File perl-study.diff of Package perl.23624
--- ./embed.fnc.orig 2017-09-07 19:19:52.000000000 +0000
+++ ./embed.fnc 2020-06-09 16:21:37.259095322 +0000
@@ -2397,7 +2397,8 @@ Es |SSize_t|study_chunk |NN RExC_state_t
|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
EsRn |U32 |add_data |NN RExC_state_t* const pRExC_state \
|NN const char* const s|const U32 n
rs |void |re_croak2 |bool utf8|NN const char* pat1|NN const char* pat2|...
--- ./embed.h.orig 2017-07-18 22:59:59.000000000 +0000
+++ ./embed.h 2020-06-09 16:21:37.263095314 +0000
@@ -1046,6 +1046,7 @@
#define output_or_return_posix_warnings(a,b,c) S_output_or_return_posix_warnings(aTHX_ a,b,c)
#define parse_lparen_question_flags(a) S_parse_lparen_question_flags(aTHX_ a)
#define populate_ANYOF_from_invlist(a,b) S_populate_ANYOF_from_invlist(aTHX_ a,b)
+#define rck_elide_nothing(a) S_rck_elide_nothing(aTHX_ a)
#define reg(a,b,c,d) S_reg(aTHX_ a,b,c,d)
#define reg2Lanode(a,b,c,d) S_reg2Lanode(aTHX_ a,b,c,d)
#define reg_node(a,b) S_reg_node(aTHX_ a,b)
@@ -1075,7 +1076,7 @@
#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)
--- ./proto.h.orig 2017-09-07 19:19:52.000000000 +0000
+++ ./proto.h 2020-06-09 16:21:52.091061856 +0000
@@ -5150,6 +5150,9 @@ STATIC void S_parse_lparen_question_flag
STATIC void S_populate_ANYOF_from_invlist(pTHX_ regnode *node, SV** invlist_ptr);
#define PERL_ARGS_ASSERT_POPULATE_ANYOF_FROM_INVLIST \
assert(node); assert(invlist_ptr)
+STATIC void S_rck_elide_nothing(pTHX_ regnode *node);
+#define PERL_ARGS_ASSERT_RCK_ELIDE_NOTHING \
+ assert(node)
PERL_STATIC_NO_RET void S_re_croak2(pTHX_ bool utf8, const char* pat1, const char* pat2, ...)
__attribute__noreturn__;
#define PERL_ARGS_ASSERT_RE_CROAK2 \
@@ -5258,7 +5261,7 @@ PERL_STATIC_INLINE void S_ssc_union(pTHX
#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
--- ./regcomp.c.orig 2020-06-09 16:21:32.515106026 +0000
+++ ./regcomp.c 2020-06-09 16:21:37.263095314 +0000
@@ -110,6 +110,7 @@ typedef struct scan_frame {
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 */
U32 is_top_frame; /* what flags do we use? */
struct scan_frame *this_prev_frame; /* this previous frame */
@@ -4093,6 +4094,44 @@ S_unwind_scan_frames(pTHX_ const void *p
} while (f);
}
+/* Follow the next-chain of the current node and optimize away
+ all the NOTHINGs from it.
+ */
+STATIC void
+S_rck_elide_nothing(pTHX_ regnode *node)
+{
+ dVAR;
+
+ PERL_ARGS_ASSERT_RCK_ELIDE_NOTHING;
+
+ if (OP(node) != CURLYX) {
+ const int max = (reg_off_by_arg[OP(node)]
+ ? I32_MAX
+ /* I32 may be smaller than U16 on CRAYs! */
+ : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
+ int off = (reg_off_by_arg[OP(node)] ? ARG(node) : NEXT_OFF(node));
+ int noff;
+ regnode *n = node;
+
+ /* Skip NOTHING and LONGJMP. */
+ while (
+ (n = regnext(n))
+ && (
+ (PL_regkind[OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
+ || ((OP(n) == LONGJMP) && (noff = ARG(n)))
+ )
+ && off + noff < max
+ ) {
+ off += noff;
+ }
+ if (reg_off_by_arg[OP(node)])
+ ARG(node) = off;
+ else
+ NEXT_OFF(node) = off;
+ }
+ return;
+}
+
STATIC SSize_t
S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
@@ -4102,7 +4141,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_
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. */
@@ -4179,6 +4218,10 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_
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);
DEBUG_PEEP("Peep", scan, depth);
@@ -4189,30 +4232,12 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_
* 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. */
- if (OP(scan) != CURLYX) {
- const int max = (reg_off_by_arg[OP(scan)]
- ? I32_MAX
- /* I32 may be smaller than U16 on CRAYs! */
- : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
- int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
- int noff;
- regnode *n = scan;
-
- /* Skip NOTHING and LONGJMP. */
- while ((n = regnext(n))
- && ((PL_regkind[OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
- || ((OP(n) == LONGJMP) && (noff = ARG(n))))
- && off + noff < max)
- off += noff;
- if (reg_off_by_arg[OP(scan)])
- ARG(scan) = off;
- else
- NEXT_OFF(scan) = off;
- }
+ rck_elide_nothing(scan);
/* The principal pseudo-switch. Cannot be a switch, since we
look into several different things. */
@@ -4238,7 +4263,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_
* NOTE we dont use the return here! */
(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
@@ -4305,7 +4330,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_
/* we suppose the run is continuous, last=next...*/
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;
@@ -4373,7 +4398,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_
}
if (PERL_ENABLE_TRIE_OPTIMISATION &&
- OP( startbranch ) == BRANCH )
+ OP( startbranch ) == BRANCH && mutate_ok)
{
/* demq.
@@ -4825,6 +4850,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_
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);
DEBUG_PEEP("fnew", scan, depth);
@@ -5043,7 +5069,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_
(mincount == 0
? (f & ~SCF_DO_SUBSTR)
: f)
- ,depth+1);
+ ,depth+1, mutate_ok);
if (flags & SCF_DO_STCLASS)
data->start_class = oclass;
@@ -5091,6 +5117,12 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_
(void)ReREFCNT_inc(RExC_rx_sv);
}
+ if ( ( minnext > 0 && mincount >= SSize_t_MAX / minnext )
+ || min >= SSize_t_MAX - minnext * mincount )
+ {
+ FAIL("Regexp out of space");
+ }
+
min += minnext * mincount;
is_inf_internal |= deltanext == SSize_t_MAX
|| (maxcount == REG_INFTY && minnext + deltanext > 0);
@@ -5105,7 +5137,8 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_
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;
@@ -5155,6 +5188,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_
/* 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. */
@@ -5201,7 +5235,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_
#endif
/* Optimize again: */
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;
@@ -5328,11 +5362,7 @@ Perl_re_printf( aTHX_ "LHS=%" UVuf " RH
if (data && (fl & SF_HAS_EVAL))
data->flags |= SF_HAS_EVAL;
optimize_curly_tail:
- if (OP(oscan) != CURLYX) {
- while (PL_regkind[OP(next = regnext(oscan))] == NOTHING
- && NEXT_OFF(next))
- NEXT_OFF(oscan) += NEXT_OFF(next);
- }
+ rck_elide_nothing(oscan);
continue;
default:
@@ -5592,7 +5622,7 @@ Perl_re_printf( aTHX_ "LHS=%" UVuf " RH
nscan = NEXTOPER(NEXTOPER(scan));
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");
@@ -5681,7 +5711,7 @@ Perl_re_printf( aTHX_ "LHS=%" UVuf " RH
*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");
@@ -5841,7 +5871,7 @@ Perl_re_printf( aTHX_ "LHS=%" UVuf " RH
branches even though they arent otherwise used. */
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);
@@ -7520,7 +7550,7 @@ Perl_re_op_compile(pTHX_ SV ** const pat
&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"));
@@ -7666,7 +7696,7 @@ Perl_re_op_compile(pTHX_ SV ** const pat
SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS|(restudied
? SCF_TRIE_DOING_RESTUDY
: 0),
- 0);
+ 0, TRUE);
CHECK_RESTUDY_GOTO_butfirst(NOOP);
