From: Dave Hansen <haveblue@us.ibm.com>
As described in this bug: http://bugme.osdl.org/show_bug.cgi?id=653 , if
you enable Summit support and NUMA Discontigmem support but boot on a
non-Summit box, the kernel will fail to boot. The problem is that the
Summit code can not correctly get the NUMA memory configuration of a
flat box. The code to do that is in get_memcfg_numa_flat(), but it
never gets called.
This patch implements a fallback to the generic NUMA code in
get_memcfg_numa_flat() if the Summit detection fails. The patch also
adds the necessary bits to the Summit code so that it *knows* when it
fails.
25-akpm/arch/i386/kernel/numaq.c | 8 +++++++-
25-akpm/arch/i386/kernel/srat.c | 33 +++++++++++++++++++++++----------
25-akpm/arch/i386/mm/discontig.c | 11 ++++++++---
25-akpm/include/asm-i386/mmzone.h | 20 +++++++++++++++++++-
25-akpm/include/asm-i386/numaq.h | 3 +--
25-akpm/include/asm-i386/srat.h | 3 +--
6 files changed, 59 insertions(+), 19 deletions(-)
diff -puN arch/i386/kernel/numaq.c~numa-detection-fail-fix arch/i386/kernel/numaq.c
--- 25/arch/i386/kernel/numaq.c~numa-detection-fail-fix Thu Sep 11 10:29:43 2003
+++ 25-akpm/arch/i386/kernel/numaq.c Thu Sep 11 10:29:43 2003
@@ -99,8 +99,14 @@ static void __init initialize_physnode_m
}
}
-void __init get_memcfg_numaq(void)
+/*
+ * Unlike Summit, we don't really care to let the NUMA-Q
+ * fall back to flat mode. Don't compile for NUMA-Q
+ * unless you really need it!
+ */
+int __init get_memcfg_numaq(void)
{
smp_dump_qct();
initialize_physnode_map();
+ return 1;
}
diff -puN arch/i386/kernel/srat.c~numa-detection-fail-fix arch/i386/kernel/srat.c
--- 25/arch/i386/kernel/srat.c~numa-detection-fail-fix Thu Sep 11 10:29:43 2003
+++ 25-akpm/arch/i386/kernel/srat.c Thu Sep 11 10:29:43 2003
@@ -239,6 +239,11 @@ static int __init acpi20_parse_srat(stru
}
}
+ if (num_memory_chunks == 0) {
+ printk("could not finy any ACPI SRAT memory areas.\n");
+ goto out_fail;
+ }
+
/* Calculate total number of nodes in system from PXM bitmap and create
* a set of sequential node IDs starting at zero. (ACPI doesn't seem
* to specify the range of _PXM values.)
@@ -295,10 +300,12 @@ static int __init acpi20_parse_srat(stru
}
}
}
+ return 1;
+out_fail:
return 0;
}
-void __init get_memcfg_from_srat(void)
+int __init get_memcfg_from_srat(void)
{
struct acpi_table_header *header = NULL;
struct acpi_table_rsdp *rsdp = NULL;
@@ -316,11 +323,11 @@ void __init get_memcfg_from_srat(void)
(u32)rsdp_address->pointer.physical;
} else {
printk("%s: rsdp_address is not a physical pointer\n", __FUNCTION__);
- return;
+ goto out_err;
}
if (!rsdp) {
printk("%s: Didn't find ACPI root!\n", __FUNCTION__);
- return;
+ goto out_err;
}
printk(KERN_INFO "%.8s v%d [%.6s]\n", rsdp->signature, rsdp->revision,
@@ -328,7 +335,7 @@ void __init get_memcfg_from_srat(void)
if (strncmp(rsdp->signature, RSDP_SIG,strlen(RSDP_SIG))) {
printk(KERN_WARNING "%s: RSDP table signature incorrect\n", __FUNCTION__);
- return;
+ goto out_err;
}
rsdt = (struct acpi_table_rsdt *)
@@ -338,14 +345,14 @@ void __init get_memcfg_from_srat(void)
printk(KERN_WARNING
"%s: ACPI: Invalid root system description tables (RSDT)\n",
__FUNCTION__);
- return;
+ goto out_err;
}
header = & rsdt->header;
if (strncmp(header->signature, RSDT_SIG, strlen(RSDT_SIG))) {
printk(KERN_WARNING "ACPI: RSDT signature incorrect\n");
- return;
+ goto out_err;
}
/*
@@ -356,15 +363,18 @@ void __init get_memcfg_from_srat(void)
*/
tables = (header->length - sizeof(struct acpi_table_header)) / 4;
+ if (!tables)
+ goto out_err;
+
memcpy(&saved_rsdt, rsdt, sizeof(saved_rsdt));
if (saved_rsdt.header.length > sizeof(saved_rsdt)) {
printk(KERN_WARNING "ACPI: Too big length in RSDT: %d\n",
saved_rsdt.header.length);
- return;
+ goto out_err;
}
-printk("Begin table scan....\n");
+ printk("Begin SRAT table scan....\n");
for (i = 0; i < tables; i++) {
/* Map in header, then map in full table length. */
@@ -379,10 +389,13 @@ printk("Begin table scan....\n");
if (strncmp((char *) &header->signature, "SRAT", 4))
continue;
- acpi20_parse_srat((struct acpi_table_srat *)header);
+
/* we've found the srat table. don't need to look at any more tables */
- break;
+ return acpi20_parse_srat((struct acpi_table_srat *)header);
}
+out_err:
+ printk("failed to get NUMA memory information from SRAT table\n");
+ return 0;
}
/* For each node run the memory list to determine whether there are
diff -puN arch/i386/mm/discontig.c~numa-detection-fail-fix arch/i386/mm/discontig.c
--- 25/arch/i386/mm/discontig.c~numa-detection-fail-fix Thu Sep 11 10:29:43 2003
+++ 25-akpm/arch/i386/mm/discontig.c Thu Sep 11 10:29:43 2003
@@ -30,6 +30,7 @@
#include <linux/initrd.h>
#include <asm/e820.h>
#include <asm/setup.h>
+#include <asm/mmzone.h>
struct pglist_data *node_data[MAX_NUMNODES];
bootmem_data_t node0_bdata;
@@ -84,7 +85,7 @@ void set_pmd_pfn(unsigned long vaddr, un
* a single node with all available processors in it with a flat
* memory map.
*/
-void __init get_memcfg_numa_flat(void)
+int __init get_memcfg_numa_flat(void)
{
int pfn;
@@ -107,6 +108,7 @@ void __init get_memcfg_numa_flat(void)
/* Indicate there is one node available. */
node_set_online(0);
numnodes = 1;
+ return 1;
}
/*
@@ -355,17 +357,20 @@ void __init zone_sizes_init(void)
unsigned long low = max_low_pfn;
unsigned long start = node_start_pfn[nid];
unsigned long high = node_end_pfn[nid];
-
+
max_dma = virt_to_phys((char *)MAX_DMA_ADDRESS) >> PAGE_SHIFT;
if (start > low) {
#ifdef CONFIG_HIGHMEM
- zones_size[ZONE_HIGHMEM] = high - start;
+ BUG_ON(start > high);
+ zones_size[ZONE_HIGHMEM] = high - start;
#endif
} else {
if (low < max_dma)
zones_size[ZONE_DMA] = low;
else {
+ BUG_ON(max_dma > low);
+ BUG_ON(low > high);
zones_size[ZONE_DMA] = max_dma;
zones_size[ZONE_NORMAL] = low - max_dma;
#ifdef CONFIG_HIGHMEM
diff -puN include/asm-i386/mmzone.h~numa-detection-fail-fix include/asm-i386/mmzone.h
--- 25/include/asm-i386/mmzone.h~numa-detection-fail-fix Thu Sep 11 10:29:43 2003
+++ 25-akpm/include/asm-i386/mmzone.h Thu Sep 11 10:29:43 2003
@@ -122,11 +122,29 @@ static inline struct pglist_data *pfn_to
#elif CONFIG_ACPI_SRAT
#include <asm/srat.h>
#elif CONFIG_X86_PC
-#define get_memcfg_numa get_memcfg_numa_flat
#define get_zholes_size(n) (0)
#else
#define pfn_to_nid(pfn) (0)
#endif /* CONFIG_X86_NUMAQ */
+extern int get_memcfg_numa_flat(void );
+/*
+ * This allows any one NUMA architecture to be compiled
+ * for, and still fall back to the flat function if it
+ * fails.
+ */
+static inline void get_memcfg_numa(void)
+{
+#ifdef CONFIG_X86_NUMAQ
+ if (get_memcfg_numaq())
+ return;
+#elif CONFIG_ACPI_SRAT
+ if (get_memcfg_from_srat())
+ return;
+#endif
+
+ get_memcfg_numa_flat();
+}
+
#endif /* CONFIG_DISCONTIGMEM */
#endif /* _ASM_MMZONE_H_ */
diff -puN include/asm-i386/numaq.h~numa-detection-fail-fix include/asm-i386/numaq.h
--- 25/include/asm-i386/numaq.h~numa-detection-fail-fix Thu Sep 11 10:29:43 2003
+++ 25-akpm/include/asm-i386/numaq.h Thu Sep 11 10:29:43 2003
@@ -29,8 +29,7 @@
#ifdef CONFIG_X86_NUMAQ
#define MAX_NUMNODES 16
-extern void get_memcfg_numaq(void);
-#define get_memcfg_numa() get_memcfg_numaq()
+extern int get_memcfg_numaq(void);
/*
* SYS_CFG_DATA_PRIV_ADDR, struct eachquadmem, and struct sys_cfg_data are the
diff -puN include/asm-i386/srat.h~numa-detection-fail-fix include/asm-i386/srat.h
--- 25/include/asm-i386/srat.h~numa-detection-fail-fix Thu Sep 11 10:29:43 2003
+++ 25-akpm/include/asm-i386/srat.h Thu Sep 11 10:29:43 2003
@@ -32,8 +32,7 @@
#endif
#define MAX_NUMNODES 8
-extern void get_memcfg_from_srat(void);
+extern int get_memcfg_from_srat(void);
extern unsigned long *get_zholes_size(int);
-#define get_memcfg_numa() get_memcfg_from_srat()
#endif /* _ASM_SRAT_H_ */
_