linux-kernel - Re: [PATCH v1 1/2] mm/memblock: expose only miminal interface to add/walk physmem

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 30 Jun 2020 18:58:41 +0200
From:   David Hildenbrand <david@...hat.com>
To:     Mike Rapoport <rppt@...ux.ibm.com>
Cc:     linux-kernel@...r.kernel.org, linux-s390@...r.kernel.org,
        linux-mm@...ck.org, Heiko Carstens <heiko.carstens@...ibm.com>,
        Vasily Gorbik <gor@...ux.ibm.com>,
        Christian Borntraeger <borntraeger@...ibm.com>,
        Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [PATCH v1 1/2] mm/memblock: expose only miminal interface to
 add/walk physmem

>>  extern struct memblock memblock;
>> @@ -114,6 +110,19 @@ int memblock_remove(phys_addr_t base, phys_addr_t size);
>>  int memblock_free(phys_addr_t base, phys_addr_t size);
>>  int memblock_reserve(phys_addr_t base, phys_addr_t size);
>>  #ifdef CONFIG_HAVE_MEMBLOCK_PHYS_MAP
>> +/**
>> + * for_each_physmem_range - iterate through physmem areas not included in type.
>> + * @i: u64 used as loop variable
>> + * @type: ptr to memblock_type which excludes from the iteration, can be %NULL
>> + * @p_start: ptr to phys_addr_t for start address of the range, can be %NULL
>> + * @p_end: ptr to phys_addr_t for end address of the range, can be %NULL
>> + */
>> +#define for_each_physmem_range(i, type, p_start, p_end)			\
>> +	for (i = 0, __next_physmem_range(&i, type, p_start, p_end);	\
>> +	     i != (u64)ULLONG_MAX;					\
>> +	     __next_physmem_range(&i, type, p_start, p_end))
>> +void __next_physmem_range(u64 *idx, struct memblock_type *type,
>> +			  phys_addr_t *out_start, phys_addr_t *out_end);
> 
> __next_physmem_range() is not really necessary, the
> for_each_physmem_range() macro can use __next_mem_range() directly, but
> I suspect it won't look nice :)
> 
> Can you please make __next_physmem_range() static inline if we are to
> keep it?

The thing is, then I have to expose "physmem" to something outside
memblock.c. That's what I wanted to avoid here. (instead, have a minimal
interface that is sufficient enough for this special case of physmem -
add memory during boot, walk memory after boot. Performance is not an
issue).

[...]

>>   * Each region is represented by :c:type:`struct memblock_region` that
>>   * defines the region extents, its attributes and NUMA node id on NUMA
>>   * systems. Every memory type is described by the :c:type:`struct
>>   * memblock_type` which contains an array of memory regions along with
>> - * the allocator metadata. The memory types are nicely wrapped with
>> - * :c:type:`struct memblock`. This structure is statically initialzed
>> - * at build time. The region arrays for the "memory" and "reserved"
>> - * types are initially sized to %INIT_MEMBLOCK_REGIONS and for the
>> - * "physmap" type to %INIT_PHYSMEM_REGIONS.
>> + * the allocator metadata. The memory types (except "physmem") are nicely
>> + * wrapped with :c:type:`struct memblock`. This structure is statically
>> + * initialized at build time. The region arrays for the "memory" and
>> + * "reserved" types are initially sized to %INIT_MEMBLOCK_REGIONS.
> 
> I'd prefer
> 
> 	... The "memory" and "reserved" types are nicely wrapped 
> 	with :c:type:`struct memblock`. This structure is statically
>         initialized at build time. 
> 
> And while on this we can update the "reserved" size to match the code:
> 
> 	The region arrays are initilily sized to %INIT_MEMBLOCK_REGIONS
>  	for "memory" and %INIT_MEMBLOCK_RESERVED_REGIONS for "reserved".

Ack to both!

-- 
Thanks,

David / dhildenb