From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43)
	id 1LChMl-0002zf-Ez
	for qemu-devel@nongnu.org; Tue, 16 Dec 2008 16:22:07 -0500
Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43)
	id 1LChMj-0002z7-JC
	for qemu-devel@nongnu.org; Tue, 16 Dec 2008 16:22:05 -0500
Received: from [199.232.76.173] (port=56712 helo=monty-python.gnu.org)
	by lists.gnu.org with esmtp (Exim 4.43) id 1LChMj-0002z4-CS
	for qemu-devel@nongnu.org; Tue, 16 Dec 2008 16:22:05 -0500
Received: from yx-out-1718.google.com ([74.125.44.153]:23823)
	by monty-python.gnu.org with esmtp (Exim 4.60)
	(envelope-from <anthony@codemonkey.ws>) id 1LChMj-0005Nd-2V
	for qemu-devel@nongnu.org; Tue, 16 Dec 2008 16:22:05 -0500
Received: by yx-out-1718.google.com with SMTP id 3so1551476yxi.82
	for <qemu-devel@nongnu.org>; Tue, 16 Dec 2008 13:22:04 -0800 (PST)
Message-ID: <49481BE2.8080205@codemonkey.ws>
Date: Tue, 16 Dec 2008 15:21:38 -0600
From: Anthony Liguori <anthony@codemonkey.ws>
MIME-Version: 1.0
References: <4947B91C.4050403@amd.com>
In-Reply-To: <4947B91C.4050403@amd.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [Qemu-devel] Re: [PATCH 7/8] v2: add numa monitor command
Reply-To: qemu-devel@nongnu.org
List-Id: qemu-devel.nongnu.org
List-Unsubscribe: <http://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.gnu.org/pipermail/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <http://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Andre Przywara <andre.przywara@amd.com>
Cc: qemu-devel@nongnu.org, Avi Kivity <avi@redhat.com>

Andre Przywara wrote:
> Signed-off-by: Andre Przywara <andre.przywara@amd.com>
>
> # HG changeset patch
> # User Andre Przywara <andre.przywara@amd.com>
> # Date 1229435568 -3600
> # Node ID 4f0a8ac2d88ffffc1dcd82785c1620553baa86da
> # Parent  5a74bd76931b79713803795b3943aa8383946521
> add -numa monitor command to repin guest nodes
>
> diff -r 5a74bd76931b -r 4f0a8ac2d88f monitor.c
> --- a/monitor.c	Tue Dec 16 14:51:24 2008 +0100
> +++ b/monitor.c	Tue Dec 16 14:52:48 2008 +0100
> @@ -39,6 +39,9 @@
>  #include "qemu-timer.h"
>  #include "migration.h"
>  #include "kvm.h"
> +#ifdef CONFIG_NUMA
> +#include <numa.h>
> +#endif
>  
>  //#define DEBUG
>  //#define DEBUG_COMPLETION
> @@ -1285,6 +1288,9 @@ static void do_info_numa(void)
>  {
>      int i, j;
>  
> +#ifndef CONFIG_NUMA
> +    term_printf("NUMA placement support: not compiled\n");
> +#endif
>      term_printf("%d nodes\n", numnumanodes);
>      for (i = 0; i < numnumanodes; i++) {
>          term_printf("node %d cpus:", i);
> @@ -1292,7 +1298,40 @@ static void do_info_numa(void)
>              if (node_to_cpus[i] & (1ULL << j)) term_printf(" %d", j);
>          term_printf("\n");
>          term_printf("node %d size: %" PRId64 " MB\n", i, node_mem[i] >> 20);
> -    }
> +        term_printf("node %d host: ", i);
> +        if (hostnodes[i] == (uint64_t)-1) term_printf("*\n"); else
> +            term_printf("%" PRId64 "\n", hostnodes[i]);
> +    }
> +}
> +
> +static void do_numa(const char *affinity)
> +{
> +#ifdef CONFIG_NUMA
> +    uint64_t newaff[MAX_NODES];
> +    int i;
> +    unsigned long offset = 0;
> +
> +    if (numnumanodes <= 0) term_printf("No NUMA nodes defined.\n"); else
> +    if (numa_available() == -1) term_printf("Not a NUMA host.\n"); else
> +    {
>   

This should be reformatted.

> +        for (i = 0; i < numnumanodes; i++) newaff[i] = hostnodes[i];
>   

This should be a separate line.

> +        parse_numa_args(affinity, newaff, NULL, NULL, MAX_NODES, 0);
> +        for (i = 0; i < numnumanodes; i++) {
> +            if (newaff[i] != hostnodes[i]) {
> +                 hostnodes[i] = newaff[i];
> +                 if (hostnodes[i] == (uint64_t)-1)
>   

I don't think using u64's for cpu=>node mappings is a reasonable thing 
to do at this stage.  64 processors systems are relatively common these 
days.

> +                     numa_tonodemask_memory (phys_ram_base + offset,
> +                         node_mem[i], &numa_all_nodes);
> +                 else
> +                     numa_tonode_memory(phys_ram_base + offset,
> +                         node_mem[i], hostnodes[i] % (numa_max_node() + 1));
> +            }
> +            offset += node_mem[i];
> +        }
> +    }
> +#else
> +    term_printf ("NUMA host affinity support not compiled in\n");
> +#endif
>  }

Regards,

Anthony Liguori