Re: [RFC PATCH] ARM: Support allocating crashkernel above 4G for LPAE

[Baoquan He <bhe@redhat.com>]

On 08/06/24 at 10:46am, Jinjie Ruan wrote:
> 
> 
> On 2024/8/5 10:56, Baoquan He wrote:
> > On 08/05/24 at 09:23am, Jinjie Ruan wrote:
> >>
> >>
> >> On 2024/8/2 19:01, Russell King (Oracle) wrote:
> >>> On Fri, Aug 02, 2024 at 05:25:10PM +0800, Jinjie Ruan wrote:
> >>>> As ARM LPAE feature support accessing memory beyond the 4G limit, define
> >>>> HAVE_ARCH_CRASHKERNEL_RESERVATION_HIGH macro to support reserving crash
> >>>> memory above 4G for ARM32 LPAE.
> >>>>
> >>>> No test because there is no LPAE ARM32 hardware.
> >>>
> >>> Why are you submitting patches for features you can't test?
> >>>
> >>> I'm not going to apply this without it being properly tested, because I
> >>> don't believe that this will work in the generic case.
> >>>
> >>> If the crash kernel is located in memory outside of the lower 4GiB of
> >>> address space, and there is no alias within physical address space
> >>> for that memory, then there is *no* *way* for such a kernel to boot.
> >>
> >> I'm sorry that I released this patch without testing it. I actually
> >> intended to bring up this issue for discussion. If anyone has the
> >> environment to test it, that would be great. In the meantime, we could
> >> have a discussion on the significance and relevance of this approach.
> > 
> > I don't know arm32 and its LPAE. I know a little about x86_32 where
> > crashkernel can only be reserved below 896M because of the virtual
> > memory layout, and all memory above that is high memory which can't be
> > used as kernel memory directly. So from this patch, arm32 is different
> > than x86_32.
> 
> Hi，Baoquan
> 
> Does the following code make sense? Now parse_crashkernel() use
> HAVE_ARCH_CRASHKERNEL_RESERVATION_HIGH macro to parse "high", but use
> CONFIG_64BIT when reserving "low" memory in reserve_crashkernel_low().

I am fine with it. BUT have you addressed Russell's concern, e.g how to
test it actually?

> 
> And when LPAE is enabled in ARM32, and "high" is reserved,
> reserve_crashkernel_low() need also function ok.
> 
> --- a/kernel/crash_reserve.c
> +++ b/kernel/crash_reserve.c
> @@ -354,7 +354,7 @@ early_param("crashkernel", parse_crashkernel_dummy);
>  #ifdef CONFIG_ARCH_HAS_GENERIC_CRASHKERNEL_RESERVATION
>  static int __init reserve_crashkernel_low(unsigned long long low_size)
>  {
> -#ifdef CONFIG_64BIT
> +#ifdef HAVE_ARCH_CRASHKERNEL_RESERVATION_HIGH
>         unsigned long long low_base;
> 
> 
> > 
> >>
> >>>
> >>> So, right now I believe this patch to be *fundamentally* wrong.
> >>>
> >>
> > 
> > 
>