public inbox for linux-efi@vger.kernel.org
 help / color / mirror / Atom feed
From: Greg KH <gregkh-hQyY1W1yCW8ekmWlsbkhG0B+6BGkLq7r@public.gmane.org>
To: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
Cc: "linux-efi-u79uwXL29TY76Z2rM5mHXA@public.gmane.org"
	<linux-efi-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>,
	linux-hardened-ZwoEplunGu1jrUoiu81ncdBPR1lH4CV8@public.gmane.org,
	Leif Lindholm
	<leif.lindholm-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>,
	Matt Fleming
	<matt-mF/unelCI9GS6iBeEJttW/XRex20P6io@public.gmane.org>,
	Lukas Wunner <lukas-JFq808J9C/izQB+pC5nmwQ@public.gmane.org>,
	Kees Cook <keescook-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org>
Subject: Re: [RFC PATCH 2/2] efi: libstub: add support for the Chaoskey RNG USB stick to the stub
Date: Sat, 2 Sep 2017 10:26:45 +0200	[thread overview]
Message-ID: <20170902082645.GB1620@kroah.com> (raw)
In-Reply-To: <CAKv+Gu-T4mUkcA-bj9Cw0GkxDQ21ayKLn7_LCbvy6ZbrNEbo1w-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>

On Sat, Sep 02, 2017 at 09:18:34AM +0100, Ard Biesheuvel wrote:
> On 2 September 2017 at 07:45, Greg KH <gregkh-hQyY1W1yCW8ekmWlsbkhG0B+6BGkLq7r@public.gmane.org> wrote:
> > On Sat, Aug 19, 2017 at 04:17:40PM +0100, Ard Biesheuvel wrote:
> >> Early entropy is hard to come by, especially on non-x86 systems that
> >> lack an architected instruction and are not as uniform as PCs.
> >> Fortunately, on UEFI systems, we can invoke the EFI_RNG_PROTOCOL,
> >> which exposes the platform specific entropy source in a generic way.
> >> We use this protocol to fill the RNG UEFI config table (which is used
> >> during early boot to seed the kernel's entropy pool), and on ARM and
> >> arm64, we invoke it to seed KASLR as well.
> >>
> >> Sadly, EFI_RNG_PROTOCOL is not widely implemented, and so it would be
> >> nice to have a fallback for UEFI systems that lack it. So implement
> >> this fallback based on the Chaoskey RNG USB stick, which should be
> >> exposed using the standard UEFI USB I/O protocol if the firmware has
> >> USB support.
> >>
> >> Signed-off-by: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
> >> ---
> >>  drivers/firmware/efi/Kconfig                   |   7 ++
> >>  drivers/firmware/efi/libstub/Makefile          |   2 +-
> >>  drivers/firmware/efi/libstub/efistub.h         |   3 +
> >>  drivers/firmware/efi/libstub/random-chaoskey.c | 126 +++++++++++++++++++++++++
> >>  4 files changed, 137 insertions(+), 1 deletion(-)
> >>  create mode 100644 drivers/firmware/efi/libstub/random-chaoskey.c
> >>
> >> diff --git a/drivers/firmware/efi/Kconfig b/drivers/firmware/efi/Kconfig
> >> index 2b4c39fdfa91..448a6186d9fb 100644
> >> --- a/drivers/firmware/efi/Kconfig
> >> +++ b/drivers/firmware/efi/Kconfig
> >> @@ -87,6 +87,13 @@ config EFI_RUNTIME_WRAPPERS
> >>  config EFI_ARMSTUB
> >>       bool
> >>
> >> +config EFI_RNG_CHAOSKEY
> >> +     bool "Chaoskey USB RNG support in the UEFI stub"
> >> +     help
> >> +       Fall back to a Chaoskey RNG USB stick in case EFI_RNG_PROTOCOL
> >> +       is not implemented by the firmware. This is used for early
> >> +       seeding of the entropy pool, and for KASLR on ARM and arm64.
> >> +
> >>  config EFI_BOOTLOADER_CONTROL
> >>       tristate "EFI Bootloader Control"
> >>       depends on EFI_VARS
> >> diff --git a/drivers/firmware/efi/libstub/Makefile b/drivers/firmware/efi/libstub/Makefile
> >> index dedf9bde44db..fad6bc1d0e2a 100644
> >> --- a/drivers/firmware/efi/libstub/Makefile
> >> +++ b/drivers/firmware/efi/libstub/Makefile
> >> @@ -40,7 +40,7 @@ $(obj)/lib-%.o: $(srctree)/lib/%.c FORCE
> >>
> >>  lib-$(CONFIG_EFI_ARMSTUB)    += arm-stub.o fdt.o string.o random.o \
> >>                                  $(patsubst %.c,lib-%.o,$(arm-deps))
> >> -
> >> +lib-$(CONFIG_EFI_RNG_CHAOSKEY)       += random-chaoskey.o
> >>  lib-$(CONFIG_ARM)            += arm32-stub.o
> >>  lib-$(CONFIG_ARM64)          += arm64-stub.o
> >>  CFLAGS_arm64-stub.o          := -DTEXT_OFFSET=$(TEXT_OFFSET)
> >> diff --git a/drivers/firmware/efi/libstub/efistub.h b/drivers/firmware/efi/libstub/efistub.h
> >> index 3afff5facd3b..9579ddb5937d 100644
> >> --- a/drivers/firmware/efi/libstub/efistub.h
> >> +++ b/drivers/firmware/efi/libstub/efistub.h
> >> @@ -62,6 +62,9 @@ efi_status_t efi_random_alloc(efi_system_table_t *sys_table_arg,
> >>                             unsigned long size, unsigned long align,
> >>                             unsigned long *addr, unsigned long random_seed);
> >>
> >> +efi_status_t efi_get_random_chaoskey(efi_system_table_t *sys_table_arg,
> >> +                                  unsigned long size, u8 *out);
> >> +
> >>  efi_status_t check_platform_features(efi_system_table_t *sys_table_arg);
> >>
> >>  efi_status_t efi_random_get_seed(efi_system_table_t *sys_table_arg);
> >> diff --git a/drivers/firmware/efi/libstub/random-chaoskey.c b/drivers/firmware/efi/libstub/random-chaoskey.c
> >> new file mode 100644
> >> index 000000000000..3f94ec3df9ff
> >> --- /dev/null
> >> +++ b/drivers/firmware/efi/libstub/random-chaoskey.c
> >> @@ -0,0 +1,126 @@
> >> +/*
> >> + * Copyright (C) 2017 Linaro Ltd;  <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
> >> + *
> >> + * This program is free software; you can redistribute it and/or modify
> >> + * it under the terms of the GNU General Public License version 2 as
> >> + * published by the Free Software Foundation.
> >> + *
> >> + */
> >> +
> >> +#include <linux/efi.h>
> >> +#include <asm/efi.h>
> >> +
> >> +#include "efistub.h"
> >> +
> >> +#define CHAOSKEY_VENDOR_ID           0x1d50  /* OpenMoko */
> >> +#define CHAOSKEY_PRODUCT_ID          0x60c6  /* ChaosKey */
> >> +
> >> +static efi_usb_io_protocol_t *chaoskey_usb_io;
> >> +static unsigned long chaoskey_usb_ep_addr, chaoskey_usb_ep_size;
> >> +
> >> +static void locate_chaoskey(efi_system_table_t *sys_table_arg)
> >> +{
> >> +     efi_guid_t usb_io_guid = EFI_USB_IO_PROTOCOL_GUID;
> >> +     unsigned long bufsize = 0, i, j;
> >> +     efi_handle_t *handles;
> >> +     efi_status_t status;
> >> +
> >> +     chaoskey_usb_io = (void *)-1;
> >> +
> >> +     /* find all USB devices that UEFI knows about */
> >> +     status = efi_call_early(locate_handle, EFI_LOCATE_BY_PROTOCOL,
> >> +                             &usb_io_guid, NULL, &bufsize, NULL);
> >> +     if (status != EFI_BUFFER_TOO_SMALL)
> >> +             return;
> >> +
> >> +     status = efi_call_early(allocate_pool, EFI_LOADER_DATA, bufsize,
> >> +                             (void **)&handles);
> >> +     if (status != EFI_SUCCESS)
> >> +             return;
> >> +
> >> +     status = efi_call_early(locate_handle, EFI_LOCATE_BY_PROTOCOL,
> >> +                             &usb_io_guid, NULL, &bufsize, handles);
> >> +     if (status != EFI_SUCCESS)
> >> +             goto out;
> >> +
> >> +     for (i = 0; i < bufsize / sizeof(efi_handle_t); i++) {
> >> +             struct usb_device_descriptor dev;
> >> +             struct usb_interface_descriptor iface;
> >> +             efi_usb_io_protocol_t *p;
> >> +
> >> +             status = efi_call_early(handle_protocol, handles[i],
> >> +                                     &usb_io_guid, (void **)&p);
> >> +             if (status != EFI_SUCCESS)
> >> +                     continue; /* shouldn't happen */
> >> +
> >> +             /* get the device descriptor so we can check the vid/pid */
> >> +             status = p->get_device_descriptor(p, &dev);
> >> +             if (status != EFI_SUCCESS)
> >> +                     continue; /* shouldn't happen either */
> >> +
> >> +             if (dev.idVendor != CHAOSKEY_VENDOR_ID ||
> >> +                 dev.idProduct != CHAOSKEY_PRODUCT_ID)
> >
> > idVendor and idProduct are little endian types.
> >
> > You have run sparse on this code, right?
> >
> >
> >> +                     continue;
> >> +
> >> +             /* get the number of endpoints from the interface descriptor */
> >> +             status = p->get_iface_descriptor(p, &iface);
> >> +             if (status != EFI_SUCCESS)
> >> +                     continue;
> >> +
> >> +             /* locate the first BULK IN endpoint */
> >> +             for (j = 0; j < iface.bNumEndpoints; j++) {
> >> +                     struct usb_endpoint_descriptor ep;
> >
> > Don't we have built-in usb functions for this now?  Why not use them
> > instead of hand-rolling your own logic?
> >
> 
> This is UEFI stub code, which is executed with the kernel image or
> decompressor loaded at a completely different address than it was
> linked at. Reusing inline functions is usually fine, but anything that
> requires linking kernel objects into the decompressor is a bit
> cumbersome.

So this is not Linux kernel code?  Or is it?  I'm totally confused now.

> In any case, these patches are RFC only, and it is quite obvious that
> having a Chaoskey USB driver in the firmware that produces the
> EFI_RNG_PROTOCOL is much better, and the UEFI stub code can just call
> into the protocol to get entropy.

Ok, why not write that and not have to deal with any Linux code at all?

thanks,

greg k-h

  parent reply	other threads:[~2017-09-02  8:26 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-08-19 15:17 [RFC PATCH 1/2] efi: import USB I/O related declarations from the UEFI spec Ard Biesheuvel
     [not found] ` <20170819151740.625-1-ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
2017-08-19 15:17   ` [RFC PATCH 2/2] efi: libstub: add support for the Chaoskey RNG USB stick to the stub Ard Biesheuvel
     [not found]     ` <20170819151740.625-2-ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
2017-09-02  6:45       ` Greg KH
     [not found]         ` <20170902064501.GB26383-U8xfFu+wG4EAvxtiuMwx3w@public.gmane.org>
2017-09-02  8:18           ` Ard Biesheuvel
     [not found]             ` <CAKv+Gu-T4mUkcA-bj9Cw0GkxDQ21ayKLn7_LCbvy6ZbrNEbo1w-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-09-02  8:26               ` Greg KH [this message]
     [not found]                 ` <20170902082645.GB1620-U8xfFu+wG4EAvxtiuMwx3w@public.gmane.org>
2017-09-02  9:13                   ` Ard Biesheuvel
2017-09-02  6:41   ` [RFC PATCH 1/2] efi: import USB I/O related declarations from the UEFI spec Greg KH
     [not found]     ` <20170902064142.GA26383-U8xfFu+wG4EAvxtiuMwx3w@public.gmane.org>
2017-09-02  8:15       ` Ard Biesheuvel
     [not found]         ` <CAKv+Gu81515CQcU89fQFpxfZCrRm0VPgfWSc_a6yEVvQ5igGyQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-09-02  8:25           ` Greg KH
     [not found]             ` <20170902082528.GA1620-U8xfFu+wG4EAvxtiuMwx3w@public.gmane.org>
2017-09-02  9:08               ` Ard Biesheuvel

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=20170902082645.GB1620@kroah.com \
    --to=gregkh-hqyy1w1ycw8ekmwlsbkhg0b+6bgklq7r@public.gmane.org \
    --cc=ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org \
    --cc=keescook-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org \
    --cc=leif.lindholm-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org \
    --cc=linux-efi-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
    --cc=linux-hardened-ZwoEplunGu1jrUoiu81ncdBPR1lH4CV8@public.gmane.org \
    --cc=lukas-JFq808J9C/izQB+pC5nmwQ@public.gmane.org \
    --cc=matt-mF/unelCI9GS6iBeEJttW/XRex20P6io@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