Section mismatches in memblock

Section mismatches in memblock

[H. Peter Anvin]

o.i386-allconfig/make.log:WARNING: mm/built-in.o(.text+0x48b6f): Section
mismatch in reference from the function memblock_is_memory() to the
variable .init.data:memblock
o.i386-allconfig/make.log:The function memblock_is_memory() references
o.i386-allconfig/make.log:This is often because memblock_is_memory lacks
a __initdata
o.i386-allconfig/make.log:WARNING: mm/built-in.o(.text+0x48bb0): Section
mismatch in reference from the function memblock_is_region_memory() to
the variable .init.data:memblock
o.i386-allconfig/make.log:The function memblock_is_region_memory()
references
o.i386-allconfig/make.log:This is often because
memblock_is_region_memory lacks a __initdata
o.i386-allconfig/make.log:WARNING: mm/built-in.o(.text+0x48bd5): Section
mismatch in reference from the function memblock_is_region_memory() to
the variable .init.data:memblock
o.i386-allconfig/make.log:The function memblock_is_region_memory()
references
o.i386-allconfig/make.log:This is often because
memblock_is_region_memory lacks a __initdata

Happens with both i386 and x86-64.

	-hpa

Re: Section mismatches in memblock

[Yinghai Lu]

On 10/05/2010 03:46 PM, H. Peter Anvin wrote:
> o.i386-allconfig/make.log:WARNING: mm/built-in.o(.text+0x48b6f): Section
> mismatch in reference from the function memblock_is_memory() to the
> variable .init.data:memblock
> o.i386-allconfig/make.log:The function memblock_is_memory() references
> o.i386-allconfig/make.log:This is often because memblock_is_memory lacks
> a __initdata
> o.i386-allconfig/make.log:WARNING: mm/built-in.o(.text+0x48bb0): Section
> mismatch in reference from the function memblock_is_region_memory() to
> the variable .init.data:memblock
> o.i386-allconfig/make.log:The function memblock_is_region_memory()
> references
> o.i386-allconfig/make.log:This is often because
> memblock_is_region_memory lacks a __initdata
> o.i386-allconfig/make.log:WARNING: mm/built-in.o(.text+0x48bd5): Section
> mismatch in reference from the function memblock_is_region_memory() to
> the variable .init.data:memblock
> o.i386-allconfig/make.log:The function memblock_is_region_memory()
> references
> o.i386-allconfig/make.log:This is often because
> memblock_is_region_memory lacks a __initdata
> 
> Happens with both i386 and x86-64.


[PATCH] memblock: Fix section mismatch warnings with memblock_search()

HPA found
o.i386-allconfig/make.log:WARNING: mm/built-in.o(.text+0x48b6f): Section
mismatch in reference from the function memblock_is_memory() to the
variable .init.data:memblock
o.i386-allconfig/make.log:The function memblock_is_memory() references
o.i386-allconfig/make.log:This is often because memblock_is_memory lacks
a __initdata
o.i386-allconfig/make.log:WARNING: mm/built-in.o(.text+0x48bb0): Section
mismatch in reference from the function memblock_is_region_memory() to
the variable .init.data:memblock
o.i386-allconfig/make.log:The function memblock_is_region_memory()
references
o.i386-allconfig/make.log:This is often because
memblock_is_region_memory lacks a __initdata

looks like gcc problem.

| static int memblock_search(struct memblock_type *type, phys_addr_t addr)
| {
|	.....
| }
|
|int __init memblock_is_reserved(phys_addr_t addr)
|{
|        return memblock_search(&memblock.reserved, addr) != -1;
|}
|
|int __init_memblock memblock_is_memory(phys_addr_t addr)
|{
|        return memblock_search(&memblock.memory, addr) != -1;
|}

Because We already have __init for those two functions.

Just make memblock_search to have __init_memblock atribute.


Reported-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Yinghai Lu <Yinghai@kernel.org>

---
 mm/memblock.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: linux-2.6/mm/memblock.c
===================================================================
--- linux-2.6.orig/mm/memblock.c
+++ linux-2.6/mm/memblock.c
@@ -653,7 +653,7 @@ void __init memblock_enforce_memory_limi
 	}
 }
 
-static int memblock_search(struct memblock_type *type, phys_addr_t addr)
+static int __init_memblock memblock_search(struct memblock_type *type, phys_addr_t addr)
 {
 	unsigned int left = 0, right = type->cnt;
 

Re: Section mismatches in memblock

[H. Peter Anvin]

On 10/05/2010 04:30 PM, Yinghai Lu wrote:
> 
> looks like gcc problem.

What makes you say that?

> | static int memblock_search(struct memblock_type *type, phys_addr_t addr)
> | {
> |	.....
> | }
> |
> |int __init memblock_is_reserved(phys_addr_t addr)
> |{
> |        return memblock_search(&memblock.reserved, addr) != -1;
> |}
> |
> |int __init_memblock memblock_is_memory(phys_addr_t addr)
> |{
> |        return memblock_search(&memblock.memory, addr) != -1;
> |}
> 
> Because We already have __init for those two functions.
> 
> Just make memblock_search to have __init_memblock atribute.
> 

Wrong functions!  Furhtermore, at least the tip tree definitely does not
have __init_memblock here:


static int memblock_search(struct memblock_type *type, phys_addr_t addr)
{
        unsigned int left = 0, right = type->cnt;

        do {
                unsigned int mid = (right + left) / 2;

                if (addr < type->regions[mid].base)
                        right = mid;
                else if (addr >= (type->regions[mid].base +
                                  type->regions[mid].size))
                        left = mid + 1;
                else
                        return mid;
        } while (left < right);
        return -1;
}

int __init memblock_is_reserved(phys_addr_t addr)
{
        return memblock_search(&memblock.reserved, addr) != -1;
}

int memblock_is_memory(phys_addr_t addr)
{
        return memblock_search(&memblock.memory, addr) != -1;
}

int memblock_is_region_memory(phys_addr_t base, phys_addr_t size)
{
        int idx = memblock_search(&memblock.reserved, base);

        if (idx == -1)
                return 0;
        return memblock.reserved.regions[idx].base <= base &&
                (memblock.reserved.regions[idx].base +
                 memblock.reserved.regions[idx].size) >= (base + size);
}


... so I don't know why you're saying that you already have them.

	-hpa

-- 
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.

Re: Section mismatches in memblock

[H. Peter Anvin]

On 10/05/2010 07:30 PM, H. Peter Anvin wrote:
> 
> Wrong functions!  Furhtermore, at least the tip tree definitely does not
> have __init_memblock here:
> 

Nevermind.  I had missed that Ingo had put stuff into core/memblock on
top of x86/memblock, just to be confusing.

	-hpa

Re: Section mismatches in memblock

[Ingo Molnar]

* H. Peter Anvin <hpa@kernel.org> wrote:

> On 10/05/2010 07:30 PM, H. Peter Anvin wrote:
> > 
> > Wrong functions!  Furhtermore, at least the tip tree definitely does not
> > have __init_memblock here:
> > 
> 
> Nevermind.  I had missed that Ingo had put stuff into core/memblock on
> top of x86/memblock, just to be confusing.

Correct - we better keep it as a generic, non-x86 topic as it affects 
other architectures. We can zap x86/memblock (it's fully contained in 
core/memblock) so that it's not confusing.

Thanks,

	Ingo