On 08/02/2022 21:31, bhe@xxxxxxxxxx wrote: > [...] >> So, what are the opinions from kdump maintainers about this idea? >> Baoquan / Vivek / Dave, does it make sense to you? Do you have any >> suggestions/concerns to add on top of Petr draft? > > Yeah, it's reasonable. As I replied to Michael in another thread, I > think splitting the current notifier list is a good idea. At least the > code to archieve hyper-V's goal with panic_notifier is a little odd and > should be taken out and execute w/o conditional before kdump, and maybe > some others Petr has combed out. > > For those which will be switched on with the need of adding panic_notifier > or panic_print into cmdline, the heavy users like HATAYAMA and Masa can > help check. > > For Petr's draft code, does it mean hyper-V need another knob to trigger > the needed notifiers? Will you go with the draft direclty? Hyper-V now > runs panic notifiers by default, just a reminder. > Hi Baoquan, thanks for your comments. I'll need to study the Hyper-V code and how it's done today - I guess most part of this implementation will be studying the notifiers we have currently, split them among the 3 new notifiers and comb them into patches, so they can be reviewed for all relevant maintainers (who know the code we are changing). I'm not sure if I go directly with the draft, likely it'll have some changes, but the draft should be the skeleton of the new implementation. Specially if you/other kdump maintainers agree it's a good idea =) Cheers, Guilherme
