| 1 |
tove 13/03/04 19:51:15 |
| 2 |
|
| 3 |
Added: perl-5.12.5-rehash-5.12.5.patch |
| 4 |
Log: |
| 5 |
Bump. Fixes CVE-2013-1667 and adds subslots. |
| 6 |
|
| 7 |
(Portage version: 2.2.0_alpha164/cvs/Linux x86_64, signed Manifest commit with key 0x5772769F4E046AEC) |
| 8 |
|
| 9 |
Revision Changes Path |
| 10 |
1.1 dev-lang/perl/files/perl-5.12.5-rehash-5.12.5.patch |
| 11 |
|
| 12 |
file : http://sources.gentoo.org/viewvc.cgi/gentoo-x86/dev-lang/perl/files/perl-5.12.5-rehash-5.12.5.patch?rev=1.1&view=markup |
| 13 |
plain: http://sources.gentoo.org/viewvc.cgi/gentoo-x86/dev-lang/perl/files/perl-5.12.5-rehash-5.12.5.patch?rev=1.1&content-type=text/plain |
| 14 |
|
| 15 |
Index: perl-5.12.5-rehash-5.12.5.patch |
| 16 |
=================================================================== |
| 17 |
From f2a571dae7d70f7e3b59022834d8003ecd2df884 Mon Sep 17 00:00:00 2001 |
| 18 |
From: Yves Orton <demerphq@×××××.com> |
| 19 |
Date: Tue, 12 Feb 2013 10:53:05 +0100 |
| 20 |
Subject: [PATCH] Prevent premature hsplit() calls, and only trigger REHASH |
| 21 |
after hsplit() |
| 22 |
|
| 23 |
Triggering a hsplit due to long chain length allows an attacker |
| 24 |
to create a carefully chosen set of keys which can cause the hash |
| 25 |
to use 2 * (2**32) * sizeof(void *) bytes ram. AKA a DOS via memory |
| 26 |
exhaustion. Doing so also takes non trivial time. |
| 27 |
|
| 28 |
Eliminating this check, and only inspecting chain length after a |
| 29 |
normal hsplit() (triggered when keys>buckets) prevents the attack |
| 30 |
entirely, and makes such attacks relatively benign. |
| 31 |
|
| 32 |
(cherry picked from commit f1220d61455253b170e81427c9d0357831ca0fac) |
| 33 |
--- |
| 34 |
ext/Hash-Util-FieldHash/t/10_hash.t | 18 ++++++++++++++++-- |
| 35 |
hv.c | 26 ++++++-------------------- |
| 36 |
t/op/hash.t | 20 +++++++++++++++++--- |
| 37 |
3 files changed, 39 insertions(+), 25 deletions(-) |
| 38 |
|
| 39 |
diff --git a/ext/Hash-Util-FieldHash/t/10_hash.t b/ext/Hash-Util-FieldHash/t/10_hash.t |
| 40 |
index 2cfb4e8..d58f053 100644 |
| 41 |
--- a/ext/Hash-Util-FieldHash/t/10_hash.t |
| 42 |
+++ b/ext/Hash-Util-FieldHash/t/10_hash.t |
| 43 |
@@ -38,15 +38,29 @@ use constant START => "a"; |
| 44 |
|
| 45 |
# some initial hash data |
| 46 |
fieldhash my %h2; |
| 47 |
-%h2 = map {$_ => 1} 'a'..'cc'; |
| 48 |
+my $counter= "a"; |
| 49 |
+$h2{$counter++}++ while $counter ne 'cd'; |
| 50 |
|
| 51 |
ok (!Internals::HvREHASH(%h2), |
| 52 |
"starting with pre-populated non-pathological hash (rehash flag if off)"); |
| 53 |
|
| 54 |
my @keys = get_keys(\%h2); |
| 55 |
+my $buckets= buckets(\%h2); |
| 56 |
$h2{$_}++ for @keys; |
| 57 |
+$h2{$counter++}++ while buckets(\%h2) == $buckets; # force a split |
| 58 |
ok (Internals::HvREHASH(%h2), |
| 59 |
- scalar(@keys) . " colliding into the same bucket keys are triggering rehash"); |
| 60 |
+ scalar(@keys) . " colliding into the same bucket keys are triggering rehash after split"); |
| 61 |
+ |
| 62 |
+# returns the number of buckets in a hash |
| 63 |
+sub buckets { |
| 64 |
+ my $hr = shift; |
| 65 |
+ my $keys_buckets= scalar(%$hr); |
| 66 |
+ if ($keys_buckets=~m!/([0-9]+)\z!) { |
| 67 |
+ return 0+$1; |
| 68 |
+ } else { |
| 69 |
+ return 8; |
| 70 |
+ } |
| 71 |
+} |
| 72 |
|
| 73 |
sub get_keys { |
| 74 |
my $hr = shift; |
| 75 |
diff --git a/hv.c b/hv.c |
| 76 |
index 89c6456..8659678 100644 |
| 77 |
--- a/hv.c |
| 78 |
+++ b/hv.c |
| 79 |
@@ -35,7 +35,8 @@ holds the key and hash value. |
| 80 |
#define PERL_HASH_INTERNAL_ACCESS |
| 81 |
#include "perl.h" |
| 82 |
|
| 83 |
-#define HV_MAX_LENGTH_BEFORE_SPLIT 14 |
| 84 |
+#define HV_MAX_LENGTH_BEFORE_REHASH 14 |
| 85 |
+#define SHOULD_DO_HSPLIT(xhv) ((xhv)->xhv_keys > (xhv)->xhv_max) /* HvTOTALKEYS(hv) > HvMAX(hv) */ |
| 86 |
|
| 87 |
static const char S_strtab_error[] |
| 88 |
= "Cannot modify shared string table in hv_%s"; |
| 89 |
@@ -818,23 +819,8 @@ Perl_hv_common(pTHX_ HV *hv, SV *keysv, const char *key, STRLEN klen, |
| 90 |
xhv->xhv_keys++; /* HvTOTALKEYS(hv)++ */ |
| 91 |
if (!counter) { /* initial entry? */ |
| 92 |
xhv->xhv_fill++; /* HvFILL(hv)++ */ |
| 93 |
- } else if (xhv->xhv_keys > (IV)xhv->xhv_max) { |
| 94 |
+ } else if ( SHOULD_DO_HSPLIT(xhv) ) { |
| 95 |
hsplit(hv); |
| 96 |
- } else if(!HvREHASH(hv)) { |
| 97 |
- U32 n_links = 1; |
| 98 |
- |
| 99 |
- while ((counter = HeNEXT(counter))) |
| 100 |
- n_links++; |
| 101 |
- |
| 102 |
- if (n_links > HV_MAX_LENGTH_BEFORE_SPLIT) { |
| 103 |
- /* Use only the old HvKEYS(hv) > HvMAX(hv) condition to limit |
| 104 |
- bucket splits on a rehashed hash, as we're not going to |
| 105 |
- split it again, and if someone is lucky (evil) enough to |
| 106 |
- get all the keys in one list they could exhaust our memory |
| 107 |
- as we repeatedly double the number of buckets on every |
| 108 |
- entry. Linear search feels a less worse thing to do. */ |
| 109 |
- hsplit(hv); |
| 110 |
- } |
| 111 |
} |
| 112 |
} |
| 113 |
|
| 114 |
@@ -1180,7 +1166,7 @@ S_hsplit(pTHX_ HV *hv) |
| 115 |
|
| 116 |
|
| 117 |
/* Pick your policy for "hashing isn't working" here: */ |
| 118 |
- if (longest_chain <= HV_MAX_LENGTH_BEFORE_SPLIT /* split worked? */ |
| 119 |
+ if (longest_chain <= HV_MAX_LENGTH_BEFORE_REHASH /* split worked? */ |
| 120 |
|| HvREHASH(hv)) { |
| 121 |
return; |
| 122 |
} |
| 123 |
@@ -2551,8 +2537,8 @@ S_share_hek_flags(pTHX_ const char *str, I32 len, register U32 hash, int flags) |
| 124 |
xhv->xhv_keys++; /* HvTOTALKEYS(hv)++ */ |
| 125 |
if (!next) { /* initial entry? */ |
| 126 |
xhv->xhv_fill++; /* HvFILL(hv)++ */ |
| 127 |
- } else if (xhv->xhv_keys > (IV)xhv->xhv_max /* HvKEYS(hv) > HvMAX(hv) */) { |
| 128 |
- hsplit(PL_strtab); |
| 129 |
+ } else if ( SHOULD_DO_HSPLIT(xhv) ) { |
| 130 |
+ hsplit(PL_strtab); |
| 131 |
} |
| 132 |
} |
| 133 |
|
| 134 |
diff --git a/t/op/hash.t b/t/op/hash.t |
| 135 |
index 9bde518..45eb782 100644 |
| 136 |
--- a/t/op/hash.t |
| 137 |
+++ b/t/op/hash.t |
| 138 |
@@ -39,22 +39,36 @@ use constant THRESHOLD => 14; |
| 139 |
use constant START => "a"; |
| 140 |
|
| 141 |
# some initial hash data |
| 142 |
-my %h2 = map {$_ => 1} 'a'..'cc'; |
| 143 |
+my %h2; |
| 144 |
+my $counter= "a"; |
| 145 |
+$h2{$counter++}++ while $counter ne 'cd'; |
| 146 |
|
| 147 |
ok (!Internals::HvREHASH(%h2), |
| 148 |
"starting with pre-populated non-pathological hash (rehash flag if off)"); |
| 149 |
|
| 150 |
my @keys = get_keys(\%h2); |
| 151 |
+my $buckets= buckets(\%h2); |
| 152 |
$h2{$_}++ for @keys; |
| 153 |
+$h2{$counter++}++ while buckets(\%h2) == $buckets; # force a split |
| 154 |
ok (Internals::HvREHASH(%h2), |
| 155 |
- scalar(@keys) . " colliding into the same bucket keys are triggering rehash"); |
| 156 |
+ scalar(@keys) . " colliding into the same bucket keys are triggering rehash after split"); |
| 157 |
+ |
| 158 |
+# returns the number of buckets in a hash |
| 159 |
+sub buckets { |
| 160 |
+ my $hr = shift; |
| 161 |
+ my $keys_buckets= scalar(%$hr); |
| 162 |
+ if ($keys_buckets=~m!/([0-9]+)\z!) { |
| 163 |
+ return 0+$1; |
| 164 |
+ } else { |
| 165 |
+ return 8; |
| 166 |
+ } |
| 167 |
+} |
| 168 |
|
| 169 |
sub get_keys { |
| 170 |
my $hr = shift; |
| 171 |
|
| 172 |
# the minimum of bits required to mount the attack on a hash |
| 173 |
my $min_bits = log(THRESHOLD)/log(2); |
| 174 |
- |
| 175 |
# if the hash has already been populated with a significant amount |
| 176 |
# of entries the number of mask bits can be higher |
| 177 |
my $keys = scalar keys %$hr; |
| 178 |
-- |
| 179 |
1.8.1.3 |
Archives