mail archive of the barebox mailing list
 help / color / mirror / Atom feed
From: Trent Piepho <tpiepho@kymetacorp.com>
To: "Uwe Kleine-König" <uwe@kleine-koenig.org>
Cc: "barebox@lists.infradead.org" <barebox@lists.infradead.org>
Subject: Re: [PATCH v2] kwboot: do a filetype check before sending the image
Date: Wed, 20 Jan 2016 19:27:00 +0000	[thread overview]
Message-ID: <1453318040.4474.287.camel@rtred1test09.kymeta.local> (raw)
In-Reply-To: <1453277701-10367-1-git-send-email-uwe@kleine-koenig.org>

On Wed, 2016-01-20 at 09:15 +0100, Uwe Kleine-König wrote:
> The images that can be sent to a Marvell CPU have a fixed format. Do
> some sanity checks before actually sending an image for easier diagnosis
> of broken files.
> 
> Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
> ---
> Changes since (implicit) v1, sent with
> Message-Id: 1453276010-4669-1-git-send-email-uwe@kleine-koenig.org:
> 
>  - whitespace fix
>  - error out if a problem is detected
>  - add a commit log
> 
>  scripts/kwboot.c | 55 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 55 insertions(+)
> 
> diff --git a/scripts/kwboot.c b/scripts/kwboot.c
> index 46328d8ed006..06e58f6a7e3b 100644
> --- a/scripts/kwboot.c
> +++ b/scripts/kwboot.c
> @@ -546,6 +546,59 @@ out:
>  	return rc;
>  }
>  
> +static int
> +kwboot_check_image(unsigned char *img, size_t size)

Make it const unsigned char?

> +{
> +	size_t i;
> +	size_t header_size, image_size;
> +	unsigned char csum = 0;
> +
> +	switch (img[0x0]) {
> +		case 0x5a: /* SPI/NOR */
> +		case 0x69: /* UART0 */
> +		case 0x78: /* SATA */
> +		case 0x8b: /* NAND */
> +		case 0x9c: /* PCIe */
> +			break;
> +		default:
> +			printf("Unknown boot source: 0x%hhx\n", img[0x0]);
> +			goto err;
> +	}
> +
> +	if (img[0x8] != 1) {
> +		printf("Unknown version: 0x%hhx\n", img[0x8]);
> +		goto err;
> +	}

If you're verifying the image, maybe also check that size > 8 before
reading img[0x8]?  Otherwise you could crash instead of rejecting a too
small image.

And having compared the version tag to 1, couldn't you then cast img
into a struct main_hdr_v1 *?  That would avoid all the hard coded magic
offsets in the rest of the code.  Unless img isn't aligned?

> +
> +	image_size = img[0x4] | (img[0x5] << 8) |
> +		(img[0x6] << 16) | (img[0x7] << 24);

struct main_hdr_v1 *hdr = (const struct main_hdr
image_size = le32_to_cpu(hdr->blocksize);
or if unaligned:
image_size = get_unaligned_le32(&img[0x4]);

> +
> +	header_size = (img[0x9] << 16) | img[0xa] | (img[0xb] << 8);

header_size = hdr->headersz_msb << 16 | le16_to_cpu(hdr->headersz_lsb);

> +
> +	if (header_size + image_size != size) {
> +		printf("Size mismatch (%zu + %zu != %zu)\n",
> +		       header_size, image_size, size);
> +		goto err;
> +	} else {

Don't really need the else block here since the failure above exits,
just like the two failure checks before this one.

> +		for (i = 0; i < header_size; ++i)
> +			csum += img[i];

csum = image_checksum8(img, header_size)

> +
> +		csum -= img[0x1f];
> +
> +		if (csum != img[0x1f]) {
> +			printf("Checksum mismatch: header: 0x%02hhx, calculated: 0x%02hhx\n",
> +			       img[0x1f], csum);
> +			goto err;
> +		}
> +	}
> +
> +	return 0;
> +
> +err:
> +	errno = EINVAL;
> +	return 1;
> +}
> +
>  static void *
>  kwboot_mmap_image(const char *path, size_t *size, int prot)
>  {
> @@ -574,6 +627,8 @@ kwboot_mmap_image(const char *path, size_t *size, int prot)
>  
>  	rc = 0;
>  	*size = st.st_size;
> +
> +	rc = kwboot_check_image(img, *size);
>  out:
>  	if (rc && img) {
>  		munmap(img, st.st_size);

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

  parent reply	other threads:[~2016-01-20 19:27 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-01-20  8:15 Uwe Kleine-König
2016-01-20 11:54 ` Sebastian Hesselbarth
2016-01-20 19:27 ` Trent Piepho [this message]
2016-01-21 10:13   ` Uwe Kleine-König
2016-01-21 20:12     ` Trent Piepho

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=1453318040.4474.287.camel@rtred1test09.kymeta.local \
    --to=tpiepho@kymetacorp.com \
    --cc=barebox@lists.infradead.org \
    --cc=uwe@kleine-koenig.org \
    /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