Simon Horman <horms at verge.net.au> writes:
> On Wed, Sep 08, 2010 at 12:44:29PM -0700, Eric W. Biederman wrote:
>>
>> I thought I had x86 biarch crashdump working on x86 but it turns
>> out was wrong. So I have merged crashdump-x86.c and crashdump-x86_64.c
>> to ensure that things will work fine.
>>
>> There are a few misc cleanups in here to support biarch crashdumps.
>>
>> There is a kexec bzImage load optimization that allows me to use
>> a 64bit crash kernel in 20M in my setup. Perhaps less. Earlier
>> because of an alighment foulup I needed 32M and the kernel was
>> loaded in the middle of the reserved crashdump region.
>>
>> There is also a new program vmcore-dmesg. Functionally it is very
>> similar to the makedumpfile's --dmesg mode but it handles all of the
>> crazy multi-arch cases so I only need to keep one binary around to strip
>> the kernel's dmesg out of a core file. Additionally vmcore-dmesg is
>> about half the size of makedumpfile and I can build it against uclibc,
>> which makes it much more usable in my constrained initrd setup.
>>
>> I would post patches but I know I would fat finger that horribly.
>>
>> git pull git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/kexec-tools-devel.git
>
> A git pull request is fine.
> Actually, with this many patches, its significantly easier for me too.
Definitely on a good day I would post the patches for review but still
send you a pull request.
>> ---
>
> [ snip ]
>
>> commit c8e346688c6358427213d9d6fdb83765e365cccd
>> Author: Eric W. Biederman <ebiederm at xmission.com>
>> Date: Tue Sep 7 12:05:40 2010 -0700
>>
>> crashdump: Move kern_vaddr_start from kexec_info into crash_elf_info
>>
>> These fields as defined in kexec_info are not biarch safe and
>> beyond that crash_elf_info is the structure for passing this kind
>> of information not kexec_info. So move them in prepartion for
>> properly cleaning up biarch x86 functionality.
>>
>> Signed-off-by: Eric W. Biederman <ebiederm at xmission.com>
>
> There appears to be a typo in the following hunk of this patch.
> I can fix this if you prefer.
If you could add a commit that fixes this that would be great. I
have an ia64 compiler around somewhere but not handy :(
> diff --git a/kexec/arch/ia64/crashdump-ia64.c b/kexec/arch/ia64/crashdump-ia64.c
> index 6b64271..5653e6e 100644
> --- a/kexec/arch/ia64/crashdump-ia64.c
> +++ b/kexec/arch/ia64/crashdump-ia64.c
> @@ -246,11 +246,11 @@ int load_crashdump_segments(struct kexec_info *info, struct mem_ehdr *ehdr,
> continue;
> if (kernel_code_start >= mstart &&
> kernel_code_start < mend) {
> - info->kern_vaddr_start = mstart + LOAD_OFFSET;
> + elf_info.kern_vaddr_start = mstart + LOAD_OFFSET;
> break;
> }
> }
> - info->kern_size = kernel_code_end - kernel_code_start + 1;
> + elf_info->kern_size = kernel_code_end - kernel_code_start + 1;
>
> -------> elf_info->kern_size should be elf_info.kern_size ?
Yes.
>
> if (crash_create_elf64_headers(info, &elf_info,
> crash_memory_range, nr_ranges,
> &tmp, &sz, EFI_PAGE_SIZE) < 0)
>
> Which results in the following build error:
>
> $ make
> ia64-unknown-linux-gnu-gcc -Wall -Wextra -O2 -fomit-frame-pointer -pipe -fno-strict-aliasing -Wall -Wstrict-prototypes -I/home/horms/local/opt/crosstool/ia64/gcc-3.4.5-glibc-2.3.6/ia64-unknown-linux-gnu/include -I./include -I./util_lib/include -Iinclude/ -I./kexec/arch/ia64/libfdt -I./kexec/arch/ia64/include -c -MD -o kexec/arch/ia64/crashdump-ia64.o
> kexec/arch/ia64/crashdump-ia64.c
> kexec/arch/ia64/crashdump-ia64.c: In function `load_crashdump_segments':
> kexec/arch/ia64/crashdump-ia64.c:253: error: invalid type argument of `->'
>
>>
>> commit d9bf000b6260ee4558b2d2eb3af08e09cd1fe794
>> Author: Eric W. Biederman <ebiederm at xmission.com>
>> Date: Tue Sep 7 11:42:39 2010 -0700
>>
>> kexec: Kill arch_init
>>
>> The function only had one user, the error checking was wrong,
>> and the functions it performed are best done elsewhere
>> so remove the tempation of a problemenatic hook.
>>
>> Signed-off-by: Eric W. Biederman <ebiederm at xmission.com>
>
> [ snip ]
Thanks,
Eric
