linux-kernel - Re: [PATCH 3/3] dma-pool: limit DMA and DMA32 zone size pools

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:   Wed, 17 Aug 2022 14:53:47 +0200
From:   Michal Hocko <mhocko@...e.com>
To:     Christoph Hellwig <hch@....de>
Cc:     iommu@...ts.linux.dev, Marek Szyprowski <m.szyprowski@...sung.com>,
        Robin Murphy <robin.murphy@....com>,
        David Rientjes <rientjes@...gle.com>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 3/3] dma-pool: limit DMA and DMA32 zone size pools

On Wed 17-08-22 08:06:47, Christoph Hellwig wrote:
> Limit the sizing of the atomic pools for allocations from
> ZONE_DMA and ZONE_DMA32 based on the number of pages actually
> present in those zones instead of the total memory.
> 
> Fixes: c84dc6e68a1d ("dma-pool: add additional coherent pools to map to gfp mask")
> Reported-by: Michal Hocko <mhocko@...e.com>
> Signed-off-by: Christoph Hellwig <hch@....de>

Yes this is much more reasonable estimation. Especially for the DMA
zone. I still believe some downscaling would make sense but as said
earlier nothing to really insist on. Also you likely want to have
pr_warn for non-poppulated DMA zone as pointed in other thread[1].

Other than that looks good to me. FWIW
Acked-by: Michal Hocko <mhocko@...e.com>

[1] https://lore.kernel.org/all/YvyN/vI7+cTv0geS@dhcp22.suse.cz/?q=http%3A%2F%2Flkml.kernel.org%2Fr%2FYvyN%2FvI7%2BcTv0geS%40dhcp22.suse.cz

> ---
>  kernel/dma/pool.c | 40 +++++++++++++++++++++++++++++-----------
>  1 file changed, 29 insertions(+), 11 deletions(-)
> 
> diff --git a/kernel/dma/pool.c b/kernel/dma/pool.c
> index 5b07b0379a501..f629c6dfd8555 100644
> --- a/kernel/dma/pool.c
> +++ b/kernel/dma/pool.c
> @@ -193,28 +193,46 @@ static unsigned long calculate_pool_size(unsigned long zone_pages)
>  	return max_t(unsigned long, nr_pages << PAGE_SHIFT, SZ_128K);
>  }
>  
> +#if defined(CONFIG_ZONE_DMA) || defined(CONFIG_ZONE_DMA32)
> +static unsigned long __init nr_managed_pages(int zone_idx)
> +{
> +	struct pglist_data *pgdat;
> +	unsigned long nr_pages = 0;
> +
> +	for_each_online_pgdat(pgdat)
> +		nr_pages += zone_managed_pages(&pgdat->node_zones[zone_idx]);
> +	return nr_pages;
> +}
> +#endif /* CONFIG_ZONE_DMA || CONFIG_ZONE_DMA32 */
> +
>  static int __init dma_atomic_pool_init(void)
>  {
> +	unsigned long nr_zone_dma_pages, nr_zone_dma32_pages;
> +
> +#ifdef CONFIG_ZONE_DMA
> +	nr_zone_dma_pages = nr_managed_pages(ZONE_DMA);
> +	if (nr_zone_dma_pages)
> +		atomic_pool_dma = __dma_atomic_pool_init(
> +				calculate_pool_size(nr_zone_dma_pages),
> +				GFP_DMA);
> +#endif
> +#ifdef CONFIG_ZONE_DMA32
> +	nr_zone_dma32_pages = nr_managed_pages(ZONE_DMA32);
> +	if (nr_zone_dma32_pages)
> +		atomic_pool_dma32 = __dma_atomic_pool_init(
> +				calculate_pool_size(nr_zone_dma32_pages),
> +				GFP_DMA32);
> +#endif
>  	/*
>  	 * If coherent_pool was not used on the command line, default the pool
>  	 * sizes to 128KB per 1GB of memory, min 128KB, max MAX_ORDER-1.
>  	 */
>  	if (!atomic_pool_size)
>  		atomic_pool_size = calculate_pool_size(totalram_pages());
> -
> -	INIT_WORK(&atomic_pool_work, atomic_pool_work_fn);
> -
>  	atomic_pool_kernel = __dma_atomic_pool_init(atomic_pool_size,
>  						    GFP_KERNEL);
> -	if (has_managed_dma()) {
> -		atomic_pool_dma = __dma_atomic_pool_init(atomic_pool_size,
> -						GFP_KERNEL | GFP_DMA);
> -	}
> -	if (IS_ENABLED(CONFIG_ZONE_DMA32)) {
> -		atomic_pool_dma32 = __dma_atomic_pool_init(atomic_pool_size,
> -						GFP_KERNEL | GFP_DMA32);
> -	}
>  
> +	INIT_WORK(&atomic_pool_work, atomic_pool_work_fn);
>  	dma_atomic_pool_debugfs_init();
>  	return 0;
>  }
> -- 
> 2.30.2

-- 
Michal Hocko
SUSE Labs