On 2022/09/16 18:18, lijiang wrote:
Hi, Kazu
Thank you for the comment.
On Fri, Sep 16, 2022 at 3:42 PM HAGIO KAZUHITO(萩尾 一仁) <k-hagio-ab(a)nec.com>
wrote:
> On 2022/09/16 15:00, Lianbo Jiang wrote:
>> Currently, crash may fail with the following error:
>>
>> # ./crash vmlinux /var/crash/127.0.0.1-2022-09-15-09:31:55/vmcore -s
>> WARNING: invalid linux_banner pointer: 65762078756e694c
>> crash: vmlinux and /var/crash/127.0.0.1-2022-09-15-09:31:55/vmcore do
> not match!
>>
>> The reason is that the type of symbols in the data segment may be
>> defined as 'D' or 'd'. The crash only handled the type
'D', but it
>> didn't deal with the type 'd'. For example:
>>
>> # nm vmlinux |grep linux_banner
>> ffffffff827cfa80 d linux_banner
>>
>> Let's add the type 'd' recognition to solve such issue.
>
> Thank you for the fix, the patch diff looks good.
>
> I've searched for a vmlinux that its symbol type is 'd', but not found.
> Do you know what makes it 'd'? It would be better to add the cause to
> the commit log, for future reference.
>
>
The kernel package(vmlinux) is built on x86 _64 machines by clang, the type
of symbol 'linux_banner' is defined as 'd' instead of 'D', but
for gcc,
usually it is the type 'D'.
Hope this helps.
Thank you for the information, applied with it:
https://github.com/crash-utility/crash/commit/4ea3a806d11f000f2eb1ddc72c2...
Thanks,
Kazu