Skip to contentSkip to conversation

mips: elfinfo

PATCH Simon Horman 3 participants 8 messages Dec 16, 2010 – Dec 21, 2010

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hi Maxim,

I noticed that your change "Patch adds kernel data section to final vmcore. This forces gdb read kernel" causes the build to break as it was applied after Eric's "crashdump: Move kern_vaddr_start from kexec_info into crash_elf_info" patch.

I have it in mind to resolve this something along the lines of the patch below. However, I now see the following warning:

kexec/arch/mips/crashdump-mips.c: In function `get_kernel_vaddr_and_size':
kexec/arch/mips/crashdump-mips.c:81: warning: integer constant is too large for "unsigned long" type

Which corresponds to this code:

        elf_info->kern_vaddr_start = elf_info->kern_paddr_start |
                                        0xffffffff80000000UL;

I am compiling for 32-bit, so at a glance that constant does seem to large.

Also, if you have any tips on a cross-compiling environment for 64bit on a x86_32 or x86_64 host I would be most grateful.

kexec/arch/mips/crashdump-mips.c +23 -32
index 5ea4f4f..b937ab5 100644
@@ -51,7 +51,7 @@ unsigned long long saved_max_mem;
 
 /* Read kernel physical load addr from the file returned by proc_iomem()
  * (Kernel Code) and store in kexec_info */
-static int get_kernel_paddr(struct kexec_info *info)
+static int get_kernel_paddr(struct crash_elf_info *elf_info)
 {
 	uint64_t start;
 
@@ -59,7 +59,7 @@ static int get_kernel_paddr(struct kexec_info *info)
 		return 0;
 
 	if (parse_iomem_single("Kernel code\n", &start, NULL) == 0) {
-		info->kern_paddr_start = start;
+		elf_info->kern_paddr_start = start;
 #ifdef DEBUG
 		printf("kernel load physical addr start = 0x%lx\n", start);
 #endif
@@ -70,21 +70,22 @@ static int get_kernel_paddr(struct kexec_info *info)
 	return -1;
 }
 
-static int get_kernel_vaddr_and_size(struct kexec_info *info)
+static int get_kernel_vaddr_and_size(struct crash_elf_info *elf_info)
 {
 	uint64_t end;
 
-	if (!info->kern_paddr_start)
+	if (!elf_info->kern_paddr_start)
 		return -1;
 
-	info->kern_vaddr_start =  info->kern_paddr_start | 0xffffffff80000000UL;
+	elf_info->kern_vaddr_start = elf_info->kern_paddr_start |
+					0xffffffff80000000UL;
 	if (parse_iomem_single("Kernel data\n", NULL, &end) == 0) {
-		info->kern_size = end - info->kern_paddr_start;
+		elf_info->kern_size = end - elf_info->kern_paddr_start;
 #ifdef DEBUG
 		printf("kernel_vaddr= 0x%llx paddr %llx\n",
-				info->kern_vaddr_start,
-				info->kern_paddr_start);
-		printf("kernel size = 0x%llx\n", info->kern_size);
+				elf_info->kern_vaddr_start,
+				elf_info->kern_paddr_start);
+		printf("kernel size = 0x%llx\n", elf_info->kern_size);
 #endif
 		return 0;
 		}
@@ -355,11 +356,20 @@ int load_crashdump_segments(struct kexec_info *info, char* mod_cmdline,
 	unsigned long sz, elfcorehdr;
 	int nr_ranges, align = 1024;
 	struct memory_range *mem_range;
+	crash_create_elf_headers_func crash_create = crash_create_elf32_headers;
+	struct crash_elf_info *elf_info = &elf_info32;
 
-	if (get_kernel_paddr(info))
+#ifdef __mips64
+	if (arch_options.core_header_type == CORE_TYPE_ELF64) {
+		elf_info = &elf_info64;
+		crash_create = crash_create_elf64;
+	}
+#endif
+
+	if (get_kernel_paddr(elf_info))
 		return -1;
 
-	if (get_kernel_vaddr_and_size(info))
+	if (get_kernel_vaddr_and_size(elf_info))
 		return -1;
 
 	if (get_crash_memory_ranges(&mem_range, &nr_ranges) < 0)
@@ -373,28 +383,9 @@ int load_crashdump_segments(struct kexec_info *info, char* mod_cmdline,
 				crash_reserved_mem.start,
 				crash_reserved_mem.end, -1);
 
-#ifdef __mips64
-	/* Create elf header segment and store crash image data. */
-	if (arch_options.core_header_type == CORE_TYPE_ELF64) {
-		if (crash_create_elf64_headers(info, &elf_info64,
-			crash_memory_range, nr_ranges,
-			&tmp, &sz,
-			ELF_CORE_HEADER_ALIGN) < 0)
-			return -1;
-	} else {
-		if (crash_create_elf32_headers(info, &elf_info32,
-			crash_memory_range, nr_ranges,
-			&tmp, &sz,
-			ELF_CORE_HEADER_ALIGN) < 0)
-			return -1;
-	}
-#else
-	if (crash_create_elf32_headers(info, &elf_info32,
-		crash_memory_range, nr_ranges,
-		&tmp, &sz,
-		ELF_CORE_HEADER_ALIGN) < 0)
+	if (crash_create(info, elf_info, crash_memory_range, nr_ranges,
+			 &tmp, &sz, ELF_CORE_HEADER_ALIGN) < 0)
 		return -1;
-#endif
 	elfcorehdr = add_buffer(info, tmp, sz, sz, align,
 		crash_reserved_mem.start,
 		crash_reserved_mem.end, -1);
kexec/crashdump.h +7 -0
index eccdb9f..060bb88 100644
@@ -35,6 +35,13 @@ struct crash_elf_info {
 	int (*get_note_info)(int cpu, uint64_t *addr, uint64_t *len);
 };
 
+typedef int(*crash_create_elf_headers_func)(struct kexec_info *info,
+					    struct crash_elf_info *elf_info,
+					    struct memory_range *range,
+					    int ranges,
+					    void **buf, unsigned long *size,
+					    unsigned long align);
+
 int crash_create_elf32_headers(struct kexec_info *info,
 			       struct crash_elf_info *elf_info,
 			       struct memory_range *range, int ranges,

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec

#1 lore raw

2010/12/16 Simon Horman <[email protected]>:

Quoted 13 lines

Hi Maxim,

I noticed that your change "Patch adds kernel data section to final vmcore. This forces gdb read kernel" causes the build to break as it was applied after Eric's "crashdump: Move kern_vaddr_start from kexec_info into crash_elf_info" patch.

I have it in mind to resolve this something along the lines of the patch below. However, I now see the following warning:

       kexec/arch/mips/crashdump-mips.c: In function `get_kernel_vaddr_and_size':
       kexec/arch/mips/crashdump-mips.c:81: warning: integer constant is too large for "unsigned long" type

Quoted 8 lines

Which corresponds to this code:

       elf_info->kern_vaddr_start = elf_info->kern_paddr_start | 0xffffffff80000000UL;

I am compiling for 32-bit, so at a glance that constant does seem to large.

Ah, yes this values does not fit to 32 bit UL.

Quoted 3 lines

Also, if you have any tips on a cross-compiling environment for 64bit on a x86_32 or x86_64 host I would be most grateful.

I tested it with building on 32 bit host. readelf -a shows that vmcore had memory segment according to kernel memory.
So that, gdb began to read real values instead of uninitialized.

Actually elf_info->kern_paddr_start | 0xffffffff80000000UL; is needed only for 64 bit mips.
Because we read kernel memory ranges from /proc/vmcore and there are values without 0xffffffff at the beginning.
But 0xfffff... are needed in case if we need access to this address from kernel.

32 bit mips kernel does not need adding 0xfffffff. But it think there will be only warning and nothing wrong.

Maxim.

Quoted 126 lines
kexec/arch/mips/crashdump-mips.c +23 -32
index 5ea4f4f..b937ab5 100644
@@ -51,7 +51,7 @@ unsigned long long saved_max_mem;
 
  /* Read kernel physical load addr from the file returned by proc_iomem()
  * (Kernel Code) and store in kexec_info */
-static int get_kernel_paddr(struct kexec_info *info)
+static int get_kernel_paddr(struct crash_elf_info *elf_info)
  {
        uint64_t start;
 
@@ -59,7 +59,7 @@ static int get_kernel_paddr(struct kexec_info *info)
                return 0;
 
        if (parse_iomem_single("Kernel code\n", &start, NULL) == 0) {
-               info->kern_paddr_start = start;
+               elf_info->kern_paddr_start = start;
  #ifdef DEBUG
                printf("kernel load physical addr start = 0x%lx\n", start);
  #endif
@@ -70,21 +70,22 @@ static int get_kernel_paddr(struct kexec_info *info)
        return -1;
  }
 
-static int get_kernel_vaddr_and_size(struct kexec_info *info)
+static int get_kernel_vaddr_and_size(struct crash_elf_info *elf_info)
  {
        uint64_t end;
 
-       if (!info->kern_paddr_start)
+       if (!elf_info->kern_paddr_start)
                return -1;
 
-       info->kern_vaddr_start =  info->kern_paddr_start | 0xffffffff80000000UL;
+       elf_info->kern_vaddr_start = elf_info->kern_paddr_start |
+                                       0xffffffff80000000UL;
        if (parse_iomem_single("Kernel data\n", NULL, &end) == 0) {
-               info->kern_size = end - info->kern_paddr_start;
+               elf_info->kern_size = end - elf_info->kern_paddr_start;
  #ifdef DEBUG
                printf("kernel_vaddr= 0x%llx paddr %llx\n",
-                               info->kern_vaddr_start,
-                               info->kern_paddr_start);
-               printf("kernel size = 0x%llx\n", info->kern_size);
+                               elf_info->kern_vaddr_start,
+                               elf_info->kern_paddr_start);
+               printf("kernel size = 0x%llx\n", elf_info->kern_size);
  #endif
                return 0;
                }
@@ -355,11 +356,20 @@ int load_crashdump_segments(struct kexec_info *info, char* mod_cmdline,
        unsigned long sz, elfcorehdr;
        int nr_ranges, align = 1024;
        struct memory_range *mem_range;
+       crash_create_elf_headers_func crash_create = crash_create_elf32_headers;
+       struct crash_elf_info *elf_info = &elf_info32;
 
-       if (get_kernel_paddr(info))
+#ifdef __mips64
+       if (arch_options.core_header_type == CORE_TYPE_ELF64) {
+               elf_info = &elf_info64;
+               crash_create = crash_create_elf64;
+       }
+#endif
+
+       if (get_kernel_paddr(elf_info))
                return -1;
 
-       if (get_kernel_vaddr_and_size(info))
+       if (get_kernel_vaddr_and_size(elf_info))
                return -1;
 
        if (get_crash_memory_ranges(&mem_range, &nr_ranges) < 0)
@@ -373,28 +383,9 @@ int load_crashdump_segments(struct kexec_info *info, char* mod_cmdline,
                                crash_reserved_mem.start,
                                crash_reserved_mem.end, -1);
 
-#ifdef __mips64
-       /* Create elf header segment and store crash image data. */
-       if (arch_options.core_header_type == CORE_TYPE_ELF64) {
-               if (crash_create_elf64_headers(info, &elf_info64,
-                       crash_memory_range, nr_ranges,
-                       &tmp, &sz,
-                       ELF_CORE_HEADER_ALIGN) < 0)
-                       return -1;
-       } else {
-               if (crash_create_elf32_headers(info, &elf_info32,
-                       crash_memory_range, nr_ranges,
-                       &tmp, &sz,
-                       ELF_CORE_HEADER_ALIGN) < 0)
-                       return -1;
-       }
-#else
-       if (crash_create_elf32_headers(info, &elf_info32,
-               crash_memory_range, nr_ranges,
-               &tmp, &sz,
-               ELF_CORE_HEADER_ALIGN) < 0)
+       if (crash_create(info, elf_info, crash_memory_range, nr_ranges,
+                        &tmp, &sz, ELF_CORE_HEADER_ALIGN) < 0)
                return -1;
-#endif
        elfcorehdr = add_buffer(info, tmp, sz, sz, align,
                crash_reserved_mem.start,
                crash_reserved_mem.end, -1);
kexec/crashdump.h +7 -0
index eccdb9f..060bb88 100644
@@ -35,6 +35,13 @@ struct crash_elf_info {
        int (*get_note_info)(int cpu, uint64_t *addr, uint64_t *len);
  };
 
+typedef int(*crash_create_elf_headers_func)(struct kexec_info *info,
+                                           struct crash_elf_info *elf_info,
+                                           struct memory_range *range,
+                                           int ranges,
+                                           void **buf, unsigned long *size,
+                                           unsigned long align);
+
  int crash_create_elf32_headers(struct kexec_info *info,
                               struct crash_elf_info *elf_info,
                               struct memory_range *range, int ranges,
Signature · 7 lines
Best regards,
Maxim Uvarov

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec
#2 lore raw
Quoted 33 lines · Maxim Uvarov

2010/12/16 Simon Horman <[email protected]>:

Quoted 13 lines

Hi Maxim,

I noticed that your change "Patch adds kernel data section to final vmcore. This forces gdb read kernel" causes the build to break as it was applied after Eric's "crashdump: Move kern_vaddr_start from kexec_info into crash_elf_info" patch.

I have it in mind to resolve this something along the lines of the patch below. However, I now see the following warning:

       kexec/arch/mips/crashdump-mips.c: In function `get_kernel_vaddr_and_size':
       kexec/arch/mips/crashdump-mips.c:81: warning: integer constant is too large for "unsigned long" type

Quoted 8 lines

Which corresponds to this code:

       elf_info->kern_vaddr_start = elf_info->kern_paddr_start | 0xffffffff80000000UL;

I am compiling for 32-bit, so at a glance that constant does seem to large.

Ah, yes this values does not fit to 32 bit UL.

Quoted 3 lines

Also, if you have any tips on a cross-compiling environment for 64bit on a x86_32 or x86_64 host I would be most grateful.

I tested it with building on 32 bit host. readelf -a shows that vmcore had memory segment according to kernel memory.
So that, gdb began to read real values instead of uninitialized.

Quoted 8 lines

Actually elf_info->kern_paddr_start | 0xffffffff80000000UL; is needed only for 64 bit mips.
Because we read kernel memory ranges from /proc/vmcore and there are values without 0xffffffff at the beginning.
But 0xfffff... are needed in case if we need access to this address from kernel.

32 bit mips kernel does not need adding 0xfffffff. But it think there will be only warning and nothing wrong.

Understood, but it would be nice to eliminate the warning. As load_crashdump_segments() is already 32/64 bit aware I wonder if we could just pass appropriate constant from there as a parameter of get_kernel_vaddr_and_size().

Perhaps something like the following which
applies on top of my previous patch.

kexec/arch/mips/crashdump-mips.c +6 -3
index b937ab5..aa50109 100644
@@ -70,7 +70,8 @@ static int get_kernel_paddr(struct crash_elf_info *elf_info)
 	return -1;
 }
 
-static int get_kernel_vaddr_and_size(struct crash_elf_info *elf_info)
+static int get_kernel_vaddr_and_size(struct crash_elf_info *elf_info,
+				     unsigned long start_offset)
 {
 	uint64_t end;
 
@@ -78,7 +79,7 @@ static int get_kernel_vaddr_and_size(struct crash_elf_info *elf_info)
 		return -1;
 
 	elf_info->kern_vaddr_start = elf_info->kern_paddr_start |
-					0xffffffff80000000UL;
+					start_offset;
 	if (parse_iomem_single("Kernel data\n", NULL, &end) == 0) {
 		elf_info->kern_size = end - elf_info->kern_paddr_start;
 #ifdef DEBUG
@@ -358,18 +359,20 @@ int load_crashdump_segments(struct kexec_info *info, char* mod_cmdline,
 	struct memory_range *mem_range;
 	crash_create_elf_headers_func crash_create = crash_create_elf32_headers;
 	struct crash_elf_info *elf_info = &elf_info32;
+	unsigned long start_offset = 0x80000000UL;
 
 #ifdef __mips64
 	if (arch_options.core_header_type == CORE_TYPE_ELF64) {
 		elf_info = &elf_info64;
 		crash_create = crash_create_elf64;
+		start_offset = 0xffffffff80000000UL;
 	}
 #endif
 
 	if (get_kernel_paddr(elf_info))
 		return -1;
 
-	if (get_kernel_vaddr_and_size(elf_info))
+	if (get_kernel_vaddr_and_size(elf_info, start_offset))
 		return -1;
 
 	if (get_crash_memory_ranges(&mem_range, &nr_ranges) < 0)

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec

#3 lore raw

2010/12/20 Simon Horman <[email protected]>:

Quoted 43 lines · Maxim Uvarov
Quoted 33 lines

2010/12/16 Simon Horman <[email protected]>:

Quoted 13 lines

Hi Maxim,

I noticed that your change "Patch adds kernel data section to final vmcore. This forces gdb read kernel" causes the build to break as it was applied after Eric's "crashdump: Move kern_vaddr_start from kexec_info into crash_elf_info" patch.

I have it in mind to resolve this something along the lines of the patch below. However, I now see the following warning:

       kexec/arch/mips/crashdump-mips.c: In function `get_kernel_vaddr_and_size':
       kexec/arch/mips/crashdump-mips.c:81: warning: integer constant is too large for "unsigned long" type

Quoted 8 lines

Which corresponds to this code:

       elf_info->kern_vaddr_start = elf_info->kern_paddr_start | 0xffffffff80000000UL;

I am compiling for 32-bit, so at a glance that constant does seem to large.

Ah, yes this values does not fit to 32 bit UL.

Quoted 3 lines

Also, if you have any tips on a cross-compiling environment for 64bit on a x86_32 or x86_64 host I would be most grateful.

I tested it with building on 32 bit host. readelf -a shows that vmcore had memory segment according to kernel memory.
So that, gdb began to read real values instead of uninitialized.

Quoted 8 lines

Actually  elf_info->kern_paddr_start | 0xffffffff80000000UL; is needed only for 64 bit mips.
Because we read kernel memory ranges from /proc/vmcore and there are values without 0xffffffff at the beginning.
But 0xfffff... are needed in case if we need access to this address from kernel.

32 bit mips kernel does not need adding 0xfffffff.  But it think there will be only warning and nothing wrong.

Quoted 8 lines

Understood, but it would be nice to eliminate the warning. As load_crashdump_segments() is already 32/64 bit aware I wonder if we could just pass appropriate constant from there as a parameter of get_kernel_vaddr_and_size().

Perhaps something like the following which
applies on top of my previous patch.

Yes, that has to work. Thanks.

Quoted 46 lines
kexec/arch/mips/crashdump-mips.c +2 -1
index b937ab5..aa50109 100644
@@ -70,7 +70,8 @@ static int get_kernel_paddr(struct crash_elf_info *elf_info)
        return -1;
  }
 
-static int get_kernel_vaddr_and_size(struct crash_elf_info *elf_info)
+static int get_kernel_vaddr_and_size(struct crash_elf_info *elf_info,
+                                    unsigned long start_offset)
  {
        uint64_t end;
 
@@ -78,7 +79,7 @@ static int get_kernel_vaddr_and_size(struct crash_elf_info *elf_info)
                return -1;
 

       elf_info->kern_vaddr_start = elf_info->kern_paddr_start |
-                                       0xffffffff80000000UL;
+                                       start_offset;
       if (parse_iomem_single("Kernel data\n", NULL, &end) == 0) { elf_info->kern_size = end - elf_info->kern_paddr_start; #ifdef DEBUG

kexec/arch/mips/crashdump-mips.c +3 -1
@@ -358,18 +359,20 @@ int load_crashdump_segments(struct kexec_info *info, char* mod_cmdline,
        struct memory_range *mem_range;
        crash_create_elf_headers_func crash_create = crash_create_elf32_headers;
        struct crash_elf_info *elf_info = &elf_info32;
+       unsigned long start_offset = 0x80000000UL;
 
  #ifdef __mips64
        if (arch_options.core_header_type == CORE_TYPE_ELF64) {
                elf_info = &elf_info64;
                crash_create = crash_create_elf64;
+               start_offset = 0xffffffff80000000UL;
        }
  #endif
 
        if (get_kernel_paddr(elf_info))
                return -1;
 
-       if (get_kernel_vaddr_and_size(elf_info))
+       if (get_kernel_vaddr_and_size(elf_info, start_offset))
                return -1;
 
        if (get_crash_memory_ranges(&mem_range, &nr_ranges) < 0)
Signature · 7 lines
Best regards,
Maxim Uvarov

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec
#4 lore raw
Quoted 49 lines · Maxim Uvarov

2010/12/20 Simon Horman <[email protected]>:

Quoted 48 lines · Maxim Uvarov
Quoted 33 lines

2010/12/16 Simon Horman <[email protected]>:

Quoted 13 lines

Hi Maxim,

I noticed that your change "Patch adds kernel data section to final vmcore. This forces gdb read kernel" causes the build to break as it was applied after Eric's "crashdump: Move kern_vaddr_start from kexec_info into crash_elf_info" patch.

I have it in mind to resolve this something along the lines of the patch below. However, I now see the following warning:

       kexec/arch/mips/crashdump-mips.c: In function `get_kernel_vaddr_and_size':
       kexec/arch/mips/crashdump-mips.c:81: warning: integer constant is too large for "unsigned long" type

Quoted 8 lines

Which corresponds to this code:

       elf_info->kern_vaddr_start = elf_info->kern_paddr_start | 0xffffffff80000000UL;

I am compiling for 32-bit, so at a glance that constant does seem to large.

Ah, yes this values does not fit to 32 bit UL.

Quoted 3 lines

Also, if you have any tips on a cross-compiling environment for 64bit on a x86_32 or x86_64 host I would be most grateful.

I tested it with building on 32 bit host. readelf -a shows that vmcore had memory segment according to kernel memory.
So that, gdb began to read real values instead of uninitialized.

Quoted 8 lines

Actually  elf_info->kern_paddr_start | 0xffffffff80000000UL; is needed only for 64 bit mips.
Because we read kernel memory ranges from /proc/vmcore and there are values without 0xffffffff at the beginning.
But 0xfffff... are needed in case if we need access to this address from kernel.

32 bit mips kernel does not need adding 0xfffffff.  But it think there will be only warning and nothing wrong.

Understood, but it would be nice to eliminate the warning. As load_crashdump_segments() is already 32/64 bit aware I wonder if we could just pass appropriate constant from there as a parameter of get_kernel_vaddr_and_size().

Quoted 4 lines
Quoted 3 lines

Perhaps something like the following which
applies on top of my previous patch.

Yes, that has to work. Thanks.

Thanks, I have pushed this and the previous change.

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec

#6 lore raw

Re: kexec fixes?

Quoted 7 lines · Eric W. Biederman

Simon did you ever get my last round of fixes for i386 after my other changes merged?

I was testing them in my tree, and I was trying to be careful and not send you something bad. I completed my testing and but I forget to send you a pull request.

Hi Eric,

I don't think that I received a pull request and in any case I don't have anything other than mips fix(es) pending. So if its not in my tree could you (rebase and) send it again?

Its been a while since there was a release, so it might be nice (for me) to get a 2.0.3 out the door some time.

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec

#7 lore raw

Re: kexec fixes?

Quoted 9 lines · Simon Horman
Quoted 7 lines · Eric W. Biederman

Simon did you ever get my last round of fixes for i386 after my other changes merged?

I was testing them in my tree, and I was trying to be careful and not send you something bad. I completed my testing and but I forget to send you a pull request.

Quoted 8 lines

Hi Eric,

I don't think that I received a pull request and in any case I don't have anything other than mips fix(es) pending. So if its not in my tree could you (rebase and) send it again?

Its been a while since there was a release, so it might be nice (for me) to get a 2.0.3 out the door some time.

Hi Eric,

do you have any pending changes?

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec