Re: [PATCH] powerpc: align syscall table for ppc32

[Christophe Leroy]

Le 20/08/2022 à 18:51, Masahiro Yamada a écrit :
> Christophe Leroy reported that commit 7b4537199a4a ("kbuild: link
> symbol CRCs at final link,  removing CONFIG_MODULE_REL_CRCS") broke
> mpc85xx_defconfig + CONFIG_RELOCATABLE=y.
> 
>      LD      vmlinux
>      SYSMAP  System.map
>      SORTTAB vmlinux
>      CHKREL  vmlinux
>    WARNING: 451 bad relocations
>    c0b312a9 R_PPC_UADDR32     .head.text-0x3ff9ed54
>    c0b312ad R_PPC_UADDR32     .head.text-0x3ffac224
>    c0b312b1 R_PPC_UADDR32     .head.text-0x3ffb09f4
>    c0b312b5 R_PPC_UADDR32     .head.text-0x3fe184dc
>    c0b312b9 R_PPC_UADDR32     .head.text-0x3fe183a8
>        ...
> 
> The compiler emits a bunch of R_PPC_UADDR32, which is not supported by
> arch/powerpc/kernel/reloc_32.S.
> 
> The reason is there exists an unaligned symbol.
> 
>    $ powerpc-linux-gnu-nm -n vmlinux
>      ...
>    c0b31258 d spe_aligninfo
>    c0b31298 d __func__.0
>    c0b312a9 D sys_call_table
>    c0b319b8 d __func__.0
> 
> Commit 7b4537199a4a is not the root cause. Even before that, I can
> reproduce the same issue for mpc85xx_defconfig + CONFIG_RELOCATABLE=y
> + CONFIG_MODVERSIONS=n.
> 
> It is just that nobody did not notice it because when CONFIG_MODVERSIONS
> is enabled, a __crc_* symbol inserted before sys_call_table was hiding
> the unalignment issue.
> 
> I checked the commit history, but I could not understand commit
> 46b45b10f142 ("[POWERPC] Align the sys_call_table").
> 
> It said 'Our _GLOBAL macro does a ".align 2" so the alignment is fine
> for 32 bit'. I checked the _GLOBAL in include/asm-powerpc/ppc_asm.h
> at that time. _GLOBAL specifies ".align 2" for ppc64, but no .align
> for ppc32.
> 
> Commit c857c43b34ec ("powerpc: Don't use a function descriptor for
> system call table") removed _GLOBAL from the syscall table.
> 
> Anyway, adding alignment to the syscall table for ppc32 fixes the issue.
> 
> I am not giving Fixes tag because I do not know since when it has been
> broken, but presumably it has been for a long while.
> 
> Link: https://lore.kernel.org/lkml/38605f6a-a568-f884-f06f-ea4da5b214f0@xxxxxxxxxx/
> Reported-by: Christophe Leroy <christophe.leroy@xxxxxxxxxx>
> Signed-off-by: Masahiro Yamada <masahiroy@xxxxxxxxxx>

Tested-by: Christophe Leroy <christophe.leroy@xxxxxxxxxx>

> ---
> 
>   arch/powerpc/kernel/systbl.S | 1 +
>   1 file changed, 1 insertion(+)
> 
> diff --git a/arch/powerpc/kernel/systbl.S b/arch/powerpc/kernel/systbl.S
> index cb3358886203..6c1db3b6de2d 100644
> --- a/arch/powerpc/kernel/systbl.S
> +++ b/arch/powerpc/kernel/systbl.S
> @@ -18,6 +18,7 @@
>   	.p2align	3
>   #define __SYSCALL(nr, entry)	.8byte entry
>   #else
> +	.p2align	2
>   #define __SYSCALL(nr, entry)	.long entry
>   #endif
>