Svcrdma currently advertises 1MB, which is too large. The correct value
is the max scatter-gather allowed in an NFSRDMA IO chunk * the host page
size. This bug is usually benign because the Linux X64 NFSRDMA client
correctly limits the payload size to the correct value (64*4096 = 256KB).
But if the Linux client is PPC64 with a 64KB page size, then the client
will indeed use a payload size that will overflow the server.
Signed-off-by: Steve Wise <[email protected]>
---
net/sunrpc/xprtrdma/svc_rdma_transport.c | 2 +-
net/sunrpc/xprtrdma/xprt_rdma.h | 2 ++
2 files changed, 3 insertions(+), 1 deletions(-)
diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
index 374feb4..4e61880 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
@@ -91,7 +91,7 @@ struct svc_xprt_class svc_rdma_class = {
.xcl_name = "rdma",
.xcl_owner = THIS_MODULE,
.xcl_ops = &svc_rdma_ops,
- .xcl_max_payload = RPCSVC_MAXPAYLOAD_TCP,
+ .xcl_max_payload = RPCSVC_MAXPAYLOAD_RDMA,
.xcl_ident = XPRT_TRANSPORT_RDMA,
};
diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h
index c419498..467a77c 100644
--- a/net/sunrpc/xprtrdma/xprt_rdma.h
+++ b/net/sunrpc/xprtrdma/xprt_rdma.h
@@ -392,4 +392,6 @@ extern struct kmem_cache *svc_rdma_ctxt_cachep;
/* Workqueue created in svc_rdma.c */
extern struct workqueue_struct *svc_rdma_wq;
+#define RPCSVC_MAXPAYLOAD_RDMA (RPCRDMA_MAX_DATA_SEGS << PAGE_SHIFT)
+
#endif /* _LINUX_SUNRPC_XPRT_RDMA_H */
On 9/22/2014 1:55 PM, J. Bruce Fields wrote:
> On Mon, Sep 22, 2014 at 01:42:07PM -0500, Steve Wise wrote:
>> On 9/22/2014 1:39 PM, J. Bruce Fields wrote:
>>> On Mon, Sep 22, 2014 at 01:36:53PM -0500, Steve Wise wrote:
>>>> Svcrdma currently advertises 1MB, which is too large. The correct value
>>>> is the max scatter-gather allowed in an NFSRDMA IO chunk * the host page
>>>> size. This bug is usually benign because the Linux X64 NFSRDMA client
>>>> correctly limits the payload size to the correct value (64*4096 = 256KB).
>>>> But if the Linux client is PPC64 with a 64KB page size, then the client
>>>> will indeed use a payload size that will overflow the server.
>>>>
>>>> Signed-off-by: Steve Wise <[email protected]>
>>>> ---
>>>>
>>>> net/sunrpc/xprtrdma/svc_rdma_transport.c | 2 +-
>>>> net/sunrpc/xprtrdma/xprt_rdma.h | 2 ++
>>>> 2 files changed, 3 insertions(+), 1 deletions(-)
>>>>
>>>> diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
>>>> index 374feb4..4e61880 100644
>>>> --- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
>>>> +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
>>>> @@ -91,7 +91,7 @@ struct svc_xprt_class svc_rdma_class = {
>>>> .xcl_name = "rdma",
>>>> .xcl_owner = THIS_MODULE,
>>>> .xcl_ops = &svc_rdma_ops,
>>>> - .xcl_max_payload = RPCSVC_MAXPAYLOAD_TCP,
>>>> + .xcl_max_payload = RPCSVC_MAXPAYLOAD_RDMA,
>>>> .xcl_ident = XPRT_TRANSPORT_RDMA,
>>>> };
>>>> diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h
>>>> index c419498..467a77c 100644
>>>> --- a/net/sunrpc/xprtrdma/xprt_rdma.h
>>>> +++ b/net/sunrpc/xprtrdma/xprt_rdma.h
>>>> @@ -392,4 +392,6 @@ extern struct kmem_cache *svc_rdma_ctxt_cachep;
>>>> /* Workqueue created in svc_rdma.c */
>>>> extern struct workqueue_struct *svc_rdma_wq;
>>>> +#define RPCSVC_MAXPAYLOAD_RDMA (RPCRDMA_MAX_DATA_SEGS << PAGE_SHIFT)
>>> Do you want to define this as the minimum of this and
>>> RPCSVC_MAXPAYLOAD_TCP, in case RPCRDMA_MAX_DATA_SEGS gets increased some
>>> day?
>> Why would it need to be limited by MAXPAYLOAD_TCP?
> Because you're also limited by the size of the rq_pages array, which is
> determined by RPCSVC_MAXPAGES, calculated from RPCSVC_MAXPAYLOAD.
>
> (Actually you probably want RPCSVC_MAXPAYLOAD, not MAXPAYLOAD_TCP.)
>
>
I see. I agree.
On Sep 22, 2014, at 11:39 AM, J. Bruce Fields <[email protected]> wrote:
> On Mon, Sep 22, 2014 at 01:36:53PM -0500, Steve Wise wrote:
>> Svcrdma currently advertises 1MB, which is too large. The correct value
>> is the max scatter-gather allowed in an NFSRDMA IO chunk * the host page
>> size. This bug is usually benign because the Linux X64 NFSRDMA client
>> correctly limits the payload size to the correct value (64*4096 = 256KB).
>> But if the Linux client is PPC64 with a 64KB page size, then the client
>> will indeed use a payload size that will overflow the server.
>>
>> Signed-off-by: Steve Wise <[email protected]>
>> ---
>>
>> net/sunrpc/xprtrdma/svc_rdma_transport.c | 2 +-
>> net/sunrpc/xprtrdma/xprt_rdma.h | 2 ++
>> 2 files changed, 3 insertions(+), 1 deletions(-)
>>
>> diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
>> index 374feb4..4e61880 100644
>> --- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
>> +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
>> @@ -91,7 +91,7 @@ struct svc_xprt_class svc_rdma_class = {
>> .xcl_name = "rdma",
>> .xcl_owner = THIS_MODULE,
>> .xcl_ops = &svc_rdma_ops,
>> - .xcl_max_payload = RPCSVC_MAXPAYLOAD_TCP,
>> + .xcl_max_payload = RPCSVC_MAXPAYLOAD_RDMA,
>> .xcl_ident = XPRT_TRANSPORT_RDMA,
>> };
>>
>> diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h
>> index c419498..467a77c 100644
>> --- a/net/sunrpc/xprtrdma/xprt_rdma.h
>> +++ b/net/sunrpc/xprtrdma/xprt_rdma.h
>> @@ -392,4 +392,6 @@ extern struct kmem_cache *svc_rdma_ctxt_cachep;
>> /* Workqueue created in svc_rdma.c */
>> extern struct workqueue_struct *svc_rdma_wq;
>>
>> +#define RPCSVC_MAXPAYLOAD_RDMA (RPCRDMA_MAX_DATA_SEGS << PAGE_SHIFT)
>
> Do you want to define this as the minimum of this and
> RPCSVC_MAXPAYLOAD_TCP, in case RPCRDMA_MAX_DATA_SEGS gets increased some
> day?
Hey Bruce-
I don?t see the _TCP and _RDMA settings are related.
> --b.
>
>> +
>> #endif /* _LINUX_SUNRPC_XPRT_RDMA_H */
>>
> --
> 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
--
Chuck Lever
chuck[dot]lever[at]oracle[dot]com
On Sep 22, 2014, at 11:55 AM, J. Bruce Fields <[email protected]> wrote:
> On Mon, Sep 22, 2014 at 01:42:07PM -0500, Steve Wise wrote:
>> On 9/22/2014 1:39 PM, J. Bruce Fields wrote:
>>> On Mon, Sep 22, 2014 at 01:36:53PM -0500, Steve Wise wrote:
>>>> Svcrdma currently advertises 1MB, which is too large. The correct value
>>>> is the max scatter-gather allowed in an NFSRDMA IO chunk * the host page
>>>> size. This bug is usually benign because the Linux X64 NFSRDMA client
>>>> correctly limits the payload size to the correct value (64*4096 = 256KB).
>>>> But if the Linux client is PPC64 with a 64KB page size, then the client
>>>> will indeed use a payload size that will overflow the server.
>>>>
>>>> Signed-off-by: Steve Wise <[email protected]>
>>>> ---
>>>>
>>>> net/sunrpc/xprtrdma/svc_rdma_transport.c | 2 +-
>>>> net/sunrpc/xprtrdma/xprt_rdma.h | 2 ++
>>>> 2 files changed, 3 insertions(+), 1 deletions(-)
>>>>
>>>> diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
>>>> index 374feb4..4e61880 100644
>>>> --- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
>>>> +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
>>>> @@ -91,7 +91,7 @@ struct svc_xprt_class svc_rdma_class = {
>>>> .xcl_name = "rdma",
>>>> .xcl_owner = THIS_MODULE,
>>>> .xcl_ops = &svc_rdma_ops,
>>>> - .xcl_max_payload = RPCSVC_MAXPAYLOAD_TCP,
>>>> + .xcl_max_payload = RPCSVC_MAXPAYLOAD_RDMA,
>>>> .xcl_ident = XPRT_TRANSPORT_RDMA,
>>>> };
>>>> diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h
>>>> index c419498..467a77c 100644
>>>> --- a/net/sunrpc/xprtrdma/xprt_rdma.h
>>>> +++ b/net/sunrpc/xprtrdma/xprt_rdma.h
>>>> @@ -392,4 +392,6 @@ extern struct kmem_cache *svc_rdma_ctxt_cachep;
>>>> /* Workqueue created in svc_rdma.c */
>>>> extern struct workqueue_struct *svc_rdma_wq;
>>>> +#define RPCSVC_MAXPAYLOAD_RDMA (RPCRDMA_MAX_DATA_SEGS << PAGE_SHIFT)
>>> Do you want to define this as the minimum of this and
>>> RPCSVC_MAXPAYLOAD_TCP, in case RPCRDMA_MAX_DATA_SEGS gets increased some
>>> day?
>>
>> Why would it need to be limited by MAXPAYLOAD_TCP?
>
> Because you're also limited by the size of the rq_pages array, which is
> determined by RPCSVC_MAXPAGES, calculated from RPCSVC_MAXPAYLOAD.
>
> (Actually you probably want RPCSVC_MAXPAYLOAD, not MAXPAYLOAD_TCP.)
Agree with this last bit.
--
Chuck Lever
chuck[dot]lever[at]oracle[dot]com
On Mon, Sep 22, 2014 at 01:42:07PM -0500, Steve Wise wrote:
> On 9/22/2014 1:39 PM, J. Bruce Fields wrote:
> >On Mon, Sep 22, 2014 at 01:36:53PM -0500, Steve Wise wrote:
> >>Svcrdma currently advertises 1MB, which is too large. The correct value
> >>is the max scatter-gather allowed in an NFSRDMA IO chunk * the host page
> >>size. This bug is usually benign because the Linux X64 NFSRDMA client
> >>correctly limits the payload size to the correct value (64*4096 = 256KB).
> >>But if the Linux client is PPC64 with a 64KB page size, then the client
> >>will indeed use a payload size that will overflow the server.
> >>
> >>Signed-off-by: Steve Wise <[email protected]>
> >>---
> >>
> >> net/sunrpc/xprtrdma/svc_rdma_transport.c | 2 +-
> >> net/sunrpc/xprtrdma/xprt_rdma.h | 2 ++
> >> 2 files changed, 3 insertions(+), 1 deletions(-)
> >>
> >>diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
> >>index 374feb4..4e61880 100644
> >>--- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
> >>+++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
> >>@@ -91,7 +91,7 @@ struct svc_xprt_class svc_rdma_class = {
> >> .xcl_name = "rdma",
> >> .xcl_owner = THIS_MODULE,
> >> .xcl_ops = &svc_rdma_ops,
> >>- .xcl_max_payload = RPCSVC_MAXPAYLOAD_TCP,
> >>+ .xcl_max_payload = RPCSVC_MAXPAYLOAD_RDMA,
> >> .xcl_ident = XPRT_TRANSPORT_RDMA,
> >> };
> >>diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h
> >>index c419498..467a77c 100644
> >>--- a/net/sunrpc/xprtrdma/xprt_rdma.h
> >>+++ b/net/sunrpc/xprtrdma/xprt_rdma.h
> >>@@ -392,4 +392,6 @@ extern struct kmem_cache *svc_rdma_ctxt_cachep;
> >> /* Workqueue created in svc_rdma.c */
> >> extern struct workqueue_struct *svc_rdma_wq;
> >>+#define RPCSVC_MAXPAYLOAD_RDMA (RPCRDMA_MAX_DATA_SEGS << PAGE_SHIFT)
> >Do you want to define this as the minimum of this and
> >RPCSVC_MAXPAYLOAD_TCP, in case RPCRDMA_MAX_DATA_SEGS gets increased some
> >day?
>
> Why would it need to be limited by MAXPAYLOAD_TCP?
Because you're also limited by the size of the rq_pages array, which is
determined by RPCSVC_MAXPAGES, calculated from RPCSVC_MAXPAYLOAD.
(Actually you probably want RPCSVC_MAXPAYLOAD, not MAXPAYLOAD_TCP.)
--b.
On Mon, Sep 22, 2014 at 01:36:53PM -0500, Steve Wise wrote:
> Svcrdma currently advertises 1MB, which is too large. The correct value
> is the max scatter-gather allowed in an NFSRDMA IO chunk * the host page
> size. This bug is usually benign because the Linux X64 NFSRDMA client
> correctly limits the payload size to the correct value (64*4096 = 256KB).
> But if the Linux client is PPC64 with a 64KB page size, then the client
> will indeed use a payload size that will overflow the server.
>
> Signed-off-by: Steve Wise <[email protected]>
> ---
>
> net/sunrpc/xprtrdma/svc_rdma_transport.c | 2 +-
> net/sunrpc/xprtrdma/xprt_rdma.h | 2 ++
> 2 files changed, 3 insertions(+), 1 deletions(-)
>
> diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
> index 374feb4..4e61880 100644
> --- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
> +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
> @@ -91,7 +91,7 @@ struct svc_xprt_class svc_rdma_class = {
> .xcl_name = "rdma",
> .xcl_owner = THIS_MODULE,
> .xcl_ops = &svc_rdma_ops,
> - .xcl_max_payload = RPCSVC_MAXPAYLOAD_TCP,
> + .xcl_max_payload = RPCSVC_MAXPAYLOAD_RDMA,
> .xcl_ident = XPRT_TRANSPORT_RDMA,
> };
>
> diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h
> index c419498..467a77c 100644
> --- a/net/sunrpc/xprtrdma/xprt_rdma.h
> +++ b/net/sunrpc/xprtrdma/xprt_rdma.h
> @@ -392,4 +392,6 @@ extern struct kmem_cache *svc_rdma_ctxt_cachep;
> /* Workqueue created in svc_rdma.c */
> extern struct workqueue_struct *svc_rdma_wq;
>
> +#define RPCSVC_MAXPAYLOAD_RDMA (RPCRDMA_MAX_DATA_SEGS << PAGE_SHIFT)
Do you want to define this as the minimum of this and
RPCSVC_MAXPAYLOAD_TCP, in case RPCRDMA_MAX_DATA_SEGS gets increased some
day?
--b.
> +
> #endif /* _LINUX_SUNRPC_XPRT_RDMA_H */
>
On 9/22/2014 1:39 PM, J. Bruce Fields wrote:
> On Mon, Sep 22, 2014 at 01:36:53PM -0500, Steve Wise wrote:
>> Svcrdma currently advertises 1MB, which is too large. The correct value
>> is the max scatter-gather allowed in an NFSRDMA IO chunk * the host page
>> size. This bug is usually benign because the Linux X64 NFSRDMA client
>> correctly limits the payload size to the correct value (64*4096 = 256KB).
>> But if the Linux client is PPC64 with a 64KB page size, then the client
>> will indeed use a payload size that will overflow the server.
>>
>> Signed-off-by: Steve Wise <[email protected]>
>> ---
>>
>> net/sunrpc/xprtrdma/svc_rdma_transport.c | 2 +-
>> net/sunrpc/xprtrdma/xprt_rdma.h | 2 ++
>> 2 files changed, 3 insertions(+), 1 deletions(-)
>>
>> diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
>> index 374feb4..4e61880 100644
>> --- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
>> +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
>> @@ -91,7 +91,7 @@ struct svc_xprt_class svc_rdma_class = {
>> .xcl_name = "rdma",
>> .xcl_owner = THIS_MODULE,
>> .xcl_ops = &svc_rdma_ops,
>> - .xcl_max_payload = RPCSVC_MAXPAYLOAD_TCP,
>> + .xcl_max_payload = RPCSVC_MAXPAYLOAD_RDMA,
>> .xcl_ident = XPRT_TRANSPORT_RDMA,
>> };
>>
>> diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h
>> index c419498..467a77c 100644
>> --- a/net/sunrpc/xprtrdma/xprt_rdma.h
>> +++ b/net/sunrpc/xprtrdma/xprt_rdma.h
>> @@ -392,4 +392,6 @@ extern struct kmem_cache *svc_rdma_ctxt_cachep;
>> /* Workqueue created in svc_rdma.c */
>> extern struct workqueue_struct *svc_rdma_wq;
>>
>> +#define RPCSVC_MAXPAYLOAD_RDMA (RPCRDMA_MAX_DATA_SEGS << PAGE_SHIFT)
> Do you want to define this as the minimum of this and
> RPCSVC_MAXPAYLOAD_TCP, in case RPCRDMA_MAX_DATA_SEGS gets increased some
> day?
Why would it need to be limited by MAXPAYLOAD_TCP?