From c0b79e3c354b25f9fb732bd49f668ae5eeaa723a Mon Sep 17 00:00:00 2001 From: Hugo van der Sanden Date: Sat, 11 Apr 2020 14:10:24 +0100 Subject: [PATCH] study_chunk: avoid mutating regexp program within GOSUB gh16947 and gh17743: studying GOSUB may restudy in an inner call (via a mix of recursion and enframing) something that an outer call is in the middle of looking at. Let the outer frame deal with it. (CVE-2020-12723) (cherry picked from commit c4033e740bd18d9fbe3456a9db2ec2053cdc5271) port from: https://github.com/perl/perl5/commit/66bbb51b93253a3f87d11c2695cfb7bdb782184a Signed-off-by: Peibao Liu --- embed.fnc | 2 +- embed.h | 2 +- proto.h | 2 +- regcomp.c | 42 ++++++++++++++++++++++++++++++------------ t/re/pat.t | 27 ++++++++++++++++++++++++++- 5 files changed, 59 insertions(+), 16 deletions(-) diff -Naur a/embed.fnc b/embed.fnc --- a/embed.fnc 2020-07-14 15:18:38.077000000 -0400 +++ b/embed.fnc 2020-07-14 15:21:38.116000000 -0400 @@ -2475,7 +2475,7 @@ |NULLOK struct scan_data_t *data \ |I32 stopparen|U32 recursed_depth \ |NULLOK regnode_ssc *and_withp \ - |U32 flags|U32 depth + |U32 flags|U32 depth|bool was_mutate_ok Es |void |rck_elide_nothing|NN regnode *node EsR |SV * |get_ANYOFM_contents|NN const regnode * n EsRn |U32 |add_data |NN RExC_state_t* const pRExC_state \ diff -Naur a/embed.h b/embed.h --- a/embed.h 2020-07-14 15:18:38.095000000 -0400 +++ b/embed.h 2020-07-14 15:21:58.416000000 -0400 @@ -1232,7 +1232,7 @@ #define ssc_is_cp_posixl_init S_ssc_is_cp_posixl_init #define ssc_or(a,b,c) S_ssc_or(aTHX_ a,b,c) #define ssc_union(a,b,c) S_ssc_union(aTHX_ a,b,c) -#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) +#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) # endif # if defined(PERL_IN_REGCOMP_C) || defined (PERL_IN_DUMP_C) #define _invlist_dump(a,b,c,d) Perl__invlist_dump(aTHX_ a,b,c,d) diff -Naur a/proto.h b/proto.h --- a/proto.h 2020-07-14 15:18:38.074000000 -0400 +++ b/proto.h 2020-07-14 15:22:41.447000000 -0400 @@ -5593,7 +5593,7 @@ #define PERL_ARGS_ASSERT_SSC_UNION \ assert(ssc); assert(invlist) #endif -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); +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); #define PERL_ARGS_ASSERT_STUDY_CHUNK \ assert(pRExC_state); assert(scanp); assert(minlenp); assert(deltap); assert(last) #endif diff -Naur a/regcomp.c b/regcomp.c --- a/regcomp.c 2020-07-14 15:18:38.091000000 -0400 +++ b/regcomp.c 2020-07-14 15:34:01.739000000 -0400 @@ -110,6 +110,7 @@ regnode *next_regnode; /* next node to process when last is reached */ U32 prev_recursed_depth; I32 stopparen; /* what stopparen do we use */ + bool in_gosub; /* this or an outer frame is for GOSUB */ struct scan_frame *this_prev_frame; /* this previous frame */ struct scan_frame *prev_frame; /* previous frame */ @@ -4225,7 +4226,7 @@ I32 stopparen, U32 recursed_depth, regnode_ssc *and_withp, - U32 flags, U32 depth) + U32 flags, U32 depth, bool was_mutate_ok) /* scanp: Start here (read-write). */ /* deltap: Write maxlen-minlen here. */ /* last: Stop before this one. */ @@ -4303,6 +4304,10 @@ node length to get a real minimum (because the folded version may be shorter) */ bool unfolded_multi_char = FALSE; + /* avoid mutating ops if we are anywhere within the recursed or + * enframed handling for a GOSUB: the outermost level will handle it. + */ + bool mutate_ok = was_mutate_ok && !(frame && frame->in_gosub); /* Peephole optimizer: */ DEBUG_STUDYDATA("Peep", data, depth, is_inf); DEBUG_PEEP("Peep", scan, depth, flags); @@ -4313,7 +4318,8 @@ * parsing code, as each (?:..) is handled by a different invocation of * reg() -- Yves */ - JOIN_EXACT(scan,&min_subtract, &unfolded_multi_char, 0); + if (mutate_ok) + JOIN_EXACT(scan,&min_subtract, &unfolded_multi_char, 0); /* Follow the next-chain of the current node and optimize away all the NOTHINGs from it. @@ -4345,7 +4351,7 @@ /* DEFINEP study_chunk() recursion */ (void)study_chunk(pRExC_state, &scan, &minlen, &deltanext, next, &data_fake, stopparen, - recursed_depth, NULL, f, depth+1); + recursed_depth, NULL, f, depth+1, mutate_ok); scan = next; } else @@ -4413,7 +4419,8 @@ /* recurse study_chunk() for each BRANCH in an alternation */ minnext = study_chunk(pRExC_state, &scan, minlenp, &deltanext, next, &data_fake, stopparen, - recursed_depth, NULL, f,depth+1); + recursed_depth, NULL, f, depth+1, + mutate_ok); if (min1 > minnext) min1 = minnext; @@ -4480,9 +4487,10 @@ } } - if (PERL_ENABLE_TRIE_OPTIMISATION && - OP( startbranch ) == BRANCH ) - { + if (PERL_ENABLE_TRIE_OPTIMISATION + && OP(startbranch) == BRANCH + && mutate_ok + ) { /* demq. Assuming this was/is a branch we are dealing with: 'scan' @@ -4933,6 +4941,9 @@ newframe->stopparen = stopparen; newframe->prev_recursed_depth = recursed_depth; newframe->this_prev_frame= frame; + newframe->in_gosub = ( + (frame && frame->in_gosub) || OP(scan) == GOSUB + ); DEBUG_STUDYDATA("frame-new", data, depth, is_inf); DEBUG_PEEP("fnew", scan, depth, flags); @@ -5153,7 +5164,7 @@ (mincount == 0 ? (f & ~SCF_DO_SUBSTR) : f) - ,depth+1); + , depth+1, mutate_ok); if (flags & SCF_DO_STCLASS) data->start_class = oclass; @@ -5221,7 +5232,9 @@ if ( OP(oscan) == CURLYX && data && data->flags & SF_IN_PAR && !(data->flags & SF_HAS_EVAL) - && !deltanext && minnext == 1 ) { + && !deltanext && minnext == 1 + && mutate_ok + ) { /* Try to optimize to CURLYN. */ regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; regnode * const nxt1 = nxt; @@ -5267,10 +5280,10 @@ && !(data->flags & SF_HAS_EVAL) && !deltanext /* atom is fixed width */ && minnext != 0 /* CURLYM can't handle zero width */ - /* Nor characters whose fold at run-time may be * multi-character */ && ! (RExC_seen & REG_UNFOLDED_MULTI_SEEN) + && mutate_ok ) { /* XXXX How to optimize if data == 0? */ /* Optimize to a simpler form. */ @@ -5318,7 +5331,8 @@ /* Optimize again: */ /* recurse study_chunk() on optimised CURLYX => CURLYM */ study_chunk(pRExC_state, &nxt1, minlenp, &deltanext, nxt, - NULL, stopparen, recursed_depth, NULL, 0,depth+1); + NULL, stopparen, recursed_depth, NULL, 0, + depth+1, mutate_ok); } else oscan->flags = 0; @@ -5735,7 +5749,8 @@ /* recurse study_chunk() for lookahead body */ minnext = study_chunk(pRExC_state, &nscan, minlenp, &deltanext, last, &data_fake, stopparen, - recursed_depth, NULL, f, depth+1); + recursed_depth, NULL, f, depth+1, + mutate_ok); if (scan->flags) { if (deltanext) { FAIL("Variable length lookbehind not implemented"); @@ -5827,7 +5842,7 @@ *minnextp = study_chunk(pRExC_state, &nscan, minnextp, &deltanext, last, &data_fake, stopparen, recursed_depth, NULL, - f,depth+1); + f, depth+1, mutate_ok); if (scan->flags) { if (deltanext) { FAIL("Variable length lookbehind not implemented"); @@ -5988,7 +6003,8 @@ /* optimise study_chunk() for TRIE */ minnext = study_chunk(pRExC_state, &scan, minlenp, &deltanext, (regnode *)nextbranch, &data_fake, - stopparen, recursed_depth, NULL, f,depth+1); + stopparen, recursed_depth, NULL, f, depth+1, + mutate_ok); } if (nextbranch && PL_regkind[OP(nextbranch)]==BRANCH) nextbranch= regnext((regnode*)nextbranch); @@ -7673,7 +7689,7 @@ &data, -1, 0, NULL, SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag | (restudied ? SCF_TRIE_DOING_RESTUDY : 0), - 0); + 0, TRUE); CHECK_RESTUDY_GOTO_butfirst(LEAVE_with_name("study_chunk")); @@ -7802,7 +7818,7 @@ SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS|(restudied ? SCF_TRIE_DOING_RESTUDY : 0), - 0); + 0, TRUE); CHECK_RESTUDY_GOTO_butfirst(NOOP); diff -Naur a/t/re/pat.t b/t/re/pat.t --- a/t/re/pat.t 2020-07-14 15:18:37.898000000 -0400 +++ b/t/re/pat.t 2020-07-14 15:40:01.252000000 -0400 @@ -23,7 +23,7 @@ skip_all('no re module') unless defined &DynaLoader::boot_DynaLoader; skip_all_without_unicode_tables(); -plan tests => 848; # Update this when adding/deleting tests. +plan tests => 852; # Update this when adding/deleting tests. run_tests() unless caller; @@ -1948,6 +1948,30 @@ fresh_perl_is('m m0*0+\Rm', "",{},"Undefined behavior in address sanitizer"); } + # gh16947: test regexp corruption (GOSUB) + { + fresh_perl_is(q{ + 'xy' =~ /x(?0)|x(?|y|y)/ && print 'ok' + }, 'ok', {}, 'gh16947: test regexp corruption (GOSUB)'); + } + # gh16947: test fix doesn't break SUSPEND + { + fresh_perl_is(q{ 'sx' =~ m{ss++}i; print 'ok' }, + 'ok', {}, "gh16947: test fix doesn't break SUSPEND"); + } + + # gh17743: more regexp corruption via GOSUB + { + fresh_perl_is(q{ + "0" =~ /((0(?0)|000(?|0000|0000)(?0))|)/; print "ok" + }, 'ok', {}, 'gh17743: test regexp corruption (1)'); + + fresh_perl_is(q{ + "000000000000" =~ /(0(())(0((?0)())|000(?|\x{ef}\x{bf}\x{bd}|\x{ef}\x{bf}\x{bd}))|)/; + print "ok" + }, 'ok', {}, 'gh17743: test regexp corruption (2)'); + } + } # End of sub run_tests 1;