Skip to contentSkip to conversation

Zero size /proc/vmcore on ia64

PATCH Horms 5 participants 16 messages Feb 5, 2007 – Feb 15, 2007 EFI

lore
Share & tools
lore

Horms

author
#0 lore raw

Hi,

I have been poking around this problem a bit over the past week, and I thought it would be a good idea to get it out in the open.

At some stage /proc/vmcore (in a crash-kernel) went from being something useful, to being zero size.

I initially thought this was because saved_max_pfn was not being set correctly. And indeed it is not set for discontig memory. But the trivial fix below has not been sufficient to resolve the problem :(

The problem seems to be along the lines of:
* kexec-tool sets up a segment to contain the elf header.
* This segment happens to be almost at the end of the crashkernel area of memory that is visible to the crash kernel.
* However, when purgatory munges the EFI map, this segment is marked as EFI_UNUSABLE_MEMORY.
* As a result of this it is not in a range covered by efi_memmap_walk()
* And thus it is outside the range of memory covered by a valid PFN (remember its at the end of memory, it turns out that the max PFN covers memory up until just before the header)
* The header can't be read by the vmcore setup code
* And vmcore is uninitialised

read_from_oldmem: error: pfn (32761) > saved_max_pfn (31744)
Kdump: vmcore not initialized

The saved_max_pfn error above is produced by debuging code that I added to read_from_oldmem().
It also uses the patch below, otherwise saved_max_pfn is 0.

For reference:
I am using today's linus tree (2.6.20)
The problem seems to have been around since at least 2.6.19-rc6 I have a Tiger2 system using disctontig memory
The problem also seems to manifest when using contig memory

Signature · 13 lines
Horms
  H: http://www.vergenet.net/~horms/
  W: http://www.valinux.co.jp/en/

Set saved_max_pfn when discontig memory is in use.

This sets up saved_max_pfn when disctontig memory is in use.
This mirrors the code for contig memory.

This patch does not entirely solve the problem of making vmcore work,
however it does appear to be neccessary. Please consider applying.

Signed-off-by: Simon Horman <[email protected]>
arch/ia64/mm/discontig.c +6 -0
index 96722cb..999cefd 100644
@@ -506,6 +509,12 @@ void __init find_memory(void)
 	max_pfn = max_low_pfn;
 
 	find_initrd();
+
+#ifdef CONFIG_CRASH_DUMP
+	/* If we are doing a crash dump, we still need to know the real mem
+	 * size before original memory map is reset. */
+        saved_max_pfn = max_pfn;
+#endif
 }
 
 #ifdef CONFIG_SMP
#1 lore raw
Quoted 95 lines

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Zou, Nanhai
Sent: 2007年2月6日 9:09
To: Horms; [email protected]; [email protected]
Cc: Khalid Aziz; Mel Gorman; Bob Picco; Magnus Damm
Subject: RE: Zero size /proc/vmcore on ia64

This seems to be a corner case which purgatory efi-memmap code does not handle correctly.

Can you print the memory ranges layout information of first and second kernel? e.g, the efi memmap, the crash notes memory address.

Thanks
Zou Nan hai

Quoted 76 lines

-----Original Message-----
From: Horms [mailto:[email protected]]
Sent: 2007年2月5日 9:59
To: [email protected]; [email protected]
Cc: Zou, Nanhai; Khalid Aziz; Mel Gorman; Bob Picco; Magnus Damm
Subject: Zero size /proc/vmcore on ia64

Hi,

I have been poking around this problem a bit over the past week, and I thought it would be a good idea to get it out in the open.

At some stage /proc/vmcore (in a crash-kernel) went from being something useful, to being zero size.

I initially thought this was because saved_max_pfn was not being set correctly. And indeed it is not set for discontig memory. But the trivial fix below has not been sufficient to resolve the problem :(

The problem seems to be along the lines of:
* kexec-tool sets up a segment to contain the elf header.
* This segment happens to be almost at the end of the crashkernel area of memory that is visible to the crash kernel.
* However, when purgatory munges the EFI map, this segment is marked as EFI_UNUSABLE_MEMORY.
* As a result of this it is not in a range covered by efi_memmap_walk()
* And thus it is outside the range of memory covered by a valid PFN (remember its at the end of memory, it turns out that the max PFN covers memory up until just before the header)
* The header can't be read by the vmcore setup code
* And vmcore is uninitialised

read_from_oldmem: error: pfn (32761) > saved_max_pfn (31744)
Kdump: vmcore not initialized

The saved_max_pfn error above is produced by debuging code that I added to read_from_oldmem().
It also uses the patch below, otherwise saved_max_pfn is 0.

For reference:
I am using today's linus tree (2.6.20)
The problem seems to have been around since at least 2.6.19-rc6 I have a Tiger2 system using disctontig memory
The problem also seems to manifest when using contig memory

--
Horms
H: http://www.vergenet.net/~horms/
W: http://www.valinux.co.jp/en/

Set saved_max_pfn when discontig memory is in use.

This sets up saved_max_pfn when disctontig memory is in use. This mirrors the code for contig memory.

This patch does not entirely solve the problem of making vmcore work, however it does appear to be neccessary. Please consider applying.

arch/ia64/mm/discontig.c +6 -0
index 96722cb..999cefd 100644
@@ -506,6 +509,12 @@ void __init find_memory(void)
 	max_pfn = max_low_pfn;
 
 	find_initrd();
+
+#ifdef CONFIG_CRASH_DUMP
+	/* If we are doing a crash dump, we still need to know the real mem
+	 * size before original memory map is reset. */
+        saved_max_pfn = max_pfn;
+#endif
 }
 
Quoted 1 line

arch/ia64/mm/discontig.c void __init find_memory(void) · line 520

Quoted 1 line
arch/ia64/mm/discontig.c +0 -0
 #ifdef CONFIG_SMP

Hi Vivek,
I have a question about why saved_max_pfn check in vmcore.c is needed.
Here is a typical memory layout of IA64 machine.

----- ==>max_pfn for first kernel
the first kernel
----- ==>max_pfn for crash dump kernel
the crash dump kernel
-----
the first kernel
-----

When crash dump kernel tries to access memory of first kernel above saved_max_pfn of him, read_from_oldmem will refuse that read.

That result an empty vmcore file. change saved_max_pfn to unsigned long(-1) will fix this issue.

However since memory ranges in vmcore is pre defined from /proc/iomem of first kernel, why do we still need to add an extra check in vmcore.c

Thanks
Zou Nan hai
-
To unsubscribe from this list: send the line "unsubscribe linux-ia64" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

author
#2 lore raw
In reply to Horms
Quoted 16 lines · Nanhai

Hi Vivek,
I have a question about why saved_max_pfn check in vmcore.c is needed.
Here is a typical memory layout of IA64 machine.

----- ==>max_pfn for first kernel
the first kernel
----- ==>max_pfn for crash dump kernel
the crash dump kernel
-----
the first kernel
-----

When crash dump kernel tries to access memory of first kernel above saved_max_pfn of him, read_from_oldmem will refuse that read.

Quoted 6 lines

That result an empty vmcore file. change saved_max_pfn to unsigned long(-1) will fix this issue.

However since memory ranges in vmcore is pre defined from /proc/iomem of first kernel, why do we still need to add an extra check in vmcore.c

Hi Nan-hai,

sorry that I did not get back to you about the information you requested about my system, I guess you have managed to reproduce the problem none the less.

I can confirm that removing the max_pfn check in vmcore.c does indeed give /proc/vmcore a non-zero (and presumably correct) size.

I wonder if the problem is that saved_max_pfn is being incorectly calculated on ia64. That it is being set to the max_pfn of the crash kernel (i.e. in the crashkernel=X@Y area), rather than the max_pfn of the physical memory of the system, which seems more sensible as the purpose of vmcore is to read memory outside of the crashkernel=X@Y area.

You may be right that we can just remove the check all together, though perhaps it is there for the case where the range information in the vmcode are corrupted. Then again, should we care about this?

Signature · 3 lines
Horms
  H: http://www.vergenet.net/~horms/
  W: http://www.valinux.co.jp/en/
#3 lore raw
In reply to Horms
Quoted 79 lines · Vivek Goyal
Quoted 4 lines · Horms
Quoted 3 lines · Nanhai

Hi Vivek,
I have a question about why saved_max_pfn check in vmcore.c is

needed.

Quoted 11 lines
Quoted 11 lines

Here is a typical memory layout of IA64 machine.

----- ==>max_pfn for first kernel
the first kernel
----- ==>max_pfn for crash dump kernel
the crash dump kernel
-----
the first kernel
-----

When crash dump kernel tries to access memory of first kernel

above

Quoted 6 lines
Quoted 6 lines

saved_max_pfn of him, read_from_oldmem will refuse that read.

That result an empty vmcore file. change saved_max_pfn to unsigned long(-1) will fix this issue.

However since memory ranges in vmcore is pre defined from

/proc/iomem

Quoted 6 lines
Quoted 2 lines

of first kernel, why do we still need to add an extra check in vmcore.c

Hi Nan-hai,

sorry that I did not get back to you about the information you

requested

Quoted 1 line

about my system, I guess you have managed to reproduce the problem

none

Quoted 5 lines

the less.

I can confirm that removing the max_pfn check in vmcore.c does indeed give /proc/vmcore a non-zero (and presumably correct) size.

Quoted 7 lines

I wonder if the problem is that saved_max_pfn is being incorectly calculated on ia64. That it is being set to the max_pfn of the crash kernel (i.e. in the crashkernel=X@Y area), rather than the max_pfn of the physical memory of the system, which seems more sensible as the purpose of vmcore is to read memory outside of the crashkernel=X@Y area.

Hi Horms/Nan-hai,

Horms, you are right. saved_max_pfn is needed to know that second kernel
is not trying to read any memory which is not present or was not being used by the crashed kernel at all. That's why in i386/x86_64, during early boot saved_max_pfn, is calculated the memory map passed to the second
kernel. This memory map is passed to second kernel by kexec through parameter
segment. So effectively saved_max_pfn will be set to max_pfn of crashed kernel.

Now this memory map is overwritten with user defined one which is basically
the memory second kernel can use to boot and max_pfn now will be maximum
pfn crash kernel can use.

Quoted 3 lines

You may be right that we can just remove the check all together, though perhaps it is there for the case where the range information in the vmcode are corrupted. Then again, should we care about this?

I think we should not remove this check because even to parse the info passed in ELF headers, you need to first read the ELF headers from crashed
kernel's memory. So if some programming error has passed wrong location of
ELF headers (elfcoreheader= invalid location) then we might try reading the
elf header from a non-existing physical page frame.

Quoted 4 lines

So the right way should be to set saved_max_pfn with right value before it
is memory map is over-written with user defined memory map.

This is reasonable.
So please apply the following patch to make saved_max_pfn point to
max_pfn of entire system.

arch/ia64/kernel/efi.c +6 -0
diff -Nraup linux-2.6.20/arch/ia64/kernel/efi.c linux-2.6.20-fix/arch/ia64/kernel/efi.c
@@ -21,6 +21,7 @@
  *	Skip non-WB memory and ignore empty memory ranges.
  */
 #include <linux/module.h>
+#include <linux/bootmem.h>
 #include <linux/kernel.h>
 #include <linux/init.h>
 #include <linux/types.h>
@@ -1010,6 +1011,11 @@ efi_memmap_init(unsigned long *s, unsign
 		} else
 			ae = efi_md_end(md);
 
+#ifdef CONFIG_CRASH_DUMP
+		/* saved_max_pfn should ignore max_addr= command line arg */ 
+		if (saved_max_pfn < (ae >> PAGE_SHIFT))
+			saved_max_pfn = (ae >> PAGE_SHIFT);
+#endif
 		/* keep within max_addr= and min_addr= command line arg */
 		as = max(as, min_addr);
 		ae = min(ae, max_addr);
arch/ia64/mm/contig.c +0 -5
diff -Nraup linux-2.6.20/arch/ia64/mm/contig.c linux-2.6.20-fix/arch/ia64/mm/contig.c
@@ -175,11 +175,6 @@ find_memory (void)
 
 	find_initrd();
 
-#ifdef CONFIG_CRASH_DUMP
-	/* If we are doing a crash dump, we still need to know the real mem
-	 * size before original memory map is * reset. */
-	saved_max_pfn = max_pfn;
-#endif
 }
 
 #ifdef CONFIG_SMP
Quoted 3 lines

Thanks
Vivek

#4 lore raw
In reply to Horms
Quoted 33 lines · Horms
Quoted 16 lines · Nanhai

Hi Vivek,
I have a question about why saved_max_pfn check in vmcore.c is needed.
Here is a typical memory layout of IA64 machine.

----- ==>max_pfn for first kernel
the first kernel
----- ==>max_pfn for crash dump kernel
the crash dump kernel
-----
the first kernel
-----

When crash dump kernel tries to access memory of first kernel above saved_max_pfn of him, read_from_oldmem will refuse that read.

Quoted 6 lines

That result an empty vmcore file. change saved_max_pfn to unsigned long(-1) will fix this issue.

However since memory ranges in vmcore is pre defined from /proc/iomem of first kernel, why do we still need to add an extra check in vmcore.c

Hi Nan-hai,

sorry that I did not get back to you about the information you requested about my system, I guess you have managed to reproduce the problem none the less.

I can confirm that removing the max_pfn check in vmcore.c does indeed give /proc/vmcore a non-zero (and presumably correct) size.

Quoted 7 lines

I wonder if the problem is that saved_max_pfn is being incorectly calculated on ia64. That it is being set to the max_pfn of the crash kernel (i.e. in the crashkernel=X@Y area), rather than the max_pfn of the physical memory of the system, which seems more sensible as the purpose of vmcore is to read memory outside of the crashkernel=X@Y area.

Hi Horms/Nan-hai,

Horms, you are right. saved_max_pfn is needed to know that second kernel is not trying to read any memory which is not present or was not being used by the crashed kernel at all. That's why in i386/x86_64, during early boot saved_max_pfn, is calculated the memory map passed to the second kernel. This memory map is passed to second kernel by kexec through parameter segment. So effectively saved_max_pfn will be set to max_pfn of crashed kernel.

Now this memory map is overwritten with user defined one which is basically the memory second kernel can use to boot and max_pfn now will be maximum pfn crash kernel can use.

Quoted 3 lines

You may be right that we can just remove the check all together, though perhaps it is there for the case where the range information in the vmcode are corrupted. Then again, should we care about this?

I think we should not remove this check because even to parse the info passed in ELF headers, you need to first read the ELF headers from crashed kernel's memory. So if some programming error has passed wrong location of ELF headers (elfcoreheader= invalid location) then we might try reading the elf header from a non-existing physical page frame.

So the right way should be to set saved_max_pfn with right value before it is memory map is over-written with user defined memory map.

Thanks
Vivek

author
#5 lore raw
In reply to Horms
Quoted 58 lines · Zou Nan hai
Quoted 57 lines · Vivek Goyal
Quoted 22 lines · Horms
Quoted 11 lines · Nanhai

When crash dump kernel tries to access memory of first kernel above saved_max_pfn of him, read_from_oldmem will refuse that read.

That result an empty vmcore file. change saved_max_pfn to unsigned long(-1) will fix this issue.

However since memory ranges in vmcore is pre defined from /proc/iomem of first kernel, why do we still need to add an extra check in vmcore.c

Hi Nan-hai,

sorry that I did not get back to you about the information you requested about my system, I guess you have managed to reproduce the problem none the less.

I can confirm that removing the max_pfn check in vmcore.c does indeed give /proc/vmcore a non-zero (and presumably correct) size.

Quoted 7 lines

I wonder if the problem is that saved_max_pfn is being incorectly calculated on ia64. That it is being set to the max_pfn of the crash kernel (i.e. in the crashkernel=X@Y area), rather than the max_pfn of the physical memory of the system, which seems more sensible as the purpose of vmcore is to read memory outside of the crashkernel=X@Y area.

Hi Horms/Nan-hai,

Horms, you are right. saved_max_pfn is needed to know that second kernel is not trying to read any memory which is not present or was not being used by the crashed kernel at all. That's why in i386/x86_64, during early boot saved_max_pfn, is calculated the memory map passed to the second kernel. This memory map is passed to second kernel by kexec through parameter segment. So effectively saved_max_pfn will be set to max_pfn of crashed kernel.

Now this memory map is overwritten with user defined one which is basically the memory second kernel can use to boot and max_pfn now will be maximum pfn crash kernel can use.

Quoted 4 lines

You may be right that we can just remove the check all together, though perhaps it is there for the case where the range information in the vmcode are corrupted. Then again, should we care about this?

I think we should not remove this check because even to parse the info passed in ELF headers, you need to first read the ELF headers from crashed kernel's memory. So if some programming error has passed wrong location of ELF headers (elfcoreheader= invalid location) then we might try reading the elf header from a non-existing physical page frame.

Quoted 7 lines
Quoted 4 lines

So the right way should be to set saved_max_pfn with right value before it is memory map is over-written with user defined memory map.

This is reasonable.
So please apply the following patch to make
saved_max_pfn point to max_pfn of entire system.

Hi Nanhai,

Although I agree with the gist of your patch, unfortunately it does not work on my system. Perhaps this is because I use discontig memory, perhaps its todo with my map. But in any case /proc/vmcore remains zero.

read_from_oldmem: error: pfn (32761) > saved_max_pfn (31744)
Kdump: vmcore not initialized

Below is your patch rediffed for Linus latest tree.
And below that is the boot log for my first and crash kernels, including the EFI map. Let me know if you need some more information or would like me to run any additional tests.

Signature · 351 lines
Horms
  H: http://www.vergenet.net/~horms/
  W: http://www.valinux.co.jp/en/

	Please apply the following patch to make saved_max_pfn point to
max_pfn of entire system.

Signed-off-by: Zou Nan hai <[email protected]>

Updated for recent changes in Linus' tree.
But it doesn't seem to work as desired on my system :(

Nacked-by: Simon Horman <[email protected]>
Index: linux-2.6/arch/ia64/kernel/efi.c
=================================--- linux-2.6.orig/arch/ia64/kernel/efi.c	2007-02-08 16:06:02.000000000 +0900
+++ linux-2.6/arch/ia64/kernel/efi.c	2007-02-08 16:06:40.000000000 +0900
@@ -21,6 +21,7 @@
  *	Skip non-WB memory and ignore empty memory ranges.
  */
 #include <linux/module.h>
+#include <linux/bootmem.h>
 #include <linux/kernel.h>
 #include <linux/init.h>
 #include <linux/types.h>
@@ -1010,6 +1011,11 @@
 		} else
 			ae = efi_md_end(md);
 
+#ifdef CONFIG_CRASH_DUMP
+		/* saved_max_pfn should ignore max_addr= command line arg */ 
+		if (saved_max_pfn < (ae >> PAGE_SHIFT))
+			saved_max_pfn = (ae >> PAGE_SHIFT);
+#endif
 		/* keep within max_addr= and min_addr= command line arg */
 		as = max(as, min_addr);
 		ae = min(ae, max_addr);
Index: linux-2.6/arch/ia64/mm/contig.c
=================================--- linux-2.6.orig/arch/ia64/mm/contig.c	2007-02-08 16:06:02.000000000 +0900
+++ linux-2.6/arch/ia64/mm/contig.c	2007-02-08 16:06:40.000000000 +0900
@@ -197,11 +197,6 @@
 
 	find_initrd();
 
-#ifdef CONFIG_CRASH_DUMP
-	/* If we are doing a crash dump, we still need to know the real mem
-	 * size before original memory map is reset. */
-	saved_max_pfn = max_pfn;
-#endif
 }
 
 #ifdef CONFIG_SMP
Index: linux-2.6/arch/ia64/mm/discontig.c
=================================--- linux-2.6.orig/arch/ia64/mm/discontig.c	2007-02-08 16:06:23.000000000 +0900
+++ linux-2.6/arch/ia64/mm/discontig.c	2007-02-08 16:06:40.000000000 +0900
@@ -478,12 +478,6 @@
 	max_pfn = max_low_pfn;
 
 	find_initrd();
-
-#ifdef CONFIG_CRASH_DUMP
-	/* If we are doing a crash dump, we still need to know the real mem
-	 * size before original memory map is reset. */
-        saved_max_pfn = max_pfn;
-#endif
 }
 
 #ifdef CONFIG_SMP

ELILO
Uncompressing Linux... done
Loading initrd people/horms/initramfs_data.cpio.gz...done
Linux version 2.6.20-kexec-g5331be09-dirty ([email protected]) (gcc version 3.4.5) #18 Thu Feb 8 16:26:47 JST 2007
EFI v1.10 by INTEL: SALsystab=0x7fe54980 ACPI=0x7ff99000 ACPI 2.0=0x7ff98000 MPS=0x7ff97000 SMBIOS=0xf0000
mem00: type=4, attr=0x9, range=[0x0000000000000000-0x0000000000001000) (0MB)
mem01: type=7, attr=0x9, range=[0x0000000000001000-0x0000000000007000) (0MB)
mem02: type=4, attr=0x9, range=[0x0000000000007000-0x0000000000009000) (0MB)
mem03: type=7, attr=0x9, range=[0x0000000000009000-0x0000000000082000) (0MB)
mem04: type=6, attr=0x8000000000000009, range=[0x0000000000082000-0x0000000000084000) (0MB)
mem05: type=7, attr=0x9, range=[0x0000000000084000-0x0000000000085000) (0MB)
mem06: type=4, attr=0x9, range=[0x0000000000085000-0x00000000000a0000) (0MB)
mem07: type=5, attr=0x8000000000000009, range=[0x00000000000c0000-0x0000000000100000) (0MB)
mem08: type=7, attr=0xb, range=[0x0000000000100000-0x0000000004000000) (63MB)
mem09: type=2, attr=0xb, range=[0x0000000004000000-0x0000000004644000) (6MB)
mem10: type=7, attr=0xb, range=[0x0000000004644000-0x000000000ffc0000) (185MB)
mem11: type=4, attr=0xb, range=[0x000000000ffc0000-0x0000000010000000) (0MB)
mem12: type=7, attr=0xb, range=[0x0000000010000000-0x000000007af6c000) (1711MB)
mem13: type=2, attr=0xb, range=[0x000000007af6c000-0x000000007c8d2000) (25MB)
mem14: type=1, attr=0xb, range=[0x000000007c8d2000-0x000000007c92e000) (0MB)
mem15: type=2, attr=0xb, range=[0x000000007c92e000-0x000000007c938000) (0MB)
mem16: type=1, attr=0xb, range=[0x000000007c938000-0x000000007c97e000) (0MB)
mem17: type=7, attr=0xb, range=[0x000000007c97e000-0x000000007ce16000) (4MB)
mem18: type=4, attr=0xb, range=[0x000000007ce16000-0x000000007ce1c000) (0MB)
mem19: type=7, attr=0xb, range=[0x000000007ce1c000-0x000000007ce20000) (0MB)
mem20: type=4, attr=0xb, range=[0x000000007ce20000-0x000000007ce22000) (0MB)
mem21: type=7, attr=0xb, range=[0x000000007ce22000-0x000000007ce2a000) (0MB)
mem22: type=4, attr=0xb, range=[0x000000007ce2a000-0x000000007d001000) (1MB)
mem23: type=7, attr=0xb, range=[0x000000007d001000-0x000000007d002000) (0MB)
mem24: type=4, attr=0xb, range=[0x000000007d002000-0x000000007d004000) (0MB)
mem25: type=7, attr=0xb, range=[0x000000007d004000-0x000000007d026000) (0MB)
mem26: type=4, attr=0xb, range=[0x000000007d026000-0x000000007d068000) (0MB)
mem27: type=7, attr=0xb, range=[0x000000007d068000-0x000000007d069000) (0MB)
mem28: type=4, attr=0xb, range=[0x000000007d069000-0x000000007d37e000) (3MB)
mem29: type=7, attr=0xb, range=[0x000000007d37e000-0x000000007d700000) (3MB)
mem30: type=3, attr=0xb, range=[0x000000007d700000-0x000000007d77e000) (0MB)
mem31: type=7, attr=0xb, range=[0x000000007d77e000-0x000000007d8b4000) (1MB)
mem32: type=6, attr=0x8000000000000009, range=[0x000000007d8b4000-0x000000007d900000) (0MB)
mem33: type=3, attr=0xb, range=[0x000000007d900000-0x000000007f980000) (32MB)
mem34: type=7, attr=0xb, range=[0x000000007f980000-0x000000007fa00000) (0MB)
mem35: type=5, attr=0x8000000000000009, range=[0x000000007fa00000-0x000000007fe00000) (4MB)
mem36: type, attr=0x8000000000000009, range=[0x000000007fe00000-0x000000007fe48000) (0MB)
mem37: type=5, attr=0x8000000000000009, range=[0x000000007fe48000-0x000000007fea0000) (0MB)
mem38: type=7, attr=0xb, range=[0x000000007fea0000-0x000000007feda000) (0MB)
mem39: type=5, attr=0x8000000000000009, range=[0x000000007feda000-0x000000007ff46000) (0MB)
mem40: type=6, attr=0x8000000000000009, range=[0x000000007ff46000-0x0000000080000000) (0MB)
mem41: type, attr=0x1, range=[0x00000000fe000000-0x00000000ff000000) (16MB)
mem42: type=6, attr=0x8000000000000001, range=[0x00000000ff000000-0x0000000100000000) (16MB)
mem43: type, attr=0x8000000000000001, range=[0x00000ffff8000000-0x00000ffffc000000) (64MB)
mem44: type, attr=0x8000000000000001, range=[0x00000ffffc000000-0x0000100000000000) (64MB)
booting generic kernel on platform dig
Early serial console at I/O port 0x2f8 (options '115200')
Initial ramdisk at: 0xe00000007af72000 (9789052 bytes)
SAL 3.20: Intel Corp                       SR870BH2                         version 3.0
SAL Platform features: BusLock
iosapic_system_init: Disabling PC-AT compatible 8259 interrupts
ACPI: Local APIC address c0000000fee00000
ACPI: [APIC:0x07] ignored 1 entries of 2 found
PLATFORM int CPEI (0x3): GSI 22 (level, low) -> CPU 0 (0x0100) vector 30
register_intr: changing vector 39 from IO-SAPIC-edge to IO-SAPIC-level
1 CPUs available, 1 CPUs total
MCA related initialization done
Virtual mem_map starts at 0xa0007fffff900000
Zone PFN ranges:
  DMA          1024 ->   262144
  Normal     262144 ->   262144
early_node_map[3] active PFN ranges
    0:     1024 ->   128557
    0:   128576 ->   130688
    0:   130984 ->   130998
Built 1 zonelists.  Total pages: 129215
Kernel command line: BOOT_IMAGE=net0:ia64/people/horms/vmlinux.gz  phys_efi console=uart,io,0x2f8,115200 crashkernel%6M loglevel=7 ro
PID hash table entries: 4096 (order: 12, 32768 bytes)
Console: colour VGA+ 80x25
Placing 64MB software IO TLB between 0x4644000 - 0x8644000
Memory: 1722416k/1796368k available (3010k code, 352128k reserved, 2124k data, 640k init)
McKinley Errata 9 workaround not needed; disabling it
Dentry cache hash table entries: 262144 (order: 7, 2097152 bytes)
Inode-cache hash table entries: 131072 (order: 6, 1048576 bytes)
Mount-cache hash table entries: 1024
ACPI: Core revision 20060707
DMI 2.3 present.
ACPI: bus type pci registered
ACPI: Interpreter enabled
ACPI: Using IOSAPIC for interrupt routing
ACPI: PCI Root Bridge [PCI0] (0000:00)
PCI quirk: region 0c00-0c7f claimed by ICH4 ACPI/GPIO/TCO
PCI quirk: region 0500-053f claimed by ICH4 GPIO
ACPI: PCI Root Bridge [PCI1] (0000:02)
ACPI: PCI Root Bridge [PCI2] (0000:05)
ACPI: Device [CSFF] status [00000008]: functional but not present; setting present
ACPI: PCI Root Bridge [CSFF] (0000:ff)
Linux Plug and Play Support v0.97 (c) Adam Belay
pnp: PnP ACPI init
pnp: PnP ACPI: found 12 devices
checking if image is initramfs... it is
Freeing initrd memory: 9536kB freed
perfmon: version 2.0 IRQ 238
perfmon: Itanium 2 PMU detected, 16 PMCs, 18 PMDs, 4 counters (47 bits)
PAL Information Facility v0.5
perfmon: added sampling format default_format
perfmon_default_smpl: default_format v2.0 registered
io scheduler noop registered
io scheduler anticipatory registered (default)
Serial: 8250/16550 driver $Revision: 1.90 $ 4 ports, IRQ sharing enabled
00:08: ttyS0 at I/O 0x3f8 (irq = 44) is a 16550A
00:09: ttyS1 at I/O 0x2f8 (irq = 45) is a 16550A
RAMDISK driver initialized: 16 RAM disks of 4096K size 1024 blocksize
mice: PS/2 mouse device common for all mice
EFI Variables Facility v0.08 2004-May-17
Adding console on ttyS1 at I/O port 0x2f8 (options '115200')
Freeing unused kernel memory: 640kB freed
init started:  BusyBox v1.2.1 (2006.09.23-05:46+0000) multi-call binary
Starting pid 772, console /dev/console: '/etc/init.d/rcS'
ifconfig: socket: Function not implemented
ifconfig: No usable address families found.
ifconfig: socket: Function not implemented
Starting pid 890, console /dev/console: '/bin/sh'


BusyBox v1.2.1 (2006.09.23-05:46+0000) Built-in shell (ash)
Enter 'help' for a list of built-in commands.

/ # do_kdump
Create ramdisk
Load kernel and ramdisk
kexec -p "/boot/vmlinux-ia64-kdump.gz" --initrd=/tmp/initramfs_data.cpio \
    --append="phys_efi clock=pit ip=on apm=power-off console=tty0 loglevel=7 console=uart,io,0x2f8,115200n8 init 1 irqpoll maxcpus=1" 
Triggering KdumpSysRq : Trigger a crashdump
Linux version 2.6.20-kexec-g5331be09-dirty ([email protected]) (gcc version 3.4.5) #18 Thu Feb 8 16:26:47 JST 2007
Ignoring memory below 256MB
Ignoring memory above 512MB
EFI v1.10 by INTEL: SALsystab=0x7fe54980 ACPI=0x7ff99000 ACPI 2.0=0x7ff98000 MPS=0x7ff97000 SMBIOS=0xf0000
mem00: type=4, attr=0x9, range=[0x0000000000000000-0x0000000000001000) (0MB)
mem01: type=7, attr=0x9, range=[0x0000000000001000-0x0000000000007000) (0MB)
mem02: type=4, attr=0x9, range=[0x0000000000007000-0x0000000000009000) (0MB)
mem03: type=7, attr=0x9, range=[0x0000000000009000-0x0000000000082000) (0MB)
mem04: type=6, attr=0x8000000000000009, range=[0x0000000000082000-0x0000000000084000) (0MB)
mem05: type=7, attr=0x9, range=[0x0000000000084000-0x0000000000085000) (0MB)
mem06: type=4, attr=0x9, range=[0x0000000000085000-0x00000000000a0000) (0MB)
mem07: type=5, attr=0x8000000000000009, range=[0x00000000000c0000-0x0000000000100000) (0MB)
mem08: type=7, attr=0xb, range=[0x0000000000100000-0x0000000004000000) (63MB)
mem09: type=7, attr=0xb, range=[0x0000000004000000-0x0000000004644000) (6MB)
mem10: type=7, attr=0xb, range=[0x0000000004644000-0x000000000ffc0000) (185MB)
mem11: type=4, attr=0xb, range=[0x000000000ffc0000-0x0000000010000000) (0MB)
mem12: type=2, attr=0xb, range=[0x0000000010000000-0x0000000010490000) (4MB)
mem13: type=2, attr=0xb, range=[0x0000000010490000-0x00000000104a0000) (0MB)
mem14: type=2, attr=0xb, range=[0x00000000104a0000-0x0000000010650000) (1MB)
mem15: type=7, attr=0xb, range=[0x0000000010650000-0x000000001ffe4000) (249MB)
mem16: type=8, attr=0x5555555555555555, range=[0x000000001ffe4000-0x600000001fff2000) (6597069766656MB)
mem17: type=7, attr=0x5555555555555555, range=[0x600000001fff2350-0x5151515151514350) (16583222432533MB)
efi_get_pal_addr: no PAL-code memory-descriptor found
No I/O port range found in EFI memory map, falling back to AR.KR0 (0xffffc000000)
booting generic kernel on platform dig
Early serial console at I/O port 0x2f8 (options '115200n8')
Initial ramdisk at: 0xe00000001f544000 (10977792 bytes)
SAL 3.20: Intel Corp                       SR870BH2                         version 3.0
SAL Platform features: BusLock
efi_get_pal_addr: no PAL-code memory-descriptor found
iosapic_system_init: Disabling PC-AT compatible 8259 interrupts
ACPI: Local APIC address c0000000fee00000
ACPI: [APIC:0x07] ignored 1 entries of 2 found
PLATFORM int CPEI (0x3): GSI 22 (level, low) -> CPU 0 (0x0100) vector 30
register_intr: changing vector 39 from IO-SAPIC-edge to IO-SAPIC-level
1 CPUs available, 1 CPUs total
MCA related initialization done
Virtual mem_map starts at 0xa0007fffffc80000
Zone PFN ranges:
  DMA         16384 ->   262144
  Normal     262144 ->   262144
early_node_map[1] active PFN ranges
    0:    16384 ->    31744
Built 1 zonelists.  Total pages: 15308
Kernel command line: phys_efi clock=pit ip=on apm=power-off console=tty0 loglevel=7 console=uart,io,0x2f8,115200n8 init 1 irqpoll maxcpus=1 elfcorehdrR4176K max_addrQ2M min_addr%6M
Warning! clock= boot option is deprecated. Use clocksource=xyz
Misrouted IRQ fixup and polling support enabled
This may significantly impact system performance
PID hash table entries: 1024 (order: 10, 8192 bytes)
Console: colour dummy device 80x25
Linux version 2.6.20-kexec-g5331be09-dirty ([email protected]) (gcc version 3.4.5) #18 Thu Feb 8 16:26:47 JST 2007
Ignoring memory below 256MB
Ignoring memory above 512MB
EFI v1.10 by INTEL: SALsystab=0x7fe54980 ACPI=0x7ff99000 ACPI 2.0=0x7ff98000 MPS=0x7ff97000 SMBIOS=0xf0000
mem00: type=4, attr=0x9, range=[0x0000000000000000-0x0000000000001000) (0MB)
mem01: type=7, attr=0x9, range=[0x0000000000001000-0x0000000000007000) (0MB)
mem02: type=4, attr=0x9, range=[0x0000000000007000-0x0000000000009000) (0MB)
mem03: type=7, attr=0x9, range=[0x0000000000009000-0x0000000000082000) (0MB)
mem04: type=6, attr=0x8000000000000009, range=[0x0000000000082000-0x0000000000084000) (0MB)
mem05: type=7, attr=0x9, range=[0x0000000000084000-0x0000000000085000) (0MB)
mem06: type=4, attr=0x9, range=[0x0000000000085000-0x00000000000a0000) (0MB)
mem07: type=5, attr=0x8000000000000009, range=[0x00000000000c0000-0x0000000000100000) (0MB)
mem08: type=7, attr=0xb, range=[0x0000000000100000-0x0000000004000000) (63MB)
mem09: type=7, attr=0xb, range=[0x0000000004000000-0x0000000004644000) (6MB)
mem10: type=7, attr=0xb, range=[0x0000000004644000-0x000000000ffc0000) (185MB)
mem11: type=4, attr=0xb, range=[0x000000000ffc0000-0x0000000010000000) (0MB)
mem12: type=2, attr=0xb, range=[0x0000000010000000-0x0000000010490000) (4MB)
mem13: type=2, attr=0xb, range=[0x0000000010490000-0x00000000104a0000) (0MB)
mem14: type=2, attr=0xb, range=[0x00000000104a0000-0x0000000010650000) (1MB)
mem15: type=7, attr=0xb, range=[0x0000000010650000-0x000000001ffe4000) (249MB)
mem16: type=8, attr=0x5555555555555555, range=[0x000000001ffe4000-0x600000001fff2000) (6597069766656MB)
mem17: type=7, attr=0x5555555555555555, range=[0x600000001fff2350-0x5151515151514350) (16583222432533MB)
efi_get_pal_addr: no PAL-code memory-descriptor found
No I/O port range found in EFI memory map, falling back to AR.KR0 (0xffffc000000)
booting generic kernel on platform dig
Early serial console at I/O port 0x2f8 (options '115200n8')
Initial ramdisk at: 0xe00000001f544000 (10977792 bytes)
SAL 3.20: Intel Corp                       SR870BH2                         version 3.0
SAL Platform features: BusLock
efi_get_pal_addr: no PAL-code memory-descriptor found
iosapic_system_init: Disabling PC-AT compatible 8259 interrupts
ACPI: Local APIC address c0000000fee00000
ACPI: [APIC:0x07] ignored 1 entries of 2 found
PLATFORM int CPEI (0x3): GSI 22 (level, low) -> CPU 0 (0x0100) vector 30
register_intr: changing vector 39 from IO-SAPIC-edge to IO-SAPIC-level
1 CPUs available, 1 CPUs total
MCA related initialization done
Virtual mem_map starts at 0xa0007fffffc80000
Zone PFN ranges:
  DMA         16384 ->   262144
  Normal     262144 ->   262144
early_node_map[1] active PFN ranges
    0:    16384 ->    31744
Built 1 zonelists.  Total pages: 15308
Kernel command line: phys_efi clock=pit ip=on apm=power-off console=tty0 loglevel=7 console=uart,io,0x2f8,115200n8 init 1 irqpoll maxcpus=1 elfcorehdrR4176K max_addrQ2M min_addr%6M
Warning! clock= boot option is deprecated. Use clocksource=xyz
Misrouted IRQ fixup and polling support enabled
This may significantly impact system performance
PID hash table entries: 1024 (order: 10, 8192 bytes)
Console: colour dummy device 80x25
Placing 64MB software IO TLB between 0x107f8000 - 0x147f8000
Memory: 171664k/239328k available (3010k code, 74096k reserved, 2124k data, 640k init)
McKinley Errata 9 workaround not needed; disabling it
Dentry cache hash table entries: 32768 (order: 4, 262144 bytes)
Inode-cache hash table entries: 16384 (order: 3, 131072 bytes)
Mount-cache hash table entries: 1024
ACPI: Core revision 20060707
DMI 2.3 present.
ACPI: bus type pci registered
ACPI: Interpreter enabled
ACPI: Using IOSAPIC for interrupt routing
ACPI: PCI Root Bridge [PCI0] (0000:00)
PCI quirk: region 0c00-0c7f claimed by ICH4 ACPI/GPIO/TCO
PCI quirk: region 0500-053f claimed by ICH4 GPIO
ACPI: PCI Root Bridge [PCI1] (0000:02)
ACPI: PCI Root Bridge [PCI2] (0000:05)
ACPI: Device [CSFF] status [00000008]: functional but not present; setting present
ACPI: PCI Root Bridge [CSFF] (0000:ff)
Linux Plug and Play Support v0.97 (c) Adam Belay
pnp: PnP ACPI init
pnp: PnP ACPI: found 12 devices
checking if image is initramfs... it is
Freeing initrd memory: 10720kB freed
perfmon: version 2.0 IRQ 238
perfmon: Itanium 2 PMU detected, 16 PMCs, 18 PMDs, 4 counters (47 bits)
PAL Information Facility v0.5
perfmon: added sampling format default_format
perfmon_default_smpl: default_format v2.0 registered
read_from_oldmem: error: pfn (32761) > saved_max_pfn (31744)
Kdump: vmcore not initialized
io scheduler noop registered
io scheduler anticipatory registered (default)
Serial: 8250/16550 driver $Revision: 1.90 $ 4 ports, IRQ sharing enabled
00:08: ttyS0 at I/O 0x3f8 (irq = 44) is a 16550A
00:09: ttyS1 at I/O 0x2f8 (irq = 45) is a 16550A
RAMDISK driver initialized: 16 RAM disks of 4096K size 1024 blocksize
mice: PS/2 mouse device common for all mice
EFI Variables Facility v0.08 2004-May-17
Adding console on ttyS1 at I/O port 0x2f8 (options '115200n8')
Freeing unused kernel memory: 640kB freed
init started:  BusyBox v1.2.1 (2006.09.23-05:46+0000) multi-call binary
Starting pid 772, console /dev/console: '/etc/init.d/rcS'
ifconfig: socket: Function not implemented
ifconfig: No usable address families found.
ifconfig: socket: Function not implemented
Starting pid 953, console /dev/console: '/bin/sh'


BusyBox v1.2.1 (2006.09.23-05:46+0000) Built-in shell (ash)
Enter 'help' for a list of built-in commands.

/ # ls -l /proc/vmcore 
-r--------    1 0        0               0 /proc/vmcore
/ # 
#6 lore raw
In reply to Horms
Quoted 341 lines

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Horms
Sent: 2007Äê2ÔÂ8ÈÕ 15:37
To: Zou, Nanhai
Cc: [email protected]; fastboot; Linux-IA64; Luck, Tony
Subject: Re: Zero size /proc/vmcore on ia64

Quoted 58 lines · Zou Nan hai
Quoted 57 lines · Vivek Goyal
Quoted 22 lines · Horms
Quoted 11 lines · Nanhai

When crash dump kernel tries to access memory of first kernel above saved_max_pfn of him, read_from_oldmem will refuse that read.

That result an empty vmcore file. change saved_max_pfn to unsigned long(-1) will fix this issue.

However since memory ranges in vmcore is pre defined from /proc/iomem of first kernel, why do we still need to add an extra check in vmcore.c

Hi Nan-hai,

sorry that I did not get back to you about the information you requested about my system, I guess you have managed to reproduce the problem none the less.

I can confirm that removing the max_pfn check in vmcore.c does indeed give /proc/vmcore a non-zero (and presumably correct) size.

Quoted 7 lines

I wonder if the problem is that saved_max_pfn is being incorectly calculated on ia64. That it is being set to the max_pfn of the crash kernel (i.e. in the crashkernel=X@Y area), rather than the max_pfn of the physical memory of the system, which seems more sensible as the purpose of vmcore is to read memory outside of the crashkernel=X@Y area.

Hi Horms/Nan-hai,

Horms, you are right. saved_max_pfn is needed to know that second kernel is not trying to read any memory which is not present or was not being used by the crashed kernel at all. That's why in i386/x86_64, during early boot saved_max_pfn, is calculated the memory map passed to the second kernel. This memory map is passed to second kernel by kexec through parameter segment. So effectively saved_max_pfn will be set to max_pfn of crashed kernel.

Now this memory map is overwritten with user defined one which is basically the memory second kernel can use to boot and max_pfn now will be maximum pfn crash kernel can use.

Quoted 4 lines

You may be right that we can just remove the check all together, though perhaps it is there for the case where the range information in the vmcode are corrupted. Then again, should we care about this?

I think we should not remove this check because even to parse the info passed in ELF headers, you need to first read the ELF headers from crashed kernel's memory. So if some programming error has passed wrong location of ELF headers (elfcoreheader= invalid location) then we might try reading the elf header from a non-existing physical page frame.

Quoted 7 lines
Quoted 4 lines

So the right way should be to set saved_max_pfn with right value before it is memory map is over-written with user defined memory map.

This is reasonable.
So please apply the following patch to make
saved_max_pfn point to max_pfn of entire system.

Hi Nanhai,

Although I agree with the gist of your patch, unfortunately it does not work on my system. Perhaps this is because I use discontig memory, perhaps its todo with my map. But in any case /proc/vmcore remains zero.

read_from_oldmem: error: pfn (32761) > saved_max_pfn (31744)
Kdump: vmcore not initialized

Below is your patch rediffed for Linus latest tree.
And below that is the boot log for my first and crash kernels, including the EFI map. Let me know if you need some more information or would like me to run any additional tests.

--
Horms
H: http://www.vergenet.net/~horms/
W: http://www.valinux.co.jp/en/

Please apply the following patch to make saved_max_pfn point to
max_pfn of entire system.

Updated for recent changes in Linus' tree. But it doesn't seem to work as desired on my system :(

Index: linux-2.6/arch/ia64/kernel/efi.c
=================================> --- linux-2.6.orig/arch/ia64/kernel/efi.c 2007-02-08 16:06:02.000000000
+0900
+++ linux-2.6/arch/ia64/kernel/efi.c 2007-02-08 16:06:40.000000000 +0900

(quoted diff) +6 -0
@@ -21,6 +21,7 @@
  *	Skip non-WB memory and ignore empty memory ranges.
  */
 #include <linux/module.h>
+#include <linux/bootmem.h>
 #include <linux/kernel.h>
 #include <linux/init.h>
 #include <linux/types.h>
@@ -1010,6 +1011,11 @@
 		} else
 			ae = efi_md_end(md);
 
+#ifdef CONFIG_CRASH_DUMP
+		/* saved_max_pfn should ignore max_addr= command line arg */
+		if (saved_max_pfn < (ae >> PAGE_SHIFT))
+			saved_max_pfn = (ae >> PAGE_SHIFT);
+#endif
 		/* keep within max_addr= and min_addr= command line arg */
 		as = max(as, min_addr);
 		ae = min(ae, max_addr);

Index: linux-2.6/arch/ia64/mm/contig.c
=================================> --- linux-2.6.orig/arch/ia64/mm/contig.c 2007-02-08 16:06:02.000000000
+0900
+++ linux-2.6/arch/ia64/mm/contig.c 2007-02-08 16:06:40.000000000 +0900

(quoted diff) +0 -5
@@ -197,11 +197,6 @@
 
 	find_initrd();
 
-#ifdef CONFIG_CRASH_DUMP
-	/* If we are doing a crash dump, we still need to know the real mem
-	 * size before original memory map is reset. */
-	saved_max_pfn = max_pfn;
-#endif
 }
 
 #ifdef CONFIG_SMP

Index: linux-2.6/arch/ia64/mm/discontig.c
=================================> --- linux-2.6.orig/arch/ia64/mm/discontig.c 2007-02-08
16:06:23.000000000 +0900
+++ linux-2.6/arch/ia64/mm/discontig.c 2007-02-08 16:06:40.000000000
+0900

arch/ia64/mm/discontig.c +0 -6
@@ -478,12 +478,6 @@
 	max_pfn = max_low_pfn;
 
 	find_initrd();
-
-#ifdef CONFIG_CRASH_DUMP
-	/* If we are doing a crash dump, we still need to know the real mem
-	 * size before original memory map is reset. */
-        saved_max_pfn = max_pfn;
-#endif
 }
 
 #ifdef CONFIG_SMP

ELILO
Uncompressing Linux... done
Loading initrd people/horms/initramfs_data.cpio.gz...done
Linux version 2.6.20-kexec-g5331be09-dirty ([email protected]) (gcc version 3.4.5) #18 Thu Feb 8 16:26:47 JST 2007
EFI v1.10 by INTEL: SALsystab=0x7fe54980 ACPI=0x7ff99000 ACPI 2.0=0x7ff98000 MPS=0x7ff97000 SMBIOS=0xf0000
mem00: type=4, attr=0x9, range=[0x0000000000000000-0x0000000000001000) (0MB)
mem01: type=7, attr=0x9, range=[0x0000000000001000-0x0000000000007000) (0MB)
mem02: type=4, attr=0x9, range=[0x0000000000007000-0x0000000000009000) (0MB)
mem03: type=7, attr=0x9, range=[0x0000000000009000-0x0000000000082000) (0MB)
mem04: type=6, attr=0x8000000000000009, range=[0x0000000000082000-0x0000000000084000) (0MB)
mem05: type=7, attr=0x9, range=[0x0000000000084000-0x0000000000085000) (0MB)
mem06: type=4, attr=0x9, range=[0x0000000000085000-0x00000000000a0000) (0MB)
mem07: type=5, attr=0x8000000000000009, range=[0x00000000000c0000-0x0000000000100000) (0MB)
mem08: type=7, attr=0xb, range=[0x0000000000100000-0x0000000004000000) (63MB)
mem09: type=2, attr=0xb, range=[0x0000000004000000-0x0000000004644000) (6MB)
mem10: type=7, attr=0xb, range=[0x0000000004644000-0x000000000ffc0000) (185MB)
mem11: type=4, attr=0xb, range=[0x000000000ffc0000-0x0000000010000000) (0MB)
mem12: type=7, attr=0xb, range=[0x0000000010000000-0x000000007af6c000) (1711MB)
mem13: type=2, attr=0xb, range=[0x000000007af6c000-0x000000007c8d2000) (25MB)
mem14: type=1, attr=0xb, range=[0x000000007c8d2000-0x000000007c92e000) (0MB)
mem15: type=2, attr=0xb, range=[0x000000007c92e000-0x000000007c938000) (0MB)
mem16: type=1, attr=0xb, range=[0x000000007c938000-0x000000007c97e000) (0MB)
mem17: type=7, attr=0xb, range=[0x000000007c97e000-0x000000007ce16000) (4MB)
mem18: type=4, attr=0xb, range=[0x000000007ce16000-0x000000007ce1c000) (0MB)
mem19: type=7, attr=0xb, range=[0x000000007ce1c000-0x000000007ce20000) (0MB)
mem20: type=4, attr=0xb, range=[0x000000007ce20000-0x000000007ce22000) (0MB)
mem21: type=7, attr=0xb, range=[0x000000007ce22000-0x000000007ce2a000) (0MB)
mem22: type=4, attr=0xb, range=[0x000000007ce2a000-0x000000007d001000) (1MB)
mem23: type=7, attr=0xb, range=[0x000000007d001000-0x000000007d002000) (0MB)
mem24: type=4, attr=0xb, range=[0x000000007d002000-0x000000007d004000) (0MB)
mem25: type=7, attr=0xb, range=[0x000000007d004000-0x000000007d026000) (0MB)
mem26: type=4, attr=0xb, range=[0x000000007d026000-0x000000007d068000) (0MB)
mem27: type=7, attr=0xb, range=[0x000000007d068000-0x000000007d069000) (0MB)
mem28: type=4, attr=0xb, range=[0x000000007d069000-0x000000007d37e000) (3MB)
mem29: type=7, attr=0xb, range=[0x000000007d37e000-0x000000007d700000) (3MB)
mem30: type=3, attr=0xb, range=[0x000000007d700000-0x000000007d77e000) (0MB)
mem31: type=7, attr=0xb, range=[0x000000007d77e000-0x000000007d8b4000) (1MB)
mem32: type=6, attr=0x8000000000000009, range=[0x000000007d8b4000-0x000000007d900000) (0MB)
mem33: type=3, attr=0xb, range=[0x000000007d900000-0x000000007f980000) (32MB)
mem34: type=7, attr=0xb, range=[0x000000007f980000-0x000000007fa00000) (0MB)
mem35: type=5, attr=0x8000000000000009, range=[0x000000007fa00000-0x000000007fe00000) (4MB)
mem36: type, attr=0x8000000000000009, range=[0x000000007fe00000-0x000000007fe48000) (0MB)
mem37: type=5, attr=0x8000000000000009, range=[0x000000007fe48000-0x000000007fea0000) (0MB)
mem38: type=7, attr=0xb, range=[0x000000007fea0000-0x000000007feda000) (0MB)
mem39: type=5, attr=0x8000000000000009, range=[0x000000007feda000-0x000000007ff46000) (0MB)
mem40: type=6, attr=0x8000000000000009, range=[0x000000007ff46000-0x0000000080000000) (0MB)
mem41: type, attr=0x1, range=[0x00000000fe000000-0x00000000ff000000) (16MB)
mem42: type=6, attr=0x8000000000000001, range=[0x00000000ff000000-0x0000000100000000) (16MB)
mem43: type, attr=0x8000000000000001, range=[0x00000ffff8000000-0x00000ffffc000000) (64MB)
mem44: type, attr=0x8000000000000001, range=[0x00000ffffc000000-0x0000100000000000) (64MB)
booting generic kernel on platform dig
Early serial console at I/O port 0x2f8 (options '115200')
Initial ramdisk at: 0xe00000007af72000 (9789052 bytes)
SAL 3.20: Intel Corp SR870BH2
version 3.0
SAL Platform features: BusLock
iosapic_system_init: Disabling PC-AT compatible 8259 interrupts
ACPI: Local APIC address c0000000fee00000
ACPI: [APIC:0x07] ignored 1 entries of 2 found
PLATFORM int CPEI (0x3): GSI 22 (level, low) -> CPU 0 (0x0100) vector 30
register_intr: changing vector 39 from IO-SAPIC-edge to IO-SAPIC-level 1 CPUs available, 1 CPUs total
MCA related initialization done
Virtual mem_map starts at 0xa0007fffff900000
Zone PFN ranges:
DMA 1024 -> 262144
Normal 262144 -> 262144
early_node_map[3] active PFN ranges

    0:     1024 ->   128557
    0:   128576 ->   130688
    0:   130984 ->   130998

Built 1 zonelists. Total pages: 129215
Kernel command line: BOOT_IMAGE=net0:ia64/people/horms/vmlinux.gz phys_efi console=uart,io,0x2f8,115200 crashkernel%6M loglevel=7 ro
PID hash table entries: 4096 (order: 12, 32768 bytes)
Console: colour VGA+ 80x25
Placing 64MB software IO TLB between 0x4644000 - 0x8644000
Memory: 1722416k/1796368k available (3010k code, 352128k reserved, 2124k data, 640k init)
McKinley Errata 9 workaround not needed; disabling it
Dentry cache hash table entries: 262144 (order: 7, 2097152 bytes) Inode-cache hash table entries: 131072 (order: 6, 1048576 bytes) Mount-cache hash table entries: 1024
ACPI: Core revision 20060707
DMI 2.3 present.
ACPI: bus type pci registered
ACPI: Interpreter enabled
ACPI: Using IOSAPIC for interrupt routing
ACPI: PCI Root Bridge [PCI0] (0000:00)
PCI quirk: region 0c00-0c7f claimed by ICH4 ACPI/GPIO/TCO
PCI quirk: region 0500-053f claimed by ICH4 GPIO
ACPI: PCI Root Bridge [PCI1] (0000:02)
ACPI: PCI Root Bridge [PCI2] (0000:05)
ACPI: Device [CSFF] status [00000008]: functional but not present; setting present
ACPI: PCI Root Bridge [CSFF] (0000:ff)
Linux Plug and Play Support v0.97 (c) Adam Belay
pnp: PnP ACPI init
pnp: PnP ACPI: found 12 devices
checking if image is initramfs... it is
Freeing initrd memory: 9536kB freed
perfmon: version 2.0 IRQ 238
perfmon: Itanium 2 PMU detected, 16 PMCs, 18 PMDs, 4 counters (47 bits) PAL Information Facility v0.5
perfmon: added sampling format default_format
perfmon_default_smpl: default_format v2.0 registered
io scheduler noop registered
io scheduler anticipatory registered (default)
Serial: 8250/16550 driver $Revision: 1.90 $ 4 ports, IRQ sharing enabled 00:08: ttyS0 at I/O 0x3f8 (irq = 44) is a 16550A
00:09: ttyS1 at I/O 0x2f8 (irq = 45) is a 16550A
RAMDISK driver initialized: 16 RAM disks of 4096K size 1024 blocksize
mice: PS/2 mouse device common for all mice
EFI Variables Facility v0.08 2004-May-17
Adding console on ttyS1 at I/O port 0x2f8 (options '115200') Freeing unused kernel memory: 640kB freed
init started: BusyBox v1.2.1 (2006.09.23-05:46+0000) multi-call binary Starting pid 772, console /dev/console: '/etc/init.d/rcS'
ifconfig: socket: Function not implemented
ifconfig: No usable address families found.
ifconfig: socket: Function not implemented
Starting pid 890, console /dev/console: '/bin/sh'

BusyBox v1.2.1 (2006.09.23-05:46+0000) Built-in shell (ash) Enter 'help' for a list of built-in commands.

/ # do_kdump
Create ramdisk
Load kernel and ramdisk
kexec -p "/boot/vmlinux-ia64-kdump.gz" --initrd=/tmp/initramfs_data.cpio \
--append="phys_efi clock=pit ip=on apm=power-off console=tty0 loglevel=7
console=uart,io,0x2f8,115200n8 init 1 irqpoll maxcpus=1" Triggering KdumpSysRq : Trigger a crashdump
Linux version 2.6.20-kexec-g5331be09-dirty ([email protected]) (gcc version 3.4.5) #18 Thu Feb 8 16:26:47 JST 2007
Ignoring memory below 256MB
Ignoring memory above 512MB
EFI v1.10 by INTEL: SALsystab=0x7fe54980 ACPI=0x7ff99000 ACPI 2.0=0x7ff98000 MPS=0x7ff97000 SMBIOS=0xf0000
mem00: type=4, attr=0x9, range=[0x0000000000000000-0x0000000000001000) (0MB)
mem01: type=7, attr=0x9, range=[0x0000000000001000-0x0000000000007000) (0MB)
mem02: type=4, attr=0x9, range=[0x0000000000007000-0x0000000000009000) (0MB)
mem03: type=7, attr=0x9, range=[0x0000000000009000-0x0000000000082000) (0MB)
mem04: type=6, attr=0x8000000000000009, range=[0x0000000000082000-0x0000000000084000) (0MB)
mem05: type=7, attr=0x9, range=[0x0000000000084000-0x0000000000085000) (0MB)
mem06: type=4, attr=0x9, range=[0x0000000000085000-0x00000000000a0000) (0MB)
mem07: type=5, attr=0x8000000000000009, range=[0x00000000000c0000-0x0000000000100000) (0MB)
mem08: type=7, attr=0xb, range=[0x0000000000100000-0x0000000004000000) (63MB)
mem09: type=7, attr=0xb, range=[0x0000000004000000-0x0000000004644000) (6MB)
mem10: type=7, attr=0xb, range=[0x0000000004644000-0x000000000ffc0000) (185MB)
mem11: type=4, attr=0xb, range=[0x000000000ffc0000-0x0000000010000000) (0MB)
mem12: type=2, attr=0xb, range=[0x0000000010000000-0x0000000010490000) (4MB)

Quoted 8 lines

mem13: type=2, attr=0xb, range=[0x0000000010490000-0x00000000104a0000) (0MB)
mem14: type=2, attr=0xb, range=[0x00000000104a0000-0x0000000010650000) (1MB)
mem15: type=7, attr=0xb, range=[0x0000000010650000-0x000000001ffe4000) (249MB)
mem16: type=8, attr=0x5555555555555555, range=[0x000000001ffe4000-0x600000001fff2000) (6597069766656MB)
mem17: type=7, attr=0x5555555555555555, range=[0x600000001fff2350-0x5151515151514350) (16583222432533MB)

Those values are wrong,
Could you test we a 2.6.20 plus the patch?
Also it will be helpful to print efi_memmap in purgatory code.

Thanks
Zou Nan hai

author
#7 lore raw
In reply to Horms
Quoted 3 lines · Nanhai

Those values are wrong,
Could you test we a 2.6.20 plus the patch?

I tried 2.6.20 + your patch.
I tried the same with the addition of Bob Picco's patch to stop the crash-kernel crashing.
And I tried with Sparse and Discontig memory.
And in all cases I get much the same result :(

I'm wondering if perhaps its got something to do with kexec-tools, I'm using kexec-tools-testing from git. Is there any possibility you could send a static binary the version that you are using (if its from Eric's old tree cross compiling doesn't really work)?

Or perhaps my kernel config is odd.

Quoted 1 line

Also it will be helpful to print efi_memmap in purgatory code.

Indeed. Do you have any way to dump purgatory's console across a serial port? The vga console and I are on opposite sides of town at the moent.

Signature · 3 lines
Horms
  H: http://www.vergenet.net/~horms/
  W: http://www.valinux.co.jp/en/
#8 lore raw
In reply to Horms
Quoted 23 lines

-----Original Message-----
From: Horms [mailto:[email protected]]
Sent: 2007Äê2ÔÂ8ÈÕ 21:08
To: Zou, Nanhai
Cc: [email protected]; fastboot; Linux-IA64; Luck, Tony
Subject: Re: Zero size /proc/vmcore on ia64

Quoted 3 lines · Nanhai

Those values are wrong,
Could you test we a 2.6.20 plus the patch?

I tried 2.6.20 + your patch.
I tried the same with the addition of Bob Picco's patch to stop the crash-kernel crashing.
And I tried with Sparse and Discontig memory.
And in all cases I get much the same result :(

I'm wondering if perhaps its got something to do with kexec-tools, I'm using kexec-tools-testing from git. Is there any possibility you could send a static binary the version that you are using (if its from Eric's old tree cross compiling doesn't really work)?

Quoted 8 lines

Or perhaps my kernel config is odd.

Quoted 1 line

Also it will be helpful to print efi_memmap in purgatory code.

Indeed. Do you have any way to dump purgatory's console across a serial port? The vga console and I are on opposite sides of town at the moent.

I have not implement serial print in purgatory code yet, see comments in purgatory/arch/ia64/console-ia64.c However from your print, I can see last 2 entries of efi mem map are corrupt.
So we need to find out if it purgatory code who corrupted them or something else overlapped with that 2 entries.

Thanks
Zou Nan hai

Quoted 5 lines

--
Horms
H: http://www.vergenet.net/~horms/
W: http://www.valinux.co.jp/en/

#9 lore raw
In reply to Horms

Hello,

Quoted 4 lines · Zou Nan hai

This is reasonable.
So please apply the following patch to make saved_max_pfn point to
max_pfn of entire system.

So, as the patch works here also and it's necessary to make kdump working, I suggest including it mainline.

Regards,
Bernhard

#10 lore raw
In reply to Horms

Vivek, everyone,

Quoted 5 lines · Vivek Goyal

I think we should not remove this check because even to parse the info passed in ELF headers, you need to first read the ELF headers from crashed kernel's memory. So if some programming error has passed wrong location of ELF headers (elfcoreheader= invalid location) then we might try reading the elf header from a non-existing physical page frame.

Are you saying that the ELF header is located in the memory space of the first kernel?

The way I read the code the ELF header is put into the reserved memory space for the secondary kernel. At least on ia64 that is true, and I think the same goes for i386.

And the fact that the ELF header is put in to the secondary kernel brings me memory setup problems on ia64.

Basically the ELF header is marked as EFI_UNUSABLE_MEMORY by the EFI mangling code in purgatory. The secondary kernel detects this while parsing the EFI tables and refuses to use/map the other memory present in the same 16M granule. And in my case the initramfs happens to be located in the same granule... boom! No good. =)

So I'm wondering about the reason why we put the ELF header in the secondary kernel. Can't we just put it in the first kernel and be done with it? We still point it out using the kernel command line, don't we?

Thanks,

/ magnus

#11 lore raw
In reply to Horms
Quoted 32 lines

-----Original Message-----
From: Magnus Damm [mailto:[email protected]]
Sent: 2007Äê2ÔÂ14ÈÕ 16:28
To: [email protected]
Cc: Horms; Zou, Nanhai; [email protected]; [email protected]
Subject: Re: Zero size /proc/vmcore on ia64

Vivek, everyone,

Quoted 5 lines · Vivek Goyal

I think we should not remove this check because even to parse the info passed in ELF headers, you need to first read the ELF headers from crashed kernel's memory. So if some programming error has passed wrong location of ELF headers (elfcoreheader= invalid location) then we might try reading the elf header from a non-existing physical page frame.

Are you saying that the ELF header is located in the memory space of the first kernel?

The way I read the code the ELF header is put into the reserved memory space for the secondary kernel. At least on ia64 that is true, and I think the same goes for i386.

And the fact that the ELF header is put in to the secondary kernel brings me memory setup problems on ia64.

Basically the ELF header is marked as EFI_UNUSABLE_MEMORY by the EFI mangling code in purgatory. The secondary kernel detects this while parsing the EFI tables and refuses to use/map the other memory present in the same 16M granule. And in my case the initramfs happens to be located in the same granule... boom! No good. =)

Quoted 4 lines

So I'm wondering about the reason why we put the ELF header in the secondary kernel. Can't we just put it in the first kernel and be done with it? We still point it out using the kernel command line, don't we?

My first design is that putting data in second kernel is easy and safer. We could put it in the first kernel if we provide an interface to reserve this area at the time of kexec -p so that nobody will touch it even at the time of crash.

Align that buffer to 16M will solve the issue but that seems to be a waste of the useful memory?
Another way is append this elf header to command line in purgatory, like we append the ummy efi function, maybe this is too hacky?

Thanks
Zou Nan hai

Quoted 4 lines

Thanks,

/ magnus

#12 lore raw
In reply to Horms
Quoted 24 lines · Nanhai
Quoted 24 lines

From: Magnus Damm [mailto:[email protected]]

Quoted 5 lines · Vivek Goyal

I think we should not remove this check because even to parse the info passed in ELF headers, you need to first read the ELF headers from crashed kernel's memory. So if some programming error has passed wrong location of ELF headers (elfcoreheader= invalid location) then we might try reading the elf header from a non-existing physical page frame.

Are you saying that the ELF header is located in the memory space of the first kernel?

The way I read the code the ELF header is put into the reserved memory space for the secondary kernel. At least on ia64 that is true, and I think the same goes for i386.

And the fact that the ELF header is put in to the secondary kernel brings me memory setup problems on ia64.

Basically the ELF header is marked as EFI_UNUSABLE_MEMORY by the EFI mangling code in purgatory. The secondary kernel detects this while parsing the EFI tables and refuses to use/map the other memory present in the same 16M granule. And in my case the initramfs happens to be located in the same granule... boom! No good. =)

Quoted 6 lines
Quoted 4 lines

So I'm wondering about the reason why we put the ELF header in the secondary kernel. Can't we just put it in the first kernel and be done with it? We still point it out using the kernel command line, don't we?

My first design is that putting data in second kernel is easy and safer. We could put it in the first kernel if we provide an interface to reserve this area at the time of kexec -p so that nobody will touch it even at the time of crash.

Maybe that's a good idea. But that would make ia64 a special case and I'd like to avoid that as long as possible.

Quoted 1 line

Align that buffer to 16M will solve the issue but that seems to be a waste of the useful memory?

Right. We could require one granule per segment or something, but at load time we don't really know if the secondary kernel is using 16M or 64M granules. A safe bet would be to always use 64M, but that would require us to use quite a lot of memory for the secondary kernel.

Quoted 1 line

Another way is append this elf header to command line in purgatory, like we append the ummy efi function, maybe this is too hacky?

Hm. I think that sounds a bit too hackish. =)

What about the option of marking the ELF headers as EFI_LOADER_DATA and let the secondary kernel allocate new space and copy the data early during boot?

Thanks!

/ magnus

#13 lore raw
In reply to Horms
Quoted 54 lines

-----Original Message-----
From: Horms [mailto:[email protected]]
Sent: 2007年2月14日 18:13
To: Zou, Nanhai
Cc: Magnus Damm; [email protected]; [email protected]; [email protected]
Subject: Re: Zero size /proc/vmcore on ia64

Quoted 5 lines · Nanhai
Quoted 5 lines

-----Original Message-----
From: Magnus Damm [mailto:[email protected]]
Sent: 2007年2月14日 16:28
To: [email protected]
Cc: Horms; Zou, Nanhai; [email protected];

[email protected]

Quoted 8 lines
Quoted 8 lines

Subject: Re: Zero size /proc/vmcore on ia64

Vivek, everyone,

Quoted 3 lines · Vivek Goyal

I think we should not remove this check because even to parse the info passed in ELF headers, you need to first read the ELF headers from crashed kernel's memory. So if some programming error has passed wrong location

of

Quoted 1 line
Quoted 1 line
Quoted 1 line

ELF headers (elfcoreheader= invalid location) then we might try reading

the

Quoted 28 lines
Quoted 18 lines
Quoted 1 line

elf header from a non-existing physical page frame.

Are you saying that the ELF header is located in the memory space of the first kernel?

The way I read the code the ELF header is put into the reserved memory space for the secondary kernel. At least on ia64 that is true, and I think the same goes for i386.

And the fact that the ELF header is put in to the secondary kernel brings me memory setup problems on ia64.

Basically the ELF header is marked as EFI_UNUSABLE_MEMORY by the EFI mangling code in purgatory. The secondary kernel detects this while parsing the EFI tables and refuses to use/map the other memory present in the same 16M granule. And in my case the initramfs happens to be located in the same granule... boom! No good. =)

Quoted 4 lines

So I'm wondering about the reason why we put the ELF header in the secondary kernel. Can't we just put it in the first kernel and be done with it? We still point it out using the kernel command line, don't we?

My first design is that putting data in second kernel is easy and
safer. We could put it in the first kernel if we provide an interface to reserve this area at the time of kexec -p so that nobody will touch it even at the time of crash.

Quoted 6 lines
Quoted 4 lines

Align that buffer to 16M will solve the issue but that seems to be a
waste of the useful memory? Another way is append this elf header to command line in purgatory, like we append the ummy efi function, maybe this is too hacky?

I think that the dummy efi function is already way to hacky.

Yes it is. However the benefit of it is that you can kexec to an old kernel even a 2.4 based kernel.

Quoted 8 lines

I'd like to work out a (good) way to get rid of it.
For starters the PAGE_OFFSET is hardcoded at kexec-tools compile time - which breaks xen as it has a different PAGE_OFFSET.

--
Horms
H: http://www.vergenet.net/~horms/
W: http://www.valinux.co.jp/en/

-
To unsubscribe from this list: send the line "unsubscribe linux-ia64" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#14 lore raw
In reply to Horms
Quoted 57 lines

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Magnus Damm
Sent: 2007Äê2ÔÂ14ÈÕ 19:47
To: Zou, Nanhai
Cc: [email protected]; Horms; [email protected]; [email protected]
Subject: Re: Zero size /proc/vmcore on ia64

Quoted 5 lines · Nanhai
Quoted 5 lines

From: Magnus Damm [mailto:[email protected]]

Quoted 3 lines · Vivek Goyal

I think we should not remove this check because even to parse the info passed in ELF headers, you need to first read the ELF headers from crashed kernel's memory. So if some programming error has passed wrong location

of

Quoted 1 line
Quoted 1 line
Quoted 1 line

ELF headers (elfcoreheader= invalid location) then we might try reading

the

Quoted 18 lines
Quoted 18 lines
Quoted 1 line

elf header from a non-existing physical page frame.

Are you saying that the ELF header is located in the memory space of the first kernel?

The way I read the code the ELF header is put into the reserved memory space for the secondary kernel. At least on ia64 that is true, and I think the same goes for i386.

And the fact that the ELF header is put in to the secondary kernel brings me memory setup problems on ia64.

Basically the ELF header is marked as EFI_UNUSABLE_MEMORY by the EFI mangling code in purgatory. The secondary kernel detects this while parsing the EFI tables and refuses to use/map the other memory present in the same 16M granule. And in my case the initramfs happens to be located in the same granule... boom! No good. =)

Quoted 6 lines
Quoted 4 lines

So I'm wondering about the reason why we put the ELF header in the secondary kernel. Can't we just put it in the first kernel and be done with it? We still point it out using the kernel command line, don't we?

My first design is that putting data in second kernel is easy and safer.

We could put it in the first kernel if we provide an interface to reserve this area at the time of kexec -p so that nobody will touch it even at the time of crash.

Maybe that's a good idea. But that would make ia64 a special case and I'd like to avoid that as long as possible.

Quoted 1 line

Align that buffer to 16M will solve the issue but that seems to be a waste

of the useful memory?

Right. We could require one granule per segment or something, but at load time we don't really know if the secondary kernel is using 16M or 64M granules. A safe bet would be to always use 64M, but that would require us to use quite a lot of memory for the secondary kernel.

Quoted 9 lines
Quoted 1 line

Another way is append this elf header to command line in purgatory, like

we append the ummy efi function, maybe this is too hacky?

Hm. I think that sounds a bit too hackish. =)

What about the option of marking the ELF headers as EFI_LOADER_DATA and let the secondary kernel allocate new space and copy the data early during boot?

Ah yes, I think vmcore.c has already done this, parse and store elf core header structure at kernel init, so mark that area as EFI_CONVENTIONAL_MEMORY should be safe. This can be done by remove the reserved = 1 setting in crashdump-ia64.c

Thanks
Zou Nan hai

Quoted 7 lines

Thanks!

/ magnus
-
To unsubscribe from this list: send the line "unsubscribe linux-ia64" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

author
#15 lore raw
In reply to Horms
Quoted 4 lines · Nanhai
Quoted 2 lines

I think that the dummy efi function is already way to hacky.

Yes it is. However the benefit of it is that you can kexec to an old kernel even a 2.4 based kernel.

That is a good point.

Signature · 3 lines
Horms
  H: http://www.vergenet.net/~horms/
  W: http://www.valinux.co.jp/en/