📄 crash5.test
字号:
# 2007 Aug 13## The author disclaims copyright to this source code. In place of# a legal notice, here is a blessing:## May you do good and not evil.# May you find forgiveness for yourself and forgive others.# May you share freely, never taking more than you give.##***********************************************************************# # This file tests aspects of recovery from a malloc() failure# in a CREATE INDEX statement.## $Id: crash5.test,v 1.3 2008/07/12 14:52:20 drh Exp $set testdir [file dirname $argv0]source $testdir/tester.tcl# Only run these tests if memory debugging is turned on.#ifcapable !memdebug||!crashtest||!memorymanage { puts "Skipping crash5 tests: not compiled with -DSQLITE_MEMDEBUG..." finish_test return}db closefor {set ii 0} {$ii < 10} {incr ii} { for {set jj 50} {$jj < 100} {incr jj} { # Set up the database so that it is an auto-vacuum database # containing a single table (root page 3) with a single row. # The row has an overflow page (page 4). file delete -force test.db test.db-journal sqlite3 db test.db set c [string repeat 3 1500] db eval { pragma auto_vacuum = 1; CREATE TABLE t1(a, b, c); INSERT INTO t1 VALUES('1111111111', '2222222222', $c); } db close do_test crash5-$ii.$jj.1 { crashsql -delay 1 -file test.db-journal -seed $ii -tclbody [join [list \ [list set iFail $jj] { sqlite3_crashparams 0 [file join [pwd] test.db-journal] # Begin a transaction and evaluate a "CREATE INDEX" statement # with the iFail'th malloc() set to fail. This operation will # have to move the current contents of page 4 (the overflow # page) to make room for the new root page. The bug is that # if malloc() fails at a particular point in sqlite3PagerMovepage(), # sqlite mistakenly thinks that the page being moved (page 4) has # been safely synced into the journal. If the page is written # to later in the transaction, it may be written out to the database # before the relevant part of the journal has been synced. # db eval BEGIN sqlite3_memdebug_fail $iFail -repeat 0 catch {db eval { CREATE UNIQUE INDEX i1 ON t1(a); }} msg # puts "$n $msg ac=[sqlite3_get_autocommit db]" # If the transaction is still active (it may not be if the malloc() # failure occured in the OS layer), write to the database. Make sure # page 4 is among those written. # if {![sqlite3_get_autocommit db]} { db eval { DELETE FROM t1; -- This will put page 4 on the free list. INSERT INTO t1 VALUES('111111111', '2222222222', '33333333'); INSERT INTO t1 SELECT * FROM t1; -- 2 INSERT INTO t1 SELECT * FROM t1; -- 4 INSERT INTO t1 SELECT * FROM t1; -- 8 INSERT INTO t1 SELECT * FROM t1; -- 16 INSERT INTO t1 SELECT * FROM t1; -- 32 INSERT INTO t1 SELECT * FROM t1 WHERE rowid%2; -- 48 } } # If the right malloc() failed during the 'CREATE INDEX' above and # the transaction was not rolled back, then the sqlite cache now # has a dirty page 4 that it incorrectly believes is already safely # in the synced part of the journal file. When # sqlite3_release_memory() is called sqlite tries to free memory # by writing page 4 out to the db file. If it crashes later on, # before syncing the journal... Corruption! # sqlite3_crashparams 1 [file join [pwd] test.db-journal] sqlite3_release_memory 8092 }]] {} expr 1 } {1} sqlite3 db test.db do_test crash5-$ii.$jj.2 { db eval {pragma integrity_check} } {ok} do_test crash5-$ii.$jj.3 { db eval {SELECT * FROM t1} } [list 1111111111 2222222222 $::c] db close }}finish_test
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -