linux-hardening - Re: [PATCH] dmaengine: dw-axi-dmac: Do not dereference NULL structure

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: <Y9egAuMYM+JyBZ3O@smile.fi.intel.com>
Date:   Mon, 30 Jan 2023 12:46:26 +0200
From:   Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
To:     Kees Cook <keescook@...omium.org>
Cc:     Eugeniy Paltsev <Eugeniy.Paltsev@...opsys.com>,
        Vinod Koul <vkoul@...nel.org>, dmaengine@...r.kernel.org,
        Sia Jee Heng <jee.heng.sia@...el.com>,
        linux-kernel@...r.kernel.org, linux-hardening@...r.kernel.org
Subject: Re: [PATCH] dmaengine: dw-axi-dmac: Do not dereference NULL structure

On Fri, Jan 27, 2023 at 02:36:27PM -0800, Kees Cook wrote:
> If "vdesc" is NULL, it cannot be used with vd_to_axi_desc(). Leave
> "bytes" unchanged at 0. Seen under GCC 13 with -Warray-bounds:
> 
> ../drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c: In function 'dma_chan_tx_status':
> ../drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c:329:46: warning: array subscript 0 is outside array bounds of 'struct
> virt_dma_desc[46116860184273879]' [-Warray-bounds=]
>   329 |                 bytes = vd_to_axi_desc(vdesc)->length;
>       |                                              ^~
> 
> Fixes: 8e55444da65c ("dmaengine: dw-axi-dmac: Support burst residue granularity")
> Cc: Eugeniy Paltsev <Eugeniy.Paltsev@...opsys.com>
> Cc: Vinod Koul <vkoul@...nel.org>
> Cc: dmaengine@...r.kernel.org
> Signed-off-by: Kees Cook <keescook@...omium.org>
> ---
>  drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c | 2 --
>  1 file changed, 2 deletions(-)
> 
> diff --git a/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c b/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
> index 08af483331fd..4169e1d7d5ca 100644
> --- a/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
> +++ b/drivers/dma/dw-axi-dmac/dw-axi-dmac-platform.c
> @@ -325,8 +325,6 @@ dma_chan_tx_status(struct dma_chan *dchan, dma_cookie_t cookie,
>  		len = vd_to_axi_desc(vdesc)->hw_desc[0].len;
>  		completed_length = completed_blocks * len;
>  		bytes = length - completed_length;
> -	} else {
> -		bytes = vd_to_axi_desc(vdesc)->length;
>  	}
>  
>  	spin_unlock_irqrestore(&chan->vc.lock, flags);

If you want to fix it properly, the code should be modified like

	status = dma_cookie_status(dchan, cookie, txstate);
	if (status == DMA_COMPLETE || !txstate) {
		if (txstate)
			goto out;
		return status;
	}

	...

out:
	dma_set_residue(txstate, bytes);

	return status;

to be in accordance with the Documentation.

**OR**

the Documentation should be fixed to tell that if status is DMA_COMPLETE,
residue is undefined and assumed to be 0.

Vinod?

-- 
With Best Regards,
Andy Shevchenko