rface to shave off another ib_modify_qp().
Fixes: ec16227e1414 ("RDS/IB: Infiniband transport")
Signed-off-by: Håkon Bugge
---
net/rds/ib_cm.c | 35 +--
net/rds/rdma_transport.c | 1 +
2 files changed, 2 insertions(+), 34 deletions(-)
diff --git a
ib_modify_qp() is an expensive operation on some HCAs running
virtualized. This series removes two ib_modify_qp() calls from RDS.
I am sending this as a v3, even though it is the first sent to
net. This because the IB Core commit has reach v3.
Håkon Bugge (2):
IB/cma: Introduce
ime value before retrying the send.
The 5-bit value to be supplied to the rdma_set_min_rnr_timer() is
documented in IBTA Table 45: "Encoding for RNR NAK Timer Field".
Signed-off-by: Håkon Bugge
Acked-by: Jason Gunthorpe
---
drivers/infiniband/core/cm
> On 31 Jul 2020, at 13:59, Greg Kroah-Hartman
> wrote:
>
> On Fri, Jul 31, 2020 at 01:14:09PM +0200, Håkon Bugge wrote:
>>
>>
>>> On 31 Jul 2020, at 11:59, Dan Carpenter wrote:
>>>
>>> On Fri, Jul 31, 2020 at 07:53:01AM +0300, Leon R
> On 31 Jul 2020, at 11:59, Dan Carpenter wrote:
>
> On Fri, Jul 31, 2020 at 07:53:01AM +0300, Leon Romanovsky wrote:
>> On Thu, Jul 30, 2020 at 03:20:26PM -0400, Peilin Ye wrote:
>>> rds_notify_queue_get() is potentially copying uninitialized kernel stack
>>> memory to userspace since the com
ous DMA coherent memory to store struct rds_header for
>>> sending/receiving packets. The memory allocated is then partitioned
>>> into struct rds_header. This is not necessary and can be costly at
>>> times when memory is fragmented. Instead, RDS should use the DMA
>
> On 6 Feb 2019, at 19:02, jackm wrote:
>
> On Wed, 6 Feb 2019 16:40:14 +0100
> Håkon Bugge wrote:
>
>> Jack,
>>
>> A major contributor to the long processing time in the PF driver
>> proxying QP1 packets is:
>>
>> c
> On 6 Feb 2019, at 09:50, Håkon Bugge wrote:
>
>
>
>> On 5 Feb 2019, at 23:36, Jason Gunthorpe wrote:
>>
>> On Thu, Jan 31, 2019 at 06:09:51PM +0100, Håkon Bugge wrote:
>>> Using CX-3 virtual functions, either from a bare-metal machine or
>
> On 5 Feb 2019, at 23:36, Jason Gunthorpe wrote:
>
> On Thu, Jan 31, 2019 at 06:09:51PM +0100, Håkon Bugge wrote:
>> Using CX-3 virtual functions, either from a bare-metal machine or
>> pass-through from a VM, MAD packets are proxied through the PF driver.
>>
>
he
numbers fell down to about one hundred for both of them.
Adjustment of the CMA timeout is _not_ part of this commit.
Signed-off-by: Håkon Bugge
---
drivers/infiniband/hw/mlx4/cm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/infiniband/hw/mlx4/cm.c b/drivers/inf
> On 14 Aug 2018, at 11:01, Yuval Shaia wrote:
>
> On Wed, Aug 08, 2018 at 03:48:08PM -0700, Saeed Mahameed wrote:
>> From: Huy Nguyen
>>
>> Remove unused netdev_registered_init/remove in en.h
>> Return ENOSUPPORT if the check MLX5_DSCP_SUPPORTED fails.
>
> s/ENOSUPPORT/EOPNOTSUPP
> (noted
> On 11 May 2018, at 01:31, Qing Huang wrote:
>
> When a system is under memory presure (high usage with fragments),
> the original 256KB ICM chunk allocations will likely trigger kernel
> memory management to enter slow path doing memory compact/migration
> ops in order to complete high order
c29 ("rds: ib: add error handle")
>
> Signed-off-by: Dag Moxnes
Reviewed-by: Håkon Bugge
Thxs, Håkon
> ---
> net/rds/ib_cm.c |3 ++-
> 1 files changed, 2 insertions(+), 1 deletions(-)
>
> diff --git a/net/rds/ib_cm.c b/net/rds/ib_cm.c
> index eea1d86..13
> On 24 Apr 2018, at 05:27, santosh.shilim...@oracle.com wrote:
>
> On 4/23/18 6:39 PM, Zhu Yanjun wrote:
>> This reverts commit 3b12f73a5c2977153f28a224392fd4729b50d1dc.
>> After long time discussion and investigations, it seems that there
>> is no mem leak. So this patch is reverted.
>> Signed
Hi Sowmini,
A little nit below. And some spelling issues in existing commentary you can
consider fixing, since you reshuffle this file considerable.
Thxs, Håkon
> On 18 Mar 2018, at 21:45, Sowmini Varadhan
> wrote:
>
> On (03/18/18 00:55), Kirill Tkhai wrote:
>>
>> I just want to make rds
02 00 00 48
89 55 d0 48 89 4d c8 85 c0 0f 84 2d 03 00 00 48 8b 87 00 03 00 00 <48>
83 b8 c0 00 00 00 00 0f 84 25 03 00 0 0 48 8b 06 48 8b 56 08
The fix is to check the existence of an underlying transport in
__rds_rdma_map().
Signed-off-by: Håkon Bugge
Reported-by: syzbot
---
net/rds/rdma
ag->f_sg),
(long) ib_sg_dma_address(
The fix is simply to move the rdsdebug() statement up before the
ib_post_recv() and remove the printing of ret, which is taken care of
anyway by the non-debug code.
Signed-off-by: Håkon Bugge
Reviewed-by: Knut Omang
Reviewed-by: Wei L
send_flags needs to be initialized before calling
rds_ib_set_wr_signal_state().
Signed-off-by: Håkon Bugge
Acked-by: Santosh Shilimkar
---
net/rds/ib_send.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/net/rds/ib_send.c b/net/rds/ib_send.c
index 6ab39db..8f46755 100644
--- a/net/rds
Ignore this, sent the old one.
Håkon
> On 24 Oct 2017, at 18:07, Håkon Bugge wrote:
>
> send_flags needs to be initialized before calling
> rds_ib_set_wr_signal_state().
>
> Signed-off-by: Håkon Bugge
> ---
> net/rds/ib_send.c | 1 +
> 1 file changed, 1 insertion(+
send_flags needs to be initialized before calling
rds_ib_set_wr_signal_state().
Signed-off-by: Håkon Bugge
---
net/rds/ib_send.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/net/rds/ib_send.c b/net/rds/ib_send.c
index 6ab39db..8f46755 100644
--- a/net/rds/ib_send.c
+++ b/net/rds
send_flags needs to be initialized before calling
rds_ib_set_wr_signal_state().
Signed-off-by: Håkon Bugge
---
net/rds/ib_send.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/net/rds/ib_send.c b/net/rds/ib_send.c
index 6ab39db..8f46755 100644
--- a/net/rds/ib_send.c
+++ b/net/rds
rds_ib_set_wr_signal_state() function.
However, this function is not always used which yields inaccurate
accounting. This commit fixes this, re-factors a code bloat related to
the matter, and makes the actual parameter type to the function
consistent.
Signed-off-by: Håkon Bugge
---
net/rds/ib_send.c | 15
counts the race condition correctly.
Changes from v1:
- Removed check for *someone_on_xmit()*
- Fixed incorrect indentation
Signed-off-by: Håkon Bugge
Reviewed-by: Knut Omang
---
net/rds/send.c | 10 +++---
1 file changed, 7 insertions(+), 3 deletions(-)
diff --git a/net/rds/send.c b/net/rds
> On 6 Sep 2017, at 17:58, Santosh Shilimkar
> wrote:
>
> On 9/6/2017 8:29 AM, Håkon Bugge wrote:
>> In rds_send_xmit() there is logic to batch the sends. However, if
>> another thread has acquired the lock, it is considered a race and we
>> yi
the race and
increments the statistics counter correctly.
Signed-off-by: Håkon Bugge
Reviewed-by: Knut Omang
---
net/rds/send.c | 16 +---
1 file changed, 13 insertions(+), 3 deletions(-)
diff --git a/net/rds/send.c b/net/rds/send.c
index 058a407..ecfe0b5 100644
--- a/net/rds/send.c
The bits in m_flags in struct rds_message are used for a plurality of
reasons, and from different contexts. To avoid any missing updates to
m_flags, use the atomic set_bit() instead of the non-atomic equivalent.
Signed-off-by: Håkon Bugge
Reviewed-by: Knut Omang
Reviewed-by: Wei Lin Guay
re-introduces filling the receive queue from rds_ib_recv(), but does
not add the statistics counter. rds_ib_recv() was later renamed to
rds_ib_recv_path().
This commit reintroduces the statistics counting of
s_ib_rx_refill_from_thread and s_ib_rx_refill_from_cq.
Signed-off-by: Håkon Bugge
Reviewed
> On 20 Jul 2017, at 13:02, Sowmini Varadhan
> wrote:
>
> On (07/20/17 12:28), H??kon Bugge wrote:
>> cp->cp_send_gen is treated as a normal variable, although it may be
>> used by different threads.
>
> I'm confused by that assertion. If you look at the comments right
> above the change in yo
del:
Loads from and stores to shared (but non-atomic) variables should
be protected with the READ_ONCE(), WRITE_ONCE(), and
ACCESS_ONCE().
Clause 5.1.2.4/25 in the C standard is also relevant.
Signed-off-by: Håkon Bugge
Reviewed-by: Knut Omang
---
net/rds/send.c | 6 +++---
1 f
29 matches
Mail list logo
