Prepare for the coming implementation by GCC and Clang of the __counted_by
attribute. Flexible array members annotated with __counted_by can have
their accesses bounds-checked at run-time via CONFIG_UBSAN_BOUNDS (for
array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family
functions).
As found with Coccinelle[1], add __counted_by for struct p54_cal_database.
Cc: Christian Lamparter <[email protected]>
Cc: Kalle Valo <[email protected]>
Cc: "Gustavo A. R. Silva" <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://github.com/kees/kernel-tools/blob/trunk/coccinelle/examples/counted_by.cocci [1]
Signed-off-by: Kees Cook <[email protected]>
---
drivers/net/wireless/intersil/p54/p54.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/wireless/intersil/p54/p54.h b/drivers/net/wireless/intersil/p54/p54.h
index 3356ea708d81..770e348d1f6c 100644
--- a/drivers/net/wireless/intersil/p54/p54.h
+++ b/drivers/net/wireless/intersil/p54/p54.h
@@ -126,7 +126,7 @@ struct p54_cal_database {
size_t entry_size;
size_t offset;
size_t len;
- u8 data[];
+ u8 data[] __counted_by(entries);
};
#define EEPROM_READBACK_LEN 0x3fc
--
2.34.1
On 10/6/23 22:17, Kees Cook wrote:
> Prepare for the coming implementation by GCC and Clang of the __counted_by
> attribute. Flexible array members annotated with __counted_by can have
> their accesses bounds-checked at run-time via CONFIG_UBSAN_BOUNDS (for
> array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family
> functions).
>
> As found with Coccinelle[1], add __counted_by for struct p54_cal_database.
>
> Cc: Christian Lamparter <[email protected]>
> Cc: Kalle Valo <[email protected]>
> Cc: "Gustavo A. R. Silva" <[email protected]>
> Cc: [email protected]
> Cc: [email protected]
> Link: https://github.com/kees/kernel-tools/blob/trunk/coccinelle/examples/counted_by.cocci [1]
> Signed-off-by: Kees Cook <[email protected]>
Reviewed-by: Gustavo A. R. Silva <[email protected]>
Thanks!
--
Gustavo
> ---
> drivers/net/wireless/intersil/p54/p54.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/net/wireless/intersil/p54/p54.h b/drivers/net/wireless/intersil/p54/p54.h
> index 3356ea708d81..770e348d1f6c 100644
> --- a/drivers/net/wireless/intersil/p54/p54.h
> +++ b/drivers/net/wireless/intersil/p54/p54.h
> @@ -126,7 +126,7 @@ struct p54_cal_database {
> size_t entry_size;
> size_t offset;
> size_t len;
> - u8 data[];
> + u8 data[] __counted_by(entries);
> };
>
> #define EEPROM_READBACK_LEN 0x3fc
Hi,
I randomly peeked at this patch. Unfortunately, I am not familiar
with the actual p54 code.
On Fri, Oct 6, 2023 at 4:17 PM Kees Cook <[email protected]> wrote:
>
> Prepare for the coming implementation by GCC and Clang of the __counted_by
> attribute. Flexible array members annotated with __counted_by can have
> their accesses bounds-checked at run-time via CONFIG_UBSAN_BOUNDS (for
> array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family
> functions).
>
> As found with Coccinelle[1], add __counted_by for struct p54_cal_database.
>
> Cc: Christian Lamparter <[email protected]>
> Cc: Kalle Valo <[email protected]>
> Cc: "Gustavo A. R. Silva" <[email protected]>
> Cc: [email protected]
> Cc: [email protected]
> Link: https://github.com/kees/kernel-tools/blob/trunk/coccinelle/examples/counted_by.cocci [1]
> Signed-off-by: Kees Cook <[email protected]>
> ---
> drivers/net/wireless/intersil/p54/p54.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/net/wireless/intersil/p54/p54.h b/drivers/net/wireless/intersil/p54/p54.h
> index 3356ea708d81..770e348d1f6c 100644
> --- a/drivers/net/wireless/intersil/p54/p54.h
> +++ b/drivers/net/wireless/intersil/p54/p54.h
> @@ -126,7 +126,7 @@ struct p54_cal_database {
> size_t entry_size;
> size_t offset;
> size_t len;
> - u8 data[];
> + u8 data[] __counted_by(entries);
This looks incorrect - I think you want __counted_by(len)? The
presence of entry_size made me suspicious.
> };
This is the function that creates struct p54_cal_database:
static struct p54_cal_database *p54_convert_db(struct pda_custom_wrapper *src,
size_t total_len)
{
struct p54_cal_database *dst;
size_t payload_len, entries, entry_size, offset;
payload_len = le16_to_cpu(src->len);
entries = le16_to_cpu(src->entries);
entry_size = le16_to_cpu(src->entry_size);
offset = le16_to_cpu(src->offset);
if (((entries * entry_size + offset) != payload_len) ||
(payload_len + sizeof(*src) != total_len))
return NULL;
dst = kmalloc(sizeof(*dst) + payload_len, GFP_KERNEL);
if (!dst)
return NULL;
dst->entries = entries;
dst->entry_size = entry_size;
dst->offset = offset;
dst->len = payload_len;
memcpy(dst->data, src->data, payload_len);
return dst;
}
You can see that kmalloc is performed with `sizeof(*dst) +
payload_len`, and payload_len is assigned to ->len.
I don't read Coccinelle, but, if this patch was auto-generated, I
wonder if the script has an error.
Regards,
Jason
On Mon, Oct 09, 2023 at 10:55:32AM -0400, Jason Andryuk wrote:
> This is the function that creates struct p54_cal_database:
>
> static struct p54_cal_database *p54_convert_db(struct pda_custom_wrapper *src,
> size_t total_len)
> {
> struct p54_cal_database *dst;
> size_t payload_len, entries, entry_size, offset;
>
> payload_len = le16_to_cpu(src->len);
> entries = le16_to_cpu(src->entries);
> entry_size = le16_to_cpu(src->entry_size);
> offset = le16_to_cpu(src->offset);
> if (((entries * entry_size + offset) != payload_len) ||
> (payload_len + sizeof(*src) != total_len))
> return NULL;
>
> dst = kmalloc(sizeof(*dst) + payload_len, GFP_KERNEL);
> if (!dst)
> return NULL;
>
> dst->entries = entries;
> dst->entry_size = entry_size;
> dst->offset = offset;
> dst->len = payload_len;
>
> memcpy(dst->data, src->data, payload_len);
> return dst;
> }
>
> You can see that kmalloc is performed with `sizeof(*dst) +
> payload_len`, and payload_len is assigned to ->len.
>
> I don't read Coccinelle, but, if this patch was auto-generated, I
> wonder if the script has an error.
It seems that my Coccinelle script got confused by this:
p54_convert_output_limits():
priv->output_limit = kmalloc(data[1] *
sizeof(struct pda_channel_output_limit) +
sizeof(*priv->output_limit), GFP_KERNEL);
...
priv->output_limit->entry_size =
sizeof(struct pda_channel_output_limit);
priv->output_limit->len = priv->output_limit->entry_size *
priv->output_limit->entries +
priv->output_limit->offset;
It thought "sizeof(struct pda_channel_output_limit)" was the element
count, since it wasn't able to identify the array member here.
Regardless, I've sent a v2 now. :)
--
Kees Cook