linux-kernel - Re: [PATCH 1/1] spi: spi-fsl-dspi: Fix crash when not using GPIO chip select

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: <20241024075203.ifih3rxwyzl7dn4s@skbuf>
Date: Thu, 24 Oct 2024 10:52:03 +0300
From: Vladimir Oltean <olteanv@...il.com>
To: Frank Li <Frank.Li@....com>
Cc: Mark Brown <broonie@...nel.org>,
	William Zhang <william.zhang@...adcom.com>,
	Heiko Stuebner <heiko@...ech.de>,
	Cédric Le Goater <clg@...d.org>,
	Dhruva Gole <d-gole@...com>,
	Amit Kumar Mahapatra <amit.kumar-mahapatra@....com>,
	"open list:FREESCALE DSPI DRIVER" <linux-spi@...r.kernel.org>,
	"open list:FREESCALE DSPI DRIVER" <imx@...ts.linux.dev>,
	open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 1/1] spi: spi-fsl-dspi: Fix crash when not using GPIO
 chip select

Hi Frank,

On Wed, Oct 23, 2024 at 04:30:32PM -0400, Frank Li wrote:
> Add check for the return value of spi_get_csgpiod() to avoid passing a NULL
> pointer to gpiod_direction_output(), preventing a crash when GPIO chip
> select is not used.
> 
> Fix below crash:
> [    4.251960] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000
> [    4.260762] Mem abort info:
> [    4.263556]   ESR = 0x0000000096000004
> [    4.267308]   EC = 0x25: DABT (current EL), IL = 32 bits
> [    4.272624]   SET = 0, FnV = 0
> [    4.275681]   EA = 0, S1PTW = 0
> [    4.278822]   FSC = 0x04: level 0 translation fault
> [    4.283704] Data abort info:
> [    4.286583]   ISV = 0, ISS = 0x00000004, ISS2 = 0x00000000
> [    4.292074]   CM = 0, WnR = 0, TnD = 0, TagAccess = 0
> [    4.297130]   GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
> [    4.302445] [0000000000000000] user address but active_mm is swapper
> [    4.308805] Internal error: Oops: 0000000096000004 [#1] PREEMPT SMP
> [    4.315072] Modules linked in:
> [    4.318124] CPU: 2 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.12.0-rc4-next-20241023-00008-ga20ec42c5fc1 #359
> [    4.328130] Hardware name: LS1046A QDS Board (DT)
> [    4.332832] pstate: 40000005 (nZcv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
> [    4.339794] pc : gpiod_direction_output+0x34/0x5c
> [    4.344505] lr : gpiod_direction_output+0x18/0x5c
> [    4.349208] sp : ffff80008003b8f0
> [    4.352517] x29: ffff80008003b8f0 x28: 0000000000000000 x27: ffffc96bcc7e9068
> [    4.359659] x26: ffffc96bcc6e00b0 x25: ffffc96bcc598398 x24: ffff447400132810
> [    4.366800] x23: 0000000000000000 x22: 0000000011e1a300 x21: 0000000000020002
> [    4.373940] x20: 0000000000000000 x19: 0000000000000000 x18: ffffffffffffffff
> [    4.381081] x17: ffff44740016e600 x16: 0000000500000003 x15: 0000000000000007
> [    4.388221] x14: 0000000000989680 x13: 0000000000020000 x12: 000000000000001e
> [    4.395362] x11: 0044b82fa09b5a53 x10: 0000000000000019 x9 : 0000000000000008
> [    4.402502] x8 : 0000000000000002 x7 : 0000000000000007 x6 : 0000000000000000
> [    4.409641] x5 : 0000000000000200 x4 : 0000000002000000 x3 : 0000000000000000
> [    4.416781] x2 : 0000000000022202 x1 : 0000000000000000 x0 : 0000000000000000
> [    4.423921] Call trace:
> [    4.426362]  gpiod_direction_output+0x34/0x5c (P)
> [    4.431067]  gpiod_direction_output+0x18/0x5c (L)
> [    4.435771]  dspi_setup+0x220/0x334
> 
> Fixes: 9e264f3f85a5 ("spi: Replace all spi->chip_select and spi->cs_gpiod references with function call")
> Cc: stable@...r.kernel.org
> Signed-off-by: Frank Li <Frank.Li@....com>
> ---

Sadly I do not reproduce this NULL pointer dereference. I also know from
hearsay that the GPIO consumer API is more or less NULL-tolerant. Could
you "make drivers/gpio/gpiolib.lst" and see what's at offsets 0x18 and
0x34 into gpiod_direction_output()? The first thing in that function is
the VALIDATE_DESC() macro, which does:

static int validate_desc(const struct gpio_desc *desc, const char *func)
{
	if (!desc)
		return 0;

	if (IS_ERR(desc)) {
		pr_warn("%s: invalid GPIO (errorpointer)\n", func);
		return PTR_ERR(desc);
	}

	return 1;
}

#define VALIDATE_DESC(desc) do { \
	int __valid = validate_desc(desc, __func__); \
	if (__valid <= 0) \
		return __valid; \
	} while (0)