linux-kernel - Re: [RFC 07/10] arm64/mm: Detect and enable FEAT

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <a2ba2c98-bcf6-85cc-bdb7-633b5e5aa756@arm.com>
Date:   Mon, 19 Jul 2021 10:17:20 +0530
From:   Anshuman Khandual <anshuman.khandual@....com>
To:     Suzuki K Poulose <suzuki.poulose@....com>,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
        linux-mm@...ck.org
Cc:     akpm@...ux-foundation.org, mark.rutland@....com, will@...nel.org,
        catalin.marinas@....com, maz@...nel.org, james.morse@....com,
        steven.price@....com
Subject: Re: [RFC 07/10] arm64/mm: Detect and enable FEAT_LPA2



On 7/16/21 1:38 PM, Suzuki K Poulose wrote:
> On 16/07/2021 08:06, Anshuman Khandual wrote:
>>
>> On 7/14/21 1:51 PM, Suzuki K Poulose wrote:
>>> On 14/07/2021 03:21, Anshuman Khandual wrote:
>>>> Detect FEAT_LPA2 implementation early enough during boot when requested via
>>>> CONFIG_ARM64_PA_BITS_52_LPA2 and remember in a variable arm64_lpa2_enabled.
>>>> This variable could then be used to turn on TCR_EL1.TCR_DS effecting the 52
>>>> bits PA range or fall back to default 48 bits PA range if FEAT_LPA2 feature
>>>> was requested but found not to be implemented.
>>>>
>>>> Signed-off-by: Anshuman Khandual <anshuman.khandual@....com>
>>>> ---
>>>>    arch/arm64/include/asm/memory.h |  1 +
>>>>    arch/arm64/kernel/head.S        | 15 +++++++++++++++
>>>>    arch/arm64/mm/mmu.c             |  3 +++
>>>>    arch/arm64/mm/proc.S            |  9 +++++++++
>>>>    4 files changed, 28 insertions(+)
>>>>
>>>> diff --git a/arch/arm64/include/asm/memory.h b/arch/arm64/include/asm/memory.h
>>>> index 824a365..d0ca002 100644
>>>> --- a/arch/arm64/include/asm/memory.h
>>>> +++ b/arch/arm64/include/asm/memory.h
>>>> @@ -178,6 +178,7 @@
>>>>    #include <asm/bug.h>
>>>>      extern u64            vabits_actual;
>>>> +extern u64            arm64_lpa2_enabled;
>>>>      extern s64            memstart_addr;
>>>>    /* PHYS_OFFSET - the physical address of the start of memory. */
>>>> diff --git a/arch/arm64/kernel/head.S b/arch/arm64/kernel/head.S
>>>> index 6444147..9cf79ea 100644
>>>> --- a/arch/arm64/kernel/head.S
>>>> +++ b/arch/arm64/kernel/head.S
>>>> @@ -94,6 +94,21 @@ SYM_CODE_START(primary_entry)
>>>>        adrp    x23, __PHYS_OFFSET
>>>>        and    x23, x23, MIN_KIMG_ALIGN - 1    // KASLR offset, defaults to 0
>>>>        bl    set_cpu_boot_mode_flag
>>>> +
>>>> +#ifdef CONFIG_ARM64_PA_BITS_52_LPA2
>>>> +    mrs     x10, ID_AA64MMFR0_EL1
>>>> +    ubfx    x10, x10, #ID_AA64MMFR0_TGRAN_SHIFT, 4
>>>> +    cmp     x10, #ID_AA64MMFR0_TGRAN_LPA2
>>>> +    b.ne    1f
>>>
>>> For the sake of forward compatibility, this should be "b.lt"
>> Right, I guess we could assume that the feature will be present from the
>> current ID_AA64MMFR0_TGRAN_LPA2 values onward in the future. But should
>> not this also be capped at ID_AA64MMFR0_TGRAN_SUPPORTED_MAX as the upper
>> limit is different for 4K and 16K page sizes.
> 
> Absolutely.

ID_AA64MMFR0_TGRAN_SUPPORTED_MAX check there is not required as __enable_mmu()
already performs the required boundary check for a given page size support.