Re: [PATCH 2.6.28-rc5 01/11] kmemleak: Add the base support

[Pekka Enberg]

Hi Catalin,

On Thu, Nov 20, 2008 at 1:30 PM, Catalin Marinas
<catalin.marinas@xxxxxxx> wrote:
> +#ifdef CONFIG_SMP
> +#define cache_line_align(x)    L1_CACHE_ALIGN(x)
> +#else
> +#define cache_line_align(x)    (x)
> +#endif

Maybe we should be put to <linux/cache.h> and call it cache_line_align_in_smp()?

> +/*
> + * Object allocation
> + */
> +static void *fast_cache_alloc(struct fast_cache *cache)
> +{
> +       unsigned int cpu = get_cpu();
> +       unsigned long flags;
> +       struct list_head *entry;
> +       struct fast_cache_page *page;
> +
> +       local_irq_save(flags);
> +
> +       if (list_empty(&cache->free_list[cpu]))
> +               __fast_cache_grow(cache, cpu);
> +
> +       entry = cache->free_list[cpu].next;
> +       page = entry_to_page(entry);
> +       list_del(entry);
> +       page->free_nr[cpu]--;
> +       BUG_ON(page->free_nr[cpu] < 0);
> +       fast_cache_dec_free(cache, cpu);
> +
> +       local_irq_restore(flags);
> +       put_cpu_no_resched();
> +
> +       return (void *)(entry + 1);
> +}

The slab allocators are pretty fast as well. Is there a reason you
can't use kmalloc() or kmem_cache_alloc() for this? You can fix the
recursion problem by adding a new GFP_NOLEAKTRACK flag that makes sure
memleak hooks are not invoked if it's set.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/