x86: add max_addr boot option

Currently, the boot option max_addr is only supported on ia64 platform. We also need it on x86 platform.
For example:
There are two nodes:
NODE#0 address range 0x00000000 00000000 - 0x00010000 00000000 NODE#1 address range 0x00010000 00000000 - 0x00020000 00000000
If we only want to use node0, we can specify the max_addr. The boot option "mem=" can do the same thing now. But the boot option "mem=" means the total memory used by the system. If we tell the user that the boot option "mem=" can do this, it will confuse the user. So we need an new boot option "max_addr" on x86 platform.

---
 Documentation/kernel-parameters.txt |    2 +-
 arch/x86/kernel/e820.c              |   36 +++++++++++++++++++++++++++++++++++
 2 files changed, 37 insertions(+), 1 deletions(-)

index c1601e5..fe80e58 100644
@@ -1409,7 +1409,7 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
 			 yeeloong laptop.
 			Example: machtype=lemote-yeeloong-2f-7inch
 
-	max_addr=nn[KMG]	[KNL,BOOT,ia64] All physical memory greater
+	max_addr=nn[KMG]	[KNL,BOOT,ia64,X86] All physical memory greater
 			than or equal to this physical address is ignored.
 
 	maxcpus=	[SMP] Maximum number of processors that	an SMP kernel

index 62d61e9..2a6bec7 100644
@@ -47,6 +47,7 @@ unsigned long pci_mem_start = 0xaeedbabe;
 #ifdef CONFIG_PCI
 EXPORT_SYMBOL(pci_mem_start);
 #endif
+static u64 max_addr = ~0ULL;
 
 /*
  * This function checks if any part of the range <start,end> is mapped

@@ -117,6 +118,20 @@ static void __init __e820_add_region(struct e820map *e820x, u64 start, u64 size,
 		return;
 	}
 
+	if (start >= max_addr) {
+		printk(KERN_INFO "Ingoring memory: %016Lx - %016Lx\n",
+		       (unsigned long long)start,
+		       (unsigned long long)(start + size));
+		return;
+	}
+
+	if (max_addr - start < size) {
+		printk(KERN_INFO "Ingoring memory: %016Lx - %016Lx\n",
+		       (unsigned long long)max_addr,
+		       (unsigned long long)(start + size));
+		size = max_addr - start;
+	}
+
 	e820x->map[x].addr = start;
 	e820x->map[x].size = size;
 	e820x->map[x].type = type;

@@ -835,6 +850,22 @@ static int __init parse_memopt(char *p)
 }
 early_param("mem", parse_memopt);
 
+static int __init parse_memmax_opt(char *p)
+{
+	char *oldp;
+
+	if (!p)
+		return -EINVAL;
+
+	oldp = p;
+	max_addr = memparse(p, &p);
+	if (p == oldp)
+		return -EINVAL;
+
+	return 0;
+}
+early_param("max_addr", parse_memmax_opt);
+
 static int __init parse_memmap_opt(char *p)
 {
 	char *oldp;

@@ -881,6 +912,11 @@ early_param("memmap", parse_memmap_opt);
 
 void __init finish_e820_parsing(void)
 {
+	if (max_addr != ~0ULL) {
+		userdef = 1;
+		e820_remove_range(max_addr, ULLONG_MAX - max_addr, E820_RAM, 1);
+	}
+
 	if (userdef) {
 		u32 nr = e820.nr_map;
 

1.7.1

The boot option "mem=" specifies the total memory that the system can use. But we implement it as max_addr.

The x86 system can be booted by EFI. If the user specify the boot option "add_efi_memmap", we add all memory map from EFI, but we donot handle the memory map according to the boot option "mem=".

This patch reimplement the boot option "mem=", and handle the memory map after calling efi_init().

---
 arch/x86/include/asm/e820.h |    1 +
 arch/x86/kernel/e820.c      |   36 +++++++++++++++++++++++++++++++-----
 arch/x86/kernel/setup.c     |    1 +
 3 files changed, 33 insertions(+), 5 deletions(-)

index 3778256..d1bb772 100644
@@ -127,6 +127,7 @@ extern void e820_reserve_resources(void);
 extern void e820_reserve_resources_late(void);
 extern void setup_memory_map(void);
 extern char *default_machine_specific_memory_setup(void);
+extern void set_memlimit(void);
 
 /*
  * Returns true iff the specified range [s,e) is completely contained inside

index 2a6bec7..0148944 100644
@@ -48,6 +48,7 @@ unsigned long pci_mem_start = 0xaeedbabe;
 EXPORT_SYMBOL(pci_mem_start);
 #endif
 static u64 max_addr = ~0ULL;
+static u64 mem_limit = ~0ULL;
 
 /*
  * This function checks if any part of the range <start,end> is mapped

@@ -824,8 +825,6 @@ static int userdef __initdata;
 /* "mem=nopentium" disables the 4MB page tables. */
 static int __init parse_memopt(char *p)
 {
-	u64 mem_size;
-
 	if (!p)
 		return -EINVAL;
 

@@ -840,16 +839,43 @@ static int __init parse_memopt(char *p)
 	}
 
 	userdef = 1;
-	mem_size = memparse(p, &p);
+	mem_limit = memparse(p, &p);
 	/* don't remove all of memory when handling "mem={invalid}" param */
-	if (mem_size == 0)
+	if (mem_limit == 0)
 		return -EINVAL;
-	e820_remove_range(mem_size, ULLONG_MAX - mem_size, E820_RAM, 1);
 
 	return 0;
 }
 early_param("mem", parse_memopt);
 
+void __init set_memlimit(void)
+{
+	u64 total_size = 0;
+	int i;
+
+	if (mem_limit == ~0ULL)
+		return;
+
+	for (i = 0; i < e820.nr_map; i++) {
+		struct e820entry *ei = &e820.map[i];
+
+		if (ei->type != E820_RAM)
+			continue;
+
+		if (total_size >= mem_limit) {
+			memset(ei, 0, sizeof(struct e820entry));
+			continue;
+		}
+
+		if (mem_limit - total_size <= ei->size)
+			ei->size = mem_limit - total_size;
+
+		total_size += ei->size;
+	}
+
+	sanitize_e820_map(e820.map, ARRAY_SIZE(e820.map), &e820.nr_map);
+}
+
 static int __init parse_memmax_opt(char *p)
 {
 	char *oldp;

index 1a29015..7938fae 100644
@@ -823,6 +823,7 @@ void __init setup_arch(char **cmdline_p)
 
 	if (efi_enabled)
 		efi_init();
+	set_memlimit();
 
 	dmi_scan_machine();
 

1.7.1

I didn't follow that reasoning at all. Care to try again?

(mem= can do this, but telling users that would confuse them? What?)

Rob

GNU/Linux isn't: Linux=GPLv2, GNU=GPLv3+, they can't share code.
Either it's "mere aggregation", or a license violation.  Pick one.

At 05/23/2012 03:51 AM, Rob Landley Wrote:

mem= means the total memory, but we implement it as max address. I donot know why we implement it as max address. The users donot know how we implement, and they only know that they can use mem= to set the total memory. If you tell the users that mem= can set max address, it will confuse them.

Thanks
Wen Congyang

Kernel doc says

mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
Amount of memory to be used when the kernel is not able to see the whole system memory or for test. [X86-32] Use together with memmap= to avoid physical address space collisions. Without memmap= PCI devices could be placed at addresses belonging to unused RAM.

max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
than or equal to this physical address is ignored.

And, now, on x86+e820, mem= option works as max_addr= option.

This caused some troubles in our customer sometimes. In many server, address range 3G-4g are reserved for PCI.

This is my host's dmesg.
==
BIOS-provided physical RAM map:
BIOS-e820: 0000000000000000 - 000000000009e800 (usable)
BIOS-e820: 000000000009e800 - 00000000000a0000 (reserved)
BIOS-e820: 00000000000ce000 - 00000000000d0000 (reserved)
BIOS-e820: 00000000000e0000 - 0000000000100000 (reserved)
BIOS-e820: 0000000000100000 - 00000000bfee0000 (usable)
BIOS-e820: 00000000bfee0000 - 00000000bfee7000 (ACPI data)
BIOS-e820: 00000000bfee7000 - 00000000bff00000 (ACPI NVS)
BIOS-e820: 00000000bff00000 - 00000000bff80000 (usable)
BIOS-e820: 00000000bff80000 - 00000000c0000000 (reserved)
BIOS-e820: 00000000e0000000 - 00000000f0000000 (reserved)
BIOS-e820: 00000000fe000000 - 00000000fed00000 (reserved)
BIOS-e820: 00000000fee00000 - 00000000fef00000 (reserved)
BIOS-e820: 00000000ffb00000 - 0000000100000000 (reserved)
BIOS-e820: 0000000100000000 - 0000000640000000 (usable)
==

So, if customer sets mem=10G, the system will boot with 9G memory. I think this is a bug and mem= should see 'amount of memory'.

For users who want to hide memory in higher address, I think max_addr= option is suitable.

Thanks,
-Kame

s/Ingoring/Ignoring/

Please use a format like "e820: ignoring [mem %#010Lx-%#010Lx]" so it matches other places where we print physical address ranges. See
https://lkml.org/lkml/2012/2/13/436

Bjorn

At 05/23/2012 10:25 PM, Bjorn Helgaas Wrote:

Hmm, no one reviews/acks patch. So I am not sure whether we need to do this. So I think there is no need to use such format now.

Thanks
Wen Congyang