[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH V3 3/3] xen/vm_event: Deny register writes if refused by vm_event reply

On 07/07/2015 03:55 PM, Lengyel, Tamas wrote:
> On Tue, Jul 7, 2015 at 5:06 AM, Razvan Cojocaru
> <rcojocaru@xxxxxxxxxxxxxxx <mailto:rcojocaru@xxxxxxxxxxxxxxx>> wrote:
>     On 07/06/2015 08:05 PM, Lengyel, Tamas wrote:
>     > @@ -410,6 +414,8 @@ void vm_event_resume(struct domain *d, struct
>     > vm_event_domain *ved)
>     >
>     >
>     >      #ifdef HAS_MEM_ACCESS
>     >              case VM_EVENT_REASON_MEM_ACCESS:
>     >     +        case VM_EVENT_REASON_MOV_TO_MSR:
>     >     +        case VM_EVENT_REASON_WRITE_CTRLREG:
>     >
>     >
>     > This doesn't really make much sense to be associated with MEM_ACCESS.
>     > I'm adding a separate arch-specific vm_event file in my other singlestep
>     > patch, I think these should trigger their appropriate handler there, not
>     > in mem_access_resume.
>     As said, I very much agree with the suggestion, but I don't see your
>     patch in staging yet.
>     Should I either (with the goal of ideally making the 4.6 release, and of
>     course unless somebody else has other issues with the patch or this
>     specific change):
>     * Add the new file your patch added again in my patch;
>     * If it's about to be commited soon (?) wait for your patch to make it
>     into staging (this I think would be the best path, if possible), or
>     * Leave it as it is for now and follow up post-4.6?
>     Thanks,
>     Razvan
> We can also just coordinate our two patch series. I'll push mine into
> github, you can rebase on top of it and submit the entire thing in one
> send. How does that sound?

That sounds very nice for an ideal scenario (thanks!), but I'm worried
that the additional synchronization overhead will have an additional
negative impact on the initial goal of getting these in before the 4.6
release. As long as moving as fast as possible is desirable, I'd prefer
to only depend on staging / master.

I guess frendliest way to your series to go about it now would be to
just add the file you've added with just my code in it (shouldn't be much).


Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.