Re: linux-next: manual merge of the akpm-current tree with the powerpc tree

[Christophe Leroy <christophe.leroy@csgroup.eu>]

Le 15/04/2021 à 12:08, Christophe Leroy a écrit :
> 
> 
> Le 15/04/2021 à 12:07, Christophe Leroy a écrit :
>>
>>
>> Le 15/04/2021 à 11:58, Stephen Rothwell a écrit :
>>> Hi all,
>>>
>>> On Thu, 15 Apr 2021 19:44:17 +1000 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
>>>>
>>>> Today's linux-next merge of the akpm-current tree got a conflict in:
>>>>
>>>>    arch/powerpc/kernel/module.c
>>>>
>>>> between commit:
>>>>
>>>>    2ec13df16704 ("powerpc/modules: Load modules closer to kernel text")
>>>>
>>>> from the powerpc tree and commit:
>>>>
>>>>    4930ba789f8d ("powerpc/64s/radix: enable huge vmalloc mappings")
>>>>
>>>> from the akpm-current tree.
>>>>
>>>> I fixed it up (I think - see below) and can carry the fix as
>>>> necessary. This is now fixed as far as linux-next is concerned, but any
>>>> non trivial conflicts should be mentioned to your upstream maintainer
>>>> when your tree is submitted for merging.  You may also want to consider
>>>> cooperating with the maintainer of the conflicting tree to minimise any
>>>> particularly complex conflicts.
>>>>
>>>> -- 
>>>> Cheers,
>>>> Stephen Rothwell
>>>>
>>>> diff --cc arch/powerpc/kernel/module.c
>>>> index fab84024650c,cdb2d88c54e7..000000000000
>>>> --- a/arch/powerpc/kernel/module.c
>>>> +++ b/arch/powerpc/kernel/module.c
>>>> @@@ -88,29 -88,26 +89,42 @@@ int module_finalize(const Elf_Ehdr *hdr
>>>>        return 0;
>>>>    }
>>>> - #ifdef MODULES_VADDR
>>>>   -void *module_alloc(unsigned long size)
>>>>   +static __always_inline void *
>>>>   +__module_alloc(unsigned long size, unsigned long start, unsigned long end)
>>>>    {
>>>>   -    unsigned long start = VMALLOC_START;
>>>>   -    unsigned long end = VMALLOC_END;
>>>>   -
>>>>   -#ifdef MODULES_VADDR
>>>>   -    BUILD_BUG_ON(TASK_SIZE > MODULES_VADDR);
>>>>   -    start = MODULES_VADDR;
>>>>   -    end = MODULES_END;
>>>>   -#endif
>>>>   -
>>>> +     /*
>>>> +      * Don't do huge page allocations for modules yet until more testing
>>>> +      * is done. STRICT_MODULE_RWX may require extra work to support this
>>>> +      * too.
>>>> +      */
>>>> +
>>>>        return __vmalloc_node_range(size, 1, start, end, GFP_KERNEL,
>>>> -                     PAGE_KERNEL_EXEC, VM_FLUSH_RESET_PERMS, NUMA_NO_NODE,
>>>> +                     PAGE_KERNEL_EXEC,
>>>> +                     VM_NO_HUGE_VMAP | VM_FLUSH_RESET_PERMS,
>>>> +                     NUMA_NO_NODE,
>>>>                        __builtin_return_address(0));
>>>>    }
>>>>   +
>>>> ++
>>>>   +void *module_alloc(unsigned long size)
>>>>   +{
>>>> ++    unsigned long start = VMALLOC_START;
>>>> ++    unsigned long end = VMALLOC_END;
>>>>   +    unsigned long limit = (unsigned long)_etext - SZ_32M;
>>>>   +    void *ptr = NULL;
>>>>   +
>>>> ++#ifdef MODULES_VADDR
>>>>   +    BUILD_BUG_ON(TASK_SIZE > MODULES_VADDR);
>>>> ++    start = MODULES_VADDR;
>>>> ++    end = MODULES_END;
>>
>> The #endif should be here.
>>
>>
>>>>   +
>>>>   +    /* First try within 32M limit from _etext to avoid branch trampolines */
>>>>   +    if (MODULES_VADDR < PAGE_OFFSET && MODULES_END > limit)
> 
> Should also use start and end here instead of MODULES_VADDR  and MODULES_END


The cleanest however should be to define MODULES_VADDR and MODULES_END all the time with a fallback 
to VMALLOC_START/VMALLOC_END, to avoid the #ifdef.

The #ifdef was OK when we wanted to define modules_alloc() only when module area was different from 
vmalloc area, but now that we want modules_alloc() at all time, MODULES_VADDR and MODULES_END should 
be defined all the time.


> 
>>>> -         ptr = __module_alloc(size, limit, MODULES_END);
>>>> ++        ptr = __module_alloc(size, limit, end);
>>>>   +
>>>>   +    if (!ptr)
>>>> -         ptr = __module_alloc(size, MODULES_VADDR, MODULES_END);
>>>> ++#endif
>>>> ++        ptr = __module_alloc(size, start, end);
>>>>   +
>>>>   +    return ptr;
>>>>   +}
>>>> - #endif
>>>
>>> Unfortunately, it also needs this:
>>
>> Before the #endif is too far.
>>
>>>
>>> From: Stephen Rothwell <sfr@canb.auug.org.au>
>>> Date: Thu, 15 Apr 2021 19:53:58 +1000
>>> Subject: [PATCH] merge fix up for powerpc merge fix
>>>
>>> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
>>> ---
>>>   arch/powerpc/kernel/module.c | 2 ++
>>>   1 file changed, 2 insertions(+)
>>>
>>> diff --git a/arch/powerpc/kernel/module.c b/arch/powerpc/kernel/module.c
>>> index d8ab1ad2eb05..c060f99afd4d 100644
>>> --- a/arch/powerpc/kernel/module.c
>>> +++ b/arch/powerpc/kernel/module.c
>>> @@ -110,7 +110,9 @@ void *module_alloc(unsigned long size)
>>>   {
>>>       unsigned long start = VMALLOC_START;
>>>       unsigned long end = VMALLOC_END;
>>> +#ifdef MODULES_VADDR
>>>       unsigned long limit = (unsigned long)_etext - SZ_32M;
>>> +#endif
>>>       void *ptr = NULL;
>>>   #ifdef MODULES_VADDR
>>>