes in our cache plus a few additional
  * bytes.  In order to stay on the safe side, and not promise more than
- * we can cache, those additional bytes must be the minimum possible: 24
+ * we can cache, those additional bytes must be the minimum possible: 20
  * bytes of rpc header (xid through accept state, with AUTH_NULL
  * verifier), 12 for the compound header (with zero-length tag), and 44
  * for the SEQUENCE op response:
  */
-#define NFSD_MIN_HDR_SEQ_SZ  (24 + 12 + 44)
+#define NFSD_MIN_HDR_SEQ_SZ  (20 + 12 + 44)
 
 /*
  * Give the client the number of ca_maxresponsesize_cached slots it
Signature · 1 line
1.6.2.5
#11 lore raw

Re: [pnfs] [PATCH 1/2] nfsd41: fix NFSD_MIN_HDR_SEQ_SZ

Quoted 4 lines · On Mon, Sep 14, 2009 at 2:51 PM, J. Bruce Fields <[email protected]>= wrote:
Quoted 3 lines · On Fri, Sep 11, 2009 at 06:52:54PM -0400, [email protected] wrote:

From: Andy Adamson <[email protected]>

The reply RPC header is 12 bytes, the NULL verifier is 8 bytes givin=

g 20

Quoted 3 lines
Quoted 1 line

not 24 bytes in the NFSD_MIN_HDR_SEQ_SZ calculation,

No, 24 is right. =A0Maybe you forgot the accept stat at the end?:

yes - missed the accept state.

-->Andy

Quoted 16 lines

=A0 =A0 =A0 =A0xid
=A0 =A0 =A0 =A0msg type
=A0 =A0 =A0 =A0reply stat
=A0 =A0 =A0 =A0verf flavor
=A0 =A0 =A0 =A0verf length
=A0 =A0 =A0 =A0accept stat

--b.

Quoted 6 lines

---
=A0fs/nfsd/nfs4state.c | =A0 =A04 ++--
=A01 files changed, 2 insertions(+), 2 deletions(-)

Quoted 6 lines
Quoted 6 lines
fs/nfsd/nfs4state.c +0 -0
index 46e9ac5..5ecb42c 100644
@@ -420,12 +420,12 @@ gen_sessionid(struct nfsd4_session *ses)

=A0 * each time. =A0Therefore we can advertise a ca_maxresponssize_c=

ached

Quoted 1 line
Quoted 1 line

=A0 * value that is the number of bytes in our cache plus a few addi=

tional

Quoted 1 line
Quoted 1 line

=A0 * bytes. =A0In order to stay on the safe side, and not promise m=

ore than

Quoted 1 line
Quoted 1 line

- * we can cache, those additional bytes must be the minimum possibl=

e: 24

Quoted 1 line
Quoted 1 line

+ * we can cache, those additional bytes must be the minimum possibl=

e: 20

Quoted 2 lines
Quoted 2 lines

=A0 * bytes of rpc header (xid through accept state, with AUTH_NULL =A0 * verifier), 12 for the compound header (with zero-length tag), =

and 44

Quoted 7 lines
Quoted 7 lines

=A0 * for the SEQUENCE op response:
=A0 */

fs/nfsd/nfs4state.c +1 -1
-#define NFSD_MIN_HDR_SEQ_SZ =A0(24 + 12 + 44)
+#define NFSD_MIN_HDR_SEQ_SZ =A0(20 + 12 + 44)

=A0/*
=A0 * Give the client the number of ca_maxresponsesize_cached slots =

it

Quoted 8 lines
Quoted 3 lines

--
1.6.2.5

_______________________________________________
pNFS mailing list
[email protected]
http://linux-nfs.org/cgi-bin/mailman/listinfo/pnfs

#12 lore raw
In reply to [email protected]
Quoted 6 lines · On Fri, Sep 11, 2009 at 06:52:53PM -0400, [email protected] wrote:

Fix forechannel attribute initialization

These two patches fix ca_maxresponsesize_cached, ca_maxresponsesize, and ca_maxrequestsize calculations for the forechannel.

Tested:

Actually the think I was interested to test was just to make sure that the client does writes of the size it should. (It should be able to do 1MB reads and writes, but with the previous code, I believe it would have been a bug for the client to do so, as we weren't advertising maxresponse/requestsizes large enough.)

--b.

Quoted 15 lines

Kerberos NFSv4.1 mount:
Passes Connectathon tests. Passes the expected pynfs tests when tests are run individually.
When the pynfs tests are run all at once, there are too many sessions created, and the server code (init_forechannel_attrs) works as planned and fails the CREATE SESSION with NFS4ERR_SERVERFAULT.

The pynfs 4.1 tests need to be fixed to clean up after themselves!

-->Andy

0001-nfsd41-fix-NFSD_MIN_HDR_SEQ_SZ.patch 0002-nfsd41-add-RPC-header-size-to-fore-channel-negotiat.patch