Thread: scanner/parser minimization
Today's b^Hdiscussion on materialized views reminded me that I spent a
little bit of time looking at gram.y and thinking about what we might
be able to do to reduce the amount of bloat it spits out. On my
system, without debugging symbols, gram.o is 1019260 bytes. Using nm
gram.o | sort | less to compare the starting addresses of each symbol
with the next symbol, I figured out the size of some of the larger
symbols: yy_transition is 516288 bytes, and yycheck and yytable are
each 145984 bytes. Thus, anything which doesn't reduce the size of
one of those three symbols isn't going to help very much.
yy_transition appears in scan.c - that is, it's part of the flex
output, not the yacc output. It is an array of 64,535 yy_trans_info
structures, each containing two flex_int32_t members. Off-hand the
values all look small enough to fit in 2-byte integers rather than
4-byte integers; it's not clear to me why flex doesn't do that. It's
possible to vastly reduce the size of the scanner output, and
therefore of gram.c, by running flex with -Cf rather than -CF, which
changes the table representation completely. I assume there is a
sound performance reason why we don't do this, but it might be worth
checking that if we haven't lately. When compiled with -Cf, the size
of gram.o drops from 1019260 bytes to 703600, which is a large
savings.
yytable and yycheck are both part of the parser; that is, they appear
in gram.c. Each is an array of 2-byte integers. I'm not clear on
exactly how these relate to the size of the state table, but it seems
that the parser has the idea that each state has a list of actions
associated with specific next-tokens, and then it also has a "default
action" which is followed for next-tokens for which no specific rule
is present. I believe that yytable and yycheck somehow encode these
transition tables, but the details are not altogether clear to me. If
that view of the situation is correct, then a reasonable way of
approaching the task of reducing the parser size is to look for states
in gram.output (generated by running bison with the -v option) that
have a lot of non-default rules and mulling over how we might reduce
that number.
A whole lot of those state transitions are attributable to states
which have separate transitions for each of many keywords. They
transition to states which then reduce the corresponding keyword to
unreserved_keyword, col_name_keyword, type_func_keyword, or
reserved_keyword. So it would seem that if we could reduce those
transitions in some way, it would help a lot. I experimented with
this a little. Consider the following patch:
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -12489,8 +12489,6 @@ SignedIconst: Iconst { $$ = $1; }/* Column identifier --- names
thatcan be column, table, etc names. */ColId: IDENT { $$ = $1; }
- | unreserved_keyword { $$ = pstrdup($1); }
- | col_name_keyword { $$ = pstrdup($1); } ;
/* Type/function identifier --- names that can be type or function names.
On my machine, that two-line patch has the effect of reducing the size
of gram.o from 1019260 bytes to 844844 bytes, a savings of 164kB.
Pretty good for ripping out two lines of code. Applying
similarly-crude hacks to type_function_name or ColLabel is not nearly
as effective. If I hack up all three, gram.o drops to 794612 bytes, a
savings of 49 additional kB. If I hack up ONLY one of the other two
and NOT ColId, the savings are much less. Clearly ColId is the big
offender by far. I suspect, but am not positive, that this is simply
because ColId is more widely-used throughout the grammar. For
example, the following patch actually makes gram.o about 4kB larger:
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -5381,7 +5381,7 @@ SecLabelStmt: } ;
-opt_provider: FOR ColId_or_Sconst { $$ = $2; }
+opt_provider: FOR ColLabel { $$ = $2; } | /* empty */ { $$ = NULL; }
;
Accepting more rather than fewer keywords in that position makes
things worse, which makes sense.
Interestingly, this analysis suggests that while reserved keywords and
type_func_name keywords are surely worse from an application
compatibility standpoint, unreserved and col_name_keywords are worse
from a parser bloat standpoint, because there are more contexts where
we have to laboriously ignore their status as keywords, via additional
parser state transitions.
I don't have a brilliant idea on what to do about this, but one
thought that did occur to me is that we might be able to use mid-rule
actions to change the scanner behavior. In other words, suppose we
reach a point in the input string where we know that we no longer care
about parsing unreserved keywords as keywords, but are instead happy
to have them returned as IDENT. We could then, at that point in the
rule, do something like { scan_unreserved_keywords = false; }, and the
{identifier} production in scan.l would then behave differently based
on the state of that flag. That would in turn allow gram.y symbols
used after that point in the rule to NOT include unreserved_keywords
alongside IDENT, which would slice out a bunch of state transitions.
There are a couple of problems with this idea. The complexity of
maintaining it is surely one, since we'd need duplicate productions
for certain things depending on the context in which they were
expected to be used. We could perhaps use Assert() against the
scan_unreserved_keywords flag (or its moral equivalent) to catch
coding mistakes. A more serious objection is that there are several
different categories of unreserved keywords and they don't all fit
neatly into this concept. For example, ENCRYPTED and UNENCRYPTED are
only used as keywords within ALTER ROLE (what a waste!), so somehow
shutting off recognition of those keywords elsewhere in the grammar
might work OK. But ZONE can appear in an arbitrary a_expr and must be
recognized there. It manages to be unreserved only because it
invariably follows TIME. If you can't flip the "don't scan unreserved
keywords" switch in any place where an a_expr might appear, it seems
to me that this idea is unlikely to apply in enough situations to
bring much benefit. Also, the mid-rule actions create new parser
states, which eats away at the gains we might otherwise make through
such tricks.
So all in all I'm not quite sure where to go with this, but I thought
that the foregoing analysis might be interesting enough to be worth
posting, in case it inspires someone else with an idea that seems
worth pursuing.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
Robert Haas <robertmhaas@gmail.com> writes:
> A whole lot of those state transitions are attributable to states
> which have separate transitions for each of many keywords.
Yeah, that's no surprise.
The idea that's been in the back of my mind for awhile is to try to
solve the problem at the lexer level not the parser level: that is,
have the lexer return IDENT whenever a keyword appears in a context
where it should be interpreted as unreserved. You suggest somehow
driving that off mid-rule actions, but that seems to be to be probably
a nonstarter from a code complexity and maintainability standpoint.
I believe however that it's possible to extract an idea of which
tokens the parser believes it can see next at any given parse state.
(I've seen code for this somewhere on the net, but am too lazy to go
searching for it again right now.) So we could imagine a rule along
the lines of "if IDENT is allowed as a next token, and $KEYWORD is
not, then return IDENT not the keyword's own token".
This might be unworkable from a speed standpoint, depending on how
expensive it is to make the determination about allowable next symbols.
But it seems worth looking into.
regards, tom lane
On 2/28/13 3:34 PM, Robert Haas wrote: > It's > possible to vastly reduce the size of the scanner output, and > therefore of gram.c, by running flex with -Cf rather than -CF, which > changes the table representation completely. I assume there is a > sound performance reason why we don't do this, but it might be worth > checking that if we haven't lately. When compiled with -Cf, the size > of gram.o drops from 1019260 bytes to 703600, which is a large > savings. The option choice is based on the recommendation in the flex manual. It wouldn't hurt to re-test it.
Regarding yytransition I think the problem is we're using flex to implement keyword recognition which is usually not what it's used for. Usually people use flex to handle syntax things like quoting and numeric formats. All identifiers are handled by flex as equivalent. Then the last step in the scanner for identifiers is to look up the identifier in a hash table and return the keyword token if it's a keyword. That would massively simplify the scanner tables. This hash table can be heavily optimized because it's a static lookup. c.f. http://www.gnu.org/software/gperf/ In theory this is more expensive since it needs to do a strcmp in addition to scanning the identifier to determine whether the token ends. But I suspect in practice the smaller tables might outweight that cost. On Thu, Feb 28, 2013 at 9:09 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote: > I believe however that it's possible to extract an idea of which > tokens the parser believes it can see next at any given parse state. > (I've seen code for this somewhere on the net, but am too lazy to go > searching for it again right now.) So we could imagine a rule along > the lines of "if IDENT is allowed as a next token, and $KEYWORD is > not, then return IDENT not the keyword's own token". That's a pretty cool idea. I'm afraid it might be kind of slow to produce that list and load it into a hash table for every ident token though. I suppose if you can request it only if the ident is a keyword of some type then it wouldn't actually kick in often. That would also imply we could simply use IDENT everywhere we currently have col_name_keyword or type_function_name. Just by having a rule that accepts a keyword that would implicitly make it not be accepted as an IDENT when unquoted in that location. That might make the documentation a bit trickier and it would make it harder for users to make their code forward-compatible. It would also make it harder for hackers to determine when we've accidentally narrowed the allowable identifiers for more cases than they expect. -- greg
Greg Stark <stark@mit.edu> writes:
> Regarding yytransition I think the problem is we're using flex to
> implement keyword recognition which is usually not what it's used for.
> Usually people use flex to handle syntax things like quoting and
> numeric formats. All identifiers are handled by flex as equivalent.
> Then the last step in the scanner for identifiers is to look up the
> identifier in a hash table and return the keyword token if it's a
> keyword. That would massively simplify the scanner tables.
Uh ... no. I haven't looked into why the flex tables are so large,
but this theory is just wrong. See ScanKeywordLookup().
regards, tom lane
On Thu, Feb 28, 2013 at 4:09 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote: > I believe however that it's possible to extract an idea of which > tokens the parser believes it can see next at any given parse state. > (I've seen code for this somewhere on the net, but am too lazy to go > searching for it again right now.) So we could imagine a rule along > the lines of "if IDENT is allowed as a next token, and $KEYWORD is > not, then return IDENT not the keyword's own token". > > This might be unworkable from a speed standpoint, depending on how > expensive it is to make the determination about allowable next symbols. > But it seems worth looking into. Interesting idea. But wouldn't that change the semantics of the grammar in some places? In particular, keywords would generally become less-reserved than they are now, but in a context-dependent way. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company
Robert Haas <robertmhaas@gmail.com> writes:
> On Thu, Feb 28, 2013 at 4:09 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> I believe however that it's possible to extract an idea of which
>> tokens the parser believes it can see next at any given parse state.
>> (I've seen code for this somewhere on the net, but am too lazy to go
>> searching for it again right now.) So we could imagine a rule along
>> the lines of "if IDENT is allowed as a next token, and $KEYWORD is
>> not, then return IDENT not the keyword's own token".
>>
>> This might be unworkable from a speed standpoint, depending on how
>> expensive it is to make the determination about allowable next symbols.
>> But it seems worth looking into.
> Interesting idea. But wouldn't that change the semantics of the
> grammar in some places? In particular, keywords would generally
> become less-reserved than they are now, but in a context-dependent
> way.
Yeah, Greg already raised that point. It shouldn't change the semantics
of the grammar *today*, but it would become much harder to tell whether
future changes create any conflicts; we'd lose the mechanical
verification we get from bison with the current method. That might be
sufficient reason not to pursue the idea.
It's possible that static analysis of the grammar could replace bison
verification; that is, assuming we have a way to identify all states in
which both IDENT and some-keyword are allowable, we could generate a
report listing which keywords are potentially going to be considered
non-reserved, and then watch for changes in that report. But this would
require writing even more code that we don't have today.
I suspect also that we'd still need a notion of fully reserved keywords
that don't get flipped to IDENT even if that's a legal next symbol.
regards, tom lane
On 02.03.2013 17:09, Tom Lane wrote: > Greg Stark<stark@mit.edu> writes: >> Regarding yytransition I think the problem is we're using flex to >> implement keyword recognition which is usually not what it's used for. >> Usually people use flex to handle syntax things like quoting and >> numeric formats. All identifiers are handled by flex as equivalent. >> Then the last step in the scanner for identifiers is to look up the >> identifier in a hash table and return the keyword token if it's a >> keyword. That would massively simplify the scanner tables. > > Uh ... no. I haven't looked into why the flex tables are so large, > but this theory is just wrong. See ScanKeywordLookup(). Interestingly, the yy_transition array generated by flex used to be much smaller: 8.3: 22072 elements 8.4: 62623 elements master: 64535 elements The big jump between 8.3 and 8.4 was caused by introduction of the unicode escapes: U&'foo' [UESCAPE 'x'] . And in particular, the "error rule" for the UESCAPE, which we use to avoid backtracking. I experimented with a patch that uses two extra flex states to shorten the error rules, see attached. The idea is that after lexing a unicode literal like "U&'foo'", you enter a new state, in which you check whether an "UESCAPE 'x'" follows. This slashes the size of the array to 36581 elements. - Heikki
Attachment
On Thu, Feb 28, 2013 at 04:09:11PM -0500, Tom Lane wrote: > Robert Haas <robertmhaas@gmail.com> writes: > > A whole lot of those state transitions are attributable to states > > which have separate transitions for each of many keywords. > > Yeah, that's no surprise. > > The idea that's been in the back of my mind for awhile is to try to > solve the problem at the lexer level not the parser level: that is, > have the lexer return IDENT whenever a keyword appears in a context > where it should be interpreted as unreserved. You suggest somehow > driving that off mid-rule actions, but that seems to be to be probably > a nonstarter from a code complexity and maintainability standpoint. > > I believe however that it's possible to extract an idea of which > tokens the parser believes it can see next at any given parse state. > (I've seen code for this somewhere on the net, but am too lazy to go > searching for it again right now.) So we could imagine a rule along I believe tokenizing of typedefs requries the lexer to peek at the parser state: http://calculist.blogspot.com/2009/02/c-typedef-parsing-problem.html The well-known "typedef problem" with parsing C is that the standard Cgrammar is ambiguous unless the lexer distinguishesidentifiers bound bytypedef and other identifiers as two separate lexical classes. Thismeans that the parserneeds to feed scope information to the lexerduring parsing. One upshot is that lexing must be done concurrently withparsing. -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://enterprisedb.com + It's impossible for everything to be true. +
On 2 March 2013 18:47, Heikki Linnakangas <hlinnakangas@vmware.com> wrote: >> Uh ... no. I haven't looked into why the flex tables are so large, >> but this theory is just wrong. See ScanKeywordLookup(). > > > Interestingly, the yy_transition array generated by flex used to be much > smaller: > > 8.3: 22072 elements > 8.4: 62623 elements > master: 64535 elements > > The big jump between 8.3 and 8.4 was caused by introduction of the unicode > escapes: U&'foo' [UESCAPE 'x'] . And in particular, the "error rule" for the > UESCAPE, which we use to avoid backtracking. > > I experimented with a patch that uses two extra flex states to shorten the > error rules, see attached. The idea is that after lexing a unicode literal > like "U&'foo'", you enter a new state, in which you check whether an > "UESCAPE 'x'" follows. This slashes the size of the array to 36581 elements. +1 to do this sooner rather than later -- Simon Riggs http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training & Services
On 13.03.2013 10:50, Simon Riggs wrote: > On 2 March 2013 18:47, Heikki Linnakangas<hlinnakangas@vmware.com> wrote: >> Interestingly, the yy_transition array generated by flex used to be much >> smaller: >> >> 8.3: 22072 elements >> 8.4: 62623 elements >> master: 64535 elements >> >> The big jump between 8.3 and 8.4 was caused by introduction of the unicode >> escapes: U&'foo' [UESCAPE 'x'] . And in particular, the "error rule" for the >> UESCAPE, which we use to avoid backtracking. >> >> I experimented with a patch that uses two extra flex states to shorten the >> error rules, see attached. The idea is that after lexing a unicode literal >> like "U&'foo'", you enter a new state, in which you check whether an >> "UESCAPE 'x'" follows. This slashes the size of the array to 36581 elements. > > +1 to do this sooner rather than later I hear no objection, so committed. (after fixing some small bugs in the patch, and adding some comments) - Heikki
Heikki Linnakangas <hlinnakangas@vmware.com> writes:
> I hear no objection, so committed. (after fixing some small bugs in the
> patch, and adding some comments)
Please keep psqlscan.l in sync with scan.l.
regards, tom lane