[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <b67798cf-17e0-444d-be9f-00071c3e32b4@huawei.com>
Date: Tue, 4 Mar 2025 20:25:28 +0800
From: Yunsheng Lin <linyunsheng@...wei.com>
To: Simon Horman <horms@...nel.org>
CC: <davem@...emloft.net>, <kuba@...nel.org>, <pabeni@...hat.com>,
<zhangkun09@...wei.com>, <liuyonglong@...wei.com>, <fanghaiqing@...wei.com>,
Robin Murphy <robin.murphy@....com>, Alexander Duyck
<alexander.duyck@...il.com>, IOMMU <iommu@...ts.linux.dev>, Eric Dumazet
<edumazet@...gle.com>, Donald Hunter <donald.hunter@...il.com>, Jesper
Dangaard Brouer <hawk@...nel.org>, Ilias Apalodimas
<ilias.apalodimas@...aro.org>, Andrew Lunn <andrew+netdev@...n.ch>,
<netdev@...r.kernel.org>, <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH net-next v10 3/4] page_pool: support unlimited number of
inflight pages
On 2025/3/4 1:59, Simon Horman wrote:
> On Wed, Feb 26, 2025 at 07:03:38PM +0800, Yunsheng Lin wrote:
>> Currently a fixed size of pre-allocated memory is used to
>> keep track of the inflight pages, in order to use the DMA
>> API correctly.
>>
>> As mentioned [1], the number of inflight pages can be up to
>> 73203 depending on the use cases. Allocate memory dynamically
>> to keep track of the inflight pages when pre-allocated memory
>> runs out.
>>
>> The overhead of using dynamic memory allocation is about 10ns~
>> 20ns, which causes 5%~10% performance degradation for the test
>> case of time_bench_page_pool03_slow() in [2].
>>
>> 1. https://lore.kernel.org/all/b8b7818a-e44b-45f5-91c2-d5eceaa5dd5b@kernel.org/
>> 2. https://github.com/netoptimizer/prototype-kernel
>> CC: Robin Murphy <robin.murphy@....com>
>> CC: Alexander Duyck <alexander.duyck@...il.com>
>> CC: IOMMU <iommu@...ts.linux.dev>
>> Fixes: f71fec47c2df ("page_pool: make sure struct device is stable")
>> Signed-off-by: Yunsheng Lin <linyunsheng@...wei.com>
>> ---
>> Documentation/netlink/specs/netdev.yaml | 16 +++++
>> include/net/page_pool/types.h | 10 ++++
>> include/uapi/linux/netdev.h | 2 +
>> net/core/page_pool.c | 79 ++++++++++++++++++++++++-
>> net/core/page_pool_priv.h | 2 +
>> net/core/page_pool_user.c | 39 ++++++++++--
>> tools/net/ynl/samples/page-pool.c | 11 ++++
>> 7 files changed, 154 insertions(+), 5 deletions(-)
>
> Hi,
>
> It looks like the header changes in this patch don't quite
> correspond to the spec changes.
>
> But if so, perhaps the spec update needs to change,
> because adding values to an enum, other than at the end,
> feels like UAPI breakage to me.
>
> I see this:
>
> $ ./tools/net/ynl/ynl-regen.sh -f
Yes, It seems I only tested the tools/net/ynl/samples/page-pool, which
doesn't seems to catch the above problem.
Will update the spec changes to the the header changes and update
tools/include/uapi/linux/netdev.h accordingly too.
Thanks for the reporting.
> $ git diff
> diff --git a/include/uapi/linux/netdev.h b/include/uapi/linux/netdev.h
> index 9309cbfeb8d2..9e02f6190b07 100644
> --- a/include/uapi/linux/netdev.h
> +++ b/include/uapi/linux/netdev.h
> @@ -100,11 +100,11 @@ enum {
> NETDEV_A_PAGE_POOL_NAPI_ID,
> NETDEV_A_PAGE_POOL_INFLIGHT,
> NETDEV_A_PAGE_POOL_INFLIGHT_MEM,
> + NETDEV_A_PAGE_POOL_ITEM_MEM_RESIDENT,
> + NETDEV_A_PAGE_POOL_ITEM_MEM_USED,
> NETDEV_A_PAGE_POOL_DETACH_TIME,
> NETDEV_A_PAGE_POOL_DMABUF,
> NETDEV_A_PAGE_POOL_IO_URING,
> - NETDEV_A_PAGE_POOL_ITEM_MEM_RESIDENT,
> - NETDEV_A_PAGE_POOL_ITEM_MEM_USED,
>
> __NETDEV_A_PAGE_POOL_MAX,
> NETDEV_A_PAGE_POOL_MAX = (__NETDEV_A_PAGE_POOL_MAX - 1)
> diff --git a/tools/include/uapi/linux/netdev.h b/tools/include/uapi/linux/netdev.h
> index 7600bf62dbdf..9e02f6190b07 100644
> --- a/tools/include/uapi/linux/netdev.h
> +++ b/tools/include/uapi/linux/netdev.h
> @@ -100,6 +100,8 @@ enum {
> NETDEV_A_PAGE_POOL_NAPI_ID,
> NETDEV_A_PAGE_POOL_INFLIGHT,
> NETDEV_A_PAGE_POOL_INFLIGHT_MEM,
> + NETDEV_A_PAGE_POOL_ITEM_MEM_RESIDENT,
> + NETDEV_A_PAGE_POOL_ITEM_MEM_USED,
> NETDEV_A_PAGE_POOL_DETACH_TIME,
> NETDEV_A_PAGE_POOL_DMABUF,
> NETDEV_A_PAGE_POOL_IO_URING,
>
Powered by blists - more mailing lists