Re: [PATCH] mmc: cavium: Fix use-after-free in of_platform_device_destroy

[Jan Glauber]

On Thu, Sep 07, 2017 at 11:58:58AM -0500, Rob Herring wrote:
> On Thu, Sep 7, 2017 at 6:24 AM, Jan Glauber <jglauber@xxxxxxxxxx> wrote:
> > KASAN reported the following:
> >
> > [   19.338655] ==================================================================
> > [   19.345946] BUG: KASAN: use-after-free in of_platform_device_destroy+0x88/0x100
> > [   19.345966] Read of size 8 at addr fffffe01aa6f1468 by task systemd-udevd/264
> >
> > [   19.345983] CPU: 1 PID: 264 Comm: systemd-udevd Not tainted 4.13.0-jang+ #737
> > [   19.345989] Hardware name: Cavium ThunderX CN81XX board (DT)
> > [   19.345995] Call trace:
> > [   19.346013] [<fffffc800808b1b0>] dump_backtrace+0x0/0x368
> > [   19.346026] [<fffffc800808b6bc>] show_stack+0x24/0x30
> > [   19.346040] [<fffffc8008cbb944>] dump_stack+0xa4/0xc8
> > [   19.346057] [<fffffc80082c2870>] print_address_description+0x68/0x258
> > [   19.346070] [<fffffc80082c2d70>] kasan_report+0x238/0x2f8
> > [   19.346082] [<fffffc80082c14a8>] __asan_load8+0x88/0xb8
> > [   19.346098] [<fffffc8008aacee0>] of_platform_device_destroy+0x88/0x100
> > [   19.346131] [<fffffc8000e02fa4>] thunder_mmc_probe+0x314/0x550 [thunderx_mmc]
> > [   19.346147] [<fffffc800879d560>] pci_device_probe+0x158/0x1f8
> > [   19.346162] [<fffffc800886e53c>] driver_probe_device+0x394/0x5f8
> > [   19.346174] [<fffffc800886e8f4>] __driver_attach+0x154/0x158
> > [   19.346185] [<fffffc800886b12c>] bus_for_each_dev+0xdc/0x140
> > [   19.346196] [<fffffc800886d9f8>] driver_attach+0x38/0x48
> > [   19.346207] [<fffffc800886d148>] bus_add_driver+0x290/0x3c8
> > [   19.346219] [<fffffc800886fc5c>] driver_register+0xbc/0x1a0
> > [   19.346232] [<fffffc800879b78c>] __pci_register_driver+0xc4/0xd8
> > [   19.346260] [<fffffc8000e80024>] thunder_mmc_driver_init+0x24/0x10000 [thunderx_mmc]
> > [   19.346273] [<fffffc8008083a80>] do_one_initcall+0x98/0x1c0
> > [   19.346289] [<fffffc8008177b54>] do_init_module+0xe0/0x2cc
> > [   19.346303] [<fffffc8008175cf0>] load_module+0x3238/0x35c0
> > [   19.346318] [<fffffc8008176438>] SyS_finit_module+0x190/0x1a0
> > [   19.346329] [<fffffc80080834a0>] __sys_trace_return+0x0/0x4
> >
> > This is caused by:
> >
> >   platform_device_register()
> >    -> platform_device_unregister(to_platform_device(dev))
> >         freeing struct device
> >    -> of_node_clear_flag(dev->of_node, ...)
> >         writing to the freed device
> >
> > The issue is solved by increasing the reference count before calling
> > of_platform_device_destroy() so freeing the device is postponed after
> > the call.
> >
> > Fixes: 8fb83b142823 ("mmc: cavium: Fix probing race with regulator")
> > Signed-off-by: Jan Glauber <jglauber@xxxxxxxxxx>
> > ---
> >  drivers/mmc/host/cavium-thunderx.c | 6 +++++-
> >  1 file changed, 5 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/mmc/host/cavium-thunderx.c b/drivers/mmc/host/cavium-thunderx.c
> > index b9cc95998799..eee08d81b242 100644
> > --- a/drivers/mmc/host/cavium-thunderx.c
> > +++ b/drivers/mmc/host/cavium-thunderx.c
> > @@ -7,6 +7,7 @@
> >   *
> >   * Copyright (C) 2016 Cavium Inc.
> >   */
> > +#include <linux/device.h>
> >  #include <linux/dma-mapping.h>
> >  #include <linux/interrupt.h>
> >  #include <linux/mmc/mmc.h>
> > @@ -149,8 +150,11 @@ static int thunder_mmc_probe(struct pci_dev *pdev,
> >         for (i = 0; i < CAVIUM_MAX_MMC; i++) {
> >                 if (host->slot[i])
> >                         cvm_mmc_of_slot_remove(host->slot[i]);
> > -               if (host->slot_pdev[i])
> > +               if (host->slot_pdev[i]) {
> > +                       get_device(&host->slot_pdev[i]->dev);
> >                         of_platform_device_destroy(&host->slot_pdev[i]->dev, NULL);
> > +                       put_device(&host->slot_pdev[i]->dev);
> 
> Why do you think this is Cavium specific?

First, the usage in the Cavium driver is quite special. The device
is not a platform device, I create this as a dummy device because
mmc_parse_of would not work otherwise.

Second, I assumed the of_node_clear_flag() after removal of the device
is a more general pattern and only not working for our case because
of the dummy device hack. I've not looked too closely at the platform
code, so I might be wrong here.

> >From my look of it, the problem is in of_platform_device_destroy. We
> should save the node ptr before the unregister call and use that to
> clear the flags.

This would be a larger patch and would need much more testing I guess.

--Jan