* Re: Enabling peer to peer device transactions for PCIe devices
From: Logan Gunthorpe @ 2016-11-30 18:01 UTC (permalink / raw)
To: Jason Gunthorpe, Haggai Eran
Cc: John.Bridgman-5C7GfCeVMHo@public.gmane.org,
linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
linux-nvdimm-y27Ovi1pjclAfugRpC6u6w@public.gmane.org,
Felix.Kuehling-5C7GfCeVMHo@public.gmane.org,
serguei.sagalovitch-5C7GfCeVMHo@public.gmane.org,
linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
dri-devel-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org,
Paul.Blinzer-5C7GfCeVMHo@public.gmane.org,
ben.sander-5C7GfCeVMHo@public.gmane.org,
Suravee.Suthikulpanit-5C7GfCeVMHo@public.gmane.org,
linux-pci-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
Alexander.Deucher-5C7GfCeVMHo@public.gmane.org, Max Gurtovoy,
christian.koenig-5C7GfCeVMHo@public.gmane.org,
Linux-media-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
In-Reply-To: <20161130162353.GA24639-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
On 30/11/16 09:23 AM, Jason Gunthorpe wrote:
>> Two cases I can think of are RDMA access to an NVMe device's controller
>> memory buffer,
>
> I'm not sure on the use model there..
The NVMe fabrics stuff could probably make use of this. It's an
in-kernel system to allow remote access to an NVMe device over RDMA. So
they ought to be able to optimize their transfers by DMAing directly to
the NVMe's CMB -- no userspace interface would be required but there
would need some kernel infrastructure.
Logan
^ permalink raw reply
* RE: [PATCH rdma-next 01/10] IB/core: Add raw packet protocol
From: Hefty, Sean @ 2016-11-30 18:17 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Liran Liss, Tom Talpey, Doug Ledford, Steve Wise,
'Leon Romanovsky',
linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
'Steve Wise', Marciniszyn, Mike, Dalessandro, Dennis,
'Lijun Ou', 'Wei Hu(Xavier)', Latif, Faisal,
Yishai Hadas, 'Selvin Xavier', 'Devesh Sharma',
'Mitesh Ahuja', 'Christian Benvenuti',
'Dave Goodell', Moni
In-Reply-To: <20161130173211.GA9067-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
> > Maybe the solution is to make the protocol selection explicit
> > throughout the APIs and associate it with a QP, rather than
> > attempting to list all transport protocols that a port can support.
>
> AH's are linked to a PD, not a QP..
But protocols are associated with QPs. I'm not sure *why* an AH is linked to a PD. Conceptually the association seems unnecessary.
> If we had to do it again, a PD centric approach would be more
> sensible:
>
> // Create a PD on 'port' using ah format 'protocol'
> pd = ibv_pd_create(port, enum ah_type protocol);
>
> // Enable APM or resource sharing on the PD across two ports
> ibv_pd_add_port(pd, alt_port);
>
> And get rid of the multi-port ib_device concept entirely.
I think no matter what option is chosen, some limitation may end up being placed on how a device is used that may not map to hardware limits. I'm personally fine with that, but we need to define the right level of abstraction (and pick the right terms to describe them).
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Re: [PATCH rdma-next 01/10] IB/core: Add raw packet protocol
From: Jason Gunthorpe @ 2016-11-30 18:34 UTC (permalink / raw)
To: Hefty, Sean
Cc: Liran Liss, Tom Talpey, Doug Ledford, Steve Wise,
'Leon Romanovsky',
linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
'Steve Wise', Marciniszyn, Mike, Dalessandro, Dennis,
'Lijun Ou', 'Wei Hu(Xavier)', Latif, Faisal,
Yishai Hadas, 'Selvin Xavier', 'Devesh Sharma',
'Mitesh Ahuja', 'Christian Benvenuti',
'Dave Goodell', Moni
In-Reply-To: <1828884A29C6694DAF28B7E6B8A82373AB0BA20F-P5GAC/sN6hkd3b2yrw5b5LfspsVTdybXVpNB7YpNyf8@public.gmane.org>
On Wed, Nov 30, 2016 at 06:17:34PM +0000, Hefty, Sean wrote:
> > > Maybe the solution is to make the protocol selection explicit
> > > throughout the APIs and associate it with a QP, rather than
> > > attempting to list all transport protocols that a port can support.
> >
> > AH's are linked to a PD, not a QP..
>
> But protocols are associated with QPs. I'm not sure *why* an AH is
> linked to a PD. Conceptually the association seems unnecessary.
The AH has to be linked to the PD because the PD specifies the
hardware target and the AH is a hardware object. All objects must be
traced back to a PD. It is linked to a PD not a QP because the AH can
be shared across all QPs, which is useful for UD applications..
The AH is really similar to the ethernet layer in an IP stack, and the QP
is akin the TCP layer.
Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* RE: [PATCH rdma-next 01/10] IB/core: Add raw packet protocol
From: Hefty, Sean @ 2016-11-30 18:46 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Liran Liss, Tom Talpey, Doug Ledford, Steve Wise,
'Leon Romanovsky',
linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
'Steve Wise', Marciniszyn, Mike, Dalessandro, Dennis,
'Lijun Ou', 'Wei Hu(Xavier)', Latif, Faisal,
Yishai Hadas, 'Selvin Xavier', 'Devesh Sharma',
'Mitesh Ahuja', 'Christian Benvenuti',
'Dave Goodell', Moni
In-Reply-To: <20161130183436.GA10057-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
> > But protocols are associated with QPs. I'm not sure *why* an AH is
> > linked to a PD. Conceptually the association seems unnecessary.
>
> The AH has to be linked to the PD because the PD specifies the
> hardware target and the AH is a hardware object. All objects must be
> traced back to a PD. It is linked to a PD not a QP because the AH can
> be shared across all QPs, which is useful for UD applications..
>
> The AH is really similar to the ethernet layer in an IP stack, and the
> QP
> is akin the TCP layer.
I always viewed the AH as specifying the route to the destination port. Locally, I still don't see why it couldn't work across the device, like the CQ does. Associating it with a PD just seems to limit which QPs, and by association the memory buffers, it can be used with. So, I'm still not seeing the reason for the linkage...
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Re: [PATCH rdma-next 01/10] IB/core: Add raw packet protocol
From: Jason Gunthorpe @ 2016-11-30 18:58 UTC (permalink / raw)
To: Hefty, Sean
Cc: Liran Liss, Tom Talpey, Doug Ledford, Steve Wise,
'Leon Romanovsky',
linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
'Steve Wise', Marciniszyn, Mike, Dalessandro, Dennis,
'Lijun Ou', 'Wei Hu(Xavier)', Latif, Faisal,
Yishai Hadas, 'Selvin Xavier', 'Devesh Sharma',
'Mitesh Ahuja', 'Christian Benvenuti',
'Dave Goodell', Moni
In-Reply-To: <1828884A29C6694DAF28B7E6B8A82373AB0BA24D-P5GAC/sN6hkd3b2yrw5b5LfspsVTdybXVpNB7YpNyf8@public.gmane.org>
On Wed, Nov 30, 2016 at 06:46:13PM +0000, Hefty, Sean wrote:
> > > But protocols are associated with QPs. I'm not sure *why* an AH is
> > > linked to a PD. Conceptually the association seems unnecessary.
> >
> > The AH has to be linked to the PD because the PD specifies the
> > hardware target and the AH is a hardware object. All objects must be
> > traced back to a PD. It is linked to a PD not a QP because the AH can
> > be shared across all QPs, which is useful for UD applications..
> >
> > The AH is really similar to the ethernet layer in an IP stack, and the
> > QP
> > is akin the TCP layer.
>
> I always viewed the AH as specifying the route to the destination
> port. Locally, I still don't see why it couldn't work across the
> device, like the CQ does. Associating it with a PD just seems to
> limit which QPs, and by association the memory buffers, it can be
> used with. So, I'm still not seeing the reason for the linkage...
Oh I see what you mean, I forgot about that - yes, I don't see any
reason why the AH is linked to the PD not the ibv_context..
Perhaps it is an artifact of the thinking when applying this to the
kernel where each ULP would allocate resources to the PD and then
destroy the PD when the connection is done? So per ulp resources were
charged to the PD..
Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Re: [PATCH rdma-next 01/10] IB/core: Add raw packet protocol
From: Doug Ledford @ 2016-11-30 19:35 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Steve Wise, 'Leon Romanovsky',
linux-rdma-u79uwXL29TY76Z2rM5mHXA, 'Steve Wise',
'Mike Marciniszyn', 'Dennis Dalessandro',
'Lijun Ou', 'Wei Hu(Xavier)',
'Faisal Latif', 'Yishai Hadas',
'Selvin Xavier', 'Devesh Sharma',
'Mitesh Ahuja', 'Christian Benvenuti',
'Dave Goodell', 'Moni Shoua',
'Or Gerlitz'
In-Reply-To: <20161130163621.GB24639-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
[-- Attachment #1.1: Type: text/plain, Size: 3248 bytes --]
On 11/30/2016 11:36 AM, Jason Gunthorpe wrote:
> On Tue, Nov 29, 2016 at 09:07:52PM -0500, Doug Ledford wrote:
>> On 11/28/2016 12:08 PM, Steve Wise wrote:
>>>>
>>>> On Sun, Nov 27, 2016 at 04:51:27PM +0200, Leon Romanovsky wrote:
>>>>
>>>>> +static inline bool rdma_protocol_raw_packet(const struct ib_device *device,
>>> u8
>>>> port_num)
>>>>> +{
>>>>> + return device->port_immutable[port_num].core_cap_flags &
>>>> RDMA_CORE_CAP_PROT_RAW_PACKET;
>>>>> +}
>>>>
>>>> Does the mlx drivers really register ports with different capabilities
>>>> as the same ib_device? I'm not sure that should be allowed.
>>>>
>>>> I keep talking about how we need to get rid of the port_num in these
>>>> sorts of places because it makes no sense...
>>>>
>>>
>>> I agree. Requiring the port number has implications that ripple up into the
>>> rdma-rw api as well...
>>>
>>>
>>
>> In all fairness, there is no requirement that any two ports on the same
>> device be the same link layer, or if the link layer is Ethernet, there
>> is no requirement that they can't support both iWARP and RoCE.
>
> There actually is a requirement. The RDMA CM hard requires all ports
> be iWARP or !iWARP at least. I'm sure there are other subtle things
> floating around.
>
> There are also things that become very confusing for user space, and
> we don't have the infrastructure to support, if ports can switch
> configurations on the fly.
>
> The simplest, approach, most in line with how verbs was designed, is
> to require each ib_device to have a single kind of AH.
Sorry, we're conflating two separate things here. I was merely
referring to the hardware. Not our implementation of a stack. I was
attempting to point out that our stack implementation is placing
artificial restrictions on the hardware that the hardware does not share.
>> The idea that the parent device defined the supported protocols for
>> all ports of a device became wrong with the first mlx4 device that
>
> Arguably it was sort of OK for roceev1, is less OK for v2, but
> shouldn't have been done anyhow.
>
> The uapi question here is do we want to double down and try and make
> this work (and what does that even *mean*) or admit mlx4 was an error
> and stop doing that going forward..
>
> Or do something else? eg Specifying the AH type when creating the PD
> could potentially solve some of the problems...
>
>> could do both IB and Ethernet. And I think I've heard rumblings of
>> a combined RoCE/iWARP device possibly in the future from someone
>> else.
>
> Two struct ib_devices for the same port then... Certainly ugly.
I'm not entirely sure I agree....I would have to think about it more,
but the underlying problem I'm concerned with is exactly what I point
out above: what restrictions would we be placing on the hardware that
are artificial and a result of our stack that the hardware itself does
not share? Could the hardware support automatic migration from iWARP to
RoCE or vice versa if both endpoints supported it? Would that work if
we required two separate ib_devices? Things like that.
--
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
GPG Key ID: 0E572FDD
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 884 bytes --]
^ permalink raw reply
* Re: [PATCH v3] ethernet :mellanox :mlx4: Replace pci_pool_alloc by pci_pool_zalloc
From: David Miller @ 2016-11-30 19:57 UTC (permalink / raw)
To: jrdr.linux-Re5JQEeQqe8AvxtiuMwx3w
Cc: sergei.shtylyov-M4DtvfQ/ZS1MRgGoP+s0PdBPR1lH4CV8,
yishaih-VPRAkNaXOzVWk0Htik3J/w, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA,
sahu.rameshwar73-Re5JQEeQqe8AvxtiuMwx3w
In-Reply-To: <20161129194611.GA4088@jordon-HP-15-Notebook-PC>
From: Souptick Joarder <jrdr.linux-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Date: Wed, 30 Nov 2016 01:16:12 +0530
> In mlx4_alloc_cmd_mailbox(), pci_pool_alloc() followed by memset will be
> replaced by pci_pool_zalloc()
>
> Signed-off-by: Souptick joarder <jrdr.linux-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Applied.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Re: [PATCH v2] ethernet :mellanox :mlx5: Replace pci_pool_alloc by pci_pool_zalloc
From: David Miller @ 2016-11-30 19:57 UTC (permalink / raw)
To: jrdr.linux
Cc: sergei.shtylyov, saeedm, matanb, leonro, netdev, linux-rdma,
sahu.rameshwar73
In-Reply-To: <20161129212018.GA5419@jordon-HP-15-Notebook-PC>
From: Souptick Joarder <jrdr.linux@gmail.com>
Date: Wed, 30 Nov 2016 02:50:18 +0530
> In alloc_cmd_box(), pci_pool_alloc() followed by memset will be
> replaced by pci_pool_zalloc()
>
> Signed-off-by: Souptick joarder <jrdr.linux@gmail.com>
Applied.
^ permalink raw reply
* [PATCH] i40iw: Utilize physically mapped memory regions
From: Henry Orosco @ 2016-11-30 20:56 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA,
e1000-rdma-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f, Henry Orosco
Add support to use physically mapped WQ's and MR's if determined
that the OS registered user-memory for the region is physically
contiguous. This feature will eliminate the need for unnecessarily
setting up and using PBL's when not required.
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Henry Orosco <henry.orosco-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/i40iw/i40iw_verbs.c | 102 +++++++++++++++++++++++++++---
1 file changed, 93 insertions(+), 9 deletions(-)
diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
index 462f5a7..59f613b 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
@@ -1358,10 +1358,62 @@ static void i40iw_copy_user_pgaddrs(struct i40iw_mr *iwmr,
}
/**
+ * i40iw_check_mem_contiguous - check if pbls stored in arr are contiguous
+ * @arr: lvl1 pbl array
+ * @npages: page count
+ * pg_size: page size
+ *
+ */
+static bool i40iw_check_mem_contiguous(u64 *arr, u32 npages, u32 pg_size)
+{
+ u32 pg_idx;
+
+ for (pg_idx = 0; pg_idx < npages; pg_idx++) {
+ if ((*arr + (pg_size * pg_idx)) != arr[pg_idx])
+ return false;
+ }
+ return true;
+}
+
+/**
+ * i40iw_check_mr_contiguous - check if MR is physically contiguous
+ * @palloc: pbl allocation struct
+ * pg_size: page size
+ */
+static bool i40iw_check_mr_contiguous(struct i40iw_pble_alloc *palloc, u32 pg_size)
+{
+ struct i40iw_pble_level2 *lvl2 = &palloc->level2;
+ struct i40iw_pble_info *leaf = lvl2->leaf;
+ u64 *arr = NULL;
+ u64 *start_addr = NULL;
+ int i;
+ bool ret;
+
+ if (palloc->level == I40IW_LEVEL_1) {
+ arr = (u64 *)palloc->level1.addr;
+ ret = i40iw_check_mem_contiguous(arr, palloc->total_cnt, pg_size);
+ return ret;
+ }
+
+ start_addr = (u64 *)leaf->addr;
+
+ for (i = 0; i < lvl2->leaf_cnt; i++, leaf++) {
+ arr = (u64 *)leaf->addr;
+ if ((*start_addr + (i * pg_size * PBLE_PER_PAGE)) != *arr)
+ return false;
+ ret = i40iw_check_mem_contiguous(arr, leaf->cnt, pg_size);
+ if (!ret)
+ return false;
+ }
+
+ return true;
+}
+
+/**
* i40iw_setup_pbles - copy user pg address to pble's
* @iwdev: iwarp device
* @iwmr: mr pointer for this memory registration
- * @use_pbles: flag if to use pble's or memory (level 0)
+ * @use_pbles: flag if to use pble's
*/
static int i40iw_setup_pbles(struct i40iw_device *iwdev,
struct i40iw_mr *iwmr,
@@ -1374,9 +1426,6 @@ static int i40iw_setup_pbles(struct i40iw_device *iwdev,
enum i40iw_status_code status;
enum i40iw_pble_level level = I40IW_LEVEL_1;
- if (!use_pbles && (iwmr->page_cnt > MAX_SAVE_PAGE_ADDRS))
- return -ENOMEM;
-
if (use_pbles) {
mutex_lock(&iwdev->pbl_mutex);
status = i40iw_get_pble(&iwdev->sc_dev, iwdev->pble_rsrc, palloc, iwmr->page_cnt);
@@ -1393,6 +1442,10 @@ static int i40iw_setup_pbles(struct i40iw_device *iwdev,
}
i40iw_copy_user_pgaddrs(iwmr, pbl, level);
+
+ if (use_pbles)
+ iwmr->pgaddrmem[0] = *pbl;
+
return 0;
}
@@ -1414,14 +1467,18 @@ static int i40iw_handle_q_mem(struct i40iw_device *iwdev,
struct i40iw_cq_mr *cqmr = &iwpbl->cq_mr;
struct i40iw_hmc_pble *hmc_p;
u64 *arr = iwmr->pgaddrmem;
+ u32 pg_size;
int err;
int total;
+ bool ret = true;
total = req->sq_pages + req->rq_pages + req->cq_pages;
+ pg_size = iwmr->region->page_size;
err = i40iw_setup_pbles(iwdev, iwmr, use_pbles);
if (err)
return err;
+
if (use_pbles && (palloc->level != I40IW_LEVEL_1)) {
i40iw_free_pble(iwdev->pble_rsrc, palloc);
iwpbl->pbl_allocated = false;
@@ -1430,26 +1487,44 @@ static int i40iw_handle_q_mem(struct i40iw_device *iwdev,
if (use_pbles)
arr = (u64 *)palloc->level1.addr;
- if (req->reg_type == IW_MEMREG_TYPE_QP) {
+
+ if (iwmr->type == IW_MEMREG_TYPE_QP) {
hmc_p = &qpmr->sq_pbl;
qpmr->shadow = (dma_addr_t)arr[total];
+
if (use_pbles) {
+ ret = i40iw_check_mem_contiguous(arr, req->sq_pages, pg_size);
+ if (ret)
+ ret = i40iw_check_mem_contiguous(&arr[req->sq_pages], req->rq_pages, pg_size);
+ }
+
+ if (!ret) {
hmc_p->idx = palloc->level1.idx;
hmc_p = &qpmr->rq_pbl;
hmc_p->idx = palloc->level1.idx + req->sq_pages;
} else {
hmc_p->addr = arr[0];
hmc_p = &qpmr->rq_pbl;
- hmc_p->addr = arr[1];
+ hmc_p->addr = arr[req->sq_pages];
}
} else { /* CQ */
hmc_p = &cqmr->cq_pbl;
cqmr->shadow = (dma_addr_t)arr[total];
+
if (use_pbles)
+ ret = i40iw_check_mem_contiguous(arr, req->cq_pages, pg_size);
+
+ if (!ret)
hmc_p->idx = palloc->level1.idx;
else
hmc_p->addr = arr[0];
}
+
+ if (use_pbles && ret) {
+ i40iw_free_pble(iwdev->pble_rsrc, palloc);
+ iwpbl->pbl_allocated = false;
+ }
+
return err;
}
@@ -1648,7 +1723,7 @@ static int i40iw_hwreg_mr(struct i40iw_device *iwdev,
stag_info->pd_id = iwpd->sc_pd.pd_id;
stag_info->addr_type = I40IW_ADDR_TYPE_VA_BASED;
- if (iwmr->page_cnt > 1) {
+ if (iwpbl->pbl_allocated) {
if (palloc->level == I40IW_LEVEL_1) {
stag_info->first_pm_pbl_index = palloc->level1.idx;
stag_info->chunk_size = 1;
@@ -1704,6 +1779,7 @@ static struct ib_mr *i40iw_reg_user_mr(struct ib_pd *pd,
bool use_pbles = false;
unsigned long flags;
int err = -ENOSYS;
+ int ret;
if (length > I40IW_MAX_MR_SIZE)
return ERR_PTR(-EINVAL);
@@ -1760,13 +1836,21 @@ static struct ib_mr *i40iw_reg_user_mr(struct ib_pd *pd,
spin_unlock_irqrestore(&ucontext->cq_reg_mem_list_lock, flags);
break;
case IW_MEMREG_TYPE_MEM:
+ use_pbles = (iwmr->page_cnt != 1);
access = I40IW_ACCESS_FLAGS_LOCALREAD;
- use_pbles = (iwmr->page_cnt != 1);
err = i40iw_setup_pbles(iwdev, iwmr, use_pbles);
if (err)
goto error;
+ if (use_pbles) {
+ ret = i40iw_check_mr_contiguous(palloc, region->page_size);
+ if (ret) {
+ i40iw_free_pble(iwdev->pble_rsrc, palloc);
+ iwpbl->pbl_allocated = false;
+ }
+ }
+
access |= i40iw_get_user_access(acc);
stag = i40iw_create_stag(iwdev);
if (!stag) {
@@ -1794,7 +1878,7 @@ static struct ib_mr *i40iw_reg_user_mr(struct ib_pd *pd,
return &iwmr->ibmr;
error:
- if (palloc->level != I40IW_LEVEL_0)
+ if (palloc->level != I40IW_LEVEL_0 && iwpbl->pbl_allocated)
i40iw_free_pble(iwdev->pble_rsrc, palloc);
ib_umem_release(region);
kfree(iwmr);
--
2.8.3
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related
* [PATCH] i40iw: Add 2MB page support
From: Henry Orosco @ 2016-11-30 20:57 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA,
e1000-rdma-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f, Henry Orosco
Add support to allow each independent memory region to
be configured for 2MB page size in addition to 4KB
page size.
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Henry Orosco <henry.orosco-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/i40iw/i40iw_verbs.c | 59 +++++++++++++++++++++++++------
drivers/infiniband/hw/i40iw/i40iw_verbs.h | 2 ++
2 files changed, 51 insertions(+), 10 deletions(-)
diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
index 59f613b..a563e1c 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
@@ -37,6 +37,7 @@
#include <linux/random.h>
#include <linux/highmem.h>
#include <linux/time.h>
+#include <linux/hugetlb.h>
#include <asm/byteorder.h>
#include <net/ip.h>
#include <rdma/ib_verbs.h>
@@ -1307,13 +1308,11 @@ static u32 i40iw_create_stag(struct i40iw_device *iwdev)
/**
* i40iw_next_pbl_addr - Get next pbl address
- * @palloc: Poiner to allocated pbles
* @pbl: pointer to a pble
* @pinfo: info pointer
* @idx: index
*/
-static inline u64 *i40iw_next_pbl_addr(struct i40iw_pble_alloc *palloc,
- u64 *pbl,
+static inline u64 *i40iw_next_pbl_addr(u64 *pbl,
struct i40iw_pble_info **pinfo,
u32 *idx)
{
@@ -1341,9 +1340,11 @@ static void i40iw_copy_user_pgaddrs(struct i40iw_mr *iwmr,
struct i40iw_pble_alloc *palloc = &iwpbl->pble_alloc;
struct i40iw_pble_info *pinfo;
struct scatterlist *sg;
+ u64 pg_addr = 0;
u32 idx = 0;
pinfo = (level == I40IW_LEVEL_1) ? NULL : palloc->level2.leaf;
+
pg_shift = ffs(region->page_size) - 1;
for_each_sg(region->sg_head.sgl, sg, region->nmap, entry) {
chunk_pages = sg_dma_len(sg) >> pg_shift;
@@ -1351,8 +1352,35 @@ static void i40iw_copy_user_pgaddrs(struct i40iw_mr *iwmr,
!iwpbl->qp_mr.sq_page)
iwpbl->qp_mr.sq_page = sg_page(sg);
for (i = 0; i < chunk_pages; i++) {
- *pbl = cpu_to_le64(sg_dma_address(sg) + region->page_size * i);
- pbl = i40iw_next_pbl_addr(palloc, pbl, &pinfo, &idx);
+ pg_addr = sg_dma_address(sg) + region->page_size * i;
+
+ if ((entry + i) == 0)
+ *pbl = cpu_to_le64(pg_addr & iwmr->page_msk);
+ else if (!(pg_addr & ~iwmr->page_msk))
+ *pbl = cpu_to_le64(pg_addr);
+ else
+ continue;
+ pbl = i40iw_next_pbl_addr(pbl, &pinfo, &idx);
+ }
+ }
+}
+
+/**
+ * i40iw_set_hugetlb_params - set MR pg size and mask to huge pg values.
+ * @addr: virtual address
+ * @iwmr: mr pointer for this memory registration
+ */
+static void i40iw_set_hugetlb_values(u64 addr, struct i40iw_mr *iwmr)
+{
+ struct vm_area_struct *vma;
+ struct hstate *h;
+
+ vma = find_vma(current->mm, addr);
+ if (vma && is_vm_hugetlb_page(vma)) {
+ h = hstate_vma(vma);
+ if (huge_page_size(h) == 0x200000) {
+ iwmr->page_size = huge_page_size(h);
+ iwmr->page_msk = huge_page_mask(h);
}
}
}
@@ -1473,7 +1501,7 @@ static int i40iw_handle_q_mem(struct i40iw_device *iwdev,
bool ret = true;
total = req->sq_pages + req->rq_pages + req->cq_pages;
- pg_size = iwmr->region->page_size;
+ pg_size = iwmr->page_size;
err = i40iw_setup_pbles(iwdev, iwmr, use_pbles);
if (err)
@@ -1722,6 +1750,7 @@ static int i40iw_hwreg_mr(struct i40iw_device *iwdev,
stag_info->access_rights = access;
stag_info->pd_id = iwpd->sc_pd.pd_id;
stag_info->addr_type = I40IW_ADDR_TYPE_VA_BASED;
+ stag_info->page_size = iwmr->page_size;
if (iwpbl->pbl_allocated) {
if (palloc->level == I40IW_LEVEL_1) {
@@ -1780,6 +1809,7 @@ static struct ib_mr *i40iw_reg_user_mr(struct ib_pd *pd,
unsigned long flags;
int err = -ENOSYS;
int ret;
+ int pg_shift;
if (length > I40IW_MAX_MR_SIZE)
return ERR_PTR(-EINVAL);
@@ -1804,9 +1834,17 @@ static struct ib_mr *i40iw_reg_user_mr(struct ib_pd *pd,
iwmr->ibmr.pd = pd;
iwmr->ibmr.device = pd->device;
ucontext = to_ucontext(pd->uobject->context);
- region_length = region->length + (start & 0xfff);
- pbl_depth = region_length >> 12;
- pbl_depth += (region_length & (4096 - 1)) ? 1 : 0;
+
+ iwmr->page_size = region->page_size;
+ iwmr->page_msk = PAGE_MASK;
+
+ if (region->hugetlb && (req.reg_type == IW_MEMREG_TYPE_MEM))
+ i40iw_set_hugetlb_values(start, iwmr);
+
+ region_length = region->length + (start & (iwmr->page_size - 1));
+ pg_shift = ffs(iwmr->page_size) - 1;
+ pbl_depth = region_length >> pg_shift;
+ pbl_depth += (region_length & (iwmr->page_size - 1)) ? 1 : 0;
iwmr->length = region->length;
iwpbl->user_base = virt;
@@ -1844,7 +1882,7 @@ static struct ib_mr *i40iw_reg_user_mr(struct ib_pd *pd,
goto error;
if (use_pbles) {
- ret = i40iw_check_mr_contiguous(palloc, region->page_size);
+ ret = i40iw_check_mr_contiguous(palloc, iwmr->page_size);
if (ret) {
i40iw_free_pble(iwdev->pble_rsrc, palloc);
iwpbl->pbl_allocated = false;
@@ -1867,6 +1905,7 @@ static struct ib_mr *i40iw_reg_user_mr(struct ib_pd *pd,
i40iw_free_stag(iwdev, stag);
goto error;
}
+
break;
default:
goto error;
diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.h b/drivers/infiniband/hw/i40iw/i40iw_verbs.h
index 0069be8..6549c93 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.h
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.h
@@ -92,6 +92,8 @@ struct i40iw_mr {
struct ib_umem *region;
u16 type;
u32 page_cnt;
+ u32 page_size;
+ u64 page_msk;
u32 npages;
u32 stag;
u64 length;
--
2.8.3
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related
* [PATCH] i40iw: Add missing cleanup on device close
From: Henry Orosco @ 2016-11-30 20:59 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA,
e1000-rdma-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f, Mustafa Ismail
From: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
On i40iw device close, disconnect all connected QPs by moving
them to error state; and block further QPs, PDs and CQs from
being created. Additionally, make sure all resources have been
freed before deallocating the ibdev as part of the device close.
Signed-off-by: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/i40iw/i40iw.h | 5 +++++
drivers/infiniband/hw/i40iw/i40iw_cm.c | 31 +++++++++++++++++++++++++++++++
drivers/infiniband/hw/i40iw/i40iw_cm.h | 2 ++
drivers/infiniband/hw/i40iw/i40iw_d.h | 2 ++
drivers/infiniband/hw/i40iw/i40iw_main.c | 4 ++++
drivers/infiniband/hw/i40iw/i40iw_utils.c | 21 +++++++++++++++++++++
drivers/infiniband/hw/i40iw/i40iw_verbs.c | 25 +++++++++++++++++++++++++
7 files changed, 90 insertions(+)
diff --git a/drivers/infiniband/hw/i40iw/i40iw.h b/drivers/infiniband/hw/i40iw/i40iw.h
index dac9a6b..c795c61 100644
--- a/drivers/infiniband/hw/i40iw/i40iw.h
+++ b/drivers/infiniband/hw/i40iw/i40iw.h
@@ -303,10 +303,13 @@ struct i40iw_device {
u32 mr_stagmask;
u32 mpa_version;
bool dcb;
+ bool closing;
u32 used_pds;
u32 used_cqs;
u32 used_mrs;
u32 used_qps;
+ wait_queue_head_t close_wq;
+ atomic64_t use_count;
};
struct i40iw_ib_device {
@@ -521,6 +524,8 @@ int i40iw_modify_qp(struct ib_qp *, struct ib_qp_attr *, int, struct ib_udata *)
void i40iw_rem_pdusecount(struct i40iw_pd *iwpd, struct i40iw_device *iwdev);
void i40iw_add_pdusecount(struct i40iw_pd *iwpd);
+void i40iw_rem_devusecount(struct i40iw_device *iwdev);
+void i40iw_add_devusecount(struct i40iw_device *iwdev);
void i40iw_hw_modify_qp(struct i40iw_device *iwdev, struct i40iw_qp *iwqp,
struct i40iw_modify_qp_info *info, bool wait);
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index b60e346..11ef0b0 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -4128,3 +4128,34 @@ static void i40iw_cm_post_event(struct i40iw_cm_event *event)
queue_work(event->cm_node->cm_core->event_wq, &event->event_work);
}
+
+/**
+ * i40iw_cm_disconnect_all - disconnect all connected qp's
+ * @iwdev: device pointer
+ */
+void i40iw_cm_disconnect_all(struct i40iw_device *iwdev)
+{
+ struct i40iw_cm_core *cm_core = &iwdev->cm_core;
+ struct list_head *list_core_temp;
+ struct list_head *list_node;
+ struct i40iw_cm_node *cm_node;
+ unsigned long flags;
+ struct list_head connected_list;
+ struct ib_qp_attr attr;
+
+ INIT_LIST_HEAD(&connected_list);
+ spin_lock_irqsave(&cm_core->ht_lock, flags);
+ list_for_each_safe(list_node, list_core_temp, &cm_core->connected_nodes) {
+ cm_node = container_of(list_node, struct i40iw_cm_node, list);
+ atomic_inc(&cm_node->ref_count);
+ list_add(&cm_node->connected_entry, &connected_list);
+ }
+ spin_unlock_irqrestore(&cm_core->ht_lock, flags);
+
+ list_for_each_safe(list_node, list_core_temp, &connected_list) {
+ cm_node = container_of(list_node, struct i40iw_cm_node, connected_entry);
+ attr.qp_state = IB_QPS_ERR;
+ i40iw_modify_qp(&cm_node->iwqp->ibqp, &attr, IB_QP_STATE, NULL);
+ i40iw_rem_ref_cm_node(cm_node);
+ }
+}
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.h b/drivers/infiniband/hw/i40iw/i40iw_cm.h
index 24615c2..0381b7f 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.h
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.h
@@ -339,6 +339,7 @@ struct i40iw_cm_node {
int accept_pend;
struct list_head timer_entry;
struct list_head reset_entry;
+ struct list_head connected_entry;
atomic_t passive_state;
bool qhash_set;
u8 user_pri;
@@ -443,4 +444,5 @@ int i40iw_arp_table(struct i40iw_device *iwdev,
u8 *mac_addr,
u32 action);
+void i40iw_cm_disconnect_all(struct i40iw_device *iwdev);
#endif /* I40IW_CM_H */
diff --git a/drivers/infiniband/hw/i40iw/i40iw_d.h b/drivers/infiniband/hw/i40iw/i40iw_d.h
index e184c0e..1bd4bad 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_d.h
+++ b/drivers/infiniband/hw/i40iw/i40iw_d.h
@@ -35,6 +35,8 @@
#ifndef I40IW_D_H
#define I40IW_D_H
+#define I40IW_FIRST_USER_QP_ID 2
+
#define I40IW_DB_ADDR_OFFSET (4 * 1024 * 1024 - 64 * 1024)
#define I40IW_VF_DB_ADDR_OFFSET (64 * 1024)
diff --git a/drivers/infiniband/hw/i40iw/i40iw_main.c b/drivers/infiniband/hw/i40iw/i40iw_main.c
index 9d3b9ee..d86bb6e 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_main.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_main.c
@@ -1546,6 +1546,7 @@ static enum i40iw_status_code i40iw_setup_init_state(struct i40iw_handler *hdl,
init_waitqueue_head(&iwdev->vchnl_waitq);
init_waitqueue_head(&dev->vf_reqs);
+ init_waitqueue_head(&iwdev->close_wq);
status = i40iw_initialize_dev(iwdev, ldev);
exit:
@@ -1748,6 +1749,9 @@ static void i40iw_close(struct i40e_info *ldev, struct i40e_client *client, bool
return;
iwdev = &hdl->device;
+ iwdev->closing = true;
+
+ i40iw_cm_disconnect_all(iwdev);
destroy_workqueue(iwdev->virtchnl_wq);
i40iw_deinit_device(iwdev, reset);
}
diff --git a/drivers/infiniband/hw/i40iw/i40iw_utils.c b/drivers/infiniband/hw/i40iw/i40iw_utils.c
index 4e880e8..5815128 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_utils.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_utils.c
@@ -392,6 +392,7 @@ static void i40iw_free_qp(struct i40iw_cqp_request *cqp_request, u32 num)
i40iw_rem_pdusecount(iwqp->iwpd, iwdev);
i40iw_free_qp_resources(iwdev, iwqp, qp_num);
+ i40iw_rem_devusecount(iwdev);
}
/**
@@ -459,6 +460,26 @@ enum i40iw_status_code i40iw_handle_cqp_op(struct i40iw_device *iwdev,
}
/**
+ * i40iw_add_devusecount - add dev refcount
+ * @iwdev: dev for refcount
+ */
+void i40iw_add_devusecount(struct i40iw_device *iwdev)
+{
+ atomic64_inc(&iwdev->use_count);
+}
+
+/**
+ * i40iw_rem_devusecount - decrement refcount for dev
+ * @iwdev: device
+ */
+void i40iw_rem_devusecount(struct i40iw_device *iwdev)
+{
+ if (!atomic64_dec_and_test(&iwdev->use_count))
+ return;
+ wake_up(&iwdev->close_wq);
+}
+
+/**
* i40iw_add_pdusecount - add pd refcount
* @iwpd: pd for refcount
*/
diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
index 6ed4f07..206d72b 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
@@ -336,6 +336,9 @@ static struct ib_pd *i40iw_alloc_pd(struct ib_device *ibdev,
u32 pd_id = 0;
int err;
+ if (iwdev->closing)
+ return ERR_PTR(-ENODEV);
+
err = i40iw_alloc_resource(iwdev, iwdev->allocated_pds,
iwdev->max_pd, &pd_id, &iwdev->next_pd);
if (err) {
@@ -601,6 +604,9 @@ static struct ib_qp *i40iw_create_qp(struct ib_pd *ibpd,
struct i40iwarp_offload_info *iwarp_info;
unsigned long flags;
+ if (iwdev->closing)
+ return ERR_PTR(-ENODEV);
+
if (init_attr->create_flags)
return ERR_PTR(-EINVAL);
if (init_attr->cap.max_inline_data > I40IW_MAX_INLINE_DATA_SIZE)
@@ -776,6 +782,7 @@ static struct ib_qp *i40iw_create_qp(struct ib_pd *ibpd,
iwqp->sig_all = (init_attr->sq_sig_type == IB_SIGNAL_ALL_WR) ? 1 : 0;
iwdev->qp_table[qp_num] = iwqp;
i40iw_add_pdusecount(iwqp->iwpd);
+ i40iw_add_devusecount(iwdev);
if (ibpd->uobject && udata) {
memset(&uresp, 0, sizeof(uresp));
uresp.actual_sq_size = sq_size;
@@ -887,6 +894,11 @@ int i40iw_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
spin_lock_irqsave(&iwqp->lock, flags);
if (attr_mask & IB_QP_STATE) {
+ if (iwdev->closing && attr->qp_state != IB_QPS_ERR) {
+ err = -EINVAL;
+ goto exit;
+ }
+
switch (attr->qp_state) {
case IB_QPS_INIT:
case IB_QPS_RTR:
@@ -1086,6 +1098,7 @@ static int i40iw_destroy_cq(struct ib_cq *ib_cq)
cq_wq_destroy(iwdev, cq);
cq_free_resources(iwdev, iwcq);
kfree(iwcq);
+ i40iw_rem_devusecount(iwdev);
return 0;
}
@@ -1116,6 +1129,9 @@ static struct ib_cq *i40iw_create_cq(struct ib_device *ibdev,
int err_code;
int entries = attr->cqe;
+ if (iwdev->closing)
+ return ERR_PTR(-ENODEV);
+
if (entries > iwdev->max_cqe)
return ERR_PTR(-EINVAL);
@@ -1233,6 +1249,7 @@ static struct ib_cq *i40iw_create_cq(struct ib_device *ibdev,
}
}
+ i40iw_add_devusecount(iwdev);
return (struct ib_cq *)iwcq;
cq_destroy:
@@ -1270,6 +1287,7 @@ static void i40iw_free_stag(struct i40iw_device *iwdev, u32 stag)
stag_idx = (stag & iwdev->mr_stagmask) >> I40IW_CQPSQ_STAG_IDX_SHIFT;
i40iw_free_resource(iwdev, iwdev->allocated_mrs, stag_idx);
+ i40iw_rem_devusecount(iwdev);
}
/**
@@ -1300,6 +1318,7 @@ static u32 i40iw_create_stag(struct i40iw_device *iwdev)
stag = stag_index << I40IW_CQPSQ_STAG_IDX_SHIFT;
stag |= driver_key;
stag += (u32)consumer_key;
+ i40iw_add_devusecount(iwdev);
}
return stag;
}
@@ -1809,6 +1828,9 @@ static struct ib_mr *i40iw_reg_user_mr(struct ib_pd *pd,
int ret;
int pg_shift;
+ if (iwdev->closing)
+ return ERR_PTR(-ENODEV);
+
if (length > I40IW_MAX_MR_SIZE)
return ERR_PTR(-EINVAL);
region = ib_umem_get(pd->uobject->context, start, length, acc, 0);
@@ -2842,6 +2864,9 @@ void i40iw_destroy_rdma_device(struct i40iw_ib_device *iwibdev)
i40iw_unregister_rdma_device(iwibdev);
kfree(iwibdev->ibdev.iwcm);
iwibdev->ibdev.iwcm = NULL;
+ wait_event_timeout(iwibdev->iwdev->close_wq,
+ !atomic64_read(&iwibdev->iwdev->use_count),
+ I40IW_EVENT_TIMEOUT);
ib_dealloc_device(&iwibdev->ibdev);
}
--
2.8.3
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related
* [PATCH net-next] net/mlx5e: skip loopback selftest with !CONFIG_INET
From: Arnd Bergmann @ 2016-11-30 21:05 UTC (permalink / raw)
To: Saeed Mahameed, Matan Barak, Leon Romanovsky
Cc: Arnd Bergmann, David S. Miller, Kamal Heib, netdev, linux-rdma,
linux-kernel
When CONFIG_INET is disabled, the new selftest results in a link
error:
drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.o: In function `mlx5e_test_loopback':
en_selftest.c:(.text.mlx5e_test_loopback+0x2ec): undefined reference to `ip_send_check'
en_selftest.c:(.text.mlx5e_test_loopback+0x34c): undefined reference to `udp4_hwcsum'
This hides the specific test in that configuration.
Fixes: 0952da791c97 ("net/mlx5e: Add support for loopback selftest")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c b/drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c
index c32af7daf3ff..65442c36a6e1 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c
@@ -39,7 +39,9 @@ enum {
MLX5E_ST_LINK_STATE,
MLX5E_ST_LINK_SPEED,
MLX5E_ST_HEALTH_INFO,
+#ifdef CONFIG_INET
MLX5E_ST_LOOPBACK,
+#endif
MLX5E_ST_NUM,
};
@@ -47,7 +49,9 @@ const char mlx5e_self_tests[MLX5E_ST_NUM][ETH_GSTRING_LEN] = {
"Link Test",
"Speed Test",
"Health Test",
+#ifdef CONFIG_INET
"Loopback Test",
+#endif
};
int mlx5e_self_test_num(struct mlx5e_priv *priv)
@@ -93,6 +97,7 @@ static int mlx5e_test_link_speed(struct mlx5e_priv *priv)
return 1;
}
+#ifdef CONFIG_INET
/* loopback test */
#define MLX5E_TEST_PKT_SIZE (MLX5_MPWRQ_SMALL_PACKET_THRESHOLD - NET_IP_ALIGN)
static const char mlx5e_test_text[ETH_GSTRING_LEN] = "MLX5E SELF TEST";
@@ -304,12 +309,15 @@ static int mlx5e_test_loopback(struct mlx5e_priv *priv)
kfree(lbtp);
return err;
}
+#endif
static int (*mlx5e_st_func[MLX5E_ST_NUM])(struct mlx5e_priv *) = {
mlx5e_test_link_state,
mlx5e_test_link_speed,
mlx5e_test_health_info,
- mlx5e_test_loopback
+#ifdef CONFIG_INET
+ mlx5e_test_loopback,
+#endif
};
void mlx5e_self_test(struct net_device *ndev, struct ethtool_test *etest,
--
2.9.0
^ permalink raw reply related
* [PATCH] i40iw: Add IP addr handling on netdev events
From: Henry Orosco @ 2016-11-30 21:07 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA,
e1000-rdma-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f, Mustafa Ismail
From: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Disable listeners and disconnect all connected QPs on
a netdev interface down event. On an interface up event,
the listeners are re-enabled.
Signed-off-by: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/i40iw/i40iw_cm.c | 138 +++++++++++++++++++++++++++++-
drivers/infiniband/hw/i40iw/i40iw_cm.h | 2 +
drivers/infiniband/hw/i40iw/i40iw_utils.c | 58 ++++---------
3 files changed, 156 insertions(+), 42 deletions(-)
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index 11ef0b0..93ae764 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -1556,9 +1556,15 @@ static enum i40iw_status_code i40iw_del_multiple_qhash(
memcpy(cm_info->loc_addr, child_listen_node->loc_addr,
sizeof(cm_info->loc_addr));
cm_info->vlan_id = child_listen_node->vlan_id;
- ret = i40iw_manage_qhash(iwdev, cm_info,
- I40IW_QHASH_TYPE_TCP_SYN,
- I40IW_QHASH_MANAGE_TYPE_DELETE, NULL, false);
+ if (child_listen_node->qhash_set) {
+ ret = i40iw_manage_qhash(iwdev, cm_info,
+ I40IW_QHASH_TYPE_TCP_SYN,
+ I40IW_QHASH_MANAGE_TYPE_DELETE,
+ NULL, false);
+ child_listen_node->qhash_set = false;
+ } else {
+ ret = I40IW_SUCCESS;
+ }
i40iw_debug(&iwdev->sc_dev,
I40IW_DEBUG_CM,
"freed pointer = %p\n",
@@ -1687,6 +1693,7 @@ static enum i40iw_status_code i40iw_add_mqh_6(struct i40iw_device *iwdev,
I40IW_QHASH_MANAGE_TYPE_ADD,
NULL, true);
if (!ret) {
+ child_listen_node->qhash_set = true;
spin_lock_irqsave(&iwdev->cm_core.listen_list_lock, flags);
list_add(&child_listen_node->child_listen_list,
&cm_parent_listen_node->child_listen_list);
@@ -1765,6 +1772,7 @@ static enum i40iw_status_code i40iw_add_mqh_4(
NULL,
true);
if (!ret) {
+ child_listen_node->qhash_set = true;
spin_lock_irqsave(&iwdev->cm_core.listen_list_lock, flags);
list_add(&child_listen_node->child_listen_list,
&cm_parent_listen_node->child_listen_list);
@@ -4130,6 +4138,73 @@ static void i40iw_cm_post_event(struct i40iw_cm_event *event)
}
/**
+ * i40iw_qhash_ctrl - enable/disable qhash for list
+ * @iwdev: device pointer
+ * @parent_listen_node: parent listen node
+ * @nfo: cm info node
+ * @ipaddr: Pointer to IPv4 or IPv6 address
+ * @ipv4: flag indicating IPv4 when true
+ * @ifup: flag indicating interface up when true
+ *
+ * Enables or disables the qhash for the node in the child
+ * listen list that matches ipaddr. If no matching IP was found
+ * it will allocate and add a new child listen node to the
+ * parent listen node. The listen_list_lock is assumed to be
+ * held when called.
+ */
+static void i40iw_qhash_ctrl(struct i40iw_device *iwdev,
+ struct i40iw_cm_listener *parent_listen_node,
+ struct i40iw_cm_info *nfo,
+ u32 *ipaddr, bool ipv4, bool ifup)
+{
+ struct list_head *child_listen_list = &parent_listen_node->child_listen_list;
+ struct i40iw_cm_listener *child_listen_node;
+ struct list_head *pos, *tpos;
+ enum i40iw_status_code ret;
+ bool node_allocated = false;
+ enum i40iw_quad_hash_manage_type op =
+ ifup ? I40IW_QHASH_MANAGE_TYPE_ADD : I40IW_QHASH_MANAGE_TYPE_DELETE;
+
+ list_for_each_safe(pos, tpos, child_listen_list) {
+ child_listen_node =
+ list_entry(pos,
+ struct i40iw_cm_listener,
+ child_listen_list);
+ if (!memcmp(child_listen_node->loc_addr, ipaddr, ipv4 ? 4 : 16))
+ goto set_qhash;
+ }
+
+ /* if not found then add a child listener if interface is going up */
+ if (!ifup)
+ return;
+ child_listen_node = kzalloc(sizeof(*child_listen_node), GFP_ATOMIC);
+ if (!child_listen_node)
+ return;
+ node_allocated = true;
+ memcpy(child_listen_node, parent_listen_node, sizeof(*child_listen_node));
+
+ memcpy(child_listen_node->loc_addr, ipaddr, ipv4 ? 4 : 16);
+
+set_qhash:
+ memcpy(nfo->loc_addr,
+ child_listen_node->loc_addr,
+ sizeof(nfo->loc_addr));
+ nfo->vlan_id = child_listen_node->vlan_id;
+ ret = i40iw_manage_qhash(iwdev, nfo,
+ I40IW_QHASH_TYPE_TCP_SYN,
+ op,
+ NULL, false);
+ if (!ret) {
+ child_listen_node->qhash_set = ifup;
+ if (node_allocated)
+ list_add(&child_listen_node->child_listen_list,
+ &parent_listen_node->child_listen_list);
+ } else if (node_allocated) {
+ kfree(child_listen_node);
+ }
+}
+
+/**
* i40iw_cm_disconnect_all - disconnect all connected qp's
* @iwdev: device pointer
*/
@@ -4159,3 +4234,60 @@ void i40iw_cm_disconnect_all(struct i40iw_device *iwdev)
i40iw_rem_ref_cm_node(cm_node);
}
}
+
+/**
+ * i40iw_ifdown_notify - process an ifdown on an interface
+ * @iwdev: device pointer
+ * @ipaddr: Pointer to IPv4 or IPv6 address
+ * @ipv4: flag indicating IPv4 when true
+ * @ifup: flag indicating interface up when true
+ */
+void i40iw_if_notify(struct i40iw_device *iwdev, struct net_device *netdev,
+ u32 *ipaddr, bool ipv4, bool ifup)
+{
+ struct i40iw_cm_core *cm_core = &iwdev->cm_core;
+ unsigned long flags;
+ struct i40iw_cm_listener *listen_node;
+ static const u32 ip_zero[4] = { 0, 0, 0, 0 };
+ struct i40iw_cm_info nfo;
+ u16 vlan_id = rdma_vlan_dev_vlan_id(netdev);
+ enum i40iw_status_code ret;
+ enum i40iw_quad_hash_manage_type op =
+ ifup ? I40IW_QHASH_MANAGE_TYPE_ADD : I40IW_QHASH_MANAGE_TYPE_DELETE;
+
+ /* Disable or enable qhash for listeners */
+ spin_lock_irqsave(&cm_core->listen_list_lock, flags);
+ list_for_each_entry(listen_node, &cm_core->listen_nodes, list) {
+ if (vlan_id == listen_node->vlan_id &&
+ (!memcmp(listen_node->loc_addr, ipaddr, ipv4 ? 4 : 16) ||
+ !memcmp(listen_node->loc_addr, ip_zero, ipv4 ? 4 : 16))) {
+ memcpy(nfo.loc_addr, listen_node->loc_addr,
+ sizeof(nfo.loc_addr));
+ nfo.loc_port = listen_node->loc_port;
+ nfo.ipv4 = listen_node->ipv4;
+ nfo.vlan_id = listen_node->vlan_id;
+ nfo.user_pri = listen_node->user_pri;
+ if (!list_empty(&listen_node->child_listen_list)) {
+ i40iw_qhash_ctrl(iwdev,
+ listen_node,
+ &nfo,
+ ipaddr, ipv4, ifup);
+ } else if (memcmp(listen_node->loc_addr, ip_zero,
+ ipv4 ? 4 : 16)) {
+ ret = i40iw_manage_qhash(iwdev,
+ &nfo,
+ I40IW_QHASH_TYPE_TCP_SYN,
+ op,
+ NULL,
+ false);
+ if (!ret)
+ listen_node->qhash_set = ifup;
+ }
+ }
+ }
+ spin_unlock_irqrestore(&cm_core->listen_list_lock, flags);
+
+ /* disconnect any connected qp's on ifdown */
+ if (!ifup)
+ i40iw_cm_disconnect_all(iwdev);
+}
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.h b/drivers/infiniband/hw/i40iw/i40iw_cm.h
index 0381b7f..49ed7a5 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.h
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.h
@@ -444,5 +444,7 @@ int i40iw_arp_table(struct i40iw_device *iwdev,
u8 *mac_addr,
u32 action);
+void i40iw_if_notify(struct i40iw_device *iwdev, struct net_device *netdev,
+ u32 *ipaddr, bool ipv4, bool ifup);
void i40iw_cm_disconnect_all(struct i40iw_device *iwdev);
#endif /* I40IW_CM_H */
diff --git a/drivers/infiniband/hw/i40iw/i40iw_utils.c b/drivers/infiniband/hw/i40iw/i40iw_utils.c
index 5815128..641f00f 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_utils.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_utils.c
@@ -153,6 +153,7 @@ int i40iw_inetaddr_event(struct notifier_block *notifier,
struct i40iw_device *iwdev;
struct i40iw_handler *hdl;
u32 local_ipaddr;
+ u32 action = I40IW_ARP_ADD;
hdl = i40iw_find_netdev(event_netdev);
if (!hdl)
@@ -164,44 +165,25 @@ int i40iw_inetaddr_event(struct notifier_block *notifier,
if (netdev != event_netdev)
return NOTIFY_DONE;
+ if (upper_dev)
+ local_ipaddr = ntohl(
+ ((struct in_device *)upper_dev->ip_ptr)->ifa_list->ifa_address);
+ else
+ local_ipaddr = ntohl(ifa->ifa_address);
switch (event) {
case NETDEV_DOWN:
- if (upper_dev)
- local_ipaddr = ntohl(
- ((struct in_device *)upper_dev->ip_ptr)->ifa_list->ifa_address);
- else
- local_ipaddr = ntohl(ifa->ifa_address);
- i40iw_manage_arp_cache(iwdev,
- netdev->dev_addr,
- &local_ipaddr,
- true,
- I40IW_ARP_DELETE);
- return NOTIFY_OK;
+ action = I40IW_ARP_DELETE;
+ /* Fall through */
case NETDEV_UP:
- if (upper_dev)
- local_ipaddr = ntohl(
- ((struct in_device *)upper_dev->ip_ptr)->ifa_list->ifa_address);
- else
- local_ipaddr = ntohl(ifa->ifa_address);
- i40iw_manage_arp_cache(iwdev,
- netdev->dev_addr,
- &local_ipaddr,
- true,
- I40IW_ARP_ADD);
- break;
+ /* Fall through */
case NETDEV_CHANGEADDR:
- /* Add the address to the IP table */
- if (upper_dev)
- local_ipaddr = ntohl(
- ((struct in_device *)upper_dev->ip_ptr)->ifa_list->ifa_address);
- else
- local_ipaddr = ntohl(ifa->ifa_address);
-
i40iw_manage_arp_cache(iwdev,
netdev->dev_addr,
&local_ipaddr,
true,
- I40IW_ARP_ADD);
+ action);
+ i40iw_if_notify(iwdev, netdev, &local_ipaddr, true,
+ (action == I40IW_ARP_ADD) ? true : false);
break;
default:
break;
@@ -225,6 +207,7 @@ int i40iw_inet6addr_event(struct notifier_block *notifier,
struct i40iw_device *iwdev;
struct i40iw_handler *hdl;
u32 local_ipaddr6[4];
+ u32 action = I40IW_ARP_ADD;
hdl = i40iw_find_netdev(event_netdev);
if (!hdl)
@@ -235,24 +218,21 @@ int i40iw_inet6addr_event(struct notifier_block *notifier,
if (netdev != event_netdev)
return NOTIFY_DONE;
+ i40iw_copy_ip_ntohl(local_ipaddr6, ifa->addr.in6_u.u6_addr32);
switch (event) {
case NETDEV_DOWN:
- i40iw_copy_ip_ntohl(local_ipaddr6, ifa->addr.in6_u.u6_addr32);
- i40iw_manage_arp_cache(iwdev,
- netdev->dev_addr,
- local_ipaddr6,
- false,
- I40IW_ARP_DELETE);
- return NOTIFY_OK;
+ action = I40IW_ARP_DELETE;
+ /* Fall through */
case NETDEV_UP:
/* Fall through */
case NETDEV_CHANGEADDR:
- i40iw_copy_ip_ntohl(local_ipaddr6, ifa->addr.in6_u.u6_addr32);
i40iw_manage_arp_cache(iwdev,
netdev->dev_addr,
local_ipaddr6,
false,
- I40IW_ARP_ADD);
+ action);
+ i40iw_if_notify(iwdev, netdev, local_ipaddr6, false,
+ (action == I40IW_ARP_ADD) ? true : false);
break;
default:
break;
--
2.8.3
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related
* [PATCH] i40iw: Replace list_for_each_entry macro with safe version
From: Henry Orosco @ 2016-11-30 21:08 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA,
e1000-rdma-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f, Mustafa Ismail
From: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Use list_for_each_entry_safe macro for the IPv6 addr list
as IPv6 addresses can be deleted while going through the
list.
Signed-off-by: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/i40iw/i40iw_cm.c | 4 ++--
drivers/infiniband/hw/i40iw/i40iw_main.c | 4 ++--
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index 93ae764..9a14880 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -1644,7 +1644,7 @@ static enum i40iw_status_code i40iw_add_mqh_6(struct i40iw_device *iwdev,
{
struct net_device *ip_dev;
struct inet6_dev *idev;
- struct inet6_ifaddr *ifp;
+ struct inet6_ifaddr *ifp, *tmp;
enum i40iw_status_code ret = 0;
struct i40iw_cm_listener *child_listen_node;
unsigned long flags;
@@ -1659,7 +1659,7 @@ static enum i40iw_status_code i40iw_add_mqh_6(struct i40iw_device *iwdev,
i40iw_pr_err("idev == NULL\n");
break;
}
- list_for_each_entry(ifp, &idev->addr_list, if_list) {
+ list_for_each_entry_safe(ifp, tmp, &idev->addr_list, if_list) {
i40iw_debug(&iwdev->sc_dev,
I40IW_DEBUG_CM,
"IP=%pI6, vlan_id=%d, MAC=%pM\n",
diff --git a/drivers/infiniband/hw/i40iw/i40iw_main.c b/drivers/infiniband/hw/i40iw/i40iw_main.c
index d86bb6e..4ce05b8 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_main.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_main.c
@@ -1164,7 +1164,7 @@ static void i40iw_add_ipv6_addr(struct i40iw_device *iwdev)
{
struct net_device *ip_dev;
struct inet6_dev *idev;
- struct inet6_ifaddr *ifp;
+ struct inet6_ifaddr *ifp, *tmp;
u32 local_ipaddr6[4];
rcu_read_lock();
@@ -1177,7 +1177,7 @@ static void i40iw_add_ipv6_addr(struct i40iw_device *iwdev)
i40iw_pr_err("ipv6 inet device not found\n");
break;
}
- list_for_each_entry(ifp, &idev->addr_list, if_list) {
+ list_for_each_entry_safe(ifp, tmp, &idev->addr_list, if_list) {
i40iw_pr_info("IP=%pI6, vlan_id=%d, MAC=%pM\n", &ifp->addr,
rdma_vlan_dev_vlan_id(ip_dev), ip_dev->dev_addr);
i40iw_copy_ip_ntohl(local_ipaddr6,
--
2.8.3
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related
* [PATCH] i40iw: Add NULL check for ibqp event handler
From: Henry Orosco @ 2016-11-30 21:09 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA,
e1000-rdma-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f, Shiraz Saleem
From: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Add NULL check for ibqp event handler before calling it to report
QP events, as it might not initialized.
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Faisal Latif <faisal.latif-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/i40iw/i40iw_cm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index 9a14880..13b6dee 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -3474,7 +3474,7 @@ static void i40iw_cm_disconn_true(struct i40iw_qp *iwqp)
/* Flush the queues */
i40iw_flush_wqes(iwdev, iwqp);
- if (qp->term_flags) {
+ if (qp->term_flags && iwqp->ibqp.event_handler) {
ibevent.device = iwqp->ibqp.device;
ibevent.event = (qp->eventtype == TERM_EVENT_QP_FATAL) ?
IB_EVENT_QP_FATAL : IB_EVENT_QP_ACCESS_ERR;
--
2.8.3
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related
* [PATCH] i40iw: Set TOS field in IP header
From: Henry Orosco @ 2016-11-30 21:09 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA,
e1000-rdma-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f, Shiraz Saleem
From: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Set the TOS field in IP header with the value passed in
from application. If there is mismatch between the remote
client's TOS and listener, set the listener Tos to the higher
of the two values.
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Faisal Latif <faisal.latif-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/i40iw/i40iw_cm.c | 25 +++++++++++++++++++------
drivers/infiniband/hw/i40iw/i40iw_cm.h | 3 +++
2 files changed, 22 insertions(+), 6 deletions(-)
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index 13b6dee..0c92037 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -445,7 +445,7 @@ static struct i40iw_puda_buf *i40iw_form_cm_frame(struct i40iw_cm_node *cm_node,
iph->version = IPVERSION;
iph->ihl = 5; /* 5 * 4Byte words, IP headr len */
- iph->tos = 0;
+ iph->tos = cm_node->tos;
iph->tot_len = htons(packetsize);
iph->id = htons(++cm_node->tcp_cntxt.loc_id);
@@ -472,7 +472,8 @@ static struct i40iw_puda_buf *i40iw_form_cm_frame(struct i40iw_cm_node *cm_node,
ethh->h_proto = htons(ETH_P_IPV6);
}
ip6h->version = 6;
- ip6h->flow_lbl[0] = 0;
+ ip6h->priority = cm_node->tos >> 4;
+ ip6h->flow_lbl[0] = cm_node->tos << 4;
ip6h->flow_lbl[1] = 0;
ip6h->flow_lbl[2] = 0;
ip6h->payload_len = htons(packetsize - sizeof(*ip6h));
@@ -2141,9 +2142,18 @@ static struct i40iw_cm_node *i40iw_make_cm_node(
cm_node->vlan_id = cm_info->vlan_id;
if ((cm_node->vlan_id == I40IW_NO_VLAN) && iwdev->dcb)
cm_node->vlan_id = 0;
+ cm_node->tos = cm_info->tos;
cm_node->user_pri = cm_info->user_pri;
- if (listener)
- cm_node->user_pri = listener->user_pri;
+ if (listener) {
+ if (listener->tos != cm_info->tos)
+ i40iw_debug(&iwdev->sc_dev, I40IW_DEBUG_DCB,
+ "application TOS[%d] and remote client TOS[%d] mismatch\n",
+ listener->tos, cm_info->tos);
+ cm_node->tos = max(listener->tos, cm_info->tos);
+ cm_node->user_pri = rt_tos2priority(cm_node->tos);
+ i40iw_debug(&iwdev->sc_dev, I40IW_DEBUG_DCB, "listener: TOS:[%d] UP:[%d]\n",
+ cm_node->tos, cm_node->user_pri);
+ }
memcpy(cm_node->loc_addr, cm_info->loc_addr, sizeof(cm_node->loc_addr));
memcpy(cm_node->rem_addr, cm_info->rem_addr, sizeof(cm_node->rem_addr));
cm_node->loc_port = cm_info->loc_port;
@@ -3092,6 +3102,7 @@ void i40iw_receive_ilq(struct i40iw_sc_dev *dev, struct i40iw_puda_buf *rbuf)
cm_info.loc_addr[0] = ntohl(iph->daddr);
cm_info.rem_addr[0] = ntohl(iph->saddr);
cm_info.ipv4 = true;
+ cm_info.tos = iph->tos;
} else {
ip6h = (struct ipv6hdr *)rbuf->iph;
i40iw_copy_ip_ntohl(cm_info.loc_addr,
@@ -3099,6 +3110,7 @@ void i40iw_receive_ilq(struct i40iw_sc_dev *dev, struct i40iw_puda_buf *rbuf)
i40iw_copy_ip_ntohl(cm_info.rem_addr,
ip6h->saddr.in6_u.u6_addr32);
cm_info.ipv4 = false;
+ cm_info.tos = (ip6h->priority << 4) | (ip6h->flow_lbl[0] >> 4);
}
cm_info.loc_port = ntohs(tcph->dest);
cm_info.rem_port = ntohs(tcph->source);
@@ -3331,6 +3343,7 @@ static void i40iw_cm_init_tsa_conn(struct i40iw_qp *iwqp,
cm_node->state = I40IW_CM_STATE_OFFLOADED;
tcp_info.tcp_state = I40IW_TCP_STATE_ESTABLISHED;
tcp_info.src_mac_addr_idx = iwdev->mac_ip_table_idx;
+ tcp_info.tos = cm_node->tos;
dev->iw_priv_qp_ops->qp_setctx(&iwqp->sc_qp, (u64 *)(iwqp->host_ctx.va), ctx_info);
@@ -3763,6 +3776,7 @@ int i40iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
i40iw_netdev_vlan_ipv6(cm_info.loc_addr, &cm_info.vlan_id, NULL);
}
cm_info.cm_id = cm_id;
+ cm_info.tos = cm_id->tos;
cm_info.user_pri = rt_tos2priority(cm_id->tos);
i40iw_debug(&iwdev->sc_dev, I40IW_DEBUG_DCB, "%s TOS:[%d] UP:[%d]\n",
__func__, cm_id->tos, cm_info.user_pri);
@@ -3911,10 +3925,9 @@ int i40iw_create_listen(struct iw_cm_id *cm_id, int backlog)
cm_id->provider_data = cm_listen_node;
+ cm_listen_node->tos = cm_id->tos;
cm_listen_node->user_pri = rt_tos2priority(cm_id->tos);
cm_info.user_pri = cm_listen_node->user_pri;
- i40iw_debug(&iwdev->sc_dev, I40IW_DEBUG_DCB, "%s TOS:[%d] UP:[%d]\n",
- __func__, cm_id->tos, cm_listen_node->user_pri);
if (!cm_listen_node->reused_node) {
if (wildcard) {
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.h b/drivers/infiniband/hw/i40iw/i40iw_cm.h
index 49ed7a5..2e52e38 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.h
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.h
@@ -297,6 +297,7 @@ struct i40iw_cm_listener {
enum i40iw_cm_listener_state listener_state;
u32 reused_node;
u8 user_pri;
+ u8 tos;
u16 vlan_id;
bool qhash_set;
bool ipv4;
@@ -343,6 +344,7 @@ struct i40iw_cm_node {
atomic_t passive_state;
bool qhash_set;
u8 user_pri;
+ u8 tos;
bool ipv4;
bool snd_mark_en;
u16 lsmm_size;
@@ -368,6 +370,7 @@ struct i40iw_cm_info {
u16 vlan_id;
int backlog;
u8 user_pri;
+ u8 tos;
bool ipv4;
};
--
2.8.3
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related
* RE: [PATCH rdma-next 01/10] IB/core: Add raw packet protocol
From: Hefty, Sean @ 2016-11-30 21:12 UTC (permalink / raw)
To: Doug Ledford, Jason Gunthorpe
Cc: Steve Wise, 'Leon Romanovsky',
linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
'Steve Wise', Marciniszyn, Mike, Dalessandro, Dennis,
'Lijun Ou', 'Wei Hu(Xavier)', Latif, Faisal,
'Yishai Hadas', 'Selvin Xavier',
'Devesh Sharma', 'Mitesh Ahuja',
'Christian Benvenuti', 'Dave Goodell',
'Moni Shoua', 'Or Gerlitz'
In-Reply-To: <0f905c52-b167-4b7d-4fd0-091056997e47-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
> I'm not entirely sure I agree....I would have to think about it more,
> but the underlying problem I'm concerned with is exactly what I point
> out above: what restrictions would we be placing on the hardware that
> are artificial and a result of our stack that the hardware itself does
> not share? Could the hardware support automatic migration from iWARP
> to
> RoCE or vice versa if both endpoints supported it? Would that work if
> we required two separate ib_devices? Things like that.
IMO, in order to create a usable software interface, we will end up defining limitations. The alternative seems to be an endless series of capability bits that expands to millions of combinations, and in the end still won't capture everything. Consider that the existing interfaces already limit what upstream drivers can expose.
The primary reason for exposing devices is for apps to associate hardware resources (e.g. CQ and QP) with each other. With software based implementations of roce/iwarp, a device is basically the system. If you connect NICs directly to the CPU, then even separate hardware devices could share state, which could complicate things even more.
I'm not sure about Jason's idea of re-defining the PD as a 'protocol domain' (my term), but I think we should seriously consider something similar as an addition to the <device, port> model. I'm still of the opinion that protocol is a property of a QP, not a port.
- Sean
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* [PATCH] i40iw: Fill in IRD value when on connect request
From: Henry Orosco @ 2016-11-30 21:12 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA,
e1000-rdma-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f, Shiraz Saleem
From: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
IRD is not populated on connect request and application is
getting 0 for the value. Fill in the correct value on
connect request.
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Faisal Latif <faisal.latif-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/i40iw/i40iw_cm.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index 0c92037..2f14de7 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -272,6 +272,7 @@ static int i40iw_send_cm_event(struct i40iw_cm_node *cm_node,
event.provider_data = (void *)cm_node;
event.private_data = (void *)cm_node->pdata_buf;
event.private_data_len = (u8)cm_node->pdata.size;
+ event.ird = cm_node->ird_size;
break;
case IW_CM_EVENT_CONNECT_REPLY:
i40iw_get_cmevent_info(cm_node, cm_id, &event);
--
2.8.3
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related
* [PATCH] i40iw: Correctly fail loopback connection if no listener
From: Henry Orosco @ 2016-11-30 21:12 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA,
e1000-rdma-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f, Shiraz Saleem
From: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Fail the connect and return the proper error code if a client
is started with local IP address and there is no corresponding
loopback listener.
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Faisal Latif <faisal.latif-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/i40iw/i40iw_cm.c | 62 +++++++++++++++++++---------------
1 file changed, 35 insertions(+), 27 deletions(-)
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index 2f14de7..25af89a 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -2878,7 +2878,7 @@ static struct i40iw_cm_node *i40iw_create_cm_node(
/* create a CM connection node */
cm_node = i40iw_make_cm_node(cm_core, iwdev, cm_info, NULL);
if (!cm_node)
- return NULL;
+ return ERR_PTR(-ENOMEM);
/* set our node side to client (active) side */
cm_node->tcp_cntxt.client = 1;
cm_node->tcp_cntxt.rcv_wscale = I40IW_CM_DEFAULT_RCV_WND_SCALE;
@@ -2891,7 +2891,8 @@ static struct i40iw_cm_node *i40iw_create_cm_node(
cm_node->vlan_id,
I40IW_CM_LISTENER_ACTIVE_STATE);
if (!loopback_remotelistener) {
- i40iw_create_event(cm_node, I40IW_CM_EVENT_ABORTED);
+ i40iw_rem_ref_cm_node(cm_node);
+ return ERR_PTR(-ECONNREFUSED);
} else {
loopback_cm_info = *cm_info;
loopback_cm_info.loc_port = cm_info->rem_port;
@@ -2904,7 +2905,7 @@ static struct i40iw_cm_node *i40iw_create_cm_node(
loopback_remotelistener);
if (!loopback_remotenode) {
i40iw_rem_ref_cm_node(cm_node);
- return NULL;
+ return ERR_PTR(-ENOMEM);
}
cm_core->stats_loopbacks++;
loopback_remotenode->loopbackpartner = cm_node;
@@ -3732,6 +3733,7 @@ int i40iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
struct sockaddr_in6 *raddr6;
bool qhash_set = false;
int apbvt_set = 0;
+ int err = 0;
enum i40iw_status_code status;
ibqp = i40iw_get_qp(cm_id->device, conn_param->qpn);
@@ -3812,8 +3814,11 @@ int i40iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
conn_param->private_data_len,
(void *)conn_param->private_data,
&cm_info);
- if (!cm_node)
- goto err;
+
+ if (IS_ERR(cm_node)) {
+ err = PTR_ERR(cm_node);
+ goto err_out;
+ }
i40iw_record_ird_ord(cm_node, (u16)conn_param->ird, (u16)conn_param->ord);
if (cm_node->send_rdma0_op == SEND_RDMA_READ_ZERO &&
@@ -3827,10 +3832,12 @@ int i40iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
iwqp->cm_id = cm_id;
i40iw_add_ref(&iwqp->ibqp);
- if (cm_node->state == I40IW_CM_STATE_SYN_SENT) {
- if (i40iw_send_syn(cm_node, 0)) {
+ if (cm_node->state != I40IW_CM_STATE_OFFLOADED) {
+ cm_node->state = I40IW_CM_STATE_SYN_SENT;
+ err = i40iw_send_syn(cm_node, 0);
+ if (err) {
i40iw_rem_ref_cm_node(cm_node);
- goto err;
+ goto err_out;
}
}
@@ -3842,24 +3849,25 @@ int i40iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
cm_node->cm_id);
return 0;
-err:
- if (cm_node) {
- if (cm_node->ipv4)
- i40iw_debug(cm_node->dev,
- I40IW_DEBUG_CM,
- "Api - connect() FAILED: dest addr=%pI4",
- cm_node->rem_addr);
- else
- i40iw_debug(cm_node->dev, I40IW_DEBUG_CM,
- "Api - connect() FAILED: dest addr=%pI6",
- cm_node->rem_addr);
- }
- i40iw_manage_qhash(iwdev,
- &cm_info,
- I40IW_QHASH_TYPE_TCP_ESTABLISHED,
- I40IW_QHASH_MANAGE_TYPE_DELETE,
- NULL,
- false);
+err_out:
+ if (cm_info.ipv4)
+ i40iw_debug(&iwdev->sc_dev,
+ I40IW_DEBUG_CM,
+ "Api - connect() FAILED: dest addr=%pI4",
+ cm_info.rem_addr);
+ else
+ i40iw_debug(&iwdev->sc_dev,
+ I40IW_DEBUG_CM,
+ "Api - connect() FAILED: dest addr=%pI6",
+ cm_info.rem_addr);
+
+ if (qhash_set)
+ i40iw_manage_qhash(iwdev,
+ &cm_info,
+ I40IW_QHASH_TYPE_TCP_ESTABLISHED,
+ I40IW_QHASH_MANAGE_TYPE_DELETE,
+ NULL,
+ false);
if (apbvt_set && !i40iw_listen_port_in_use(&iwdev->cm_core,
cm_info.loc_port))
@@ -3868,7 +3876,7 @@ int i40iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
I40IW_MANAGE_APBVT_DEL);
cm_id->rem_ref(cm_id);
iwdev->cm_core.stats_connect_errs++;
- return -ENOMEM;
+ return err;
}
/**
--
2.8.3
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related
* [PATCH] i40iw: Code cleanup, remove check of PBLE pages
From: Henry Orosco @ 2016-11-30 21:13 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA,
e1000-rdma-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f, Henry Orosco
Remove check for zero 'pages' of unallocated pbles calculated in
add_pble_pool(); as it can never be true.
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Henry Orosco <henry.orosco-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/i40iw/i40iw_pble.c | 4 ----
1 file changed, 4 deletions(-)
diff --git a/drivers/infiniband/hw/i40iw/i40iw_pble.c b/drivers/infiniband/hw/i40iw/i40iw_pble.c
index 85993dc..c87ba16 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_pble.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_pble.c
@@ -353,10 +353,6 @@ static enum i40iw_status_code add_pble_pool(struct i40iw_sc_dev *dev,
pages = (idx->rel_pd_idx) ? (I40IW_HMC_PD_CNT_IN_SD -
idx->rel_pd_idx) : I40IW_HMC_PD_CNT_IN_SD;
pages = min(pages, pble_rsrc->unallocated_pble >> PBLE_512_SHIFT);
- if (!pages) {
- ret_code = I40IW_ERR_NO_PBLCHUNKS_AVAILABLE;
- goto error;
- }
info.chunk = chunk;
info.hmc_info = hmc_info;
info.pages = pages;
--
2.8.3
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related
* [PATCH] i40iw: Add request for reset on CQP timeout
From: Henry Orosco @ 2016-11-30 21:14 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA,
e1000-rdma-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f, Henry Orosco
When CQP times out, send a request to LAN driver for reset.
Signed-off-by: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Henry Orosco <henry.orosco-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/i40iw/i40iw.h | 1 +
drivers/infiniband/hw/i40iw/i40iw_utils.c | 10 +++++++++-
2 files changed, 10 insertions(+), 1 deletion(-)
diff --git a/drivers/infiniband/hw/i40iw/i40iw.h b/drivers/infiniband/hw/i40iw/i40iw.h
index c795c61..ef188e6 100644
--- a/drivers/infiniband/hw/i40iw/i40iw.h
+++ b/drivers/infiniband/hw/i40iw/i40iw.h
@@ -304,6 +304,7 @@ struct i40iw_device {
u32 mpa_version;
bool dcb;
bool closing;
+ bool reset;
u32 used_pds;
u32 used_cqs;
u32 used_mrs;
diff --git a/drivers/infiniband/hw/i40iw/i40iw_utils.c b/drivers/infiniband/hw/i40iw/i40iw_utils.c
index 641f00f..4a08ffb 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_utils.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_utils.c
@@ -396,7 +396,10 @@ static int i40iw_wait_event(struct i40iw_device *iwdev,
i40iw_pr_err("error cqp command 0x%x timed out ret = %d\n",
info->cqp_cmd, timeout_ret);
err_code = -ETIME;
- i40iw_request_reset(iwdev);
+ if (!iwdev->reset) {
+ iwdev->reset = true;
+ i40iw_request_reset(iwdev);
+ }
goto done;
}
cqp_error = cqp_request->compl_info.error;
@@ -426,6 +429,11 @@ enum i40iw_status_code i40iw_handle_cqp_op(struct i40iw_device *iwdev,
struct cqp_commands_info *info = &cqp_request->info;
int err_code = 0;
+ if (iwdev->reset) {
+ i40iw_free_cqp_request(&iwdev->cqp, cqp_request);
+ return I40IW_ERR_CQP_COMPL_ERROR;
+ }
+
status = i40iw_process_cqp_cmd(dev, info);
if (status) {
i40iw_pr_err("error cqp command 0x%x failed\n", info->cqp_cmd);
--
2.8.3
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related
* Re: [PATCH v3 00/10] Server-side NFS/RDMA patches proposed for v4.10
From: J. Bruce Fields @ 2016-11-30 22:30 UTC (permalink / raw)
To: Chuck Lever
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA,
linux-nfs-u79uwXL29TY76Z2rM5mHXA
In-Reply-To: <20161129155521.4477.53561.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org>
On Tue, Nov 29, 2016 at 11:04:16AM -0500, Chuck Lever wrote:
> These patches are ready for you to consider for v4.10. Changes proposed
> for v4.10 include:
They look fine to me, thanks; applying for 4.10.--b.
>
> - Drop connection on GSS sequence window overflow
> - Remove unnecessary spin lock in the svc_rdma_send path
> - A number of minor clean-ups
>
> Available in the "nfsd-rdma-for-4.10" topic branch of this git repo:
>
> git://git.linux-nfs.org/projects/cel/cel-2.6.git
>
>
> Or for browsing:
>
> http://git.linux-nfs.org/?p=cel/cel-2.6.git;a=log;h=refs/heads/nfsd-rdma-for-4.10
>
>
> Changes since v2:
> - Rebased on v4.9-rc7
>
>
> Changes since v1:
> - Rebased on v4.9-rc5
> - Address checkpatch.pl warnings
> - Address compiler warnings
> - Remove a dprintk in svc_rdma_get_inv_rkey()
> - Patches reordered so fixes come before clean-ups
>
> ---
>
> Chuck Lever (10):
> svcrdma: Clear xpt_bc_xps in xprt_setup_rdma_bc() error exit arm
> svcauth_gss: Close connection when dropping an incoming message
> svcrdma: Renovate sendto chunk list parsing
> svcrdma: Remove BH-disabled spin locking in svc_rdma_send()
> svcrdma: Remove DMA map accounting
> svcrdma: Remove svc_rdma_op_ctxt::wc_status
> svcrdma: Remove unused variables in xprt_rdma_bc_allocate()
> svcrdma: Remove unused variable in rdma_copy_tail()
> svcrdma: Break up dprintk format in svc_rdma_accept()
> svcrdma: Further clean-up of svc_rdma_get_inv_rkey()
>
>
> include/linux/sunrpc/svc_rdma.h | 7 --
> net/sunrpc/auth_gss/svcauth_gss.c | 2
> net/sunrpc/svc.c | 14 ++-
> net/sunrpc/xprtrdma/svc_rdma_backchannel.c | 5 -
> net/sunrpc/xprtrdma/svc_rdma_recvfrom.c | 21 ++++-
> net/sunrpc/xprtrdma/svc_rdma_sendto.c | 116 ++++++++--------------------
> net/sunrpc/xprtrdma/svc_rdma_transport.c | 94 +++++++----------------
> 7 files changed, 94 insertions(+), 165 deletions(-)
>
> --
> Chuck Lever
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* RE: [1/1] ibacm: Modify comment related to default pkey in ibacm_addr.cfg
From: Hefty, Sean @ 2016-11-30 23:36 UTC (permalink / raw)
To: Patel, Jay P; +Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
In-Reply-To: <1480445462-88653-2-git-send-email-jay.p.patel-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> Comment in ibacm_addr.cfg mentions that default pkey is 0xffff while
> the default pkey is the first entry in pkey table which may be
> different from 0xffff. Hence, comment specifying default key as 0xffff
> is misleading.
Please add a signed-off-by line. Opening a pull request directly against the github tree would simplify merging this.
- Sean
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Re: [PATCH rdma-next 1/4] IB/mlx4: Fix out-of-range array index in destroy qp flow
From: Leon Romanovsky @ 2016-12-01 5:31 UTC (permalink / raw)
To: Bart Van Assche
Cc: dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org,
linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
Jack Morgenstein
In-Reply-To: <bdd9788e-d393-b23e-876c-ae84774fdccd-XdAiOPVOjttBDgjK7y7TUQ@public.gmane.org>
[-- Attachment #1: Type: text/plain, Size: 651 bytes --]
On Mon, Nov 28, 2016 at 08:29:55AM -0800, Bart Van Assche wrote:
> On 11/28/2016 12:31 AM, Leon Romanovsky wrote:
> >I'm extra cautions with stable tags and prefer to finalize my stable
> >checker in my submissions scripts first, before adding it manually and
> >I have plans to use it next kernel release.
>
> Hello Leon,
>
> Thanks for explaining your workflow. However, the question remains whether
> or not stable tags should be added to the patches in this series?
It can be added but I don't see any real advantage of it. Do you know
about real users who run RDMA stack from stable trees and don't rely on
distro kernel and/or OFED?
>
> Bart.
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]
^ permalink raw reply
* Re: [PATCH rdma-next 1/4] IB/mlx4: Fix out-of-range array index in destroy qp flow
From: Bart Van Assche @ 2016-12-01 5:35 UTC (permalink / raw)
To: Leon Romanovsky
Cc: dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org,
linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
Jack Morgenstein
In-Reply-To: <20161201053105.GA4497@leon.nu>
On 11/30/16 21:31, Leon Romanovsky wrote:
> On Mon, Nov 28, 2016 at 08:29:55AM -0800, Bart Van Assche wrote:
>> On 11/28/2016 12:31 AM, Leon Romanovsky wrote:
>>> I'm extra cautions with stable tags and prefer to finalize my stable
>>> checker in my submissions scripts first, before adding it manually and
>>> I have plans to use it next kernel release.
>>
>> Thanks for explaining your workflow. However, the question remains whether
>> or not stable tags should be added to the patches in this series?
>
> It can be added but I don't see any real advantage of it. Do you know
> about real users who run RDMA stack from stable trees and don't rely on
> distro kernel and/or OFED?
Hello Leon,
Many users who build an iSER or SRP storage target themselves use a
recent stable kernel without OFED.
Bart.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox