Bug 945842 (CVE-2015-5276) - VUL-1: CVE-2015-5276 gcc: Predictable randomness from std::random_device
Summary: VUL-1: CVE-2015-5276 gcc: Predictable randomness from std::random_device
Status: RESOLVED FIXED
Alias: CVE-2015-5276
Product: SUSE Security Incidents
Classification: Novell Products
Component: Incidents (show other bugs)
Version: unspecified
Hardware: Other Other
: P4 - Low : Minor
Target Milestone: ---
Assignee: Richard Biener
QA Contact: Security Team bot
URL: https://smash.suse.de/issue/156558/
Whiteboard: CVSSv2:RedHat:CVE-2015-5276:3.3:(AV:L...
Keywords:
Depends on:
Blocks:
 
Reported: 2015-09-15 10:10 UTC by Victor Pereira
Modified: 2017-09-07 22:37 UTC (History)
5 users (show)

See Also:
Found By: Security Response Team
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Victor Pereira 2015-09-15 10:10:08 UTC 
rh#1262846

It was discovered that the std::random_device class in libstdc++ would
not properly detect short reads and could return return predictable
values if applications used it to obtain randomness from a blocking
source such as /dev/random.


References:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65142
https://gcc.gnu.org/viewcvs/gcc?view=revision&revision=227687 (upstream commit)
https://bugzilla.redhat.com/show_bug.cgi?id=1262846
http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2015-5276
http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2015-5276
Comment 1 Richard Biener 2015-09-15 10:40:08 UTC 
Note the upstream fix isn't complete/correct, waiting for sth better to appear.
Comment 2 Swamp Workflow Management 2015-09-15 22:00:36 UTC 
bugbot adjusting priority
Comment 3 Andreas Stieger 2015-10-02 12:48:57 UTC 
(In reply to Richard Biener from comment #1)
> Note the upstream fix isn't complete/correct, waiting for sth better to
> appear.

A second commit appeared:

https://gcc.gnu.org/viewcvs/gcc?view=revision&revision=227687
https://gcc.gnu.org/viewcvs/gcc?view=revision&revision=227872
( https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65142#c11 )

Does that cover the issue?
Comment 4 Richard Biener 2015-10-02 13:04:18 UTC 
I've asked for confirmation (and backport).  A pending update would needs to
cover the system C++ runtime which is the one from GCC 5 on SLE11 SP4 + Updates (soon) and SLE12 + Updates.  SLE11 SP3 + Update will have GCC 4.8 (soon).
Remaining is LTSS SLE11 where some customers have the 4.8 runtime and some
the original one.
Comment 6 Bernhard Wiedemann 2015-10-08 13:00:16 UTC 
This is an autogenerated message for OBS integration:
This bug (945842) was mentioned in
https://build.opensuse.org/request/show/337212 Factory / gcc5
Comment 7 Swamp Workflow Management 2015-10-27 16:10:03 UTC 
SUSE-SU-2015:1833-1: An update that solves one vulnerability and has four fixes is now available.

Category: security (moderate)
Bug References: 945842,947772,947791,948168,949000
CVE References: CVE-2015-5276
Sources used:
SUSE Linux Enterprise Workstation Extension 12 (src):    libgcj48-4.8.5-24.1
SUSE Linux Enterprise Software Development Kit 12 (src):    gcc48-4.8.5-24.1, libffi48-4.8.5-24.1, libgcj48-4.8.5-24.1
SUSE Linux Enterprise Server 12 (src):    gcc48-4.8.5-24.1, libffi48-4.8.5-24.1
SUSE Linux Enterprise Desktop 12 (src):    gcc48-4.8.5-24.1, libgcj48-4.8.5-24.1
Comment 8 Swamp Workflow Management 2015-11-10 10:11:52 UTC 
openSUSE-SU-2015:1946-1: An update that solves one vulnerability and has four fixes is now available.

Category: security (moderate)
Bug References: 945842,947772,947791,948168,949000
CVE References: CVE-2015-5276
Sources used:
openSUSE Leap 42.1 (src):    cross-aarch64-gcc48-icecream-backend-4.8.5-18.2, cross-armv6hl-gcc48-icecream-backend-4.8.5-18.2, cross-armv7hl-gcc48-icecream-backend-4.8.5-18.2, cross-hppa-gcc48-icecream-backend-4.8.5-18.2, cross-i386-gcc48-icecream-backend-4.8.5-18.2, cross-ia64-gcc48-icecream-backend-4.8.5-18.2, cross-ppc-gcc48-icecream-backend-4.8.5-18.2, cross-ppc64-gcc48-icecream-backend-4.8.5-18.2, cross-ppc64le-gcc48-icecream-backend-4.8.5-18.2, cross-s390-gcc48-icecream-backend-4.8.5-18.2, cross-s390x-gcc48-icecream-backend-4.8.5-18.2, gcc48-4.8.5-18.1, gcc48-testresults-4.8.5-18.3, gdb-7.9.1-7.1, libffi48-4.8.5-18.1, libgcj48-4.8.5-18.1
Comment 11 Marcus Meissner 2016-03-21 14:32:50 UTC 
Richi, can you provide a small testprogram that runs the code so we can verify it is fixed?

(if it prints numbers and we can see different behaviuour before/after this would be ok.)
Comment 12 Richard Biener 2016-03-21 14:59:03 UTC 
There is no way to do a small test program.

#include <iostream>
#include <string>
#include <map>
#include <random>
#include <cstdlib>
int main(int argc, char **argv)
{
  if (argc < 3)
    return 1;
  std::random_device rd (argv[1]);
  std::map<int, int> hist;
  int num = std::atoi(argv[2]);
  for(int n=0; n<num; ++n)
    ++hist[rd()%10];
  for(auto p : hist)
    std::cout << p.first << " : " << std::string(p.second/(num < 100 ? 1 : num / 100), '*') << '\n';
}

if you call that with /dev/random 1000 and make the program interrupt itself
reliably during each rd() call then you'd see a non-even distribution.  But
of course after the fix you simply see a not handled exception.

The question is how to reliably get the test program interrupted.  Eventually
it is enough for you to see the unhandled exception?

Anyway, a simple alarm() doesn't work:

read(3, "\22\374\325\336", 4)           = 4
read(3, 0x7ffd6a98aebc, 4)              = ? ERESTARTSYS (To be restarted if SA_RESTART is set)
--- SIGALRM {si_signo=SIGALRM, si_code=SI_KERNEL} ---
rt_sigreturn()                          = 0
read(3, "C<\17\4", 4)                   = 4

it looks like all fopen files are opened blocking and read is always
restartable?

at least the reads from /dev/random block reliably...
Comment 13 Michael Matz 2016-03-21 15:20:42 UTC 
One could establish the signal handler for SIGALRM via sigaction(2) without
the SA_RESTART flag.  Then read() should return with -1 with EINTR.  But I don't
see how this can be relieably tested.  Throwing signals in quick succession maybe, and then measuring the histogram for skew.
Comment 14 Marcus Meissner 2016-03-21 15:55:11 UTC 
i was hoping for something simpler.

then we will probably not verify the fix currently and trust upstream fixed it correctly.
Comment 15 Swamp Workflow Management 2016-04-01 10:14:24 UTC 
SUSE-SU-2016:0908-1: An update that solves one vulnerability and has 7 fixes is now available.

Category: security (moderate)
Bug References: 939460,945842,953831,955382,962765,964468,966220,968771
CVE References: CVE-2015-5276
Sources used:
SUSE Linux Enterprise Debuginfo 11-SP4 (src):    gcc5-5.3.1+r233831-10.1, libffi-gcc5-5.3.1+r233831-10.1

Product List: SUSE Linux Enterprise Debuginfo 11-SP4
Comment 16 Swamp Workflow Management 2016-04-01 10:58:40 UTC 
SUSE-SU-2016:0908-2: An update that solves one vulnerability and has 7 fixes is now available.

Category: security (moderate)
Bug References: 939460,945842,953831,955382,962765,964468,966220,968771
CVE References: CVE-2015-5276
Sources used:
SUSE Linux Enterprise Software Development Kit 11-SP4 (src):    gcc5-5.3.1+r233831-10.1, libffi-gcc5-5.3.1+r233831-10.1
SUSE Linux Enterprise Server 11-SP4 (src):    gcc5-5.3.1+r233831-10.1, libffi-gcc5-5.3.1+r233831-10.1
SUSE Linux Enterprise Desktop 11-SP4 (src):    gcc5-5.3.1+r233831-10.1
SUSE Linux Enterprise Debuginfo 11-SP4 (src):    gcc5-5.3.1+r233831-10.1, libffi-gcc5-5.3.1+r233831-10.1

Product List: SUSE Linux Enterprise Software Development Kit 11-SP4
SUSE Linux Enterprise Server 11-SP4
SUSE Linux Enterprise Desktop 11-SP4
SUSE Linux Enterprise Debuginfo 11-SP4
Comment 17 Richard Biener 2016-04-01 11:50:34 UTC 
Fixed where C++11 is supported.
Comment 18 Swamp Workflow Management 2016-04-06 18:08:18 UTC 
SUSE-SU-2016:0963-1: An update that solves one vulnerability and has 9 fixes is now available.

Category: security (moderate)
Bug References: 939460,945842,952151,953831,954002,955382,962765,964468,966220,968771
CVE References: CVE-2015-5276
Sources used:
SUSE Linux Enterprise Software Development Kit 12-SP1 (src):    gcc5-5.3.1+r233831-9.1
SUSE Linux Enterprise Server 12-SP1 (src):    gcc5-5.3.1+r233831-9.1, libffi-gcc5-5.3.1+r233831-9.1
SUSE Linux Enterprise Server 12 (src):    gcc5-5.3.1+r233831-9.1, libffi-gcc5-5.3.1+r233831-9.1
SUSE Linux Enterprise Module for Toolchain 12 (src):    gcc5-5.3.1+r233831-9.1, libffi-gcc5-5.3.1+r233831-9.1
SUSE Linux Enterprise Desktop 12-SP1 (src):    gcc5-5.3.1+r233831-9.1, libffi-gcc5-5.3.1+r233831-9.1
SUSE Linux Enterprise Desktop 12 (src):    gcc5-5.3.1+r233831-9.1, libffi-gcc5-5.3.1+r233831-9.1
Comment 19 Swamp Workflow Management 2016-04-17 15:09:24 UTC 
openSUSE-SU-2016:1069-1: An update that solves one vulnerability and has 9 fixes is now available.

Category: security (moderate)
Bug References: 939460,945842,952151,953831,954002,955382,962765,964468,966220,968771
CVE References: CVE-2015-5276
Sources used:
openSUSE Leap 42.1 (src):    gcc5-5.3.1+r233831-6.1, gcc5-testresults-5.3.1+r233831-6.2, libffi-gcc5-5.3.1+r233831-6.1
Comment 20 Swamp Workflow Management 2017-02-03 23:09:51 UTC 
SUSE-SU-2017:0379-1: An update that solves one vulnerability and has 21 fixes is now available.

Category: security (moderate)
Bug References: 1011348,889990,899871,917169,919274,922534,924525,924687,930176,934689,945842,947772,947791,948168,949000,951644,955382,970009,976627,977654,981311,988274
CVE References: CVE-2015-5276
Sources used:
SUSE Linux Enterprise Software Development Kit 11-SP4 (src):    gcc48-4.8.5-4.2
SUSE Linux Enterprise Debuginfo 11-SP4 (src):    gcc48-4.8.5-4.2
Comment 21 Swamp Workflow Management 2017-08-22 19:09:49 UTC 
SUSE-SU-2017:2235-1: An update that fixes 51 vulnerabilities is now available.

Category: security (important)
Bug References: 1025108,1031485,1035082,1043960,930392,930496,935510,939460,945842,953831,954002,955382,962765,964468,966220,968771
CVE References: CVE-2015-5276,CVE-2016-10196,CVE-2017-5429,CVE-2017-5430,CVE-2017-5432,CVE-2017-5433,CVE-2017-5434,CVE-2017-5435,CVE-2017-5436,CVE-2017-5438,CVE-2017-5439,CVE-2017-5440,CVE-2017-5441,CVE-2017-5442,CVE-2017-5443,CVE-2017-5444,CVE-2017-5445,CVE-2017-5446,CVE-2017-5447,CVE-2017-5448,CVE-2017-5449,CVE-2017-5451,CVE-2017-5454,CVE-2017-5455,CVE-2017-5456,CVE-2017-5459,CVE-2017-5460,CVE-2017-5461,CVE-2017-5462,CVE-2017-5464,CVE-2017-5465,CVE-2017-5466,CVE-2017-5467,CVE-2017-5469,CVE-2017-5470,CVE-2017-5472,CVE-2017-7749,CVE-2017-7750,CVE-2017-7751,CVE-2017-7752,CVE-2017-7754,CVE-2017-7755,CVE-2017-7756,CVE-2017-7757,CVE-2017-7758,CVE-2017-7761,CVE-2017-7763,CVE-2017-7764,CVE-2017-7765,CVE-2017-7768,CVE-2017-7778
Sources used:
SUSE Linux Enterprise Software Development Kit 11-SP4 (src):    MozillaFirefox-52.2.0esr-72.5.2, mozilla-nss-3.29.5-47.3.2
SUSE Linux Enterprise Server 11-SP4 (src):    MozillaFirefox-52.2.0esr-72.5.2, MozillaFirefox-branding-SLED-52-24.3.44, firefox-gcc5-5.3.1+r233831-7.1, firefox-libffi-gcc5-5.3.1+r233831-7.1, mozilla-nss-3.29.5-47.3.2
SUSE Linux Enterprise Server 11-SP3-LTSS (src):    MozillaFirefox-52.2.0esr-72.5.2, MozillaFirefox-branding-SLED-52-24.3.44, firefox-gcc5-5.3.1+r233831-7.1, firefox-libffi-gcc5-5.3.1+r233831-7.1, mozilla-nss-3.29.5-47.3.2
SUSE Linux Enterprise Point of Sale 11-SP3 (src):    MozillaFirefox-52.2.0esr-72.5.2, MozillaFirefox-branding-SLED-52-24.3.44, firefox-gcc5-5.3.1+r233831-7.1, firefox-libffi-gcc5-5.3.1+r233831-7.1, mozilla-nss-3.29.5-47.3.2
SUSE Linux Enterprise Debuginfo 11-SP4 (src):    MozillaFirefox-52.2.0esr-72.5.2, mozilla-nss-3.29.5-47.3.2
SUSE Linux Enterprise Debuginfo 11-SP3 (src):    MozillaFirefox-52.2.0esr-72.5.2, firefox-gcc5-5.3.1+r233831-7.1, firefox-libffi-gcc5-5.3.1+r233831-7.1, mozilla-nss-3.29.5-47.3.2