pedroediaz/lexvm
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

pike: improve size calculations

Browse Source
This commit is contained in:
Kyryl Melekhin
2022-02-17 18:20:06 +00:00
parent 281820d7c9
commit eb01f29134
2 changed files with 131 additions and 144 deletions
Download Patch File Download Diff File Expand all files Collapse all files

192
README
View File

@@ -23,7 +23,7 @@ Features
* UnLike re1.5, here is only pikevm, one file easy to use. * UnLike re1.5, here is only pikevm, one file easy to use.
* Unlike re1.5, regexes is compiled to type sized code rather than bytecode, * Unlike re1.5, regexes is compiled to type sized code rather than bytecode,
eliviating the problem of byte overflow in splits/jmps on large regexes. eliviating the problem of byte overflow in splits/jmps on large regexes.
Currently the type used is int, and every atom in compiled code is aligned Currently the type used is int, and every atom in compiled code is aligned
to that. to that.
* Matcher does not take size of string as param, it checks for '\0' instead, * Matcher does not take size of string as param, it checks for '\0' instead,
@@ -54,121 +54,103 @@ NOTES
The problem described in this paper has been fixed. Ambiguous matching is correct. The problem described in this paper has been fixed. Ambiguous matching is correct.
HISTORY: HISTORY:
https://re2c.org/2019_borsotti_trofimovich_efficient_posix_submatch_extraction_on_nfa.pdf https://re2c.org/2019_borsotti_trofimovich_efficient_posix_submatch_extraction_on_nfa.pdf
"Cox, 2009 (incorrect). Cox came up with the idea of backward POSIX matching, "Cox, 2009 (incorrect). Cox came up with the idea of backward POSIX matching,
which is based on the observation that reversing the longest-match rule which is based on the observation that reversing the longest-match rule
simplifies the handling of iteration subexpressions: instead of maximizing simplifies the handling of iteration subexpressions: instead of maximizing
submatch from the first to the last iteration, one needs to maximize the submatch from the first to the last iteration, one needs to maximize the
iterations in reverse order. This means that the disambiguation is always iterations in reverse order. This means that the disambiguation is always
based on the most recent iteration, removing the need to remember all previous based on the most recent iteration, removing the need to remember all previous
iterations (except for the backwards-first, i.e. the last one, which contains iterations (except for the backwards-first, i.e. the last one, which contains
submatch result). The algorithm tracks two pairs of offsets per each submatch submatch result). The algorithm tracks two pairs of offsets per each submatch
group: the active pair (used for disambiguation) and the result pair. It gives group: the active pair (used for disambiguation) and the result pair. It gives
incorrect results under two conditions: (1) ambiguous matches have equal incorrect results under two conditions: (1) ambiguous matches have equal
offsets on some iteration, and (2) disambiguation happens too late, when offsets on some iteration, and (2) disambiguation happens too late, when
the active offsets have already been updated and the difference between the active offsets have already been updated and the difference between
ambiguous matches is erased. We found that such situations may occur for two ambiguous matches is erased. We found that such situations may occur for two
reasons. First, the ε-closure algorithm may compare ambiguous paths after reasons. First, the ε-closure algorithm may compare ambiguous paths after
their join point, when both paths have a common suffix with tagged their join point, when both paths have a common suffix with tagged
transitions. This is the case with the Cox prototype implementation; for transitions. This is the case with the Cox prototype implementation; for
example, it gives incorrect results for (aa|a)* and string aaaaa. Most of such example, it gives incorrect results for (aa|a)* and string aaaaa. Most of such
failures can be repaired by exploring states in topological order, but a failures can be repaired by exploring states in topological order, but a
topological order does not exist in the presence of ε-loops. The second reason topological order does not exist in the presence of ε-loops. The second reason
is bounded repetition: ambiguous paths may not have an intermediate join point is bounded repetition: ambiguous paths may not have an intermediate join point
at all. For example, in the case of (aaaa|aaa|a){3,4} and string aaaaaaaaaa we at all. For example, in the case of (aaaa|aaa|a){3,4} and string aaaaaaaaaa we
have matches (aaaa)(aaaa)(a)(a) and (aaaa)(aaa)(aaa) with a different number have matches (aaaa)(aaaa)(a)(a) and (aaaa)(aaa)(aaa) with a different number
of iterations. Assuming that the bounded repetition is unrolled by chaining of iterations. Assuming that the bounded repetition is unrolled by chaining
three sub-automata for (aaaa|aaa|a) and an optional fourth one, by the time three sub-automata for (aaaa|aaa|a) and an optional fourth one, by the time
ambiguous paths meet both have active offsets (0,4). Despite the flaw, Cox ambiguous paths meet both have active offsets (0,4). Despite the flaw, Cox
algorithm is interesting: if somehow the delayed comparison problem was fixed, algorithm is interesting: if somehow the delayed comparison problem was fixed,
it would work. The algorithm requires O(mt) memory and O(nm^2t) time it would work. The algorithm requires O(mt) memory and O(nm^2t) time
(assuming a worst-case optimal closure algorithm), where n is the (assuming a worst-case optimal closure algorithm), where n is the
length of input, m it the size of RE and t is the number of submatch groups length of input, m it the size of RE and t is the number of submatch groups
and subexpressions that contain them." and subexpressions that contain them."
Research has shown that it is possible to disambiguate NFA in polynomial time Research has shown that it is possible to disambiguate NFA in polynomial time
but it brings serious performance issues on non ambiguous inputs. but it brings serious performance issues on non ambiguous inputs. See the
See the branch "disambiguate_paths" on this repo shows what is being branch "disambiguate_paths" on this repo shows what is being done to solve it
done to solve it and the potential performance costs. In short it and the potential performance costs. In short it requires tracking the parent
requires tracking the parent of every state added on nlist from clist. of every state added on nlist from clist. If the state from nlist matches
If the state from nlist matches the consumer, the alternative clist the consumer, the alternative clist state related to that nlist state gets
state related to that nlist state gets discarded and the nsub ref discarded and the nsub ref can be decremented (freed). The reason why this
can be decremented (freed). The reason why this problem does not problem does not exist for non ambiguous regexes is because the alternative
exist for non ambiguous regexes is because the alternative clist clist state will never match due to the next state having a different consumer
state will never match due to the next state having a different . There is no need for any extra handling it gets freed normally. I decided
consumer. There is no need for any extra handling it gets freed normally. to not apply this solution here because I think most use cases for regex are
I decided to not apply this solution here because I think not ambiguious like say regex: "a{10000}". If you try matching 10000 'a'
most use cases for regex are not ambiguious like say regex: characters in a row like that you will have a problem where the stack usage
"a{10000}". If you try matching 10000 'a' characters in a row will jump up to 10000*(subsize) but it will never exceed the size of regex
like that you will have a problem where the stack usage will though, but the number of NFA states will also increase by the same amount,
jump up to 10000*(subsize) but it will never exceed the size so at the charater 9999 you will find 9999 redundant nlist states, that will
of regex though, but the number of NFA states will also increase degrade performance linearly, however it will be very slow compared to
by the same amount, so at the charater 9999 you will find uplimited regex like a+. The cost of this solution is somewhere around 2%
9999 redundant nlist states, that will degrade performance general performance decrease (broadly), but a magnitude of complexity
linearly, however it will be very slow compared to uplimited decrease for ambiguous cases, for example matching 64 characters went down
regex like a+. The cost of this solution is somewhere around from 30 to 9 microseconds. Another solution to this problem can be to
2% general performance decrease (broadly), but a magnitude of determine the ambiguous paths at compile time and flag the inner states as
complexity decrease for ambiguous cases, for example ambiguous ahead of time, still this can't avoid having a loop though the alt
matching 64 characters went down from 30 to 9 microseconds. states as their positioning in clist can't be precomputed due to the dynamic
Another solution to this problem can be to determine the changes.
ambiguous paths at compile time and flag the inner
states as ambiguous ahead of time, still this can't avoid
having a loop though the alt states as their positioning
in clist can't be precomputed due to the dynamic changes.
(Comment about O(mt) memory complexity) (Comment about O(mt) memory complexity)
This worst case scenario can only happen on ambiguous input, that is why nsubs This worst case scenario can only happen on ambiguous input. Ambiguous
size is set to half a MB just in case, this can match 5000000 consumers (char, class, any) assuming t is 1. In practice there is almost
ambiguous consumers (char, class, any) assuming t is 1. In practice there never a situation where someone wants to search using regex this large. Most
is almost never a situation where someone wants to search using regex this of the time memory usage is very low and the space complexity for non
large. Use of alloca() instead of VLA, could remove this limit, I just wish ambigious regex is O(nt) where n is the number of currently considering
it was standardized. If you ever wondered about a situation where alloca alternate paths in the regex and t is the number of submatch groups.
is a must, this is the algorithm.
Most of the time memory usage is very low and the space
complexity for non ambigious regex is O(nt) where n is
the number of currently considering alternate paths
in the regex and t is the number of submatch groups.
This pikevm features an improved submatch extraction This pikevm implementation features an improved submatch extraction algorithm
algorithm based on Russ Cox's original design. based on Russ Cox's original design. I - Kyryl Melekhin have found a way to
I - Kyryl Melekhin have found a way to optimize the tracking optimize the tracking properly of 1st number in the submatch pair. Based on
properly of 1st number in the submatch pair. Based on simple simple observation of how the NFA is constructed I noticed that there is no
observation of how the NFA is constructed I noticed that way for addthread1() to ever reach inner SAVE instructions in the regex, so
there is no way for addthread1() to ever reach inner SAVE that leaves tracking 2nd pairs by addthread1() irrelevant to the final
instructions in the regex, so that leaves tracking 2nd pairs by results (except the need to initialize the sub after allocation). This
addthread1() irrelevant to the final results (except the need to improved the overall performance by 25% which is massive considering that at
initialize the sub after allocation). This improved the overall the time there was nothing else left to can be done to make it faster.
performance by 25% which is massive considering that at the
time there was nothing else left to can be done to make it faster.
What are on##list macros? What are on##list macros?
Redundant state inside nlist can happen in couple of Redundant state inside nlist can happen in couple of ways, and has to do with
ways, and has to do with the (closure) a* (star) operations and the (closure) a* (star) operations and also +. Due to the automata machine
also +. Due to the automata machine design split happens design split happens to be above the next consumed instruction and if that
to be above the next consumed instruction and if that state gets added onto the list we may segfault or give wrong submatch result.
state gets added onto the list we may segfault or give Rsplit does not have this problem because it is generated below the consumer
wrong submatch result. Rsplit does not have this problem instruction, but it can still add redundant states. Overall this is extremely
because it is generated below the consumer instruction, but difficult to understand or explain, but this is just something we have to
it can still add redundant states. Overall this is extremely check for. We checked for this using extra int inside the split instructions,
difficult to understand or explain, but this is just something so this left some global state inside the machine insts. Most of the time we
we have to check for. We checked for this using extra int inside just added to the next gen number and kept incrementing it forever. This
the split instructions, so this left some global state inside the leaves a small chance of overflowing the int and getting a run on a false
machine insts. Most of the time we just added to the next state left from previous use of the regex. Though if overflow never happens
gen number and kept incrementing it forever. This leaves a small there is no chance of getting a false state. Overflows like this pose a high
chance of overflowing the int and getting a run on a false state security threat, if the hacker knows how many cycles he needs to overflow the
left from previous use of the regex. Though if overflow never gen variable and get inconsistent result. It is possible to reset the marks
happens there is no chance of getting a false state. Overflows if we near the overflow, but as you may guess that does not come for free.
like this pose a high security threat, if the hacker knows
how many cycles he needs to overflow the gen variable and get
inconsistent result. It is possible to reset the marks if we
near the overflow, but as you may guess that does not come
for free.
Currently I removed all dynamic global state from the instructions Currently I removed all dynamic global state from the instructions fixing any
fixing any overlow issue utilizing a sparse set datastructure trick overlow issue utilizing a sparse set datastructure trick which abuses the
which abuses the uninitialized varibles. This allows the redundant uninitialized varibles. This allows the redundant states to be excluded in
states to be excluded in O(1) operation. That said, don't run O(1) operation. That said, don't run valgrind on pikevm as it will go crazy, or
valgrind on pikevm as it will go crazy, or find a way to surpress find a way to surpress errors from pikevm.
errors from pikevm.
Further reading Further reading
=============== ===============

83
pike.c
View File

@@ -1,5 +1,8 @@
// Copyright 2007-2009 Russ Cox. All Rights Reserved. /*
// Use of this source code is governed by a BSD-style Copyright 2007-2009 Russ Cox. All Rights Reserved.
Copyright 2020-2021 Kyryl Melekhin. All Rights Reserved.
Use of this source code is governed by a BSD-style
*/
#include <stdio.h> #include <stdio.h>
#include <stdlib.h> #include <stdlib.h>
@@ -51,35 +54,36 @@ static int isword(const char *s)
typedef struct rcode rcode; typedef struct rcode rcode;
struct rcode struct rcode
{ {
int unilen; int unilen; /* number of integers in insts */
int len; int len; /* number of atoms/instructions */
int sub; int sub; /* interim val = save count; final val = nsubs size */
int presub; int presub; /* interim val = save count; final val = 1 rsub size */
int splits; int splits; /* number of split insts */
int insts[]; int sparsesz; /* sdense size */
int insts[]; /* re code */
}; };
enum enum
{ {
// Instructions which consume input bytes (and thus fail if none left) /* Instructions which consume input bytes */
CHAR = 1, CHAR = 1,
CLASS, CLASS,
MATCH, MATCH,
ANY, ANY,
// Assert position /* Assert position */
WBEG, WBEG,
WEND, WEND,
BOL, BOL,
EOL, EOL,
// Other (special) instructions /* Other (special) instructions */
SAVE, SAVE,
// Instructions which take relative offset as arg /* Instructions which take relative offset as arg */
JMP, JMP,
SPLIT, SPLIT,
RSPLIT, RSPLIT,
}; };
// Return codes for re_sizecode() and re_comp() /* Return codes for re_sizecode() and re_comp() */
enum { enum {
RE_SUCCESS = 0, RE_SUCCESS = 0,
RE_SYNTAX_ERROR = -2, RE_SYNTAX_ERROR = -2,
@@ -111,7 +115,7 @@ pc += num;
static int re_classmatch(const int *pc, int c) static int re_classmatch(const int *pc, int c)
{ {
// pc points to "classnot" byte after opcode /* pc points to "classnot" byte after opcode */
int is_positive = *pc++; int is_positive = *pc++;
int cnt = *pc++; int cnt = *pc++;
while (cnt--) { while (cnt--) {
@@ -176,7 +180,7 @@ void re_dumpcode(rcode *prog)
break; break;
} }
} }
printf("Unilen: %d, insts: %d, splits: %d, counted insts: %d\n", printf("unilen: %d, insts: %d, splits: %d, counted insts: %d\n",
prog->unilen, prog->len, prog->splits, i); prog->unilen, prog->len, prog->splits, i);
} }
@@ -196,7 +200,7 @@ static int _compilecode(const char **re_loc, rcode *prog, int sizecode)
switch (*re) { switch (*re) {
case '\\': case '\\':
re++; re++;
if (!*re) goto syntax_error; // Trailing backslash if (!*re) goto syntax_error; /* Trailing backslash */
if (*re == '<' || *re == '>') { if (*re == '<' || *re == '>') {
if (re - *re_loc > 2 && re[-2] == '\\') if (re - *re_loc > 2 && re[-2] == '\\')
break; break;
@@ -223,7 +227,7 @@ static int _compilecode(const char **re_loc, rcode *prog, int sizecode)
re++; re++;
} else } else
EMIT(PC++, 1); EMIT(PC++, 1);
PC++; // Skip "# of pairs" byte PC++; /* Skip "# of pairs" byte */
for (cnt = 0; *re != ']'; cnt++) { for (cnt = 0; *re != ']'; cnt++) {
if (*re == '\\') re++; if (*re == '\\') re++;
if (!*re) goto syntax_error; if (!*re) goto syntax_error;
@@ -372,9 +376,8 @@ int re_sizecode(const char *re, int *nsub)
dummyprog.unilen = 3; dummyprog.unilen = 3;
dummyprog.sub = 0; dummyprog.sub = 0;
int res = _compilecode(&re, &dummyprog, /*sizecode*/1); int res = _compilecode(&re, &dummyprog, 1);
if (res < 0) return res; if (res < 0) return res;
// If unparsed chars left
if (*re) return RE_SYNTAX_ERROR; if (*re) return RE_SYNTAX_ERROR;
*nsub = dummyprog.sub; *nsub = dummyprog.sub;
return dummyprog.unilen; return dummyprog.unilen;
@@ -388,9 +391,8 @@ int re_comp(rcode *prog, const char *re, int nsubs)
prog->presub = nsubs; prog->presub = nsubs;
prog->splits = 0; prog->splits = 0;
int res = _compilecode(&re, prog, /*sizecode*/0); int res = _compilecode(&re, prog, 0);
if (res < 0) return res; if (res < 0) return res;
// If unparsed chars left
if (*re) return RE_SYNTAX_ERROR; if (*re) return RE_SYNTAX_ERROR;
int icnt = 0, scnt = SPLIT; int icnt = 0, scnt = SPLIT;
for (int i = 0; i < prog->unilen; i++) for (int i = 0; i < prog->unilen; i++)
@@ -417,8 +419,11 @@ int re_comp(rcode *prog, const char *re, int nsubs)
prog->insts[prog->unilen++] = SAVE; prog->insts[prog->unilen++] = SAVE;
prog->insts[prog->unilen++] = prog->sub + 1; prog->insts[prog->unilen++] = prog->sub + 1;
prog->insts[prog->unilen++] = MATCH; prog->insts[prog->unilen++] = MATCH;
prog->splits = (scnt - SPLIT) / 2 + SPLIT; prog->splits = (scnt - SPLIT) / 2;
prog->len = icnt+2; prog->len = icnt + 2;
prog->presub = sizeof(rsub)+(sizeof(char*) * (nsubs + 1) * 2);
prog->sub = prog->presub * (prog->len - prog->splits + 4);
prog->sparsesz = (scnt - 2) * 2;
return RE_SUCCESS; return RE_SUCCESS;
} }
@@ -434,8 +439,14 @@ if (--csub->ref == 0) { \
freesub = csub; \ freesub = csub; \
} \ } \
#define deccheck(nn) \ #define rec_check(nn) \
{ decref(nsub) goto rec_check##nn; } \ if (si) { \
npc = pcs[--si]; \
nsub = subs[si]; \
goto rec##nn; \
} \
#define deccheck(nn) { decref(nsub) rec_check(nn) continue; } \
#define onclist(nn) #define onclist(nn)
#define onnlist(nn) \ #define onnlist(nn) \
@@ -493,19 +504,13 @@ if (spc == MATCH) \
} \ } \
#define addthread(nn, list, listidx) \ #define addthread(nn, list, listidx) \
si = 0; \
rec##nn: \ rec##nn: \
spc = *npc; \ spc = *npc; \
if ((unsigned int)spc < WBEG) { \ if ((unsigned int)spc < WBEG) { \
list[listidx].sub = nsub; \ list[listidx].sub = nsub; \
list[listidx++].pc = npc; \ list[listidx++].pc = npc; \
rec_check(nn) \
list##match() \ list##match() \
rec_check##nn: \
if (si) { \
npc = pcs[--si]; \
nsub = subs[si]; \
goto rec##nn; \
} \
continue; \ continue; \
} \ } \
next##nn: \ next##nn: \
@@ -557,18 +562,18 @@ clistidx = nlistidx; \
int re_pikevm(rcode *prog, const char *s, const char **subp, int nsubp) int re_pikevm(rcode *prog, const char *s, const char **subp, int nsubp)
{ {
int rsubsize = sizeof(rsub)+(sizeof(char*)*nsubp); int rsubsize = prog->presub, suboff = rsubsize;
int si, i, j, c, suboff = rsubsize, *npc, osubp = nsubp * sizeof(char*); int spc, i, j, c, *npc, osubp = nsubp * sizeof(char*);
int clistidx = 0, nlistidx, spc, mcont = MATCH; int si = 0, clistidx = 0, nlistidx, mcont = MATCH;
const char *sp = s, *_sp = s; const char *sp = s, *_sp = s;
int *insts = prog->insts; int *insts = prog->insts;
int *pcs[prog->splits]; int *pcs[prog->splits];
unsigned int sdense[prog->splits * 2], sparsesz;
rsub *subs[prog->splits]; rsub *subs[prog->splits];
char nsubs[rsubsize * (prog->len-prog->splits+14)]; unsigned int sdense[prog->sparsesz], sparsesz;
rsub *nsub, *s1, *matched = NULL, *freesub = NULL; rsub *nsub, *s1, *matched = NULL, *freesub = NULL;
rthread _clist[prog->len], _nlist[prog->len]; rthread _clist[prog->len], _nlist[prog->len];
rthread *clist = _clist, *nlist = _nlist, *tmp; rthread *clist = _clist, *nlist = _nlist, *tmp;
char nsubs[prog->sub];
goto jmp_start; goto jmp_start;
for (;; sp = _sp) { for (;; sp = _sp) {
uc_len(i, sp) uc_code(c, sp) uc_len(i, sp) uc_code(c, sp)
@@ -651,10 +656,10 @@ int main(int argc, char *argv[])
printf("Done in %f seconds\n", elapsed_time); printf("Done in %f seconds\n", elapsed_time);
if (!sz) if (!sz)
{ printf("-nomatch-\n"); continue; } { printf("-nomatch-\n"); continue; }
for (int k=sub_els; k>0; k--) for (int k = sub_els; k > 0; k--)
if (sub[k-1]) if (sub[k-1])
break; break;
for (int l=0; l<sub_els; l+=2) { for (int l = 0; l < sub_els; l+=2) {
printf("("); printf("(");
if (sub[l] == NULL || sub[l+1] == NULL) if (sub[l] == NULL || sub[l+1] == NULL)
printf("?"); printf("?");