[PATCH linux-next 1/3] x86, crash: don't nest CONFIG_CRASH_DUMP ifdef inside CONFIG_KEXEC_CODE ifdef scope

Baoquan He bhe at redhat.com
Mon Jan 29 19:58:08 PST 2024


On 01/30/24 at 01:39am, Michael Kelley wrote:
> From: Baoquan He <bhe at redhat.com>
> > 
> > On 01/29/24 at 06:27pm, Michael Kelley wrote:
> > > From: Baoquan He <bhe at redhat.com> Sent: Monday, January 29, 2024
> > 5:51 AM
> > > >
> > > > Michael pointed out that the #ifdef CONFIG_CRASH_DUMP is nested inside
> > > > arch/x86/xen/enlighten_hvm.c.
> > >
> > > Did some words get left out in the above sentence?  It mentions the Xen
> > > case, but not the Hyper-V case.  I'm not sure what you intended.
> > 
> > Thanks a lot for your careful reviewing.
> > 
> > Yeah, I tried to list all affected file names, seems my vim editor threw
> > away some words. And I forgot mentioning the change in reboot.c.
> > 
> > I adjusted log as below according to your comments, do you think it's OK
> > now?
> 
> Yes -- looks like everything is included and clear up my confusion.  But
> I still have two small nits per below. :-)

Right, I will grabbed them into v2. Thanks again.

> 
> > 
> > ===
> > Michael pointed out that the #ifdef CONFIG_CRASH_DUMP is nested inside
> > CONFIG_KEXEC_CODE ifdef scope in some XEN, HyperV codes.
> 
> s/Hyper-V/HyperV/
> 
> > 
> > Although the nesting works well too since CONFIG_CRASH_DUMP has
> > dependency on CONFIG_KEXEC_CORE, it may cause confusion because there
> > are places where it's not nested, and people may think it needs be nested
> 
> s/needs to be/needs be/
> 
> > even though it doesn't have to.
> > 
> > Fix that by moving  CONFIG_CRASH_DUMP ifdeffery of codes out of
> > CONFIG_KEXEC_CODE ifdeffery scope.
> > 
> > And also put function machine_crash_shutdown() definition inside
> > CONFIG_CRASH_DUMP ifdef scope instead of CONFIG_KEXEC_CORE ifdef.
> > 
> > And also fix a building error Nathan reported as below by replacing
> > CONFIG_KEXEC_CORE ifdef with CONFIG_VMCORE_INFO ifdef.
> > ......
> > ===
> > 
> > Thanks
> > Baoquan
> 




More information about the kexec mailing list