postgresql/src/include/utils/plancache.h

186 lines
8.5 KiB
C
Raw Normal View History

/*-------------------------------------------------------------------------
*
* plancache.h
* Plan cache definitions.
*
* See plancache.c for comments.
*
2017-01-03 19:48:53 +01:00
* Portions Copyright (c) 1996-2017, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
2010-09-20 22:08:53 +02:00
* src/include/utils/plancache.h
*
*-------------------------------------------------------------------------
*/
#ifndef PLANCACHE_H
#define PLANCACHE_H
#include "access/tupdesc.h"
#include "nodes/params.h"
#include "utils/queryenvironment.h"
Change representation of statement lists, and add statement location info. This patch makes several changes that improve the consistency of representation of lists of statements. It's always been the case that the output of parse analysis is a list of Query nodes, whatever the types of the individual statements in the list. This patch brings similar consistency to the outputs of raw parsing and planning steps: * The output of raw parsing is now always a list of RawStmt nodes; the statement-type-dependent nodes are one level down from that. * The output of pg_plan_queries() is now always a list of PlannedStmt nodes, even for utility statements. In the case of a utility statement, "planning" just consists of wrapping a CMD_UTILITY PlannedStmt around the utility node. This list representation is now used in Portal and CachedPlan plan lists, replacing the former convention of intermixing PlannedStmts with bare utility-statement nodes. Now, every list of statements has a consistent head-node type depending on how far along it is in processing. This allows changing many places that formerly used generic "Node *" pointers to use a more specific pointer type, thus reducing the number of IsA() tests and casts needed, as well as improving code clarity. Also, the post-parse-analysis representation of DECLARE CURSOR is changed so that it looks more like EXPLAIN, PREPARE, etc. That is, the contained SELECT remains a child of the DeclareCursorStmt rather than getting flipped around to be the other way. It's now true for both Query and PlannedStmt that utilityStmt is non-null if and only if commandType is CMD_UTILITY. That allows simplifying a lot of places that were testing both fields. (I think some of those were just defensive programming, but in many places, it was actually necessary to avoid confusing DECLARE CURSOR with SELECT.) Because PlannedStmt carries a canSetTag field, we're also able to get rid of some ad-hoc rules about how to reconstruct canSetTag for a bare utility statement; specifically, the assumption that a utility is canSetTag if and only if it's the only one in its list. While I see no near-term need for relaxing that restriction, it's nice to get rid of the ad-hocery. The API of ProcessUtility() is changed so that what it's passed is the wrapper PlannedStmt not just the bare utility statement. This will affect all users of ProcessUtility_hook, but the changes are pretty trivial; see the affected contrib modules for examples of the minimum change needed. (Most compilers should give pointer-type-mismatch warnings for uncorrected code.) There's also a change in the API of ExplainOneQuery_hook, to pass through cursorOptions instead of expecting hook functions to know what to pick. This is needed because of the DECLARE CURSOR changes, but really should have been done in 9.6; it's unlikely that any extant hook functions know about using CURSOR_OPT_PARALLEL_OK. Finally, teach gram.y to save statement boundary locations in RawStmt nodes, and pass those through to Query and PlannedStmt nodes. This allows more intelligent handling of cases where a source query string contains multiple statements. This patch doesn't actually do anything with the information, but a follow-on patch will. (Passing this information through cleanly is the true motivation for these changes; while I think this is all good cleanup, it's unlikely we'd have bothered without this end goal.) catversion bump because addition of location fields to struct Query affects stored rules. This patch is by me, but it owes a good deal to Fabien Coelho who did a lot of preliminary work on the problem, and also reviewed the patch. Discussion: https://postgr.es/m/alpine.DEB.2.20.1612200926310.29821@lancre
2017-01-14 22:02:35 +01:00
/* Forward declaration, to avoid including parsenodes.h here */
struct RawStmt;
#define CACHEDPLANSOURCE_MAGIC 195726186
#define CACHEDPLAN_MAGIC 953717834
/*
* CachedPlanSource (which might better have been called CachedQuery)
* represents a SQL query that we expect to use multiple times. It stores
* the query source text, the raw parse tree, and the analyzed-and-rewritten
* query tree, as well as adjunct data. Cache invalidation can happen as a
* result of DDL affecting objects used by the query. In that case we discard
* the analyzed-and-rewritten query tree, and rebuild it when next needed.
*
* An actual execution plan, represented by CachedPlan, is derived from the
* CachedPlanSource when we need to execute the query. The plan could be
* either generic (usable with any set of plan parameters) or custom (for a
* specific set of parameters). plancache.c contains the logic that decides
* which way to do it for any particular execution. If we are using a generic
* cached plan then it is meant to be re-used across multiple executions, so
* callers must always treat CachedPlans as read-only.
*
* Once successfully built and "saved", CachedPlanSources typically live
* for the life of the backend, although they can be dropped explicitly.
* CachedPlans are reference-counted and go away automatically when the last
* reference is dropped. A CachedPlan can outlive the CachedPlanSource it
* was created from.
*
* An "unsaved" CachedPlanSource can be used for generating plans, but it
* lives in transient storage and will not be updated in response to sinval
* events.
*
* CachedPlans made from saved CachedPlanSources are likewise in permanent
* storage, so to avoid memory leaks, the reference-counted references to them
* must be held in permanent data structures or ResourceOwners. CachedPlans
* made from unsaved CachedPlanSources are in children of the caller's
* memory context, so references to them should not be longer-lived than
* that context. (Reference counting is somewhat pro forma in that case,
* though it may be useful if the CachedPlan can be discarded early.)
*
* A CachedPlanSource has two associated memory contexts: one that holds the
* struct itself, the query source text and the raw parse tree, and another
* context that holds the rewritten query tree and associated data. This
* allows the query tree to be discarded easily when it is invalidated.
*
Invent a "one-shot" variant of CachedPlans for better performance. SPI_execute() and related functions create a CachedPlan, execute it once, and immediately discard it, so that the functionality offered by plancache.c is of no value in this code path. And performance measurements show that the extra data copying and invalidation checking done by plancache.c slows down simple queries by 10% or more compared to 9.1. However, enough of the SPI code is shared with functions that do need plan caching that it seems impractical to bypass plancache.c altogether. Instead, let's invent a variant version of cached plans that preserves 99% of the API but doesn't offer any of the actual functionality, nor the overhead. This puts SPI_execute() performance back on par, or maybe even slightly better, than it was before. This change should resolve recent complaints of performance degradation from Dong Ye, Pavel Stehule, and others. By avoiding data copying, this change also reduces the amount of memory needed to execute many-statement SPI_execute() strings, as for instance in a recent complaint from Tomas Vondra. An additional benefit of this change is that multi-statement SPI_execute() query strings are now processed fully serially, that is we complete execution of earlier statements before running parse analysis and planning on following ones. This eliminates a long-standing POLA violation, in that DDL that affects the behavior of a later statement will now behave as expected. Back-patch to 9.2, since this was a performance regression compared to 9.1. (In 9.2, place the added struct fields so as to avoid changing the offsets of existing fields.) Heikki Linnakangas and Tom Lane
2013-01-04 23:42:19 +01:00
* Some callers wish to use the CachedPlan API even with one-shot queries
* that have no reason to be saved at all. We therefore support a "oneshot"
* variant that does no data copying or invalidation checking. In this case
Invent a "one-shot" variant of CachedPlans for better performance. SPI_execute() and related functions create a CachedPlan, execute it once, and immediately discard it, so that the functionality offered by plancache.c is of no value in this code path. And performance measurements show that the extra data copying and invalidation checking done by plancache.c slows down simple queries by 10% or more compared to 9.1. However, enough of the SPI code is shared with functions that do need plan caching that it seems impractical to bypass plancache.c altogether. Instead, let's invent a variant version of cached plans that preserves 99% of the API but doesn't offer any of the actual functionality, nor the overhead. This puts SPI_execute() performance back on par, or maybe even slightly better, than it was before. This change should resolve recent complaints of performance degradation from Dong Ye, Pavel Stehule, and others. By avoiding data copying, this change also reduces the amount of memory needed to execute many-statement SPI_execute() strings, as for instance in a recent complaint from Tomas Vondra. An additional benefit of this change is that multi-statement SPI_execute() query strings are now processed fully serially, that is we complete execution of earlier statements before running parse analysis and planning on following ones. This eliminates a long-standing POLA violation, in that DDL that affects the behavior of a later statement will now behave as expected. Back-patch to 9.2, since this was a performance regression compared to 9.1. (In 9.2, place the added struct fields so as to avoid changing the offsets of existing fields.) Heikki Linnakangas and Tom Lane
2013-01-04 23:42:19 +01:00
* there are no separate memory contexts: the CachedPlanSource struct and
* all subsidiary data live in the caller's CurrentMemoryContext, and there
* is no way to free memory short of clearing that entire context. A oneshot
Invent a "one-shot" variant of CachedPlans for better performance. SPI_execute() and related functions create a CachedPlan, execute it once, and immediately discard it, so that the functionality offered by plancache.c is of no value in this code path. And performance measurements show that the extra data copying and invalidation checking done by plancache.c slows down simple queries by 10% or more compared to 9.1. However, enough of the SPI code is shared with functions that do need plan caching that it seems impractical to bypass plancache.c altogether. Instead, let's invent a variant version of cached plans that preserves 99% of the API but doesn't offer any of the actual functionality, nor the overhead. This puts SPI_execute() performance back on par, or maybe even slightly better, than it was before. This change should resolve recent complaints of performance degradation from Dong Ye, Pavel Stehule, and others. By avoiding data copying, this change also reduces the amount of memory needed to execute many-statement SPI_execute() strings, as for instance in a recent complaint from Tomas Vondra. An additional benefit of this change is that multi-statement SPI_execute() query strings are now processed fully serially, that is we complete execution of earlier statements before running parse analysis and planning on following ones. This eliminates a long-standing POLA violation, in that DDL that affects the behavior of a later statement will now behave as expected. Back-patch to 9.2, since this was a performance regression compared to 9.1. (In 9.2, place the added struct fields so as to avoid changing the offsets of existing fields.) Heikki Linnakangas and Tom Lane
2013-01-04 23:42:19 +01:00
* plan is always treated as unsaved.
*
* Note: the string referenced by commandTag is not subsidiary storage;
* it is assumed to be a compile-time-constant string. As with portals,
* commandTag shall be NULL if and only if the original query string (before
* rewriting) was an empty string.
*/
typedef struct CachedPlanSource
{
int magic; /* should equal CACHEDPLANSOURCE_MAGIC */
Change representation of statement lists, and add statement location info. This patch makes several changes that improve the consistency of representation of lists of statements. It's always been the case that the output of parse analysis is a list of Query nodes, whatever the types of the individual statements in the list. This patch brings similar consistency to the outputs of raw parsing and planning steps: * The output of raw parsing is now always a list of RawStmt nodes; the statement-type-dependent nodes are one level down from that. * The output of pg_plan_queries() is now always a list of PlannedStmt nodes, even for utility statements. In the case of a utility statement, "planning" just consists of wrapping a CMD_UTILITY PlannedStmt around the utility node. This list representation is now used in Portal and CachedPlan plan lists, replacing the former convention of intermixing PlannedStmts with bare utility-statement nodes. Now, every list of statements has a consistent head-node type depending on how far along it is in processing. This allows changing many places that formerly used generic "Node *" pointers to use a more specific pointer type, thus reducing the number of IsA() tests and casts needed, as well as improving code clarity. Also, the post-parse-analysis representation of DECLARE CURSOR is changed so that it looks more like EXPLAIN, PREPARE, etc. That is, the contained SELECT remains a child of the DeclareCursorStmt rather than getting flipped around to be the other way. It's now true for both Query and PlannedStmt that utilityStmt is non-null if and only if commandType is CMD_UTILITY. That allows simplifying a lot of places that were testing both fields. (I think some of those were just defensive programming, but in many places, it was actually necessary to avoid confusing DECLARE CURSOR with SELECT.) Because PlannedStmt carries a canSetTag field, we're also able to get rid of some ad-hoc rules about how to reconstruct canSetTag for a bare utility statement; specifically, the assumption that a utility is canSetTag if and only if it's the only one in its list. While I see no near-term need for relaxing that restriction, it's nice to get rid of the ad-hocery. The API of ProcessUtility() is changed so that what it's passed is the wrapper PlannedStmt not just the bare utility statement. This will affect all users of ProcessUtility_hook, but the changes are pretty trivial; see the affected contrib modules for examples of the minimum change needed. (Most compilers should give pointer-type-mismatch warnings for uncorrected code.) There's also a change in the API of ExplainOneQuery_hook, to pass through cursorOptions instead of expecting hook functions to know what to pick. This is needed because of the DECLARE CURSOR changes, but really should have been done in 9.6; it's unlikely that any extant hook functions know about using CURSOR_OPT_PARALLEL_OK. Finally, teach gram.y to save statement boundary locations in RawStmt nodes, and pass those through to Query and PlannedStmt nodes. This allows more intelligent handling of cases where a source query string contains multiple statements. This patch doesn't actually do anything with the information, but a follow-on patch will. (Passing this information through cleanly is the true motivation for these changes; while I think this is all good cleanup, it's unlikely we'd have bothered without this end goal.) catversion bump because addition of location fields to struct Query affects stored rules. This patch is by me, but it owes a good deal to Fabien Coelho who did a lot of preliminary work on the problem, and also reviewed the patch. Discussion: https://postgr.es/m/alpine.DEB.2.20.1612200926310.29821@lancre
2017-01-14 22:02:35 +01:00
struct RawStmt *raw_parse_tree; /* output of raw_parser(), or NULL */
Invent a "one-shot" variant of CachedPlans for better performance. SPI_execute() and related functions create a CachedPlan, execute it once, and immediately discard it, so that the functionality offered by plancache.c is of no value in this code path. And performance measurements show that the extra data copying and invalidation checking done by plancache.c slows down simple queries by 10% or more compared to 9.1. However, enough of the SPI code is shared with functions that do need plan caching that it seems impractical to bypass plancache.c altogether. Instead, let's invent a variant version of cached plans that preserves 99% of the API but doesn't offer any of the actual functionality, nor the overhead. This puts SPI_execute() performance back on par, or maybe even slightly better, than it was before. This change should resolve recent complaints of performance degradation from Dong Ye, Pavel Stehule, and others. By avoiding data copying, this change also reduces the amount of memory needed to execute many-statement SPI_execute() strings, as for instance in a recent complaint from Tomas Vondra. An additional benefit of this change is that multi-statement SPI_execute() query strings are now processed fully serially, that is we complete execution of earlier statements before running parse analysis and planning on following ones. This eliminates a long-standing POLA violation, in that DDL that affects the behavior of a later statement will now behave as expected. Back-patch to 9.2, since this was a performance regression compared to 9.1. (In 9.2, place the added struct fields so as to avoid changing the offsets of existing fields.) Heikki Linnakangas and Tom Lane
2013-01-04 23:42:19 +01:00
const char *query_string; /* source text of query */
const char *commandTag; /* command tag (a constant!), or NULL */
Oid *param_types; /* array of parameter type OIDs, or NULL */
int num_params; /* length of param_types array */
ParserSetupHook parserSetup; /* alternative parameter spec method */
2010-02-26 03:01:40 +01:00
void *parserSetupArg;
2007-11-15 22:14:46 +01:00
int cursor_options; /* cursor options used for planning */
bool fixed_result; /* disallow change in result tupdesc? */
TupleDesc resultDesc; /* result type; NULL = doesn't return tuples */
MemoryContext context; /* memory context holding all above */
/* These fields describe the current analyzed-and-rewritten query tree: */
List *query_list; /* list of Query nodes, or NIL if not valid */
List *relationOids; /* OIDs of relations the queries depend on */
List *invalItems; /* other dependencies, as PlanInvalItems */
struct OverrideSearchPath *search_path; /* search_path used for
* parsing and planning */
MemoryContext query_context; /* context holding the above, or NULL */
Avoid invalidating all foreign-join cached plans when user mappings change. We must not push down a foreign join when the foreign tables involved should be accessed under different user mappings. Previously we tried to enforce that rule literally during planning, but that meant that the resulting plans were dependent on the current contents of the pg_user_mapping catalog, and we had to blow away all cached plans containing any remote join when anything at all changed in pg_user_mapping. This could have been improved somewhat, but the fact that a syscache inval callback has very limited info about what changed made it hard to do better within that design. Instead, let's change the planner to not consider user mappings per se, but to allow a foreign join if both RTEs have the same checkAsUser value. If they do, then they necessarily will use the same user mapping at runtime, and we don't need to know specifically which one that is. Post-plan-time changes in pg_user_mapping no longer require any plan invalidation. This rule does give up some optimization ability, to wit where two foreign table references come from views with different owners or one's from a view and one's directly in the query, but nonetheless the same user mapping would have applied. We'll sacrifice the first case, but to not regress more than we have to in the second case, allow a foreign join involving both zero and nonzero checkAsUser values if the nonzero one is the same as the prevailing effective userID. In that case, mark the plan as only runnable by that userID. The plancache code already had a notion of plans being userID-specific, in order to support RLS. It was a little confused though, in particular lacking clarity of thought as to whether it was the rewritten query or just the finished plan that's dependent on the userID. Rearrange that code so that it's clearer what depends on which, and so that the same logic applies to both RLS-injected role dependency and foreign-join-injected role dependency. Note that this patch doesn't remove the other issue mentioned in the original complaint, which is that while we'll reliably stop using a foreign join if it's disallowed in a new context, we might fail to start using a foreign join if it's now allowed, but we previously created a generic cached plan that didn't use one. It was agreed that the chance of winning that way was not high enough to justify the much larger number of plan invalidations that would have to occur if we tried to cause it to happen. In passing, clean up randomly-varying spelling of EXPLAIN commands in postgres_fdw.sql, and fix a COSTS ON example that had been allowed to leak into the committed tests. This reverts most of commits fbe5a3fb7 and 5d4171d1c, which were the previous attempt at ensuring we wouldn't push down foreign joins that span permissions contexts. Etsuro Fujita and Tom Lane Discussion: <d49c1e5b-f059-20f4-c132-e9752ee0113e@lab.ntt.co.jp>
2016-07-15 23:22:56 +02:00
Oid rewriteRoleId; /* Role ID we did rewriting for */
bool rewriteRowSecurity; /* row_security used during rewrite */
bool dependsOnRLS; /* is rewritten query specific to the above? */
/* If we have a generic plan, this is a reference-counted link to it: */
struct CachedPlan *gplan; /* generic plan, or NULL if not valid */
/* Some state flags: */
Invent a "one-shot" variant of CachedPlans for better performance. SPI_execute() and related functions create a CachedPlan, execute it once, and immediately discard it, so that the functionality offered by plancache.c is of no value in this code path. And performance measurements show that the extra data copying and invalidation checking done by plancache.c slows down simple queries by 10% or more compared to 9.1. However, enough of the SPI code is shared with functions that do need plan caching that it seems impractical to bypass plancache.c altogether. Instead, let's invent a variant version of cached plans that preserves 99% of the API but doesn't offer any of the actual functionality, nor the overhead. This puts SPI_execute() performance back on par, or maybe even slightly better, than it was before. This change should resolve recent complaints of performance degradation from Dong Ye, Pavel Stehule, and others. By avoiding data copying, this change also reduces the amount of memory needed to execute many-statement SPI_execute() strings, as for instance in a recent complaint from Tomas Vondra. An additional benefit of this change is that multi-statement SPI_execute() query strings are now processed fully serially, that is we complete execution of earlier statements before running parse analysis and planning on following ones. This eliminates a long-standing POLA violation, in that DDL that affects the behavior of a later statement will now behave as expected. Back-patch to 9.2, since this was a performance regression compared to 9.1. (In 9.2, place the added struct fields so as to avoid changing the offsets of existing fields.) Heikki Linnakangas and Tom Lane
2013-01-04 23:42:19 +01:00
bool is_oneshot; /* is it a "oneshot" plan? */
bool is_complete; /* has CompleteCachedPlan been done? */
bool is_saved; /* has CachedPlanSource been "saved"? */
bool is_valid; /* is the query_list currently valid? */
int generation; /* increments each time we create a plan */
/* If CachedPlanSource has been saved, it is a member of a global list */
struct CachedPlanSource *next_saved; /* list link, if so */
/* State kept to help decide whether to use custom or generic plans: */
double generic_cost; /* cost of generic plan, or -1 if not known */
double total_custom_cost; /* total cost of custom plans so far */
int num_custom_plans; /* number of plans included in total */
} CachedPlanSource;
/*
* CachedPlan represents an execution plan derived from a CachedPlanSource.
* The reference count includes both the link from the parent CachedPlanSource
* (if any), and any active plan executions, so the plan can be discarded
* exactly when refcount goes to zero. Both the struct itself and the
* subsidiary data live in the context denoted by the context field.
Invent a "one-shot" variant of CachedPlans for better performance. SPI_execute() and related functions create a CachedPlan, execute it once, and immediately discard it, so that the functionality offered by plancache.c is of no value in this code path. And performance measurements show that the extra data copying and invalidation checking done by plancache.c slows down simple queries by 10% or more compared to 9.1. However, enough of the SPI code is shared with functions that do need plan caching that it seems impractical to bypass plancache.c altogether. Instead, let's invent a variant version of cached plans that preserves 99% of the API but doesn't offer any of the actual functionality, nor the overhead. This puts SPI_execute() performance back on par, or maybe even slightly better, than it was before. This change should resolve recent complaints of performance degradation from Dong Ye, Pavel Stehule, and others. By avoiding data copying, this change also reduces the amount of memory needed to execute many-statement SPI_execute() strings, as for instance in a recent complaint from Tomas Vondra. An additional benefit of this change is that multi-statement SPI_execute() query strings are now processed fully serially, that is we complete execution of earlier statements before running parse analysis and planning on following ones. This eliminates a long-standing POLA violation, in that DDL that affects the behavior of a later statement will now behave as expected. Back-patch to 9.2, since this was a performance regression compared to 9.1. (In 9.2, place the added struct fields so as to avoid changing the offsets of existing fields.) Heikki Linnakangas and Tom Lane
2013-01-04 23:42:19 +01:00
* This makes it easy to free a no-longer-needed cached plan. (However,
* if is_oneshot is true, the context does not belong solely to the CachedPlan
* so no freeing is possible.)
*/
typedef struct CachedPlan
{
int magic; /* should equal CACHEDPLAN_MAGIC */
Change representation of statement lists, and add statement location info. This patch makes several changes that improve the consistency of representation of lists of statements. It's always been the case that the output of parse analysis is a list of Query nodes, whatever the types of the individual statements in the list. This patch brings similar consistency to the outputs of raw parsing and planning steps: * The output of raw parsing is now always a list of RawStmt nodes; the statement-type-dependent nodes are one level down from that. * The output of pg_plan_queries() is now always a list of PlannedStmt nodes, even for utility statements. In the case of a utility statement, "planning" just consists of wrapping a CMD_UTILITY PlannedStmt around the utility node. This list representation is now used in Portal and CachedPlan plan lists, replacing the former convention of intermixing PlannedStmts with bare utility-statement nodes. Now, every list of statements has a consistent head-node type depending on how far along it is in processing. This allows changing many places that formerly used generic "Node *" pointers to use a more specific pointer type, thus reducing the number of IsA() tests and casts needed, as well as improving code clarity. Also, the post-parse-analysis representation of DECLARE CURSOR is changed so that it looks more like EXPLAIN, PREPARE, etc. That is, the contained SELECT remains a child of the DeclareCursorStmt rather than getting flipped around to be the other way. It's now true for both Query and PlannedStmt that utilityStmt is non-null if and only if commandType is CMD_UTILITY. That allows simplifying a lot of places that were testing both fields. (I think some of those were just defensive programming, but in many places, it was actually necessary to avoid confusing DECLARE CURSOR with SELECT.) Because PlannedStmt carries a canSetTag field, we're also able to get rid of some ad-hoc rules about how to reconstruct canSetTag for a bare utility statement; specifically, the assumption that a utility is canSetTag if and only if it's the only one in its list. While I see no near-term need for relaxing that restriction, it's nice to get rid of the ad-hocery. The API of ProcessUtility() is changed so that what it's passed is the wrapper PlannedStmt not just the bare utility statement. This will affect all users of ProcessUtility_hook, but the changes are pretty trivial; see the affected contrib modules for examples of the minimum change needed. (Most compilers should give pointer-type-mismatch warnings for uncorrected code.) There's also a change in the API of ExplainOneQuery_hook, to pass through cursorOptions instead of expecting hook functions to know what to pick. This is needed because of the DECLARE CURSOR changes, but really should have been done in 9.6; it's unlikely that any extant hook functions know about using CURSOR_OPT_PARALLEL_OK. Finally, teach gram.y to save statement boundary locations in RawStmt nodes, and pass those through to Query and PlannedStmt nodes. This allows more intelligent handling of cases where a source query string contains multiple statements. This patch doesn't actually do anything with the information, but a follow-on patch will. (Passing this information through cleanly is the true motivation for these changes; while I think this is all good cleanup, it's unlikely we'd have bothered without this end goal.) catversion bump because addition of location fields to struct Query affects stored rules. This patch is by me, but it owes a good deal to Fabien Coelho who did a lot of preliminary work on the problem, and also reviewed the patch. Discussion: https://postgr.es/m/alpine.DEB.2.20.1612200926310.29821@lancre
2017-01-14 22:02:35 +01:00
List *stmt_list; /* list of PlannedStmts */
Invent a "one-shot" variant of CachedPlans for better performance. SPI_execute() and related functions create a CachedPlan, execute it once, and immediately discard it, so that the functionality offered by plancache.c is of no value in this code path. And performance measurements show that the extra data copying and invalidation checking done by plancache.c slows down simple queries by 10% or more compared to 9.1. However, enough of the SPI code is shared with functions that do need plan caching that it seems impractical to bypass plancache.c altogether. Instead, let's invent a variant version of cached plans that preserves 99% of the API but doesn't offer any of the actual functionality, nor the overhead. This puts SPI_execute() performance back on par, or maybe even slightly better, than it was before. This change should resolve recent complaints of performance degradation from Dong Ye, Pavel Stehule, and others. By avoiding data copying, this change also reduces the amount of memory needed to execute many-statement SPI_execute() strings, as for instance in a recent complaint from Tomas Vondra. An additional benefit of this change is that multi-statement SPI_execute() query strings are now processed fully serially, that is we complete execution of earlier statements before running parse analysis and planning on following ones. This eliminates a long-standing POLA violation, in that DDL that affects the behavior of a later statement will now behave as expected. Back-patch to 9.2, since this was a performance regression compared to 9.1. (In 9.2, place the added struct fields so as to avoid changing the offsets of existing fields.) Heikki Linnakangas and Tom Lane
2013-01-04 23:42:19 +01:00
bool is_oneshot; /* is it a "oneshot" plan? */
bool is_saved; /* is CachedPlan in a long-lived context? */
bool is_valid; /* is the stmt_list currently valid? */
Avoid invalidating all foreign-join cached plans when user mappings change. We must not push down a foreign join when the foreign tables involved should be accessed under different user mappings. Previously we tried to enforce that rule literally during planning, but that meant that the resulting plans were dependent on the current contents of the pg_user_mapping catalog, and we had to blow away all cached plans containing any remote join when anything at all changed in pg_user_mapping. This could have been improved somewhat, but the fact that a syscache inval callback has very limited info about what changed made it hard to do better within that design. Instead, let's change the planner to not consider user mappings per se, but to allow a foreign join if both RTEs have the same checkAsUser value. If they do, then they necessarily will use the same user mapping at runtime, and we don't need to know specifically which one that is. Post-plan-time changes in pg_user_mapping no longer require any plan invalidation. This rule does give up some optimization ability, to wit where two foreign table references come from views with different owners or one's from a view and one's directly in the query, but nonetheless the same user mapping would have applied. We'll sacrifice the first case, but to not regress more than we have to in the second case, allow a foreign join involving both zero and nonzero checkAsUser values if the nonzero one is the same as the prevailing effective userID. In that case, mark the plan as only runnable by that userID. The plancache code already had a notion of plans being userID-specific, in order to support RLS. It was a little confused though, in particular lacking clarity of thought as to whether it was the rewritten query or just the finished plan that's dependent on the userID. Rearrange that code so that it's clearer what depends on which, and so that the same logic applies to both RLS-injected role dependency and foreign-join-injected role dependency. Note that this patch doesn't remove the other issue mentioned in the original complaint, which is that while we'll reliably stop using a foreign join if it's disallowed in a new context, we might fail to start using a foreign join if it's now allowed, but we previously created a generic cached plan that didn't use one. It was agreed that the chance of winning that way was not high enough to justify the much larger number of plan invalidations that would have to occur if we tried to cause it to happen. In passing, clean up randomly-varying spelling of EXPLAIN commands in postgres_fdw.sql, and fix a COSTS ON example that had been allowed to leak into the committed tests. This reverts most of commits fbe5a3fb7 and 5d4171d1c, which were the previous attempt at ensuring we wouldn't push down foreign joins that span permissions contexts. Etsuro Fujita and Tom Lane Discussion: <d49c1e5b-f059-20f4-c132-e9752ee0113e@lab.ntt.co.jp>
2016-07-15 23:22:56 +02:00
Oid planRoleId; /* Role ID the plan was created for */
bool dependsOnRole; /* is plan specific to that role? */
TransactionId saved_xmin; /* if valid, replan when TransactionXmin
* changes from this value */
int generation; /* parent's generation number for this plan */
int refcount; /* count of live references to this struct */
MemoryContext context; /* context containing this CachedPlan */
} CachedPlan;
extern void InitPlanCache(void);
extern void ResetPlanCache(void);
Change representation of statement lists, and add statement location info. This patch makes several changes that improve the consistency of representation of lists of statements. It's always been the case that the output of parse analysis is a list of Query nodes, whatever the types of the individual statements in the list. This patch brings similar consistency to the outputs of raw parsing and planning steps: * The output of raw parsing is now always a list of RawStmt nodes; the statement-type-dependent nodes are one level down from that. * The output of pg_plan_queries() is now always a list of PlannedStmt nodes, even for utility statements. In the case of a utility statement, "planning" just consists of wrapping a CMD_UTILITY PlannedStmt around the utility node. This list representation is now used in Portal and CachedPlan plan lists, replacing the former convention of intermixing PlannedStmts with bare utility-statement nodes. Now, every list of statements has a consistent head-node type depending on how far along it is in processing. This allows changing many places that formerly used generic "Node *" pointers to use a more specific pointer type, thus reducing the number of IsA() tests and casts needed, as well as improving code clarity. Also, the post-parse-analysis representation of DECLARE CURSOR is changed so that it looks more like EXPLAIN, PREPARE, etc. That is, the contained SELECT remains a child of the DeclareCursorStmt rather than getting flipped around to be the other way. It's now true for both Query and PlannedStmt that utilityStmt is non-null if and only if commandType is CMD_UTILITY. That allows simplifying a lot of places that were testing both fields. (I think some of those were just defensive programming, but in many places, it was actually necessary to avoid confusing DECLARE CURSOR with SELECT.) Because PlannedStmt carries a canSetTag field, we're also able to get rid of some ad-hoc rules about how to reconstruct canSetTag for a bare utility statement; specifically, the assumption that a utility is canSetTag if and only if it's the only one in its list. While I see no near-term need for relaxing that restriction, it's nice to get rid of the ad-hocery. The API of ProcessUtility() is changed so that what it's passed is the wrapper PlannedStmt not just the bare utility statement. This will affect all users of ProcessUtility_hook, but the changes are pretty trivial; see the affected contrib modules for examples of the minimum change needed. (Most compilers should give pointer-type-mismatch warnings for uncorrected code.) There's also a change in the API of ExplainOneQuery_hook, to pass through cursorOptions instead of expecting hook functions to know what to pick. This is needed because of the DECLARE CURSOR changes, but really should have been done in 9.6; it's unlikely that any extant hook functions know about using CURSOR_OPT_PARALLEL_OK. Finally, teach gram.y to save statement boundary locations in RawStmt nodes, and pass those through to Query and PlannedStmt nodes. This allows more intelligent handling of cases where a source query string contains multiple statements. This patch doesn't actually do anything with the information, but a follow-on patch will. (Passing this information through cleanly is the true motivation for these changes; while I think this is all good cleanup, it's unlikely we'd have bothered without this end goal.) catversion bump because addition of location fields to struct Query affects stored rules. This patch is by me, but it owes a good deal to Fabien Coelho who did a lot of preliminary work on the problem, and also reviewed the patch. Discussion: https://postgr.es/m/alpine.DEB.2.20.1612200926310.29821@lancre
2017-01-14 22:02:35 +01:00
extern CachedPlanSource *CreateCachedPlan(struct RawStmt *raw_parse_tree,
Invent a "one-shot" variant of CachedPlans for better performance. SPI_execute() and related functions create a CachedPlan, execute it once, and immediately discard it, so that the functionality offered by plancache.c is of no value in this code path. And performance measurements show that the extra data copying and invalidation checking done by plancache.c slows down simple queries by 10% or more compared to 9.1. However, enough of the SPI code is shared with functions that do need plan caching that it seems impractical to bypass plancache.c altogether. Instead, let's invent a variant version of cached plans that preserves 99% of the API but doesn't offer any of the actual functionality, nor the overhead. This puts SPI_execute() performance back on par, or maybe even slightly better, than it was before. This change should resolve recent complaints of performance degradation from Dong Ye, Pavel Stehule, and others. By avoiding data copying, this change also reduces the amount of memory needed to execute many-statement SPI_execute() strings, as for instance in a recent complaint from Tomas Vondra. An additional benefit of this change is that multi-statement SPI_execute() query strings are now processed fully serially, that is we complete execution of earlier statements before running parse analysis and planning on following ones. This eliminates a long-standing POLA violation, in that DDL that affects the behavior of a later statement will now behave as expected. Back-patch to 9.2, since this was a performance regression compared to 9.1. (In 9.2, place the added struct fields so as to avoid changing the offsets of existing fields.) Heikki Linnakangas and Tom Lane
2013-01-04 23:42:19 +01:00
const char *query_string,
const char *commandTag);
Change representation of statement lists, and add statement location info. This patch makes several changes that improve the consistency of representation of lists of statements. It's always been the case that the output of parse analysis is a list of Query nodes, whatever the types of the individual statements in the list. This patch brings similar consistency to the outputs of raw parsing and planning steps: * The output of raw parsing is now always a list of RawStmt nodes; the statement-type-dependent nodes are one level down from that. * The output of pg_plan_queries() is now always a list of PlannedStmt nodes, even for utility statements. In the case of a utility statement, "planning" just consists of wrapping a CMD_UTILITY PlannedStmt around the utility node. This list representation is now used in Portal and CachedPlan plan lists, replacing the former convention of intermixing PlannedStmts with bare utility-statement nodes. Now, every list of statements has a consistent head-node type depending on how far along it is in processing. This allows changing many places that formerly used generic "Node *" pointers to use a more specific pointer type, thus reducing the number of IsA() tests and casts needed, as well as improving code clarity. Also, the post-parse-analysis representation of DECLARE CURSOR is changed so that it looks more like EXPLAIN, PREPARE, etc. That is, the contained SELECT remains a child of the DeclareCursorStmt rather than getting flipped around to be the other way. It's now true for both Query and PlannedStmt that utilityStmt is non-null if and only if commandType is CMD_UTILITY. That allows simplifying a lot of places that were testing both fields. (I think some of those were just defensive programming, but in many places, it was actually necessary to avoid confusing DECLARE CURSOR with SELECT.) Because PlannedStmt carries a canSetTag field, we're also able to get rid of some ad-hoc rules about how to reconstruct canSetTag for a bare utility statement; specifically, the assumption that a utility is canSetTag if and only if it's the only one in its list. While I see no near-term need for relaxing that restriction, it's nice to get rid of the ad-hocery. The API of ProcessUtility() is changed so that what it's passed is the wrapper PlannedStmt not just the bare utility statement. This will affect all users of ProcessUtility_hook, but the changes are pretty trivial; see the affected contrib modules for examples of the minimum change needed. (Most compilers should give pointer-type-mismatch warnings for uncorrected code.) There's also a change in the API of ExplainOneQuery_hook, to pass through cursorOptions instead of expecting hook functions to know what to pick. This is needed because of the DECLARE CURSOR changes, but really should have been done in 9.6; it's unlikely that any extant hook functions know about using CURSOR_OPT_PARALLEL_OK. Finally, teach gram.y to save statement boundary locations in RawStmt nodes, and pass those through to Query and PlannedStmt nodes. This allows more intelligent handling of cases where a source query string contains multiple statements. This patch doesn't actually do anything with the information, but a follow-on patch will. (Passing this information through cleanly is the true motivation for these changes; while I think this is all good cleanup, it's unlikely we'd have bothered without this end goal.) catversion bump because addition of location fields to struct Query affects stored rules. This patch is by me, but it owes a good deal to Fabien Coelho who did a lot of preliminary work on the problem, and also reviewed the patch. Discussion: https://postgr.es/m/alpine.DEB.2.20.1612200926310.29821@lancre
2017-01-14 22:02:35 +01:00
extern CachedPlanSource *CreateOneShotCachedPlan(struct RawStmt *raw_parse_tree,
const char *query_string,
const char *commandTag);
extern void CompleteCachedPlan(CachedPlanSource *plansource,
List *querytree_list,
MemoryContext querytree_context,
Oid *param_types,
int num_params,
ParserSetupHook parserSetup,
void *parserSetupArg,
int cursor_options,
bool fixed_result);
extern void SaveCachedPlan(CachedPlanSource *plansource);
extern void DropCachedPlan(CachedPlanSource *plansource);
extern void CachedPlanSetParentContext(CachedPlanSource *plansource,
MemoryContext newcontext);
extern CachedPlanSource *CopyCachedPlan(CachedPlanSource *plansource);
extern bool CachedPlanIsValid(CachedPlanSource *plansource);
extern List *CachedPlanGetTargetList(CachedPlanSource *plansource,
QueryEnvironment *queryEnv);
extern CachedPlan *GetCachedPlan(CachedPlanSource *plansource,
ParamListInfo boundParams,
bool useResOwner,
QueryEnvironment *queryEnv);
extern void ReleaseCachedPlan(CachedPlan *plan, bool useResOwner);
#endif /* PLANCACHE_H */