On 06/27/2013 04:54 PM, Stefan Hajnoczi wrote:
The interesting question is how effective this approach is. If it's good enough then it would be a fairly simple modification to dump.c.
I see, if excluding zero page in ELF can make a lot of size reduce, it's better to choose this method. But think over the situation that kernel is on for a long time, then few zero pages will be in memory, compression will do more work to reduce size not excluding zero pages. So the approach is not always effective. A test on a 1GB memory, and the machine is just on: size format method for reducing memory 1.1GB ELF no 1.1GB kdump no 227MB kdump with all zero pages excluded 96MB kdump compressed with zero pages remained 88MB kdump compressed with zero pages excluded excluding zero pages does some work, but compression seems to be more effective.
I meant simply compressing the ELF output during creation. But I guess the tools need random access to the ELF file, which is inefficient with zlib and friends.
It is the reason of why I don't like entirely compression. It will make crash work in terrific bad performance.
I'm not happy with duplicating the kdump FILEDUMP format code into QEMU because it is relatively big and ugly (temporary files make up a large part of that).
If it makes the code simpler and smaller it would be nice.
That's the point. I will make the code simpler. Thanks for your comments! -- Regards Qiao Nuohan