kdump x86: fix total mem size calculation for reservation

PATCH Superseded Dave Young 3 participants 6 messages Mar 9, 2012 – Mar 10, 2012 x86

Share & tools

Dave Young

author

2012-03-09 08:27 UTC #0 lore raw

PATCH 1/1 kdump x86: fix total mem size calculation for reservation

crashkernel reservation need know the total memory size. Current get_total_mem simply use max_pfn - min_low_pfn. It is wrong because it will including memory holes in the middle.

Especially for kvm guest with memory > 0xe0000000, there's below in qemu code:
qemu split memory as below:

    if (ram_size >= 0xe0000000 ) {
        above_4g_mem_size = ram_size - 0xe0000000;
        below_4g_mem_size = 0xe0000000;
    } else {
        below_4g_mem_size = ram_size;
    }

So for 4G mem guest, seabios will insert a 512M usable region beyond of 4G. Thus in above case max_pfn - min_low_pfn will be more than original memsize.

Fixing this issue by using memblock_phys_mem_size() to get the total memsize.

Signed-off-by Dave Young <[email protected]>

---
[Ther's similar code in arm, I'm not sure whether the memblock_phys_mem_size works with arm or not. If it works I can also update that part of code.]

 arch/x86/kernel/setup.c |   12 +-----------
 1 file changed, 1 insertion(+), 11 deletions(-)

arch/x86/kernel/setup.c +1 ~~-11~~

@@ -509,16 +509,6 @@ static void __init memblock_x86_reserve_
 
 #ifdef CONFIG_KEXEC
 
-static inline unsigned long long get_total_mem(void)
-{
-	unsigned long long total;
-
-	total = max_pfn - min_low_pfn;
-	printk("hidave: memsize=%llu\n", memblock_phys_mem_size());
-
-	return total << PAGE_SHIFT;
-}
-
 /*
  * Keep the crash kernel below this limit.  On 32 bits earlier kernels
  * would limit the kernel to the low 512 MiB due to mapping restrictions.

@@ -537,7 +527,7 @@ static void __init reserve_crashkernel(v
 	unsigned long long crash_size, crash_base;
 	int ret;
 
-	total_mem = get_total_mem();
+	total_mem = memblock_phys_mem_size();
 
 	ret = parse_crashkernel(boot_command_line, total_mem,
 			&crash_size, &crash_base);

Cong Wang

2012-03-10 03:27 UTC #1 lore raw

Quoted 12 lines · Dave Young

crashkernel reservation need know the total memory size. Current get_total_mem simply use max_pfn - min_low_pfn. It is wrong because it will including memory holes in the middle.
Especially for kvm guest with memory> 0xe0000000, there's below in qemu code:
qemu split memory as below:
     if (ram_size>= 0xe0000000 ) {
         above_4g_mem_size = ram_size - 0xe0000000;
         below_4g_mem_size = 0xe0000000;
     } else {
         below_4g_mem_size = ram_size;
     }

Quoted 5 lines

So for 4G mem guest, seabios will insert a 512M usable region beyond of 4G. Thus in above case max_pfn - min_low_pfn will be more than original memsize.
Fixing this issue by using memblock_phys_mem_size() to get the total memsize.

Makes sense for me!

arch/x86/kernel/setup.c static void __init memblock_x86_reserve_ · line 517

Quoted 6 lines

arch/x86/kernel/setup.c +0 -6

-static inline unsigned long long get_total_mem(void)
-{
-	unsigned long long total;
-
-	total = max_pfn - min_low_pfn;
-	printk("hidave: memsize=%llu\n", memblock_phys_mem_size());

This debugging line does not exist... so this patch can't be applied.

Dave Young

author

2012-03-10 04:34 UTC #2 lore raw

Quoted 26 lines · Cong Wang

Quoted 14 lines · Dave Young
crashkernel reservation need know the total memory size. Current get_total_mem
simply use max_pfn - min_low_pfn. It is wrong because it will including memory holes in the middle.
Especially for kvm guest with memory> 0xe0000000, there's below in qemu code:
qemu split memory as below:
     if (ram_size>= 0xe0000000 ) {
         above_4g_mem_size = ram_size - 0xe0000000;
         below_4g_mem_size = 0xe0000000;
     } else {
         below_4g_mem_size = ram_size;
     }
Quoted 8 lines
So for 4G mem guest, seabios will insert a 512M usable region beyond of 4G.
Thus in above case max_pfn - min_low_pfn will be more than original memsize.
Fixing this issue by using memblock_phys_mem_size() to get the total memsize.
Makes sense for me!

Quoted 8 lines

arch/x86/kernel/setup.c static void __init memblock_x86_reserve_ · line 517
Quoted 6 lines
arch/x86/kernel/setup.c +0 -6
-static inline unsigned long long get_total_mem(void)
-{
-    unsigned long long total;
-
-    total = max_pfn - min_low_pfn;
-    printk("hidave: memsize=%llu\n", memblock_phys_mem_size());
This debugging line does not exist... so this patch can't be applied.

Yes, good catch, will remove the line and resend.

Signature · 2 lines

Thanks
Dave

Dave Young

author

2012-03-10 04:58 UTC #3 lore raw

PATCH v2 1/1 kdump x86: fix total mem size calculation for reservation

In reply to Cong Wang

crashkernel reservation need know the total memory size. Current get_total_mem simply use max_pfn - min_low_pfn. It is wrong because it will including memory holes in the middle.

Especially for kvm guest with memory > 0xe0000000, there's below in qemu code:
qemu split memory as below:

    if (ram_size >= 0xe0000000 ) {
        above_4g_mem_size = ram_size - 0xe0000000;
        below_4g_mem_size = 0xe0000000;
    } else {
        below_4g_mem_size = ram_size;
    }

So for 4G mem guest, seabios will insert a 512M usable region beyond of 4G. Thus in above case max_pfn - min_low_pfn will be more than original memsize.

Fixing this issue by using memblock_phys_mem_size() to get the total memsize.

[v1 -> v2]: refresh the patch based on latest linus tree

Signed-off-by Dave Young <[email protected]>

---
[Ther's similar code in arm, I'm not sure whether the memblock_phys_mem_size works with arm or not. If it works I can also update that part of code.]

 arch/x86/kernel/setup.c |   11 +----------
 1 file changed, 1 insertion(+), 10 deletions(-)

arch/x86/kernel/setup.c +1 ~~-10~~

@@ -509,15 +509,6 @@ static void __init memblock_x86_reserve_
 
 #ifdef CONFIG_KEXEC
 
-static inline unsigned long long get_total_mem(void)
-{
-	unsigned long long total;
-
-	total = max_pfn - min_low_pfn;
-
-	return total << PAGE_SHIFT;
-}
-
 /*
  * Keep the crash kernel below this limit.  On 32 bits earlier kernels
  * would limit the kernel to the low 512 MiB due to mapping restrictions.

@@ -536,7 +527,7 @@ static void __init reserve_crashkernel(v
 	unsigned long long crash_size, crash_base;
 	int ret;
 
-	total_mem = get_total_mem();
+	total_mem = memblock_phys_mem_size();
 
 	ret = parse_crashkernel(boot_command_line, total_mem,
 			&crash_size, &crash_base);

Cong Wang

Reviewed-by on v2 1/1

2012-03-10 08:38 UTC #4 lore raw

Quoted 12 lines · Dave Young

crashkernel reservation need know the total memory size. Current get_total_mem simply use max_pfn - min_low_pfn. It is wrong because it will including memory holes in the middle.
Especially for kvm guest with memory> 0xe0000000, there's below in qemu code:
qemu split memory as below:
     if (ram_size>= 0xe0000000 ) {
         above_4g_mem_size = ram_size - 0xe0000000;
         below_4g_mem_size = 0xe0000000;
     } else {
         below_4g_mem_size = ram_size;
     }

Quoted 7 lines

So for 4G mem guest, seabios will insert a 512M usable region beyond of 4G. Thus in above case max_pfn - min_low_pfn will be more than original memsize.
Fixing this issue by using memblock_phys_mem_size() to get the total memsize.
[v1 -> v2]: refresh the patch based on latest linus tree

Looks good to my eyes,

Reviewed-by WANG Cong <[email protected]>

Can you post the result of crashkernel after this patch to convince people?

Thanks.

Dave Young

2012-03-10 16:06 UTC #5 lore raw

Quoted 24 lines · Cong Wang

Quoted 23 lines · Dave Young
crashkernel reservation need know the total memory size. Current get_total_mem
simply use max_pfn - min_low_pfn. It is wrong because it will including memory holes in the middle.
Especially for kvm guest with memory> 0xe0000000, there's below in qemu code:
qemu split memory as below:
if (ram_size>= 0xe0000000 ) {
above_4g_mem_size = ram_size - 0xe0000000; below_4g_mem_size = 0xe0000000;
} else {
below_4g_mem_size = ram_size;
}
So for 4G mem guest, seabios will insert a 512M usable region beyond of 4G.
Thus in above case max_pfn - min_low_pfn will be more than original memsize.
Fixing this issue by using memblock_phys_mem_size() to get the total memsize.

Quoted 8 lines

Quoted 2 lines
[v1 -> v2]: refresh the patch based on latest linus tree
Looks good to my eyes,
Reviewed-by WANG Cong <[email protected]>
Can you post the result of crashkernel after this patch to convince people?

Sure, for the qemu 4G vm results below:
-[ 0.000000] Reserving 128MB of memory at 768MB for crashkernel
(System RAM: 4608MB)
+[ 0.000000] Reserving 128MB of memory at 768MB for crashkernel
(System RAM: 4095MB)

Quoted 8 lines

Thanks.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 2 lines

Regards
Dave