Re: [PATCH v3] acpi: nfit: vmalloc-out-of-bounds Read in acpi_nfit_ctl

[Dave Jiang <dave.jiang@intel.com>]

On 11/13/24 5:51 AM, Suraj Sonawane wrote:
> Fix an issue detected by syzbot with KASAN:
> 
> BUG: KASAN: vmalloc-out-of-bounds in cmd_to_func drivers/acpi/nfit/
> core.c:416 [inline]
> BUG: KASAN: vmalloc-out-of-bounds in acpi_nfit_ctl+0x20e8/0x24a0
> drivers/acpi/nfit/core.c:459
> 
> The issue occurs in cmd_to_func when the call_pkg->nd_reserved2
> array is accessed without verifying that call_pkg points to a buffer
> that is appropriately sized as a struct nd_cmd_pkg. This can lead
> to out-of-bounds access and undefined behavior if the buffer does not
> have sufficient space.
> 
> To address this, a check was added in acpi_nfit_ctl() to ensure that
> buf is not NULL and that buf_len is greater than sizeof(*call_pkg)
> before casting buf to struct nd_cmd_pkg *. This ensures safe access
> to the members of call_pkg, including the nd_reserved2 array.
> 
> Reported-by: syzbot+7534f060ebda6b8b51b3@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=7534f060ebda6b8b51b3
> Tested-by: syzbot+7534f060ebda6b8b51b3@syzkaller.appspotmail.com
> Fixes: ebe9f6f19d80 ("acpi/nfit: Fix bus command validation")
> Signed-off-by: Suraj Sonawane <surajsonawane0215@gmail.com>
> ---
> V1: https://lore.kernel.org/lkml/20241111080429.9861-1-surajsonawane0215@gmail.com/
> V2: Initialized `out_obj` to `NULL` in `acpi_nfit_ctl()` to prevent
> potential uninitialized variable usage if condition is true.
> V3: Changed the condition to if (!buf || buf_len < sizeof(*call_pkg))
> and updated the Fixes tag to reference the correct commit.
> 
>  drivers/acpi/nfit/core.c | 12 +++++++++---
>  1 file changed, 9 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/acpi/nfit/core.c b/drivers/acpi/nfit/core.c
> index 5429ec9ef..eb5349606 100644
> --- a/drivers/acpi/nfit/core.c
> +++ b/drivers/acpi/nfit/core.c
> @@ -439,7 +439,7 @@ int acpi_nfit_ctl(struct nvdimm_bus_descriptor *nd_desc, struct nvdimm *nvdimm,
>  {
>  	struct acpi_nfit_desc *acpi_desc = to_acpi_desc(nd_desc);
>  	struct nfit_mem *nfit_mem = nvdimm_provider_data(nvdimm);
> -	union acpi_object in_obj, in_buf, *out_obj;
> +	union acpi_object in_obj, in_buf, *out_obj = NULL;

Looking at the code later, out_obj is always assigned before access. I'm not seeing a path where out_obj would be accessed unitialized...

https://elixir.bootlin.com/linux/v6.12-rc7/source/drivers/acpi/nfit/core.c#L538
 
>  	const struct nd_cmd_desc *desc = NULL;
>  	struct device *dev = acpi_desc->dev;
>  	struct nd_cmd_pkg *call_pkg = NULL;
> @@ -454,8 +454,14 @@ int acpi_nfit_ctl(struct nvdimm_bus_descriptor *nd_desc, struct nvdimm *nvdimm,
>  	if (cmd_rc)
>  		*cmd_rc = -EINVAL;
>  
> -	if (cmd == ND_CMD_CALL)
> -		call_pkg = buf;
> +	if (cmd == ND_CMD_CALL) {
> +		if (!buf || buf_len < sizeof(*call_pkg)) {
> +			rc = -EINVAL;
> +			goto out;
> +		}
> +		call_pkg = (struct nd_cmd_pkg *)buf;

Is the casting needed? It wasn't in the old code.

> +	}
> +
>  	func = cmd_to_func(nfit_mem, cmd, call_pkg, &family);
>  	if (func < 0)
>  		return func;