From: Simon Glass <sjg-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org>
To: U-Boot Mailing List <u-boot-0aAXYlwwYIKGBzrmiIFOJg@public.gmane.org>
Cc: Devicetree Discuss
<devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org>,
Gabe Black <gabeblack-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>,
Jerry Van Baren
<vanbaren-He//nVnquyzQT0dZR+AlfA@public.gmane.org>,
Tom Rini <trini-l0cyMroinI0@public.gmane.org>
Subject: [PATCH 12/14] fdt: Allow device tree to specify secure booting
Date: Thu, 25 Oct 2012 19:31:09 -0700 [thread overview]
Message-ID: <1351218671-15228-13-git-send-email-sjg@chromium.org> (raw)
In-Reply-To: <1351218671-15228-1-git-send-email-sjg-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org>
From: Doug Anderson <dianders-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org>
When secure booting is chosen:
* The u-boot shell is never invoked during boot--we just do a simple
table lookup to find the command. This means we could even remove
the shell parsing from u-boot and still be able to boot.
* The boot command can't be interruped.
* Failure doesn't cause us to fall back to the shell.
Signed-off-by: Gabe Black <gabeblack-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Doug Anderson <dianders-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org>
Signed-off-by: Simon Glass <sjg-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org>
---
common/main.c | 62 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
1 files changed, 62 insertions(+), 0 deletions(-)
diff --git a/common/main.c b/common/main.c
index cf1b5f9..03c63b4 100644
--- a/common/main.c
+++ b/common/main.c
@@ -283,6 +283,59 @@ int abortboot(int bootdelay)
# endif /* CONFIG_AUTOBOOT_KEYED */
#endif /* CONFIG_BOOTDELAY >= 0 */
+/*
+ * Runs the given boot command securely. Specifically:
+ * - Doesn't run the command with the shell (run_command or parse_string_outer),
+ * since that's a lot of code surface that an attacker might exploit.
+ * Because of this, we don't do any argument parsing--the secure boot command
+ * has to be a full-fledged u-boot command.
+ * - Doesn't check for keypresses before booting, since that could be a
+ * security hole; also disables Ctrl-C.
+ * - Doesn't allow the command to return.
+ *
+ * Upon any failures, this function will drop into an infinite loop after
+ * printing the error message to console.
+ */
+
+#if defined(CONFIG_BOOTDELAY) && (CONFIG_BOOTDELAY >= 0) && \
+ defined(CONFIG_OF_CONTROL)
+static void secure_boot_cmd(char *cmd)
+{
+ cmd_tbl_t *cmdtp;
+ int rc;
+
+ if (!cmd) {
+ printf("## Error: Secure boot command not specified\n");
+ goto err;
+ }
+
+ /* Disable Ctrl-C just in case some command is used that checks it. */
+ disable_ctrlc(1);
+
+ /* Find the command directly. */
+ cmdtp = find_cmd(cmd);
+ if (!cmdtp) {
+ printf("## Error: \"%s\" not defined\n", cmd);
+ goto err;
+ }
+
+ /* Run the command, forcing no flags and faking argc and argv. */
+ rc = (cmdtp->cmd)(cmdtp, 0, 1, &cmd);
+
+ /* Shouldn't ever return from boot command. */
+ printf("## Error: \"%s\" returned (code %d)\n", cmd, rc);
+
+err:
+ /*
+ * Not a whole lot to do here. Rebooting won't help much, since we'll
+ * just end up right back here. Just loop.
+ */
+ hang();
+}
+
+#endif /* CONFIG_OF_CONTROL */
+
+
/****************************************************************************/
void main_loop (void)
@@ -397,6 +450,15 @@ void main_loop (void)
env = fdtdec_get_config_string(gd->fdt_blob, "bootcmd");
if (env)
s = env;
+
+ /*
+ * If the bootsecure option was chosen, use secure_boot_cmd().
+ * Always use 'env' in this case, since bootsecure requres that the
+ * bootcmd was specified in the FDT too.
+ */
+ if (fdtdec_get_config_int(gd->fdt_blob, "bootsecure", 0))
+ secure_boot_cmd(env);
+
#endif /* CONFIG_OF_CONTROL */
debug ("### main_loop: bootcmd=\"%s\"\n", s ? s : "<UNDEFINED>");
--
1.7.7.3
next prev parent reply other threads:[~2012-10-26 2:31 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-10-26 2:30 [PATCH 0/14] fdt: Add various device tree utilities and features Simon Glass
[not found] ` <1351218671-15228-1-git-send-email-sjg-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org>
2012-10-26 2:30 ` [PATCH 01/14] fdt: Add function to get config int from device tree Simon Glass
2012-10-26 2:30 ` [PATCH 02/14] fdt: Add function to get a config string " Simon Glass
2012-10-26 2:31 ` [PATCH 03/14] fdt: Add fdtdec_decode_region() to decode memory region Simon Glass
2012-10-26 2:31 ` [PATCH 04/14] fdt: Add function for decoding multiple gpios globally available Simon Glass
2012-10-26 2:31 ` [PATCH 05/14] fdt: Export fdtdec_find_alias_node() function Simon Glass
[not found] ` <1351218671-15228-6-git-send-email-sjg-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org>
2012-10-26 4:24 ` David Gibson
2012-10-31 23:50 ` Simon Glass
2012-10-26 2:31 ` [PATCH 06/14] fdt: Export fdtdec_lookup() and fix the name Simon Glass
2012-10-26 2:31 ` [PATCH 07/14] fdt: Add function to read boolean property Simon Glass
2012-10-26 2:31 ` [PATCH 08/14] fdt: Add fdtdec_get_uint64 to decode a 64-bit value from a property Simon Glass
2012-10-26 2:31 ` [PATCH 09/14] fdt: Add polarity-aware gpio functions to fdtdec Simon Glass
2012-10-26 7:17 ` Lucas Stach
2012-10-31 23:59 ` Simon Glass
2012-11-01 4:50 ` Stephen Warren
2012-11-15 23:31 ` Simon Glass
2012-11-15 23:46 ` Stephen Warren
2012-11-16 0:01 ` Simon Glass
2012-10-26 2:31 ` [PATCH 11/14] fdt: Tell the FDT library where the device tree is Simon Glass
2012-10-26 2:31 ` Simon Glass [this message]
2012-10-26 2:31 ` [PATCH 10/14] fdt: Load boot command from device tree Simon Glass
2012-10-26 2:31 ` [PATCH 13/14] fdt: Add option to default to most compatible conf in a fit image Simon Glass
2012-10-26 2:31 ` [PATCH 14/14] fdt: Set kernaddr if fdt indicates a kernel is present Simon Glass
[not found] ` <1351218671-15228-15-git-send-email-sjg-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org>
2012-11-28 14:30 ` [U-Boot] " Dennis Lan (dlan)
2012-11-18 1:35 ` [PATCH 0/14] fdt: Add various device tree utilities and features Jerry Van Baren
2012-11-19 17:08 ` Simon Glass
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=1351218671-15228-13-git-send-email-sjg@chromium.org \
--to=sjg-f7+t8e8rja9g9huczpvpmw@public.gmane.org \
--cc=devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org \
--cc=gabeblack-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org \
--cc=trini-l0cyMroinI0@public.gmane.org \
--cc=u-boot-0aAXYlwwYIKGBzrmiIFOJg@public.gmane.org \
--cc=vanbaren-He//nVnquyzQT0dZR+AlfA@public.gmane.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;
as well as URLs for NNTP newsgroup(s).