Christophe Leroy <christophe.leroy@xxxxxxxxxx> writes:
> Hi Ariel
>
> Le 09/04/2022 à 15:47, Ariel Marcovitch a écrit :
>> Hi Christophe, did you get the chance to look at this?
>
> I tested something this morning, it works for me, see below
>
>>
>> On 23/03/2022 21:06, Mike Rapoport wrote:
>>> Hi Catalin,
>>>
>>> On Wed, Mar 23, 2022 at 05:22:38PM +0000, Catalin Marinas wrote:
>>>> Hi Ariel,
>>>>
>>>> On Fri, Feb 18, 2022 at 09:45:51PM +0200, Ariel Marcovitch wrote:
>>>>> I was running a powerpc 32bit kernel (built using
>>>>> qemu_ppc_mpc8544ds_defconfig
>>>>> buildroot config, with enabling DEBUGFS+KMEMLEAK+HIGHMEM in the kernel
>>>>> config)
>
> ...
>
>>>>> I don't suppose I can just shuffle the calls in setup_arch() around,
>>>>> so I
>>>>> wanted to hear your opinions first
>>>> I think it's better if we change the logic than shuffling the calls.
>>>> IIUC MEMBLOCK_ALLOC_ACCESSIBLE means that __va() works on the phys
>>>> address return by memblock, so something like below (untested):
>>> MEMBLOCK_ALLOC_ACCESSIBLE means "anywhere", see commit e63075a3c937
>>> ("memblock: Introduce default allocation limit and use it to replace
>>> explicit ones"), so it won't help to detect high memory.
>>>
>>> If I remember correctly, ppc initializes memblock *very* early, so
>>> setting
>>> max_low_pfn along with lowmem_end_addr in
>>> arch/powerpc/mm/init_32::MMU_init() makes sense to me.
>>>
>>> Maybe ppc folks have other ideas...
>>> I've added Christophe who works on ppc32 these days.
>
> I think memblock is already available at the end of MMU_init() on PPC32
> and at the end of early_setup() on PPC64. It means it is ready when we
> enter setup_arch().
>
> I tested the change below, it works for me, I don't get any kmemleak
> report anymore.
>
> diff --git a/arch/powerpc/kernel/setup-common.c
> b/arch/powerpc/kernel/setup-common.c
> index 518ae5aa9410..9f4e50b176c9 100644
> --- a/arch/powerpc/kernel/setup-common.c
> +++ b/arch/powerpc/kernel/setup-common.c
> @@ -840,6 +840,9 @@ void __init setup_arch(char **cmdline_p)
> /* Set a half-reasonable default so udelay does something sensible */
> loops_per_jiffy = 500000000 / HZ;
>
> + /* Parse memory topology */
> + mem_topology_setup();
> +
> /* Unflatten the device-tree passed by prom_init or kexec */
> unflatten_device_tree();
The 64-bit/NUMA version of mem_topology_setup() requires the device tree
to be unflattened, so I don't think that can work.
Setting max_low_pfn etc in MMU_init() as Mike suggested seems more
likely to work.
But we might need to set it again in mem_topology_setup() though, so
that things that change memblock_end_of_DRAM() are reflected, eg. memory
limit or crash dump?
cheers
