[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [XEN RFC PATCH 11/40] xen/x86: Move NUMA nodes and memory block ranges to common


  • To: Julien Grall <julien@xxxxxxx>, "xen-devel@xxxxxxxxxxxxxxxxxxxx" <xen-devel@xxxxxxxxxxxxxxxxxxxx>, "sstabellini@xxxxxxxxxx" <sstabellini@xxxxxxxxxx>
  • From: Wei Chen <Wei.Chen@xxxxxxx>
  • Date: Wed, 25 Aug 2021 00:57:48 +0000
  • Accept-language: en-US
  • Arc-authentication-results: i=1; mx.microsoft.com 1; spf=pass smtp.mailfrom=arm.com; dmarc=pass action=none header.from=arm.com; dkim=pass header.d=arm.com; arc=none
  • Arc-message-signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=microsoft.com; s=arcselector9901; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-SenderADCheck; bh=lIqq4yEPMX26m7Mci8vNfighAE3Jc/IuTMLJDUSRzxE=; b=KVfXFSyMdsXRSsXtLfwAyYsa9G+wsY/72TvEr7Y46UQW3/Je4vPh6mRe8DdjllH0p7OzV74yEYi4519F87H9UR7ooBbO2YJCmIHg68bNYQjvrORps20hdu5bpPwiixFcKRaRiHbkfSUfmO+hWtgVyCWEItBAbEVcZJp8EACT0KX7eyh+UFXE+IhTvbmV2QLy+yYDD7T7UDg/HYof57ZykORY+Yy2xNsXnz18RXgwx0OdKMnlozAYXtOSTiAuln78JjXfhYMh5ZHx4hN/SQCMs1TnyBhYQtWPsGuWuuDOf6nPPj/gtGEo+FPs+P3Q2XvH2CK+IcE2YgYRvHS3tYDFJw==
  • Arc-seal: i=1; a=rsa-sha256; s=arcselector9901; d=microsoft.com; cv=none; b=OU2SRDcR5vRE+yJ7fnNk5x4UMarGzZuN+Hy29/+EcfdLOHwnMAjEb2/RLcrvn8aX1FwZZRAhcYnykY7TZZigCeDylaHJcMOp8KU/zjL72vsUq/tnWmlP2TCt1vSrQmSLyG5ulrGC7Ww7shW3IMLv5Fquh6aDODRfykB9FFaQQkY6y8NCXpnWFefhJSHwnLIhFrpk43QFgKEmMqkJuN8fDfHM2/Pa6JJno6MDqN/utCXoPzwnsoQMwUG0d009YXStwNKZzlCZOe1JHuUyaxbqrd9r4Yaq5ECOlBjqFdR9w2kFMHQuXmEnyLbLhYjdEuQP9JxTRGhYoQ29fFGBwYDckQ==
  • Authentication-results-original: xen.org; dkim=none (message not signed) header.d=none;xen.org; dmarc=none action=none header.from=arm.com;
  • Cc: Bertrand Marquis <Bertrand.Marquis@xxxxxxx>
  • Delivery-date: Wed, 25 Aug 2021 00:58:25 +0000
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>
  • Nodisclaimer: true
  • Original-authentication-results: xen.org; dkim=none (message not signed) header.d=none;xen.org; dmarc=none action=none header.from=arm.com;
  • Thread-index: AQHXjps8s5lstNyuz0u+0iYcW58qA6uDARgAgABxTaA=
  • Thread-topic: [XEN RFC PATCH 11/40] xen/x86: Move NUMA nodes and memory block ranges to common

Hi Julien,

> -----Original Message-----
> From: Julien Grall <julien@xxxxxxx>
> Sent: 2021年8月25日 1:41
> To: Wei Chen <Wei.Chen@xxxxxxx>; xen-devel@xxxxxxxxxxxxxxxxxxxx;
> sstabellini@xxxxxxxxxx; jbeulich@xxxxxxxx
> Cc: Bertrand Marquis <Bertrand.Marquis@xxxxxxx>
> Subject: Re: [XEN RFC PATCH 11/40] xen/x86: Move NUMA nodes and memory
> block ranges to common
> 
> Hi Wei,
> 
> On 11/08/2021 11:23, Wei Chen wrote:
> > These data structures and functions are used to create the
> > mapping between node and memory blocks. In device tree based
> > NUMA, we will reuse these data structures and functions, so
> > we move this part of code from x86 to common.
> >
> > Signed-off-by: Wei Chen <wei.chen@xxxxxxx>
> > ---
> >   xen/arch/x86/srat.c        | 50 -------------------------------------
> >   xen/common/numa.c          | 51 ++++++++++++++++++++++++++++++++++++++
> >   xen/include/asm-x86/numa.h |  8 ------
> >   xen/include/xen/numa.h     | 15 +++++++++++
> >   4 files changed, 66 insertions(+), 58 deletions(-)
> >
> > diff --git a/xen/arch/x86/srat.c b/xen/arch/x86/srat.c
> > index 6b77b98201..6d68b8a614 100644
> > --- a/xen/arch/x86/srat.c
> > +++ b/xen/arch/x86/srat.c
> > @@ -26,7 +26,6 @@ static struct acpi_table_slit *__read_mostly acpi_slit;
> >
> >   static nodemask_t memory_nodes_parsed __initdata;
> >   static nodemask_t processor_nodes_parsed __initdata;
> > -static struct node nodes[MAX_NUMNODES] __initdata;
> >
> >   struct pxm2node {
> >     unsigned pxm;
> > @@ -37,9 +36,6 @@ static struct pxm2node __read_mostly
> pxm2node[MAX_NUMNODES] =
> >
> >   static unsigned node_to_pxm(nodeid_t n);
> >
> > -static int num_node_memblks;
> > -static struct node node_memblk_range[NR_NODE_MEMBLKS];
> > -static nodeid_t memblk_nodeid[NR_NODE_MEMBLKS];
> >   static __initdata DECLARE_BITMAP(memblk_hotplug, NR_NODE_MEMBLKS);
> >
> >   static inline bool node_found(unsigned idx, unsigned pxm)
> > @@ -104,52 +100,6 @@ nodeid_t setup_node(unsigned pxm)
> >     return node;
> >   }
> >
> > -int valid_numa_range(u64 start, u64 end, nodeid_t node)
> > -{
> > -   int i;
> > -
> > -   for (i = 0; i < num_node_memblks; i++) {
> > -           struct node *nd = &node_memblk_range[i];
> > -
> > -           if (nd->start <= start && nd->end >= end &&
> > -                   memblk_nodeid[i] == node)
> > -                   return 1;
> > -   }
> > -
> > -   return 0;
> > -}
> > -
> > -static __init int conflicting_memblks(u64 start, u64 end)
> > -{
> > -   int i;
> > -
> > -   for (i = 0; i < num_node_memblks; i++) {
> > -           struct node *nd = &node_memblk_range[i];
> > -           if (nd->start == nd->end)
> > -                   continue;
> > -           if (nd->end > start && nd->start < end)
> > -                   return i;
> > -           if (nd->end == end && nd->start == start)
> > -                   return i;
> > -   }
> > -   return -1;
> > -}
> > -
> > -static __init void cutoff_node(int i, u64 start, u64 end)
> > -{
> > -   struct node *nd = &nodes[i];
> > -   if (nd->start < start) {
> > -           nd->start = start;
> > -           if (nd->end < nd->start)
> > -                   nd->start = nd->end;
> > -   }
> > -   if (nd->end > end) {
> > -           nd->end = end;
> > -           if (nd->start > nd->end)
> > -                   nd->start = nd->end;
> > -   }
> > -}
> > -
> >   static __init void bad_srat(void)
> >   {
> >     int i;
> > diff --git a/xen/common/numa.c b/xen/common/numa.c
> > index 9b6f23dfc1..1facc8fe2b 100644
> > --- a/xen/common/numa.c
> > +++ b/xen/common/numa.c
> > @@ -29,6 +29,11 @@ nodeid_t cpu_to_node[NR_CPUS] __read_mostly = {
> >
> >   cpumask_t node_to_cpumask[MAX_NUMNODES] __read_mostly;
> >
> > +struct node nodes[MAX_NUMNODES] __initdata;
> > +int num_node_memblks;
> > +struct node node_memblk_range[NR_NODE_MEMBLKS];
> > +nodeid_t memblk_nodeid[NR_NODE_MEMBLKS];
> > +
> >   /*
> >    * Given a shift value, try to populate memnodemap[]
> >    * Returns :
> > @@ -136,6 +141,52 @@ int __init compute_hash_shift(struct node *nodes,
> int numnodes,
> >       return shift;
> >   }
> >
> > +int valid_numa_range(u64 start, u64 end, nodeid_t node)
> > +{
> > +   int i;
> > +
> > +   for (i = 0; i < num_node_memblks; i++) {
> > +           struct node *nd = &node_memblk_range[i];
> > +
> > +           if (nd->start <= start && nd->end >= end &&
> > +                   memblk_nodeid[i] == node)
> > +                   return 1;
> > +   }
> > +
> > +   return 0;
> > +}
> > +
> > +int __init conflicting_memblks(u64 start, u64 end)
> > +{
> > +   int i;
> > +
> > +   for (i = 0; i < num_node_memblks; i++) {
> > +           struct node *nd = &node_memblk_range[i];
> > +           if (nd->start == nd->end)
> > +                   continue;
> > +           if (nd->end > start && nd->start < end)
> > +                   return i;
> > +           if (nd->end == end && nd->start == start)
> > +                   return i;
> > +   }
> > +   return -1;
> > +}
> > +
> > +void __init cutoff_node(int i, u64 start, u64 end)
> > +{
> > +   struct node *nd = &nodes[i];
> > +   if (nd->start < start) {
> > +           nd->start = start;
> > +           if (nd->end < nd->start)
> > +                   nd->start = nd->end;
> > +   }
> > +   if (nd->end > end) {
> > +           nd->end = end;
> > +           if (nd->start > nd->end)
> > +                   nd->start = nd->end;
> > +   }
> > +}
> > +
> >   void numa_add_cpu(int cpu)
> >   {
> >       cpumask_set_cpu(cpu, &node_to_cpumask[cpu_to_node(cpu)]);
> > diff --git a/xen/include/asm-x86/numa.h b/xen/include/asm-x86/numa.h
> > index 07ff78ea1b..e8a92ad9df 100644
> > --- a/xen/include/asm-x86/numa.h
> > +++ b/xen/include/asm-x86/numa.h
> > @@ -17,12 +17,6 @@ extern cpumask_t     node_to_cpumask[];
> >   #define node_to_first_cpu(node)  (__ffs(node_to_cpumask[node]))
> >   #define node_to_cpumask(node)    (node_to_cpumask[node])
> >
> > -struct node {
> > -   u64 start,end;
> > -};
> > -
> > -extern int compute_hash_shift(struct node *nodes, int numnodes,
> > -                         nodeid_t *nodeids);
> >   extern nodeid_t pxm_to_node(unsigned int pxm);
> >
> >   #define ZONE_ALIGN (1UL << (MAX_ORDER+PAGE_SHIFT))
> > @@ -45,8 +39,6 @@ static inline void clear_node_cpumask(int cpu)
> >     cpumask_clear_cpu(cpu, &node_to_cpumask[cpu_to_node(cpu)]);
> >   }
> >
> > -extern int valid_numa_range(u64 start, u64 end, nodeid_t node);
> > -
> >   void srat_parse_regions(u64 addr);
> >   extern u8 __node_distance(nodeid_t a, nodeid_t b);
> >   unsigned int arch_get_dma_bitsize(void);
> > diff --git a/xen/include/xen/numa.h b/xen/include/xen/numa.h
> > index 5af74b357f..67b79a73a3 100644
> > --- a/xen/include/xen/numa.h
> > +++ b/xen/include/xen/numa.h
> > @@ -54,6 +54,21 @@ static inline __attribute__((pure)) nodeid_t
> phys_to_nid(paddr_t addr)
> >
> >   extern void numa_add_cpu(int cpu);
> >
> > +struct node {
> > +   u64 start,end;
> > +};
> > +
> > +extern struct node nodes[MAX_NUMNODES];
> > +extern int num_node_memblks;
> > +extern struct node node_memblk_range[NR_NODE_MEMBLKS];
> > +extern nodeid_t memblk_nodeid[NR_NODE_MEMBLKS];
> 
> I am not overly happy that the 4 ariables above are now exported.
> Looking at the code, they are (only?) used in arch specific code for
> acpi_numa_memory_affinity_init() and dtb_numa_memory_affinity_init().
> 
> There bits touching the variables looks quite similar between the two
> functions. The main differences seems to be the messages in printk() and
> the hotplug bits.
> 
> So I think we should attempt to abstract the code. IIRC, we discussed
> some of the way to abstract when Vijay Kilari attempted to add support
> for NUMA (see [1]). It might be worth to have a look and see if you can
> re-use some of the ideas.

Ok, I will look at that thread. If it's useful, I will do it in next
version.

> 
> > +
> > +extern int compute_hash_shift(struct node *nodes, int numnodes,
> > +                         nodeid_t *nodeids);
> > +extern int conflicting_memblks(u64 start, u64 end);
> > +extern void cutoff_node(int i, u64 start, u64 end);
> > +extern int valid_numa_range(u64 start, u64 end, nodeid_t node);
> > +
> >   #endif /* CONFIG_NUMA */
> >
> >   #endif /* _XEN_NUMA_H */
> >
> 
> [1]
> https://lore.kernel.org/xen-devel/1500378106-2620-1-git-send-email-
> vijay.kilari@xxxxxxxxx/
> 
> --
> Julien Grall

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.