-----Original Message-----
> Hi, Kazu.
>
> For now, I would keep Pingfan's patch as is. I decided to test some other
> kernels, like 3.16 and 4.9 without commit
> 83e3c48729d9ebb7af5a31a504f3fd6aff0348c4.
>
> So, they would be valid for the first iteration and invalid for the second,
> just like Ubuntu's 4.4 kernel. As I couldn't reproduce the problem, I
> investigated further and realized I was testing without your commit for
> makedumpfile: 8f1aafa1643532ece86cba22f2187d0f42fb7ca3 ("PATCH Fix
> validate_mem_section()").
>
> With that one, all of Ubuntu's 4.4 kernel and Debian's 3.16 and 4.9 kernels
> dump correctly. Without that one, all fail, and I suppose it would be easily
> reproducible.
>
> So, in effect, we are looking for any entry with the present bit, and then
> checking it is valid kernel address. That seems to work just fine.
OK, good to turn out Pingfan's patch is enough. Applied.
https://sourceforge.net/p/makedumpfile/code/ci/e113f1c974c820f9633dc0073eda525d7575f365/
(Oops, forgot to add your ack. sorry about that..)
>
> As you mention, the only case we do the second check is for some downstream
> kernels (though I would argue we should care about those the most), but at
(IMHO, upstream makedumpfile can support downstream kernels but if a patch
to support them has a risk to upstream kernels, we should be careful.)
> least from the Ubuntu side, those should not be around in the field anymore,
> and, by default, those should be the rare exception anyway. So, I agree with
> your follow-up commit in your branch, as it also simplifies the code a lot.
>
> If you care, feel free to add my Ack to the two patches.
>
> Acked-by: Thadeu Lima de Souza Cascardo <cascardo@xxxxxxxxxxxxx>
Thank you for looking into those.
As for the follow-up commit, I'll test and consider whether to merge it
a little more. (Any comments or cons are welcome.)
Kazu
>
> Thanks for your patience.
> Thadeu Cascardo.
_______________________________________________
kexec mailing list
kexec@xxxxxxxxxxxxxxxxxxx
http://lists.infradead.org/mailman/listinfo/kexec
