[OE-core] [PATCH 1/1] dropbear: a fix for hang in dropbearkey, built for x32

Kamble, Nitin A nitin.a.kamble at intel.com
Tue May 28 16:10:33 UTC 2013


Saul, RP,
  FYI, this fix is still pending for oecore layer. 
Paul,
  This is for the 1.4.1 bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=4496 . Once Saul or RP pulls it in master, you can pull it in 1.4.1 release branch.

Thanks,
Nitin


> -----Original Message-----
> From: Kamble, Nitin A
> Sent: Thursday, May 23, 2013 9:09 AM
> To: Openembedded-core at lists.openembedded.org
> Cc: Kamble, Nitin A
> Subject: [PATCH 1/1] dropbear: a fix for hang in dropbearkey, built for x32
> 
> From: Nitin A Kamble <nitin.a.kamble at intel.com>
> 
> This commit fixes runtime hang of 'dropbearkey' utility, built for a x32 target
> abi system. The hang was observed while generating ssh keys, with this
> command:
>   dropbearkey -t dss -f private
> 
> The issue is fixed by changing the code, where 'long' in x86_64 mode is
> assumed as 64bit quantity. With the x32 abi, the processor is in x86_64 mode,
> but the 'long' is a 32bit quantity. Hence the fix uses 'long long'
> instead of 'long' to define/access 64bit data variables.
> 
> Fixes bug:
> [YOCTO #4496]
> 
> Signed-off-by: Nitin A Kamble <nitin.a.kamble at intel.com>
> ---
>  .../0007-dropbear-fix-for-x32-abi.patch            | 140
> +++++++++++++++++++++
>  meta/recipes-core/dropbear/dropbear.inc            |   3 +-
>  2 files changed, 142 insertions(+), 1 deletion(-)  create mode 100644
> meta/recipes-core/dropbear/dropbear-2013.58/0007-dropbear-fix-for-x32-
> abi.patch
> 
> diff --git a/meta/recipes-core/dropbear/dropbear-2013.58/0007-dropbear-
> fix-for-x32-abi.patch b/meta/recipes-core/dropbear/dropbear-
> 2013.58/0007-dropbear-fix-for-x32-abi.patch
> new file mode 100644
> index 0000000..b450121
> --- /dev/null
> +++ b/meta/recipes-core/dropbear/dropbear-2013.58/0007-dropbear-fix-
> for-
> +++ x32-abi.patch
> @@ -0,0 +1,140 @@
> +Upstream-Status: Pending
> +
> +The dropbearkey utility built in x32 abi format, when generating ssh
> +keys, was getting lost in the infinite loop.
> +
> +This patch fixes the issue by fixing types of variables and parameters
> +of functions used in the code, which were getting undesired size, when
> +compiled with the x32 abi toolchain.
> +
> +2013/05/23
> +Received this fix from H J Lu.
> +
> +Signed-Off-By: Nitin A Kamble <nitin.a.kamble at intel.com>
> +
> +# HG changeset patch
> +# User H.J. Lu <hjl.tools at gmail.com>
> +# Date 1369344079 25200
> +# Node ID a10a1c46b857cc8a3923c3bb6d1504aa25b6052f
> +# Parent  e76614145aea67f66e4a4257685c771efba21aa1
> +Typdef mp_digit to unsigned long long for MP_64BIT
> +
> +When GCC is used with MP_64BIT, we should typedef mp_digit to unsigned
> +long long instead of unsigned long since for x32, unsigned long is
> +32-bit and unsigned long long is 64-bit and it is safe to use unsigned
> +long long for 64-bit integer with GCC.
> +
> +diff -r e76614145aea -r a10a1c46b857 libtommath/tommath.h
> +--- a/libtommath/tommath.h	Thu Apr 18 22:57:47 2013 +0800
> ++++ b/libtommath/tommath.h	Thu May 23 14:21:19 2013 -0700
> +@@ -73,7 +73,7 @@
> +    typedef signed long long   long64;
> + #endif
> +
> +-   typedef unsigned long      mp_digit;
> ++   typedef unsigned long long mp_digit;
> +    typedef unsigned long      mp_word __attribute__ ((mode(TI)));
> +
> +    #define DIGIT_BIT          60
> +# HG changeset patch
> +# User H.J. Lu <hjl.tools at gmail.com>
> +# Date 1369344241 25200
> +# Node ID c7555a4cb7ded3a88409ba85f4027baa7af5f536
> +# Parent  a10a1c46b857cc8a3923c3bb6d1504aa25b6052f
> +Cast to mp_digit when updating *rho
> +
> +There is
> +
> +int
> +mp_montgomery_setup (mp_int * n, mp_digit * rho)
> +
> +We should cast to mp_digit instead of unsigned long when updating *rho
> +since mp_digit may be unsigned long long and unsigned long long may be
> +different from unsigned long, like in x32.
> +
> +diff -r a10a1c46b857 -r c7555a4cb7de
> libtommath/bn_mp_montgomery_setup.c
> +--- a/libtommath/bn_mp_montgomery_setup.c	Thu May 23 14:21:19
> 2013 -0700
> ++++ b/libtommath/bn_mp_montgomery_setup.c	Thu May 23 14:24:01
> 2013 -0700
> +@@ -48,7 +48,7 @@
> + #endif
> +
> +   /* rho = -1/m mod b */
> +-  *rho = (unsigned long)(((mp_word)1 << ((mp_word) DIGIT_BIT)) - x) &
> +MP_MASK;
> ++  *rho = (mp_digit)(((mp_word)1 << ((mp_word) DIGIT_BIT)) - x) &
> ++ MP_MASK;
> +
> +   return MP_OKAY;
> + }
> +# HG changeset patch
> +# User H.J. Lu <hjl.tools at gmail.com>
> +# Date 1369344541 25200
> +# Node ID 7c656e7071a6412688b2f30a529a9afac6c7bf5a
> +# Parent  c7555a4cb7ded3a88409ba85f4027baa7af5f536
> +Define LTC_FAST_TYPE to unsigned long long for __x86_64__
> +
> +We should define LTC_FAST_TYPE to unsigned long long instead of
> +unsigned long if __x86_64__ to support x32 where unsigned long long is
> +64-bit and unsigned long is 32-bit.
> +
> +diff -r c7555a4cb7de -r 7c656e7071a6
> libtomcrypt/src/headers/tomcrypt_cfg.h
> +--- a/libtomcrypt/src/headers/tomcrypt_cfg.h	Thu May 23 14:24:01
> 2013 -0700
> ++++ b/libtomcrypt/src/headers/tomcrypt_cfg.h	Thu May 23 14:29:01
> 2013 -0700
> +@@ -74,7 +74,7 @@
> +    #define ENDIAN_LITTLE
> +    #define ENDIAN_64BITWORD
> +    #define LTC_FAST
> +-   #define LTC_FAST_TYPE    unsigned long
> ++   #define LTC_FAST_TYPE    unsigned long long
> + #endif
> +
> + /* detect PPC32 */
> +# HG changeset patch
> +# User H.J. Lu <hjl.tools at gmail.com>
> +# Date 1369344730 25200
> +# Node ID a7d4690158fae4ede2c4e5b56233e83730bf38ee
> +# Parent  7c656e7071a6412688b2f30a529a9afac6c7bf5a
> +Use unsigned long long aas unsigned 64-bit integer for x86-64 GCC
> +
> +We should use unsigned long long instead of unsigned long as unsigned
> +64-bit integer for x86-64 GCC to support x32 where unsigned long is
> +32-bit.
> +
> +diff -r 7c656e7071a6 -r a7d4690158fa
> libtomcrypt/src/headers/tomcrypt_macros.h
> +--- a/libtomcrypt/src/headers/tomcrypt_macros.h	Thu May 23 14:29:01
> 2013 -0700
> ++++ b/libtomcrypt/src/headers/tomcrypt_macros.h	Thu May 23 14:32:10
> 2013 -0700
> +@@ -343,7 +343,7 @@
> + /* 64-bit Rotates */
> + #if !defined(__STRICT_ANSI__) && defined(__GNUC__) &&
> +defined(__x86_64__) && !defined(LTC_NO_ASM)
> +
> +-static inline unsigned long ROL64(unsigned long word, int i)
> ++static inline unsigned long long ROL64(unsigned long long word, int i)
> + {
> +    asm("rolq %%cl,%0"
> +       :"=r" (word)
> +@@ -351,7 +351,7 @@
> +    return word;
> + }
> +
> +-static inline unsigned long ROR64(unsigned long word, int i)
> ++static inline unsigned long long ROR64(unsigned long long word, int i)
> + {
> +    asm("rorq %%cl,%0"
> +       :"=r" (word)
> +@@ -361,7 +361,7 @@
> +
> + #ifndef LTC_NO_ROLC
> +
> +-static inline unsigned long ROL64c(unsigned long word, const int i)
> ++static inline unsigned long long ROL64c(unsigned long long word, const
> ++int i)
> + {
> +    asm("rolq %2,%0"
> +       :"=r" (word)
> +@@ -369,7 +369,7 @@
> +    return word;
> + }
> +
> +-static inline unsigned long ROR64c(unsigned long word, const int i)
> ++static inline unsigned long long ROR64c(unsigned long long word, const
> ++int i)
> + {
> +    asm("rorq %2,%0"
> +       :"=r" (word)
> +
> diff --git a/meta/recipes-core/dropbear/dropbear.inc b/meta/recipes-
> core/dropbear/dropbear.inc
> index 9864ae8..be93d60 100644
> --- a/meta/recipes-core/dropbear/dropbear.inc
> +++ b/meta/recipes-core/dropbear/dropbear.inc
> @@ -2,7 +2,7 @@ DESCRIPTION = "Dropbear is a lightweight SSH and SCP
> implementation"
>  HOMEPAGE = "http://matt.ucc.asn.au/dropbear/dropbear.html"
>  SECTION = "console/network"
> 
> -INC_PR = "r0"
> +INC_PR = "r1"
> 
>  # some files are from other projects and have others license terms:
>  #   public domain, OpenSSH 3.5p1, OpenSSH3.6.1p2, PuTTY
> @@ -19,6 +19,7 @@ SRC_URI =
> "http://matt.ucc.asn.au/dropbear/releases/dropbear-${PV}.tar.bz2 \
>             file://0002-static_build_fix.patch \
>             file://0003-configure.patch \
>             file://0004-fix-2kb-keys.patch \
> +           file://0007-dropbear-fix-for-x32-abi.patch \
>             file://init \
>             ${@base_contains('DISTRO_FEATURES', 'pam', '${PAM_SRC_URI}', '',
> d)} "
> 
> --
> 1.8.1.4




More information about the Openembedded-core mailing list