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]
Message-ID: <757018c2d9bd235cd2dad4363fb9a54354c9a372.camel@redhat.com>
Date: Tue, 30 Jan 2024 13:37:25 +0100
From: Paolo Abeni <pabeni@...hat.com>
To: Geoff Levand <geoff@...radead.org>, sambat goson <sombat3960@...il.com>,
  Christophe Leroy <christophe.leroy@...roup.eu>, "David S. Miller"
 <davem@...emloft.net>, "netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Re: [PATCH net] ps3/gelic: Fix SKB allocation

Hi,

On Fri, 2024-01-26 at 18:25 +0900, Geoff Levand wrote:
> Commit 3ce4f9c3fbb3 ("net/ps3_gelic_net: Add gelic_descr structures")
> of 6.8-rc1 did not set up the ps3 gelic network SKB's correctly,
> resulting in a kernel panic.
>     
> This fix changes the way the napi buffer and corresponding SKB are
> allocated and managed.
>     
> Reported-by: sambat goson <sombat3960@...il.com>
> Fixes: 3ce4f9c3fbb3 ("net/ps3_gelic_net: Add gelic_descr structures")
> Signed-off-by: Geoff Levand <geoff@...radead.org>

The patch overall looks correct to me, but there are a few formal
issues worthy to be addressed, see below.

> 
> diff --git a/drivers/net/ethernet/toshiba/ps3_gelic_net.c b/drivers/net/ethernet/toshiba/ps3_gelic_net.c
> index d5b75af163d3..1870f173e850 100644
> --- a/drivers/net/ethernet/toshiba/ps3_gelic_net.c
> +++ b/drivers/net/ethernet/toshiba/ps3_gelic_net.c
> @@ -375,20 +375,16 @@ static int gelic_card_init_chain(struct gelic_card *card,
>  static int gelic_descr_prepare_rx(struct gelic_card *card,
>  				  struct gelic_descr *descr)
>  {
> -	static const unsigned int rx_skb_size =
> -		ALIGN(GELIC_NET_MAX_FRAME, GELIC_NET_RXBUF_ALIGN) +
> -		GELIC_NET_RXBUF_ALIGN - 1;
> +	static const unsigned int napi_buff_size =
> +		round_up(GELIC_NET_MAX_FRAME, GELIC_NET_RXBUF_ALIGN);
> +

No empty line in the declaration area.

> +	struct device *dev = ctodev(card);
>  	dma_addr_t cpu_addr;
> -	int offset;
> +	void *napi_buff;
>  
>  	if (gelic_descr_get_status(descr) !=  GELIC_DESCR_DMA_NOT_IN_USE)
> -		dev_info(ctodev(card), "%s: ERROR status\n", __func__);
> +		dev_info(dev, "%s: ERROR status\n", __func__);
>  
> -	descr->skb = netdev_alloc_skb(*card->netdev, rx_skb_size);
> -	if (!descr->skb) {
> -		descr->hw_regs.payload.dev_addr = 0; /* tell DMAC don't touch memory */
> -		return -ENOMEM;
> -	}
>  	descr->hw_regs.dmac_cmd_status = 0;
>  	descr->hw_regs.result_size = 0;
>  	descr->hw_regs.valid_size = 0;
> @@ -397,24 +393,33 @@ static int gelic_descr_prepare_rx(struct gelic_card *card,
>  	descr->hw_regs.payload.size = 0;
>  	descr->skb = NULL;
>  
> -	offset = ((unsigned long)descr->skb->data) &
> -		(GELIC_NET_RXBUF_ALIGN - 1);
> -	if (offset)
> -		skb_reserve(descr->skb, GELIC_NET_RXBUF_ALIGN - offset);
> -	/* io-mmu-map the skb */
> -	cpu_addr = dma_map_single(ctodev(card), descr->skb->data,
> -				  GELIC_NET_MAX_FRAME, DMA_FROM_DEVICE);
> -	descr->hw_regs.payload.dev_addr = cpu_to_be32(cpu_addr);
> -	if (dma_mapping_error(ctodev(card), cpu_addr)) {
> -		dev_kfree_skb_any(descr->skb);
> +	napi_buff = napi_alloc_frag_align(napi_buff_size,
> +		GELIC_NET_RXBUF_ALIGN);

Please align with the above bracket:

	napi_buff = napi_alloc_frag_align(napi_buff_size,
					  GELIC_NET_RXBUF_ALIGN);

> +
> +	if (unlikely(!napi_buff)) {
> +		return -ENOMEM;
> +	}

The brackets are not needed here.

> +
> +	descr->skb = napi_build_skb(napi_buff, napi_buff_size);
> +
> +	if (unlikely(!descr->skb)) {
> +		skb_free_frag(napi_buff);
> +		return -ENOMEM;
> +	}
> +
> +	cpu_addr = dma_map_single(dev, napi_buff, napi_buff_size,
> +		DMA_FROM_DEVICE);

Please align with the above bracket

> +
> +	if (dma_mapping_error(dev, cpu_addr)) {
> +		skb_free_frag(napi_buff);
>  		descr->skb = NULL;
> -		dev_info(ctodev(card),
> -			 "%s:Could not iommu-map rx buffer\n", __func__);
> +		dev_err_once(dev, "%s:Could not iommu-map rx buffer\n",
> +			__func__);

Same here.

Side note: as a nice net-next follow-up you could move the skb
allocation into gelic_net_pass_skb_up() - so that the skbuff is hot in
the cache at rx time.


Cheers,

Paolo


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ