Question about the size of VMCOREINFO

Baoquan He bhe at redhat.com
Wed Aug 10 17:45:54 PDT 2022


Hi,

On 08/10/22 at 03:39pm, Donny Xia wrote:
> Hello Kdump Maintainers,
> 
> We utilize the kexec and kdump feature heavily to collect the crash dump
> file when the kernel crashes. We write some customized values into
> VMCOREINFO and we find that eventually the size of the data we write
> exceeds the size of VMCOREINFO. I find that the size of VMCOREINFO is one
> page, which I believe is 4k. I am wondering is there any specific reason to
> choose this size? Is it possible to use a bigger number? Any suggestions
> about it would be appreciated.

No specific reason. Just our current usage of vmcoreinfo_data is enough to
cover our need. I don't see why you can't enlarge it to, e.g 2 pages or
more. Have you tried? E.g allocate more pages and change
VMCOREINFO_BYTES to the corresponding value.




More information about the kexec mailing list