On 10/07/2014 06:02 PM, Dave Hansen wrote:
> On 09/29/2014 06:47 PM, Sasha Levin wrote:
>> struct page {
>> +#ifdef CONFIG_DEBUG_VM_POISON
>> + u32 poison_start;
>> +#endif
>> /* First double word block */
>> unsigned long flags; /* Atomic flags, some possibly
>> * updated asynchronously */
>> @@ -196,6 +199,9 @@ struct page {
>> #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
>> int _last_cpupid;
>> #endif
>> +#ifdef CONFIG_DEBUG_VM_POISON
>> + u32 poison_end;
>> +#endif
>> }
>
> Does this break slub's __cmpxchg_double_slab trick? I thought it
> required page->freelist and page->counters to be doubleword-aligned.
I'll probably have to switch it to 8 bytes anyways to make it work with
kasan. This should take care of the slub optimization as well.
> It's not like we really require this optimization when we're debugging,
> but trying to use it will unnecessarily slow things down.
>
> FWIW, if you're looking to trim down the number of lines of code, you
> could certainly play some macro tricks and #ifdef tricks.
>
> struct vm_poison {
> #ifdef CONFIG_DEBUG_VM_POISON
> u32 val;
> #endif
> };
>
> Then, instead of #ifdefs in each structure, you do:
>
> struct page {
> struct vm_poison poison_start;
> ... other gunk
> struct vm_poison poison_end;
> };
Agreed, I'll reword that in the next version.
Thanks,
Sasha
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>