From: Lukas Wunner <lukas@wunner.de>
To: Florian Fainelli <f.fainelli@gmail.com>
Cc: Mark Brown <broonie@kernel.org>,
Vladimir Oltean <olteanv@gmail.com>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
linux-spi <linux-spi@vger.kernel.org>
Subject: Re: Use after free in bcm2835_spi_remove()
Date: Thu, 22 Oct 2020 14:12:54 +0200 [thread overview]
Message-ID: <20201022121254.GA3847@wunner.de> (raw)
In-Reply-To: <b094c266-99ce-4462-9041-7d1659b13300@gmail.com>
On Wed, Oct 14, 2020 at 02:20:16PM -0700, Florian Fainelli wrote:
> In bcm2835_spi_remove(), spi_controller_unregister() will free the ctlr
> reference which will lead to an use after free in bcm2835_release_dma().
>
> To avoid this use after free, allocate the bcm2835_spi structure with a
> different lifecycle than the spi_controller structure such that we
> unregister the SPI controller, free up all the resources and finally let
> device managed allocations free the bcm2835_spi structure.
[...]
> - if (ctlr->dma_tx) {
> - dmaengine_terminate_sync(ctlr->dma_tx);
> + if (dma_tx) {
> + dmaengine_terminate_sync(dma_tx);
>
> if (bs->fill_tx_desc)
> dmaengine_desc_free(bs->fill_tx_desc);
>
> if (bs->fill_tx_addr)
> - dma_unmap_page_attrs(ctlr->dma_tx->device->dev,
> + dma_unmap_page_attrs(dma_tx->device->dev,
> bs->fill_tx_addr, sizeof(u32),
> DMA_TO_DEVICE,
> DMA_ATTR_SKIP_CPU_SYNC);
>
> - dma_release_channel(ctlr->dma_tx);
> - ctlr->dma_tx = NULL;
> + dma_release_channel(dma_tx);
> }
You must set ctlr->dma_tx and ctlr->dma_rx to NULL because the driver
checks their value in a couple of places.
E.g. bcm2835_spi_setup() checks ctlr->dma_rx.
Likewise, the error paths of bcm2835_dma_init() and bcm2835_spi_probe()
call bcm2835_dma_release() and the latter checks ctlr->dma_tx and
ctlr->dma_rx to determine whether DMA was set up, hence needs to be
torn down.
> + bs = devm_kzalloc(&pdev->dev, sizeof(*bs), GFP_KERNEL);
> + if (!bs)
> + return -ENOMEM;
> +
> ctlr = spi_alloc_master(&pdev->dev, ALIGN(sizeof(*bs),
> dma_get_cache_alignment()));
You can set the second argument to spi_alloc_master() to 0
to conserve memory.
Thanks,
Lukas
next prev parent reply other threads:[~2020-10-22 12:13 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-13 23:48 Use after free in bcm2835_spi_remove() Florian Fainelli
2020-10-14 14:09 ` Lukas Wunner
2020-10-14 19:40 ` Vladimir Oltean
2020-10-14 20:25 ` Mark Brown
2020-10-14 21:20 ` Florian Fainelli
2020-10-22 12:12 ` Lukas Wunner [this message]
2020-10-15 5:38 ` Lukas Wunner
2020-10-15 12:53 ` Mark Brown
2020-10-28 9:59 ` Lukas Wunner
2020-10-29 22:24 ` Mark Brown
2020-11-11 19:07 ` [PATCH 0/4] Use-after-free be gone Lukas Wunner
2020-11-11 19:07 ` [PATCH 1/4] spi: Introduce device-managed SPI controller allocation Lukas Wunner
2020-11-11 19:07 ` [PATCH 2/4] spi: bcm2835: Fix use-after-free on unbind Lukas Wunner
2020-11-11 20:18 ` Florian Fainelli
2020-11-11 19:07 ` [PATCH 3/4] spi: bcm2835aux: " Lukas Wunner
2020-11-11 19:07 ` [PATCH 4/4] spi: bcm-qspi: " Lukas Wunner
2020-11-11 21:12 ` Florian Fainelli
2020-11-12 13:50 ` [PATCH 0/4] Use-after-free be gone Mark Brown
2020-11-12 19:39 ` Mark Brown
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20201022121254.GA3847@wunner.de \
--to=lukas@wunner.de \
--cc=broonie@kernel.org \
--cc=f.fainelli@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-spi@vger.kernel.org \
--cc=olteanv@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).