On 2024/01/30 9:24, [email protected] wrote:
> Folks
> 
> The Linux kernel commit b9f174c811e3ae4ae8959dc57e6adb9990e913f4 
> (https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=b9f174c811e3ae4ae8959dc57e6adb9990e913f4)
>  added an ELF section for the ORC version identifier.  I think the crash 
> utility should use this section to identify the ORC version in use.
> 
> I want to implement this feature in the crash utility.  As I understand it, 
> we would need to read the kernel binary first, find out whether it has 
> .orc_header section and then read the section to figure out the version of 
> the ORC format in use.

thanks for the information.

I can see that there is the orc_header symbol and data also in a vmcore, 
so maybe it can be read easily:

crash> whatis orc_header
const u8 orc_header[20];
crash> rd -8 orc_header 20
ffffffffb718ef94:  fe 5d 32 bf 58 1b d6 3b 2c a9 a5 c6 5b a5 a6 25   ...
ffffffffb718efa4:  ea b3 fe 24

but it's a hash, not a version number, how are you going to use this?

Does crash have to have all historical hash values?  all of format 
changes do not affect crash's unwinder though.

Thanks,
Kazu

> 
> I do not have familiarity with the code for the crash utility.  Is it 
> possible for someone to advise on whether the above sounds reasonable?  If 
> yes, then how to divide the functionality among the files?  If no, then what 
> would a reasonable to have this functionality?
> 
> Thanks!
> --
> Crash-utility mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
> https://${domain_name}/admin/lists/devel.lists.crash-utility.osci.io/
> Contribution Guidelines: https://github.com/crash-utility/crash/wiki
--
Crash-utility mailing list -- [email protected]
To unsubscribe send an email to [email protected]
https://${domain_name}/admin/lists/devel.lists.crash-utility.osci.io/
Contribution Guidelines: https://github.com/crash-utility/crash/wiki

Reply via email to