public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Tejun Heo <tj@kernel.org>
To: mingo@redhat.com, yinghai@kernel.org, rientjes@google.com,
	tglx@linutronix.de, hpa@zytor.com, x86@kernel.org,
	linux-kernel@vger.kernel.org
Cc: Tejun Heo <tj@kernel.org>
Subject: [PATCH 25/25] x86, NUMA: Enable emulation on 32bit too
Date: Fri, 29 Apr 2011 17:28:44 +0200	[thread overview]
Message-ID: <1304090924-8197-26-git-send-email-tj@kernel.org> (raw)
In-Reply-To: <1304090924-8197-1-git-send-email-tj@kernel.org>

Now that NUMA init path is unified, NUMA emulation can be enabled on
32bit.  Make numa_emluation.c safe on 32bit by doing the followings.

* Define MAX_DMA32_PFN on 32bit too.

* Include bootmem.h for max_pfn declaration.

* Use u64 explicitly and always use PFN_PHYS() when converting page
  number to address.

* Avoid __udivdi3() generation on 32bit by doing number of pages
  calculation instead in split_nodes_interleave().

And drop X86_64 dependency from Kconfig.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
---
 arch/x86/Kconfig             |    2 +-
 arch/x86/include/asm/dma.h   |   10 +++-------
 arch/x86/mm/numa_emulation.c |   16 +++++++++++-----
 3 files changed, 15 insertions(+), 13 deletions(-)

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 50cb68d..648fca4 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -1201,7 +1201,7 @@ config NODES_SPAN_OTHER_NODES
 
 config NUMA_EMU
 	bool "NUMA emulation"
-	depends on X86_64 && NUMA
+	depends on NUMA
 	---help---
 	  Enable NUMA emulation. A flat machine will be split
 	  into virtual nodes when booted with "numa=fake=N", where N is the
diff --git a/arch/x86/include/asm/dma.h b/arch/x86/include/asm/dma.h
index d1a314b..0bdb0c5 100644
--- a/arch/x86/include/asm/dma.h
+++ b/arch/x86/include/asm/dma.h
@@ -72,19 +72,15 @@
 /* 16MB ISA DMA zone */
 #define MAX_DMA_PFN   ((16 * 1024 * 1024) >> PAGE_SHIFT)
 
-#ifdef CONFIG_X86_32
+/* 4GB broken PCI/AGP hardware bus master zone */
+#define MAX_DMA32_PFN ((4UL * 1024 * 1024 * 1024) >> PAGE_SHIFT)
 
+#ifdef CONFIG_X86_32
 /* The maximum address that we can perform a DMA transfer to on this platform */
 #define MAX_DMA_ADDRESS      (PAGE_OFFSET + 0x1000000)
-
 #else
-
-/* 4GB broken PCI/AGP hardware bus master zone */
-#define MAX_DMA32_PFN ((4UL * 1024 * 1024 * 1024) >> PAGE_SHIFT)
-
 /* Compat define for old dma zone */
 #define MAX_DMA_ADDRESS ((unsigned long)__va(MAX_DMA_PFN << PAGE_SHIFT))
-
 #endif
 
 /* 8237 DMA controllers */
diff --git a/arch/x86/mm/numa_emulation.c b/arch/x86/mm/numa_emulation.c
index ad091e4..c0040f8 100644
--- a/arch/x86/mm/numa_emulation.c
+++ b/arch/x86/mm/numa_emulation.c
@@ -5,6 +5,7 @@
 #include <linux/errno.h>
 #include <linux/topology.h>
 #include <linux/memblock.h>
+#include <linux/bootmem.h>
 #include <asm/dma.h>
 
 #include "numa_internal.h"
@@ -84,7 +85,13 @@ static int __init split_nodes_interleave(struct numa_meminfo *ei,
 		nr_nodes = MAX_NUMNODES;
 	}
 
-	size = (max_addr - addr - memblock_x86_hole_size(addr, max_addr)) / nr_nodes;
+	/*
+	 * Calculate target node size.  x86_32 freaks on __udivdi3() so do
+	 * the division in ulong number of pages and convert back.
+	 */
+	size = max_addr - addr - memblock_x86_hole_size(addr, max_addr);
+	size = PFN_PHYS((unsigned long)(size >> PAGE_SHIFT) / nr_nodes);
+
 	/*
 	 * Calculate the number of big nodes that can be allocated as a result
 	 * of consolidating the remainder.
@@ -226,7 +233,7 @@ static int __init split_nodes_size_interleave(struct numa_meminfo *ei,
 	 */
 	while (nodes_weight(physnode_mask)) {
 		for_each_node_mask(i, physnode_mask) {
-			u64 dma32_end = MAX_DMA32_PFN << PAGE_SHIFT;
+			u64 dma32_end = PFN_PHYS(MAX_DMA32_PFN);
 			u64 start, limit, end;
 			int phys_blk;
 
@@ -298,7 +305,7 @@ void __init numa_emulation(struct numa_meminfo *numa_meminfo, int numa_dist_cnt)
 {
 	static struct numa_meminfo ei __initdata;
 	static struct numa_meminfo pi __initdata;
-	const u64 max_addr = max_pfn << PAGE_SHIFT;
+	const u64 max_addr = PFN_PHYS(max_pfn);
 	u8 *phys_dist = NULL;
 	size_t phys_size = numa_dist_cnt * numa_dist_cnt * sizeof(phys_dist[0]);
 	int max_emu_nid, dfl_phys_nid;
@@ -342,8 +349,7 @@ void __init numa_emulation(struct numa_meminfo *numa_meminfo, int numa_dist_cnt)
 	if (numa_dist_cnt) {
 		u64 phys;
 
-		phys = memblock_find_in_range(0,
-					      (u64)max_pfn_mapped << PAGE_SHIFT,
+		phys = memblock_find_in_range(0, PFN_PHYS(max_pfn_mapped),
 					      phys_size, PAGE_SIZE);
 		if (phys == MEMBLOCK_ERROR) {
 			pr_warning("NUMA: Warning: can't allocate copy of distance table, disabling emulation\n");
-- 
1.7.1


  parent reply	other threads:[~2011-04-29 15:29 UTC|newest]

Thread overview: 43+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-04-29 15:28 [PATCHSET tip] x86, NUMA: Unify 32 and 64bit NUMA initialization Tejun Heo
2011-04-29 15:28 ` [PATCH 01/25] x86-64, NUMA: Simplify hotadd memory handling Tejun Heo
2011-04-29 15:28 ` [PATCH 02/25] x86-64, NUMA: trivial cleanups for setup_node_bootmem() Tejun Heo
2011-04-29 15:28 ` [PATCH 03/25] x86-64, NUMA: simplify nodedata allocation Tejun Heo
2011-04-29 17:23   ` Yinghai Lu
2011-04-30 12:02     ` Tejun Heo
2011-04-29 15:28 ` [PATCH 04/25] x86-32, NUMA: Automatically set apicid -> node in setup_local_APIC() Tejun Heo
2011-04-29 15:28 ` [PATCH 05/25] x86, NUMA: Unify 32/64bit numa_cpu_node() implementation Tejun Heo
2011-04-29 15:28 ` [PATCH 06/25] x86-32, NUMA: Make apic->x86_32_numa_cpu_node() optional Tejun Heo
2011-04-29 15:28 ` [PATCH 07/25] x86-32, NUMA: use sparse_memory_present_with_active_regions() Tejun Heo
2011-04-29 15:28 ` [PATCH 08/25] x86, NUMA: trivial cleanups Tejun Heo
2011-04-29 17:25   ` Yinghai Lu
2011-04-30 12:03     ` Tejun Heo
2011-04-30 16:24       ` Yinghai Lu
2011-04-30 18:00         ` Tejun Heo
2011-04-30 23:10           ` Yinghai Lu
2011-04-30 23:11           ` [PATCH] x86, numa: Rename setup_node_bootmem to setup_node_data Yinghai Lu
2011-04-29 15:28 ` [PATCH 09/25] x86, NUMA: rename srat_64.c to srat.c Tejun Heo
2011-04-29 15:28 ` [PATCH 10/25] x86, NUMA: make srat.c 32bit safe Tejun Heo
2011-04-29 15:28 ` [PATCH 11/25] x86-32, NUMA: Move get_memcfg_numa() into numa_32.c Tejun Heo
2011-04-29 15:28 ` [PATCH 12/25] x86, NUMA: Move numa_nodes_parsed to numa.[hc] Tejun Heo
2011-04-29 15:28 ` [PATCH 13/25] x86-32, NUMA: implement temporary NUMA init shims Tejun Heo
2011-04-29 15:28 ` [PATCH 14/25] x86-32, NUMA: Replace srat_32.c with srat.c Tejun Heo
2011-04-29 15:28 ` [PATCH 15/25] x86-32, NUMA: Update numaq to use new NUMA init protocol Tejun Heo
2011-04-29 15:28 ` [PATCH 16/25] x86, NUMA: Move NUMA init logic from numa_64.c to numa.c Tejun Heo
2011-04-29 15:28 ` [PATCH 17/25] x86, NUMA: Enable build of generic NUMA init code on 32bit Tejun Heo
2011-04-29 15:28 ` [PATCH 18/25] x86, NUMA: Remove long 64bit assumption from numa.c Tejun Heo
2011-04-29 15:28 ` [PATCH 19/25] x86-32, NUMA: Add @start and @end to init_alloc_remap() Tejun Heo
2011-04-29 15:28 ` [PATCH 20/25] x86, NUMA: Initialize and use remap allocator from setup_node_bootmem() Tejun Heo
2011-04-29 15:28 ` [PATCH 21/25] x86, NUMA: Make 32bit use common NUMA init path Tejun Heo
2011-04-29 15:28 ` [PATCH 22/25] x86, NUMA: Make numa_init_array() static Tejun Heo
2011-04-29 15:28 ` [PATCH 23/25] x86, NUMA: Rename amdtopology_64.c to amdtopology.c Tejun Heo
2011-04-29 15:28 ` [PATCH 24/25] x86, NUMA: Enable CONFIG_AMD_NUMA on 32bit too Tejun Heo
2011-04-29 15:28 ` Tejun Heo [this message]
2011-04-29 18:15 ` [PATCHSET tip] x86, NUMA: Unify 32 and 64bit NUMA initialization Ingo Molnar
2011-04-29 20:14 ` Yinghai Lu
2011-04-30 12:17   ` Tejun Heo
2011-04-30 12:33     ` [PATCH] x86, NUMA: Fix empty memblk detection in numa_cleanup_meminfo() Tejun Heo
2011-04-30 12:35       ` Tejun Heo
2011-05-01  0:43       ` Yinghai Lu
2011-05-01 10:20         ` Tejun Heo
2011-05-01 19:44           ` [PATCH] x86, numa: Trim numa meminfo with max_pfn in separated loop Yinghai Lu
2011-04-30 16:31     ` [PATCHSET tip] x86, NUMA: Unify 32 and 64bit NUMA initialization Yinghai Lu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1304090924-8197-26-git-send-email-tj@kernel.org \
    --to=tj@kernel.org \
    --cc=hpa@zytor.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=rientjes@google.com \
    --cc=tglx@linutronix.de \
    --cc=x86@kernel.org \
    --cc=yinghai@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox