On Fri, 10 Jun 2022 at 11:27, Kefeng Wang <[email protected]> wrote:
>
>
> On 2022/6/7 17:38, Wupeng Ma wrote:
> > From: Ma Wupeng <[email protected]>
> >
> > Commit b05b9f5f9dcf ("x86, mirror: x86 enabling - find mirrored memory ranges")
> > introduced mirrored memory support for x86 and this could be used on arm64.
> >
> > Since we only support this feature on arm64, efi_find_mirror() won't be placed
> > into efi_init(), which is used by riscv/arm/arm64, it is added in setup_arch()
> > to scan the memory map and mark mirrored memory in memblock.
>
> Move into efi_init() looks better, it won't bring negative effects on
> arm/riscv.
>
> but let's maintainer to make a decision.
>
> Reviewed-by: Kefeng Wang <[email protected]>
>
I don't disagree with that in principle, but x86 calls the efi fake
memmap stuff between efi_init() and efi_find_mirror(), and I already
indicated that I don't want to enable fake memmap on !X86.
But I do think there is some room for improvement here: we could move
things like efi_mokvar_table_init() and efi_esrt_init() into
efi_init() as well, and make efi_fake_memmap() do nothing on !X86 so
we can move it into efi_init() too.
> >
> > Signed-off-by: Ma Wupeng <[email protected]>
> > ---
> > Documentation/admin-guide/kernel-parameters.txt | 2 +-
> > arch/arm64/kernel/setup.c | 1 +
> > 2 files changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> > index 8090130b544b..e3537646b6f7 100644
> > --- a/Documentation/admin-guide/kernel-parameters.txt
> > +++ b/Documentation/admin-guide/kernel-parameters.txt
> > @@ -2301,7 +2301,7 @@
> >
> > keepinitrd [HW,ARM]
> >
> > - kernelcore= [KNL,X86,IA-64,PPC]
> > + kernelcore= [KNL,X86,IA-64,PPC,ARM64]
> > Format: nn[KMGTPE] | nn% | "mirror"
> > This parameter specifies the amount of memory usable by
> > the kernel for non-movable allocations. The requested
> > diff --git a/arch/arm64/kernel/setup.c b/arch/arm64/kernel/setup.c
> > index cf3a759f10d4..6e9acd7ecf0f 100644
> > --- a/arch/arm64/kernel/setup.c
> > +++ b/arch/arm64/kernel/setup.c
> > @@ -328,6 +328,7 @@ void __init __no_sanitize_address setup_arch(char **cmdline_p)
> >
> > xen_early_init();
> > efi_init();
> > + efi_find_mirror();
> >
> > if (!efi_enabled(EFI_BOOT) && ((u64)_text % MIN_KIMG_ALIGN) != 0)
> > pr_warn(FW_BUG "Kernel image misaligned at boot, please fix your bootloader!");
On 2022/6/10 17:34, Ard Biesheuvel wrote:
> On Fri, 10 Jun 2022 at 11:27, Kefeng Wang <[email protected]> wrote:
>>
>> On 2022/6/7 17:38, Wupeng Ma wrote:
>>> From: Ma Wupeng <[email protected]>
>>>
>>> Commit b05b9f5f9dcf ("x86, mirror: x86 enabling - find mirrored memory ranges")
>>> introduced mirrored memory support for x86 and this could be used on arm64.
>>>
>>> Since we only support this feature on arm64, efi_find_mirror() won't be placed
>>> into efi_init(), which is used by riscv/arm/arm64, it is added in setup_arch()
>>> to scan the memory map and mark mirrored memory in memblock.
>> Move into efi_init() looks better, it won't bring negative effects on
>> arm/riscv.
>>
>> but let's maintainer to make a decision.
>>
>> Reviewed-by: Kefeng Wang <[email protected]>
>>
> I don't disagree with that in principle, but x86 calls the efi fake
> memmap stuff between efi_init() and efi_find_mirror(), and I already
> indicated that I don't want to enable fake memmap on !X86.
I mean move into efi_init() in drivers/firmware/efi/efi-init.c which
only used
in arm32/arm64/riscv.
>
> But I do think there is some room for improvement here: we could move
> things like efi_mokvar_table_init() and efi_esrt_init() into
> efi_init() as well, and make efi_fake_memmap() do nothing on !X86 so
> we can move it into efi_init() too.
x86 has own efi_init() in arch/x86/platform/efi/efi.c, it seams that all
the above
function could move into x86's own efi_init(), not sure, but we could do
it later
if it possible.
>
>>> Signed-off-by: Ma Wupeng <[email protected]>
>>> ---
>>> Documentation/admin-guide/kernel-parameters.txt | 2 +-
>>> arch/arm64/kernel/setup.c | 1 +
>>> 2 files changed, 2 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
>>> index 8090130b544b..e3537646b6f7 100644
>>> --- a/Documentation/admin-guide/kernel-parameters.txt
>>> +++ b/Documentation/admin-guide/kernel-parameters.txt
>>> @@ -2301,7 +2301,7 @@
>>>
>>> keepinitrd [HW,ARM]
>>>
>>> - kernelcore= [KNL,X86,IA-64,PPC]
>>> + kernelcore= [KNL,X86,IA-64,PPC,ARM64]
>>> Format: nn[KMGTPE] | nn% | "mirror"
>>> This parameter specifies the amount of memory usable by
>>> the kernel for non-movable allocations. The requested
>>> diff --git a/arch/arm64/kernel/setup.c b/arch/arm64/kernel/setup.c
>>> index cf3a759f10d4..6e9acd7ecf0f 100644
>>> --- a/arch/arm64/kernel/setup.c
>>> +++ b/arch/arm64/kernel/setup.c
>>> @@ -328,6 +328,7 @@ void __init __no_sanitize_address setup_arch(char **cmdline_p)
>>>
>>> xen_early_init();
>>> efi_init();
>>> + efi_find_mirror();
>>>
>>> if (!efi_enabled(EFI_BOOT) && ((u64)_text % MIN_KIMG_ALIGN) != 0)
>>> pr_warn(FW_BUG "Kernel image misaligned at boot, please fix your bootloader!");
> .
On Fri, 10 Jun 2022 at 12:24, Kefeng Wang <[email protected]> wrote:
>
>
> On 2022/6/10 17:34, Ard Biesheuvel wrote:
> > On Fri, 10 Jun 2022 at 11:27, Kefeng Wang <[email protected]> wrote:
> >>
> >> On 2022/6/7 17:38, Wupeng Ma wrote:
> >>> From: Ma Wupeng <[email protected]>
> >>>
> >>> Commit b05b9f5f9dcf ("x86, mirror: x86 enabling - find mirrored memory ranges")
> >>> introduced mirrored memory support for x86 and this could be used on arm64.
> >>>
> >>> Since we only support this feature on arm64, efi_find_mirror() won't be placed
> >>> into efi_init(), which is used by riscv/arm/arm64, it is added in setup_arch()
> >>> to scan the memory map and mark mirrored memory in memblock.
> >> Move into efi_init() looks better, it won't bring negative effects on
> >> arm/riscv.
> >>
> >> but let's maintainer to make a decision.
> >>
> >> Reviewed-by: Kefeng Wang <[email protected]>
> >>
> > I don't disagree with that in principle, but x86 calls the efi fake
> > memmap stuff between efi_init() and efi_find_mirror(), and I already
> > indicated that I don't want to enable fake memmap on !X86.
>
> I mean move into efi_init() in drivers/firmware/efi/efi-init.c which
> only used
>
> in arm32/arm64/riscv.
>
> >
> > But I do think there is some room for improvement here: we could move
> > things like efi_mokvar_table_init() and efi_esrt_init() into
> > efi_init() as well, and make efi_fake_memmap() do nothing on !X86 so
> > we can move it into efi_init() too.
>
> x86 has own efi_init() in arch/x86/platform/efi/efi.c, it seams that all
> the above
>
> function could move into x86's own efi_init(), not sure, but we could do
> it later
>
> if it possible.
>
Yes, I see what you mean now. I agree that would be better.