Re: [PATCH for-next v2 1/2] RDMA/efa: Add initialization of AH cache rhashtable

[Zhu Yanjun <yanjun.zhu@linux.dev>]

在 2026/5/14 3:02, Yonatan Nachum 写道:
> On Wed, May 13, 2026 at 10:12:01PM -0700, Zhu Yanjun wrote:
>> 在 2026/5/11 23:11, Yonatan Nachum 写道:
>>> New EFA devices don't support the creation of multiple address handles
>>> to the same remote on the same PD.
>>> To overcome this limitation, introduce an AH cache rhashtable which will
>>> store the refcounts of the same AH creation on the same PD and will
>>> allow the driver to manage AH reuse. The hashtable key is the
>>> combination of PD and GID. Add initialization and teardown logic for the
>>> rhashtable.
>>>
>>> Reviewed-by: Firas Jahjah <firasj@amazon.com>
>>> Reviewed-by: Michael Margolin <mrgolin@amazon.com>
>>> Signed-off-by: Yonatan Nachum <ynachum@amazon.com>
>>> ---
>>>   drivers/infiniband/hw/efa/Makefile       |  4 +--
>>>   drivers/infiniband/hw/efa/efa_ah_cache.c | 30 ++++++++++++++++++++
>>>   drivers/infiniband/hw/efa/efa_ah_cache.h | 36 ++++++++++++++++++++++++
>>>   drivers/infiniband/hw/efa/efa_com.c      | 12 +++++++-
>>>   drivers/infiniband/hw/efa/efa_com.h      |  5 +++-
>>>   5 files changed, 83 insertions(+), 4 deletions(-)
>>>   create mode 100644 drivers/infiniband/hw/efa/efa_ah_cache.c
>>>   create mode 100644 drivers/infiniband/hw/efa/efa_ah_cache.h
>>>
>>> diff --git a/drivers/infiniband/hw/efa/efa_ah_cache.h b/drivers/infiniband/hw/efa/efa_ah_cache.h
>>> new file mode 100644
>>> index 000000000000..25288fdf778a
>>> --- /dev/null
>>> +++ b/drivers/infiniband/hw/efa/efa_ah_cache.h
>>> @@ -0,0 +1,36 @@
>>> +/* SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause */
>>> +/*
>>> + * Copyright 2026 Amazon.com, Inc. or its affiliates. All rights reserved.
>>> + */
>>> +
>>> +#ifndef _EFA_AH_CACHE_H_
>>> +#define _EFA_AH_CACHE_H_
>>> +
>>> +#include <linux/refcount.h>
>>> +#include <linux/rhashtable.h>
>>> +
>>> +#define EFA_AH_GID_SIZE 16
>>> +
>>> +struct efa_ah_cache_key {
>>> +	u8 gid[EFA_AH_GID_SIZE];
>>> +	u16 pd;
>>> +};
>> I am not sure if we add __packed to avoid memory hole.
>>
>> Zhu Yanjun
> Currently there is no holes in the struct (verified using pahole) and we
> Zero-initialize AH cache key so I don't think packed is really needed.

I’m wondering whether you’ve tested this on all architectures.

Explicitly using `__packed` or ensuring proper memory alignment

would likely be a better solution to avoid memory holes in the key 
structure.

Zhu Yanjun


>
> Thanks

-- 
Best Regards,
Yanjun.Zhu