Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Daniel Axtens <dja@axtens.net>
To:     Michael Ellerman <mpe@ellerman.id.au>,
        Christophe Leroy <christophe.leroy@csgroup.eu>,
        Benjamin Herrenschmidt <benh@kernel.crashing.org>,
        Paul Mackerras <paulus@samba.org>
Cc:     linux-kernel@vger.kernel.org, linuxppc-dev@lists.ozlabs.org
Subject: Re: [PATCH 2/3] powerpc: Define swapper_pg_dir[] in C
In-Reply-To: <87czsc21st.fsf@mpe.ellerman.id.au>
References: <5838caffa269e0957c5a50cc85477876220298b0.1623063174.git.christophe.leroy@csgroup.eu>
 <5e3f1b8a4695c33ccc80aa3870e016bef32b85e1.1623063174.git.christophe.leroy@csgroup.eu>
 <871r8siyqm.fsf@dja-thinkpad.axtens.net>
 <87czsc21st.fsf@mpe.ellerman.id.au>
Date:   Thu, 24 Jun 2021 12:36:43 +1000
Message-ID: <874kdoyon8.fsf@dja-thinkpad.axtens.net>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk

Michael Ellerman <mpe@ellerman.id.au> writes:

> Daniel Axtens <dja@axtens.net> writes:
>> Hi Christophe,
>>
>> This breaks booting a radix KVM guest with 4k pages for me:
>>
>> make pseries_le_defconfig
>> scripts/config -d CONFIG_PPC_64K_PAGES
>> scripts/config -e CONFIG_PPC_4K_PAGES
>> make vmlinux
>> sudo qemu-system-ppc64 -enable-kvm -M pseries -m 1G -nographic -vga none -smp 4 -cpu host -kernel vmlinux
>>
>> Boot hangs after printing 'Booting Linux via __start()' and qemu's 'info
>> registers' reports that it's stuck at the instruction fetch exception.
>>
>> My host is Power9, 64k page size radix, and
>> gcc (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0, GNU ld (GNU Binutils for Ubuntu) 2.34
>>
>
> ...
>>> diff --git a/arch/powerpc/kernel/head_64.S b/arch/powerpc/kernel/head_64.S
>>> index 730838c7ca39..79f2d1e61abd 100644
>>> --- a/arch/powerpc/kernel/head_64.S
>>> +++ b/arch/powerpc/kernel/head_64.S
>>> @@ -997,18 +997,3 @@ start_here_common:
>>>  0:	trap
>>>  	EMIT_BUG_ENTRY 0b, __FILE__, __LINE__, 0
>>>  	.previous
>>> -
>>> -/*
>>> - * We put a few things here that have to be page-aligned.
>>> - * This stuff goes at the beginning of the bss, which is page-aligned.
>>> - */
>>> -	.section ".bss"
>>> -/*
>>> - * pgd dir should be aligned to PGD_TABLE_SIZE which is 64K.
>>> - * We will need to find a better way to fix this
>>> - */
>>> -	.align	16
>>> -
>>> -	.globl	swapper_pg_dir
>>> -swapper_pg_dir:
>>> -	.space	PGD_TABLE_SIZE
>
> This is now 4K aligned whereas it used to be 64K.
>
> This fixes it and is not completely ugly?
>
> diff --git a/arch/powerpc/mm/pgtable.c b/arch/powerpc/mm/pgtable.c
> index 1707ab580ee2..298469beaa90 100644
> --- a/arch/powerpc/mm/pgtable.c
> +++ b/arch/powerpc/mm/pgtable.c
> @@ -28,7 +28,13 @@
>  #include <asm/hugetlb.h>
>  #include <asm/pte-walk.h>
>  
> -pgd_t swapper_pg_dir[MAX_PTRS_PER_PGD] __page_aligned_bss;
> +#ifdef CONFIG_PPC64
> +#define PGD_ALIGN 0x10000
> +#else
> +#define PGD_ALIGN PAGE_SIZE
> +#endif
> +
> +pgd_t swapper_pg_dir[MAX_PTRS_PER_PGD] __section(".bss..page_aligned") __aligned(PGD_ALIGN);

The fix works for me, thank you.

Kind regards,
Daniel
>  
>  static inline int is_exec_fault(void)
>  {
>
>
> cheers