|
 |
f84e6a |
From c031e3ec7c713077659f5f7dc6638d926c69d7b2 Mon Sep 17 00:00:00 2001
|
|
|
f84e6a |
From: Hugo van der Sanden <hv@crypt.org>
|
|
|
f84e6a |
Date: Sat, 11 Apr 2020 14:10:24 +0100
|
|
|
f84e6a |
Subject: [PATCH v528 3/3] study_chunk: avoid mutating regexp program within
|
|
|
f84e6a |
GOSUB
|
|
|
f84e6a |
|
|
|
f84e6a |
gh16947 and gh17743: studying GOSUB may restudy in an inner call
|
|
|
f84e6a |
(via a mix of recursion and enframing) something that an outer call
|
|
|
f84e6a |
is in the middle of looking at. Let the outer frame deal with it.
|
|
|
f84e6a |
|
|
|
f84e6a |
(CVE-2020-12723)
|
|
|
f84e6a |
---
|
|
|
f84e6a |
embed.fnc | 2 +-
|
|
|
f84e6a |
embed.h | 2 +-
|
|
|
f84e6a |
proto.h | 2 +-
|
|
|
f84e6a |
regcomp.c | 48 ++++++++++++++++++++++++++++++++----------------
|
|
|
f84e6a |
t/re/pat.t | 26 +++++++++++++++++++++++++-
|
|
|
f84e6a |
5 files changed, 60 insertions(+), 20 deletions(-)
|
|
|
f84e6a |
|
|
|
f84e6a |
diff --git a/embed.fnc b/embed.fnc
|
|
|
f84e6a |
index cf89277163..4b1ba28277 100644
|
|
|
f84e6a |
--- a/embed.fnc
|
|
|
f84e6a |
+++ b/embed.fnc
|
|
|
f84e6a |
@@ -2397,7 +2397,7 @@ Es |SSize_t|study_chunk |NN RExC_state_t *pRExC_state \
|
|
|
f84e6a |
|NULLOK struct scan_data_t *data \
|
|
|
f84e6a |
|I32 stopparen|U32 recursed_depth \
|
|
|
f84e6a |
|NULLOK regnode_ssc *and_withp \
|
|
|
f84e6a |
- |U32 flags|U32 depth
|
|
|
f84e6a |
+ |U32 flags|U32 depth|bool was_mutate_ok
|
|
|
f84e6a |
EsRn |U32 |add_data |NN RExC_state_t* const pRExC_state \
|
|
|
f84e6a |
|NN const char* const s|const U32 n
|
|
|
f84e6a |
rs |void |re_croak2 |bool utf8|NN const char* pat1|NN const char* pat2|...
|
|
|
f84e6a |
diff --git a/embed.h b/embed.h
|
|
|
f84e6a |
index 886551ce5c..50fcabc140 100644
|
|
|
f84e6a |
--- a/embed.h
|
|
|
f84e6a |
+++ b/embed.h
|
|
|
f84e6a |
@@ -1075,7 +1075,7 @@
|
|
|
f84e6a |
#define ssc_is_cp_posixl_init S_ssc_is_cp_posixl_init
|
|
|
f84e6a |
#define ssc_or(a,b,c) S_ssc_or(aTHX_ a,b,c)
|
|
|
f84e6a |
#define ssc_union(a,b,c) S_ssc_union(aTHX_ a,b,c)
|
|
|
f84e6a |
-#define study_chunk(a,b,c,d,e,f,g,h,i,j,k) S_study_chunk(aTHX_ a,b,c,d,e,f,g,h,i,j,k)
|
|
|
f84e6a |
+#define study_chunk(a,b,c,d,e,f,g,h,i,j,k,l) S_study_chunk(aTHX_ a,b,c,d,e,f,g,h,i,j,k,l)
|
|
|
f84e6a |
# endif
|
|
|
f84e6a |
# if defined(PERL_IN_REGCOMP_C) || defined (PERL_IN_DUMP_C)
|
|
|
f84e6a |
#define _invlist_dump(a,b,c,d) Perl__invlist_dump(aTHX_ a,b,c,d)
|
|
|
f84e6a |
diff --git a/proto.h b/proto.h
|
|
|
f84e6a |
index d3f8802c1d..e276f69bd1 100644
|
|
|
f84e6a |
--- a/proto.h
|
|
|
f84e6a |
+++ b/proto.h
|
|
|
f84e6a |
@@ -5258,7 +5258,7 @@ PERL_STATIC_INLINE void S_ssc_union(pTHX_ regnode_ssc *ssc, SV* const invlist, c
|
|
|
f84e6a |
#define PERL_ARGS_ASSERT_SSC_UNION \
|
|
|
f84e6a |
assert(ssc); assert(invlist)
|
|
|
f84e6a |
#endif
|
|
|
f84e6a |
-STATIC SSize_t S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp, SSize_t *minlenp, SSize_t *deltap, regnode *last, struct scan_data_t *data, I32 stopparen, U32 recursed_depth, regnode_ssc *and_withp, U32 flags, U32 depth);
|
|
|
f84e6a |
+STATIC SSize_t S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp, SSize_t *minlenp, SSize_t *deltap, regnode *last, struct scan_data_t *data, I32 stopparen, U32 recursed_depth, regnode_ssc *and_withp, U32 flags, U32 depth, bool was_mutate_ok);
|
|
|
f84e6a |
#define PERL_ARGS_ASSERT_STUDY_CHUNK \
|
|
|
f84e6a |
assert(pRExC_state); assert(scanp); assert(minlenp); assert(deltap); assert(last)
|
|
|
f84e6a |
#endif
|
|
|
f84e6a |
diff --git a/regcomp.c b/regcomp.c
|
|
|
f84e6a |
index 0a9c6a8085..e66032a16a 100644
|
|
|
f84e6a |
--- a/regcomp.c
|
|
|
f84e6a |
+++ b/regcomp.c
|
|
|
f84e6a |
@@ -110,6 +110,7 @@ typedef struct scan_frame {
|
|
|
f84e6a |
regnode *next_regnode; /* next node to process when last is reached */
|
|
|
f84e6a |
U32 prev_recursed_depth;
|
|
|
f84e6a |
I32 stopparen; /* what stopparen do we use */
|
|
|
f84e6a |
+ bool in_gosub; /* this or an outer frame is for GOSUB */
|
|
|
f84e6a |
U32 is_top_frame; /* what flags do we use? */
|
|
|
f84e6a |
|
|
|
f84e6a |
struct scan_frame *this_prev_frame; /* this previous frame */
|
|
|
f84e6a |
@@ -4102,7 +4103,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
|
|
|
f84e6a |
I32 stopparen,
|
|
|
f84e6a |
U32 recursed_depth,
|
|
|
f84e6a |
regnode_ssc *and_withp,
|
|
|
f84e6a |
- U32 flags, U32 depth)
|
|
|
f84e6a |
+ U32 flags, U32 depth, bool was_mutate_ok)
|
|
|
f84e6a |
/* scanp: Start here (read-write). */
|
|
|
f84e6a |
/* deltap: Write maxlen-minlen here. */
|
|
|
f84e6a |
/* last: Stop before this one. */
|
|
|
f84e6a |
@@ -4179,6 +4180,10 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
|
|
|
f84e6a |
node length to get a real minimum (because
|
|
|
f84e6a |
the folded version may be shorter) */
|
|
|
f84e6a |
bool unfolded_multi_char = FALSE;
|
|
|
f84e6a |
+ /* avoid mutating ops if we are anywhere within the recursed or
|
|
|
f84e6a |
+ * enframed handling for a GOSUB: the outermost level will handle it.
|
|
|
f84e6a |
+ */
|
|
|
f84e6a |
+ bool mutate_ok = was_mutate_ok && !(frame && frame->in_gosub);
|
|
|
f84e6a |
/* Peephole optimizer: */
|
|
|
f84e6a |
DEBUG_STUDYDATA("Peep:", data, depth);
|
|
|
f84e6a |
DEBUG_PEEP("Peep", scan, depth);
|
|
|
f84e6a |
@@ -4189,7 +4194,8 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
|
|
|
f84e6a |
* parsing code, as each (?:..) is handled by a different invocation of
|
|
|
f84e6a |
* reg() -- Yves
|
|
|
f84e6a |
*/
|
|
|
f84e6a |
- JOIN_EXACT(scan,&min_subtract, &unfolded_multi_char, 0);
|
|
|
f84e6a |
+ if (mutate_ok)
|
|
|
f84e6a |
+ JOIN_EXACT(scan,&min_subtract, &unfolded_multi_char, 0);
|
|
|
f84e6a |
|
|
|
f84e6a |
/* Follow the next-chain of the current node and optimize
|
|
|
f84e6a |
away all the NOTHINGs from it. */
|
|
|
f84e6a |
@@ -4238,7 +4244,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
|
|
|
f84e6a |
* NOTE we dont use the return here! */
|
|
|
f84e6a |
(void)study_chunk(pRExC_state, &scan, &minlen,
|
|
|
f84e6a |
&deltanext, next, &data_fake, stopparen,
|
|
|
f84e6a |
- recursed_depth, NULL, f, depth+1);
|
|
|
f84e6a |
+ recursed_depth, NULL, f, depth+1, mutate_ok);
|
|
|
f84e6a |
|
|
|
f84e6a |
scan = next;
|
|
|
f84e6a |
} else
|
|
|
f84e6a |
@@ -4305,7 +4311,8 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
|
|
|
f84e6a |
/* we suppose the run is continuous, last=next...*/
|
|
|
f84e6a |
minnext = study_chunk(pRExC_state, &scan, minlenp,
|
|
|
f84e6a |
&deltanext, next, &data_fake, stopparen,
|
|
|
f84e6a |
- recursed_depth, NULL, f,depth+1);
|
|
|
f84e6a |
+ recursed_depth, NULL, f, depth+1,
|
|
|
f84e6a |
+ mutate_ok);
|
|
|
f84e6a |
|
|
|
f84e6a |
if (min1 > minnext)
|
|
|
f84e6a |
min1 = minnext;
|
|
|
f84e6a |
@@ -4372,9 +4379,10 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
|
|
|
f84e6a |
}
|
|
|
f84e6a |
}
|
|
|
f84e6a |
|
|
|
f84e6a |
- if (PERL_ENABLE_TRIE_OPTIMISATION &&
|
|
|
f84e6a |
- OP( startbranch ) == BRANCH )
|
|
|
f84e6a |
- {
|
|
|
f84e6a |
+ if (PERL_ENABLE_TRIE_OPTIMISATION
|
|
|
f84e6a |
+ && OP(startbranch) == BRANCH
|
|
|
f84e6a |
+ && mutate_ok
|
|
|
f84e6a |
+ ) {
|
|
|
f84e6a |
/* demq.
|
|
|
f84e6a |
|
|
|
f84e6a |
Assuming this was/is a branch we are dealing with: 'scan'
|
|
|
f84e6a |
@@ -4825,6 +4833,9 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
|
|
|
f84e6a |
newframe->stopparen = stopparen;
|
|
|
f84e6a |
newframe->prev_recursed_depth = recursed_depth;
|
|
|
f84e6a |
newframe->this_prev_frame= frame;
|
|
|
f84e6a |
+ newframe->in_gosub = (
|
|
|
f84e6a |
+ (frame && frame->in_gosub) || OP(scan) == GOSUB
|
|
|
f84e6a |
+ );
|
|
|
f84e6a |
|
|
|
f84e6a |
DEBUG_STUDYDATA("frame-new:",data,depth);
|
|
|
f84e6a |
DEBUG_PEEP("fnew", scan, depth);
|
|
|
f84e6a |
@@ -5043,7 +5054,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
|
|
|
f84e6a |
(mincount == 0
|
|
|
f84e6a |
? (f & ~SCF_DO_SUBSTR)
|
|
|
f84e6a |
: f)
|
|
|
f84e6a |
- ,depth+1);
|
|
|
f84e6a |
+ , depth+1, mutate_ok);
|
|
|
f84e6a |
|
|
|
f84e6a |
if (flags & SCF_DO_STCLASS)
|
|
|
f84e6a |
data->start_class = oclass;
|
|
|
f84e6a |
@@ -5105,7 +5116,9 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
|
|
|
f84e6a |
if ( OP(oscan) == CURLYX && data
|
|
|
f84e6a |
&& data->flags & SF_IN_PAR
|
|
|
f84e6a |
&& !(data->flags & SF_HAS_EVAL)
|
|
|
f84e6a |
- && !deltanext && minnext == 1 ) {
|
|
|
f84e6a |
+ && !deltanext && minnext == 1
|
|
|
f84e6a |
+ && mutate_ok
|
|
|
f84e6a |
+ ) {
|
|
|
f84e6a |
/* Try to optimize to CURLYN. */
|
|
|
f84e6a |
regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
|
|
|
f84e6a |
regnode * const nxt1 = nxt;
|
|
|
f84e6a |
@@ -5151,10 +5164,10 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
|
|
|
f84e6a |
&& !(data->flags & SF_HAS_EVAL)
|
|
|
f84e6a |
&& !deltanext /* atom is fixed width */
|
|
|
f84e6a |
&& minnext != 0 /* CURLYM can't handle zero width */
|
|
|
f84e6a |
-
|
|
|
f84e6a |
/* Nor characters whose fold at run-time may be
|
|
|
f84e6a |
* multi-character */
|
|
|
f84e6a |
&& ! (RExC_seen & REG_UNFOLDED_MULTI_SEEN)
|
|
|
f84e6a |
+ && mutate_ok
|
|
|
f84e6a |
) {
|
|
|
f84e6a |
/* XXXX How to optimize if data == 0? */
|
|
|
f84e6a |
/* Optimize to a simpler form. */
|
|
|
f84e6a |
@@ -5201,7 +5214,8 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
|
|
|
f84e6a |
#endif
|
|
|
f84e6a |
/* Optimize again: */
|
|
|
f84e6a |
study_chunk(pRExC_state, &nxt1, minlenp, &deltanext, nxt,
|
|
|
f84e6a |
- NULL, stopparen, recursed_depth, NULL, 0,depth+1);
|
|
|
f84e6a |
+ NULL, stopparen, recursed_depth, NULL, 0,
|
|
|
f84e6a |
+ depth+1, mutate_ok);
|
|
|
f84e6a |
}
|
|
|
f84e6a |
else
|
|
|
f84e6a |
oscan->flags = 0;
|
|
|
f84e6a |
@@ -5592,7 +5606,8 @@ Perl_re_printf( aTHX_ "LHS=%" UVuf " RHS=%" UVuf "\n",
|
|
|
f84e6a |
nscan = NEXTOPER(NEXTOPER(scan));
|
|
|
f84e6a |
minnext = study_chunk(pRExC_state, &nscan, minlenp, &deltanext,
|
|
|
f84e6a |
last, &data_fake, stopparen,
|
|
|
f84e6a |
- recursed_depth, NULL, f, depth+1);
|
|
|
f84e6a |
+ recursed_depth, NULL, f, depth+1,
|
|
|
f84e6a |
+ mutate_ok);
|
|
|
f84e6a |
if (scan->flags) {
|
|
|
f84e6a |
if (deltanext) {
|
|
|
f84e6a |
FAIL("Variable length lookbehind not implemented");
|
|
|
f84e6a |
@@ -5681,7 +5696,7 @@ Perl_re_printf( aTHX_ "LHS=%" UVuf " RHS=%" UVuf "\n",
|
|
|
f84e6a |
*minnextp = study_chunk(pRExC_state, &nscan, minnextp,
|
|
|
f84e6a |
&deltanext, last, &data_fake,
|
|
|
f84e6a |
stopparen, recursed_depth, NULL,
|
|
|
f84e6a |
- f,depth+1);
|
|
|
f84e6a |
+ f, depth+1, mutate_ok);
|
|
|
f84e6a |
if (scan->flags) {
|
|
|
f84e6a |
if (deltanext) {
|
|
|
f84e6a |
FAIL("Variable length lookbehind not implemented");
|
|
|
f84e6a |
@@ -5841,7 +5856,8 @@ Perl_re_printf( aTHX_ "LHS=%" UVuf " RHS=%" UVuf "\n",
|
|
|
f84e6a |
branches even though they arent otherwise used. */
|
|
|
f84e6a |
minnext = study_chunk(pRExC_state, &scan, minlenp,
|
|
|
f84e6a |
&deltanext, (regnode *)nextbranch, &data_fake,
|
|
|
f84e6a |
- stopparen, recursed_depth, NULL, f,depth+1);
|
|
|
f84e6a |
+ stopparen, recursed_depth, NULL, f, depth+1,
|
|
|
f84e6a |
+ mutate_ok);
|
|
|
f84e6a |
}
|
|
|
f84e6a |
if (nextbranch && PL_regkind[OP(nextbranch)]==BRANCH)
|
|
|
f84e6a |
nextbranch= regnext((regnode*)nextbranch);
|
|
|
f84e6a |
@@ -7524,7 +7540,7 @@ Perl_re_op_compile(pTHX_ SV ** const patternp, int pat_count,
|
|
|
f84e6a |
&data, -1, 0, NULL,
|
|
|
f84e6a |
SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag
|
|
|
f84e6a |
| (restudied ? SCF_TRIE_DOING_RESTUDY : 0),
|
|
|
f84e6a |
- 0);
|
|
|
f84e6a |
+ 0, TRUE);
|
|
|
f84e6a |
|
|
|
f84e6a |
|
|
|
f84e6a |
CHECK_RESTUDY_GOTO_butfirst(LEAVE_with_name("study_chunk"));
|
|
|
f84e6a |
@@ -7670,7 +7686,7 @@ Perl_re_op_compile(pTHX_ SV ** const patternp, int pat_count,
|
|
|
f84e6a |
SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS|(restudied
|
|
|
f84e6a |
? SCF_TRIE_DOING_RESTUDY
|
|
|
f84e6a |
: 0),
|
|
|
f84e6a |
- 0);
|
|
|
f84e6a |
+ 0, TRUE);
|
|
|
f84e6a |
|
|
|
f84e6a |
CHECK_RESTUDY_GOTO_butfirst(NOOP);
|
|
|
f84e6a |
|
|
|
f84e6a |
diff --git a/t/re/pat.t b/t/re/pat.t
|
|
|
f84e6a |
index 1d98fe77d7..1488259b02 100644
|
|
|
f84e6a |
--- a/t/re/pat.t
|
|
|
f84e6a |
+++ b/t/re/pat.t
|
|
|
f84e6a |
@@ -23,7 +23,7 @@ BEGIN {
|
|
|
f84e6a |
skip_all('no re module') unless defined &DynaLoader::boot_DynaLoader;
|
|
|
f84e6a |
skip_all_without_unicode_tables();
|
|
|
f84e6a |
|
|
|
f84e6a |
-plan tests => 840; # Update this when adding/deleting tests.
|
|
|
f84e6a |
+plan tests => 844; # Update this when adding/deleting tests.
|
|
|
f84e6a |
|
|
|
f84e6a |
run_tests() unless caller;
|
|
|
f84e6a |
|
|
|
f84e6a |
@@ -1929,6 +1929,30 @@ EOP
|
|
|
f84e6a |
fresh_perl_is('"AA" =~ m/AA{1,0}/','',{},"handle OPFAIL insert properly");
|
|
|
f84e6a |
}
|
|
|
f84e6a |
|
|
|
f84e6a |
+ # gh16947: test regexp corruption (GOSUB)
|
|
|
f84e6a |
+ {
|
|
|
f84e6a |
+ fresh_perl_is(q{
|
|
|
f84e6a |
+ 'xy' =~ /x(?0)|x(?|y|y)/ && print 'ok'
|
|
|
f84e6a |
+ }, 'ok', {}, 'gh16947: test regexp corruption (GOSUB)');
|
|
|
f84e6a |
+ }
|
|
|
f84e6a |
+ # gh16947: test fix doesn't break SUSPEND
|
|
|
f84e6a |
+ {
|
|
|
f84e6a |
+ fresh_perl_is(q{ 'sx' =~ m{ss++}i; print 'ok' },
|
|
|
f84e6a |
+ 'ok', {}, "gh16947: test fix doesn't break SUSPEND");
|
|
|
f84e6a |
+ }
|
|
|
f84e6a |
+
|
|
|
f84e6a |
+ # gh17743: more regexp corruption via GOSUB
|
|
|
f84e6a |
+ {
|
|
|
f84e6a |
+ fresh_perl_is(q{
|
|
|
f84e6a |
+ "0" =~ /((0(?0)|000(?|0000|0000)(?0))|)/; print "ok"
|
|
|
f84e6a |
+ }, 'ok', {}, 'gh17743: test regexp corruption (1)');
|
|
|
f84e6a |
+
|
|
|
f84e6a |
+ fresh_perl_is(q{
|
|
|
f84e6a |
+ "000000000000" =~ /(0(())(0((?0)())|000(?|\x{ef}\x{bf}\x{bd}|\x{ef}\x{bf}\x{bd}))|)/;
|
|
|
f84e6a |
+ print "ok"
|
|
|
f84e6a |
+ }, 'ok', {}, 'gh17743: test regexp corruption (2)');
|
|
|
f84e6a |
+ }
|
|
|
f84e6a |
+
|
|
|
f84e6a |
} # End of sub run_tests
|
|
|
f84e6a |
|
|
|
f84e6a |
1;
|
|
|
f84e6a |
--
|
|
|
f84e6a |
2.20.1
|
|
|
f84e6a |
|