Re: kern/57560: xennetback_xenbus.c rev 1.109 broke guests, causes DOM0 reboot

[Brad Spencer <brad%anduin.eldar.org@localhost>]

Taylor R Campbell <riastradh%NetBSD.org@localhost> writes:

> The following reply was made to PR port-xen/57560; it has been noted by GNATS.
>
> From: Taylor R Campbell <riastradh%NetBSD.org@localhost>
> To: Brad Spencer <brad%anduin.eldar.org@localhost>
> Cc: gnats-bugs%NetBSD.org@localhost, martin%NetBSD.org@localhost
> Subject: Re: kern/57560: xennetback_xenbus.c rev 1.109 broke guests, causes DOM0 reboot
> Date: Fri, 4 Aug 2023 20:19:33 +0000
>
>  I reverted the commits on HEAD and pulled the revert up to netbsd-10,
>  so the immediate problem should be addressed.
>  
>  However, the code is also slightly problematic because it's still
>  missing a barrier in one place, and still using a much-too-expensive
>  barrier (even on x86!) in the loop in another place, and still has
>  several nonsensical barriers in other places that can't hurt
>  correctness but do hurt understandability and maintainability.
>  
>  So I went through to try to redo the changes more carefully.  If you
>  have the opportunity to test or bisect on another patch, I've drafted
>  a series of changes, attached:
>  
>  - The .diff file is a single diff of the whole changes.  If it works,
>    great -- I can commit.  But if not:
>  
>  - The .patch file is a series of individual changes for bisection to
>    see exactly which one breaks it.
>  


I can help test this, but I am really out of time in the next 24 - 48
hours at this point.  I expect that I may be able to do something in the
next week at some point, just don't know when.


Thanks very much for looking into this.


-- 
Brad Spencer - brad%anduin.eldar.org@localhost - KC8VKS - http://anduin.eldar.org