Date:   Thu, 9 Sep 2021 18:35:09 +0800
From:   zhenguo yao <yaozhenguo1@...il.com>
To:     Mike Kravetz <mike.kravetz@...cle.com>
Cc:     corbet@....net, Andrew Morton <akpm@...ux-foundation.org>,
        yaozhenguo@...com, Matthew Wilcox <willy@...radead.org>,
        linux-kernel@...r.kernel.org, linux-doc@...r.kernel.org,
        Linux Memory Management List <linux-mm@...ck.org>
Subject: Re: [v3] hugetlbfs: Extend the definition of hugepages parameter to
 support node allocation

Thanks for your careful review. The processing of those parameters are
really complicated. I will fix all you mentioned soon

Mike Kravetz <mike.kravetz@...cle.com> 于2021年9月9日周四 上午7:13写道：
>
> On 9/1/21 11:59 PM, yaozhenguo wrote:
> > We can specify the number of hugepages to allocate at boot. But the
> > hugepages is balanced in all nodes at present. In some scenarios,
> > we only need hugepages in one node. For example: DPDK needs hugepages
> > which is in the same node as NIC. if DPDK needs four hugepages of 1G
> > size in node1 and system has 16 numa nodes. We must reserve 64 hugepages
> > in kernel cmdline. But, only four hugepages are used. The others should
> > be free after boot.If the system memory is low(for example: 64G), it will
>                     ^ add space after . please
> > be an impossible task. So, extend hugepages kernel parameter to specify
> > node number of hugepages to allocate at boot.
> > For example add following parameter:
> >
> > hugepagesz=1G hugepages=0:1,1:3
> >
> > It will allocate 1 hugepages in node0 and 3 hugepages in node1.
> >
> > Signed-off-by: yaozhenguo <yaozhenguo1@...il.com>
>
> Thank you for continuing to work this!  This kernel command line parsing
> code is tricky, and there is still something missing in your changes.
> I also added some suggestions to change working of documentation.
>
> > ---
> > v3:   1. Skip gigantic hugepages allocation if hugetlb_cma is enabled.
> >       2. Fix wrong behavior for parameter: hugepagesz=2M hugepages=2 hugepages=5.
> >       3. Update hugetlbpage.rst.
> >       4. Fix side effects which v2 brings in.
> >       5. add cond_resched in hugetlb_hstate_alloc_pages_onenode.
> > ---
> >  .../admin-guide/kernel-parameters.txt         |   8 +-
> >  Documentation/admin-guide/mm/hugetlbpage.rst  |  12 +-
> >  include/linux/hugetlb.h                       |   1 +
> >  mm/hugetlb.c                                  | 116 ++++++++++++++++--
> >  4 files changed, 126 insertions(+), 11 deletions(-)
> >
> > diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> > index bdb22006f..64a128924 100644
> > --- a/Documentation/admin-guide/kernel-parameters.txt
> > +++ b/Documentation/admin-guide/kernel-parameters.txt
> > @@ -1588,9 +1588,11 @@
> >                       the number of pages of hugepagesz to be allocated.
> >                       If this is the first HugeTLB parameter on the command
> >                       line, it specifies the number of pages to allocate for
> > -                     the default huge page size.  See also
> > -                     Documentation/admin-guide/mm/hugetlbpage.rst.
> > -                     Format: <integer>
> > +                     the default huge page size. If using node format, It
> > +                     specifies numbers of hugepage in a specific node.
>
> Perhaps rewrite as?
>                         the default huge page size. If using node format, the
>                         number of pages to allocate per-node can be specified.
>
> > +                     See also Documentation/admin-guide/mm/hugetlbpage.rst.
> > +                     Format: <integer> or (node format)
> > +                             <node>:<numbers>[,<node>:<numbers>]
>
> Perhaps node format should be written as?
>                                 <node>:<integer>[,<node>:<integer>]
>
> >
> >       hugepagesz=
> >                       [HW] The size of the HugeTLB pages.  This is used in
> > diff --git a/Documentation/admin-guide/mm/hugetlbpage.rst b/Documentation/admin-guide/mm/hugetlbpage.rst
> > index 8abaeb144..bc5f674ff 100644
> > --- a/Documentation/admin-guide/mm/hugetlbpage.rst
> > +++ b/Documentation/admin-guide/mm/hugetlbpage.rst
> > @@ -128,7 +128,9 @@ hugepages
> >       implicitly specifies the number of huge pages of default size to
> >       allocate.  If the number of huge pages of default size is implicitly
> >       specified, it can not be overwritten by a hugepagesz,hugepages
> > -     parameter pair for the default size.
> > +     parameter pair for the default size. This parameter also has node
> > +     format. It specifies numbers of hugepage in a specific node when
> > +     using node format.
>
> Perhaps rewrite as?
>
>         parameter pair for the default size. This parameter also has a
>         node format.  The node format specifies the number of huge pages
>         to allocate on specific nodes.
> >
> >       For example, on an architecture with 2M default huge page size::
> >
> > @@ -138,6 +140,14 @@ hugepages
> >       indicating that the hugepages=512 parameter is ignored.  If a hugepages
> >       parameter is preceded by an invalid hugepagesz parameter, it will
> >       be ignored.
> > +
> > +     Node format example::
> > +
> > +             hugepagesz=2M hugepages=0:1,1:2
> > +
> > +     It will allocate 1 2M hugepages in node0 and 2 2M hugepages in node1.
> > +     If the node number exceeds the maximum node, the parameter will be
> > +     ignored.
>
> Perhaps rewrite as?
>
>         It will allocate 1 2M hugepage on node0 and 2 2M hugepages on node1.
>         If the node number is invalid,  the parameter will be ignored.
>
> >  default_hugepagesz
> >       Specify the default huge page size.  This parameter can
> >       only be specified once on the command line.  default_hugepagesz can
> <snip>
> > diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> <snip>
> > @@ -3580,6 +3641,10 @@ static int __init hugetlb_init(void)
> >                               default_hstate_max_huge_pages;
> >               }
> >       }
> > +     for (i = 0; i < nodes_weight(node_states[N_MEMORY]); i++)
> > +             if (default_hugepages_in_node[i] > 0)
> > +                     default_hstate.max_huge_pages_node[i] =
> > +                                     default_hugepages_in_node[i];
>
> Logic like the above for loop also needs to be added to the routine
> default_hugepagesz_setup.  See the code block starting with:
>
>         if (default_hstate_max_huge_pages) {
>
> The kernel command line parameters,
>         hugepages=0:1,1:3 default_hugepagesz=1G
>
> should allocate 1 1G page on node 0 and 3 1G pages on node 1.  With this
> patch it will allocate 2 1G pages on node 0 and 2 1G pages on node 1.
> --
> Mike Kravetz

Powered by blists - more mailing lists