caleb-brown/redict
1
0
Fork 0
mirror of https://codeberg.org/redict/redict.git synced 2025-01-22 16:18:28 -05:00
Code Issues Actions Packages Projects Releases Wiki Activity
51887e61b8
redict/tests/integration/corrupt-dump.tcl

815 lines
58 KiB
Tcl
Raw Normal View History

Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
# tests of corrupt ziplist payload with valid CRC
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
# * setting crash-memcheck-enabled to no to avoid issues with valgrind
# * setting use-exit-on-panic to yes so that valgrind can search for leaks
Fixed some typos, add a spell check ci and others minor fix (#8890) This PR adds a spell checker CI action that will fail future PRs if they introduce typos and spelling mistakes. This spell checker is based on blacklist of common spelling mistakes, so it will not catch everything, but at least it is also unlikely to cause false positives. Besides that, the PR also fixes many spelling mistakes and types, not all are a result of the spell checker we use. Here's a summary of other changes: 1. Scanned the entire source code and fixes all sorts of typos and spelling mistakes (including missing or extra spaces). 2. Outdated function / variable / argument names in comments 3. Fix outdated keyspace masks error log when we check `config.notify-keyspace-events` in loadServerConfigFromString. 4. Trim the white space at the end of line in `module.c`. Check: https://github.com/redis/redis/pull/7751 5. Some outdated https link URLs. 6. Fix some outdated comment. Such as: - In README: about the rdb, we used to said create a `thread`, change to `process` - dbRandomKey function coment (about the dictGetRandomKey, change to dictGetFairRandomKey) - notifyKeyspaceEvent fucntion comment (add type arg) - Some others minor fix in comment (Most of them are incorrectly quoted by variable names) 7. Modified the error log so that users can easily distinguish between TCP and TLS in `changeBindAddr`
2021-06-10 08:39:33 -04:00
# * setting debug set-skip-checksum-validation to 1 on some tests for which we
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
# didn't bother to fake a valid checksum
# * some tests set sanitize-dump-payload to no and some to yet, depending on
# what we want to test
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
Improve test suite to handle external servers better. (#9033) This commit revives the improves the ability to run the test suite against external servers, instead of launching and managing `redis-server` processes as part of the test fixture. This capability existed in the past, using the `--host` and `--port` options. However, it was quite limited and mostly useful when running a specific tests. Attempting to run larger chunks of the test suite experienced many issues: * Many tests depend on being able to start and control `redis-server` themselves, and there's no clear distinction between external server compatible and other tests. * Cluster mode is not supported (resulting with `CROSSSLOT` errors). This PR cleans up many things and makes it possible to run the entire test suite against an external server. It also provides more fine grained controls to handle cases where the external server supports a subset of the Redis commands, limited number of databases, cluster mode, etc. The tests directory now contains a `README.md` file that describes how this works. This commit also includes additional cleanups and fixes: * Tests can now be tagged. * Tag-based selection is now unified across `start_server`, `tags` and `test`. * More information is provided about skipped or ignored tests. * Repeated patterns in tests have been extracted to common procedures, both at a global level and on a per-test file basis. * Cleaned up some cases where test setup was based on a previous test executing (a major anti-pattern that repeats itself in many places). * Cleaned up some cases where test teardown was not part of a test (in the future we should have dedicated teardown code that executes even when tests fail). * Fixed some tests that were flaky running on external servers.
2021-06-09 08:13:24 -04:00
tags {"dump" "corruption" "external:skip"} {
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
Skip OOM-related tests on incompatible platforms. (#9386) We only run OOM related tests on x86_64 and aarch64, as jemalloc on other platforms (notably s390x) may actually succeed very large allocations. As a result the test may hang for a very long time at the cleanup phase, iterating as many as 2^61 hash table slots.
2021-08-18 09:00:22 -04:00
# We only run OOM related tests on x86_64 and aarch64, as jemalloc on other
# platforms (notably s390x) may actually succeed very large allocations. As
# a result the test may hang for a very long time at the cleanup phase,
# iterating as many as 2^61 hash table slots.
set arch_name [exec uname -m]
set run_oom_tests [expr {$arch_name == "x86_64" || $arch_name == "aarch64"}]
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
set corrupt_payload_7445 "\x0E\x01\x1D\x1D\x00\x00\x00\x16\x00\x00\x00\x03\x00\x00\x04\x43\x43\x43\x43\x06\x04\x42\x42\x42\x42\x06\x3F\x41\x41\x41\x41\xFF\x09\x00\x88\xA5\xCA\xA8\xC5\x41\xF4\x35"
test {corrupt payload: #7445 - with sanitize} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
catch {
r restore key 0 $corrupt_payload_7445
} err
assert_match "*Bad data format*" $err
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
}
}
test {corrupt payload: hash with valid zip list header, invalid entry len} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
Replace all usage of ziplist with listpack for t_hash (#8887) Part one of implementing #8702 (taking hashes first before other types) ## Description of the feature 1. Change ziplist encoded hash objects to listpack encoding. 2. Convert existing ziplists on RDB loading time. an O(n) operation. ## Rdb format changes 1. Add RDB_TYPE_HASH_LISTPACK rdb type. 2. Bump RDB_VERSION to 10 ## Interface changes 1. New `hash-max-listpack-entries` config is an alias for `hash-max-ziplist-entries` (same with `hash-max-listpack-value`) 2. OBJECT ENCODING will return `listpack` instead of `ziplist` ## Listpack improvements: 1. Support direct insert, replace integer element (rather than convert back and forth from string) 3. Add more listpack capabilities to match the ziplist ones (like `lpFind`, `lpRandomPairs` and such) 4. Optimize element length fetching, avoid multiple calculations 5. Use inline to avoid function call overhead. ## Tests 1. Add a new test to the RDB load time conversion 2. Adding the listpack unit tests. (based on the one in ziplist.c) 3. Add a few "corrupt payload: fuzzer findings" tests, and slightly modify existing ones. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-10 02:18:49 -04:00
catch {
r restore key 0 "\x0D\x1B\x1B\x00\x00\x00\x16\x00\x00\x00\x04\x00\x00\x02\x61\x00\x04\x02\x62\x00\x04\x14\x63\x00\x04\x02\x64\x00\xFF\x09\x00\xD9\x10\x54\x92\x15\xF5\x5F\x52"
} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
}
}
test {corrupt payload: invalid zlbytes header} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
catch {
r restore key 0 "\x0D\x1B\x25\x00\x00\x00\x16\x00\x00\x00\x04\x00\x00\x02\x61\x00\x04\x02\x62\x00\x04\x02\x63\x00\x04\x02\x64\x00\xFF\x09\x00\xB7\xF7\x6E\x9F\x43\x43\x14\xC6"
} err
assert_match "*Bad data format*" $err
Replace all usage of ziplist with listpack for t_hash (#8887) Part one of implementing #8702 (taking hashes first before other types) ## Description of the feature 1. Change ziplist encoded hash objects to listpack encoding. 2. Convert existing ziplists on RDB loading time. an O(n) operation. ## Rdb format changes 1. Add RDB_TYPE_HASH_LISTPACK rdb type. 2. Bump RDB_VERSION to 10 ## Interface changes 1. New `hash-max-listpack-entries` config is an alias for `hash-max-ziplist-entries` (same with `hash-max-listpack-value`) 2. OBJECT ENCODING will return `listpack` instead of `ziplist` ## Listpack improvements: 1. Support direct insert, replace integer element (rather than convert back and forth from string) 3. Add more listpack capabilities to match the ziplist ones (like `lpFind`, `lpRandomPairs` and such) 4. Optimize element length fetching, avoid multiple calculations 5. Use inline to avoid function call overhead. ## Tests 1. Add a new test to the RDB load time conversion 2. Adding the listpack unit tests. (based on the one in ziplist.c) 3. Add a few "corrupt payload: fuzzer findings" tests, and slightly modify existing ones. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-10 02:18:49 -04:00
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
}
}
test {corrupt payload: valid zipped hash header, dup records} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
Replace all usage of ziplist with listpack for t_hash (#8887) Part one of implementing #8702 (taking hashes first before other types) ## Description of the feature 1. Change ziplist encoded hash objects to listpack encoding. 2. Convert existing ziplists on RDB loading time. an O(n) operation. ## Rdb format changes 1. Add RDB_TYPE_HASH_LISTPACK rdb type. 2. Bump RDB_VERSION to 10 ## Interface changes 1. New `hash-max-listpack-entries` config is an alias for `hash-max-ziplist-entries` (same with `hash-max-listpack-value`) 2. OBJECT ENCODING will return `listpack` instead of `ziplist` ## Listpack improvements: 1. Support direct insert, replace integer element (rather than convert back and forth from string) 3. Add more listpack capabilities to match the ziplist ones (like `lpFind`, `lpRandomPairs` and such) 4. Optimize element length fetching, avoid multiple calculations 5. Use inline to avoid function call overhead. ## Tests 1. Add a new test to the RDB load time conversion 2. Adding the listpack unit tests. (based on the one in ziplist.c) 3. Add a few "corrupt payload: fuzzer findings" tests, and slightly modify existing ones. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-10 02:18:49 -04:00
catch {
r restore key 0 "\x0D\x1B\x1B\x00\x00\x00\x16\x00\x00\x00\x04\x00\x00\x02\x61\x00\x04\x02\x62\x00\x04\x02\x61\x00\x04\x02\x64\x00\xFF\x09\x00\xA1\x98\x36\x78\xCC\x8E\x93\x2E"
} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
}
}
test {corrupt payload: quicklist big ziplist prev len} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
r config set sanitize-dump-payload no
Replace ziplist with listpack in quicklist (#9740) Part three of implementing #8702, following #8887 and #9366 . ## Description of the feature 1. Replace the ziplist container of quicklist with listpack. 2. Convert existing quicklist ziplists on RDB loading time. an O(n) operation. ## Interface changes 1. New `list-max-listpack-size` config is an alias for `list-max-ziplist-size`. 2. Replace `debug ziplist` command with `debug listpack`. ## Internal changes 1. Add `lpMerge` to merge two listpacks . (same as `ziplistMerge`) 2. Add `lpRepr` to print info of listpack which is used in debugCommand and `quicklistRepr`. (same as `ziplistRepr`) 3. Replace `QUICKLIST_NODE_CONTAINER_ZIPLIST` with `QUICKLIST_NODE_CONTAINER_PACKED`(following #9357 ). It represent that a quicklistNode is a packed node, as opposed to a plain node. 4. Remove `createZiplistObject` method, which is never used. 5. Calculate listpack entry size using overhead overestimation in `quicklistAllowInsert`. We prefer an overestimation, which would at worse lead to a few bytes below the lowest limit of 4k. ## Improvements 1. Calling `lpShrinkToFit` after converting Ziplist to listpack, which was missed at #9366. 2. Optimize `quicklistAppendPlainNode` to avoid memcpy data. ## Bugfix 1. Fix crash in `quicklistRepr` when ziplist is compressed, introduced from #9366. ## Test 1. Add unittest for `lpMerge`. 2. Modify the old quicklist ziplist corrupt dump test. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-11-24 06:34:13 -05:00
catch {r restore key 0 "\x0E\x01\x13\x13\x00\x00\x00\x0E\x00\x00\x00\x02\x00\x00\x02\x61\x00\x0E\x02\x62\x00\xFF\x09\x00\x49\x97\x30\xB2\x0D\xA1\xED\xAA"} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
}
}
test {corrupt payload: quicklist small ziplist prev len} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
catch {
r restore key 0 "\x0E\x01\x13\x13\x00\x00\x00\x0E\x00\x00\x00\x02\x00\x00\x02\x61\x00\x02\x02\x62\x00\xFF\x09\x00\xC7\x71\x03\x97\x07\x75\xB0\x63"
} err
assert_match "*Bad data format*" $err
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
}
}
test {corrupt payload: quicklist ziplist wrong count} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
r config set sanitize-dump-payload no
Replace ziplist with listpack in quicklist (#9740) Part three of implementing #8702, following #8887 and #9366 . ## Description of the feature 1. Replace the ziplist container of quicklist with listpack. 2. Convert existing quicklist ziplists on RDB loading time. an O(n) operation. ## Interface changes 1. New `list-max-listpack-size` config is an alias for `list-max-ziplist-size`. 2. Replace `debug ziplist` command with `debug listpack`. ## Internal changes 1. Add `lpMerge` to merge two listpacks . (same as `ziplistMerge`) 2. Add `lpRepr` to print info of listpack which is used in debugCommand and `quicklistRepr`. (same as `ziplistRepr`) 3. Replace `QUICKLIST_NODE_CONTAINER_ZIPLIST` with `QUICKLIST_NODE_CONTAINER_PACKED`(following #9357 ). It represent that a quicklistNode is a packed node, as opposed to a plain node. 4. Remove `createZiplistObject` method, which is never used. 5. Calculate listpack entry size using overhead overestimation in `quicklistAllowInsert`. We prefer an overestimation, which would at worse lead to a few bytes below the lowest limit of 4k. ## Improvements 1. Calling `lpShrinkToFit` after converting Ziplist to listpack, which was missed at #9366. 2. Optimize `quicklistAppendPlainNode` to avoid memcpy data. ## Bugfix 1. Fix crash in `quicklistRepr` when ziplist is compressed, introduced from #9366. ## Test 1. Add unittest for `lpMerge`. 2. Modify the old quicklist ziplist corrupt dump test. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-11-24 06:34:13 -05:00
catch {r restore key 0 "\x0E\x01\x13\x13\x00\x00\x00\x0E\x00\x00\x00\x03\x00\x00\x02\x61\x00\x04\x02\x62\x00\xFF\x09\x00\x4D\xE2\x0A\x2F\x08\x25\xDF\x91"} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
}
}
test {corrupt payload: #3080 - quicklist} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
r config set sanitize-dump-payload no
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
catch {
r RESTORE key 0 "\x0E\x01\x80\x00\x00\x00\x10\x41\x41\x41\x41\x41\x41\x41\x41\x02\x00\x00\x80\x41\x41\x41\x41\x07\x00\x03\xC7\x1D\xEF\x54\x68\xCC\xF3"
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
r DUMP key ;# DUMP was used in the original issue, but now even with shallow sanitization restore safely fails, so this is dead code
} err
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
assert_match "*Bad data format*" $err
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
}
}
Sanitize dump payload: handle remaining empty key when RDB loading and restore command (#9349) This commit mainly fixes empty keys due to RDB loading and restore command, which was omitted in #9297. 1) When loading quicklsit, if all the ziplists in the quicklist are empty, NULL will be returned. If only some of the ziplists are empty, then we will skip the empty ziplists silently. 2) When loading hash zipmap, if zipmap is empty, sanitization check will fail. 3) When loading hash ziplist, if ziplist is empty, NULL will be returned. 4) Add RDB loading test with sanitize.
2021-08-09 10:13:46 -04:00
test {corrupt payload: quicklist with empty ziplist} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch {r restore key 0 "\x0E\x01\x0B\x0B\x00\x00\x00\x0A\x00\x00\x00\x00\x00\xFF\x09\x00\xC2\x69\x37\x83\x3C\x7F\xFE\x6F" replace} err
assert_match "*Bad data format*" $err
r ping
}
}
Add support for list type to store elements larger than 4GB (#9357) Redis lists are stored in quicklist, which is currently a linked list of ziplists. Ziplists are limited to storing elements no larger than 4GB, so when bigger items are added they're getting truncated. This PR changes quicklists so that they're capable of storing large items in quicklist nodes that are plain string buffers rather than ziplist. As part of the PR there were few other changes in redis: 1. new DEBUG sub-commands: - QUICKLIST-PACKED-THRESHOLD - set the threshold of for the node type to be plan or ziplist. default (1GB) - QUICKLIST <key> - Shows low level info about the quicklist encoding of <key> 2. rdb format change: - A new type was added - RDB_TYPE_LIST_QUICKLIST_2 . - container type (packed / plain) was added to the beginning of the rdb object (before the actual node list). 3. testing: - Tests that requires over 100MB will be by default skipped. a new flag was added to 'runtest' to run the large memory tests (not used by default) Co-authored-by: sundb <sundbcn@gmail.com> Co-authored-by: Oran Agra <oran@redislabs.com>
2021-11-03 14:47:18 -04:00
test {corrupt payload: quicklist encoded_len is 0} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
fix string escaping in corrupt-dump test to support TCL8.5 (#9824) TCL8.5 can't handle cases where part of the string is escaped and part of it isn't, if there's a single char that needs escaping, we need to escape the whole string.
2021-11-22 05:30:06 -05:00
catch { r restore _list 0 "\x12\x01\x01\x00\x0a\x00\x8f\xc6\xc0\x57\x1c\x0a\xb3\x3c" replace } err
Add support for list type to store elements larger than 4GB (#9357) Redis lists are stored in quicklist, which is currently a linked list of ziplists. Ziplists are limited to storing elements no larger than 4GB, so when bigger items are added they're getting truncated. This PR changes quicklists so that they're capable of storing large items in quicklist nodes that are plain string buffers rather than ziplist. As part of the PR there were few other changes in redis: 1. new DEBUG sub-commands: - QUICKLIST-PACKED-THRESHOLD - set the threshold of for the node type to be plan or ziplist. default (1GB) - QUICKLIST <key> - Shows low level info about the quicklist encoding of <key> 2. rdb format change: - A new type was added - RDB_TYPE_LIST_QUICKLIST_2 . - container type (packed / plain) was added to the beginning of the rdb object (before the actual node list). 3. testing: - Tests that requires over 100MB will be by default skipped. a new flag was added to 'runtest' to run the large memory tests (not used by default) Co-authored-by: sundb <sundbcn@gmail.com> Co-authored-by: Oran Agra <oran@redislabs.com>
2021-11-03 14:47:18 -04:00
assert_match "*Bad data format*" $err
r ping
}
}
Santize dump payload: fix invalid listpack entry start with EOF (#9889) When an invalid listpack entry starts with EOF, we will skip it when we verify it in the loop.
2021-12-04 09:43:08 -05:00
test {corrupt payload: quicklist listpack entry start with EOF} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
catch { r restore _list 0 "\x12\x01\x02\x0b\x0b\x00\x00\x00\x01\x00\x81\x61\x02\xff\xff\x0a\x00\x7e\xd8\xde\x5b\x0d\xd7\x70\xb8" replace } err
assert_match "*Bad data format*" $err
r ping
}
}
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
test {corrupt payload: #3080 - ziplist} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
# shallow sanitization is enough for restore to safely reject the payload with wrong size
r config set sanitize-dump-payload no
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
catch {
r RESTORE key 0 "\x0A\x80\x00\x00\x00\x10\x41\x41\x41\x41\x41\x41\x41\x41\x02\x00\x00\x80\x41\x41\x41\x41\x07\x00\x39\x5B\x49\xE0\xC1\xC6\xDD\x76"
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
} err
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
assert_match "*Bad data format*" $err
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
}
}
test {corrupt payload: load corrupted rdb with no CRC - #3505} {
set server_path [tmpdir "server.rdb-corruption-test"]
exec cp tests/assets/corrupt_ziplist.rdb $server_path
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
set srv [start_server [list overrides [list "dir" $server_path "dbfilename" "corrupt_ziplist.rdb" loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no sanitize-dump-payload no]]]
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
# wait for termination
wait_for_condition 100 50 {
! [is_alive $srv]
} else {
fail "rdb loading didn't fail"
}
set stdout [dict get $srv stdout]
assert_equal [count_message_lines $stdout "Terminating server after rdb file reading failure."] 1
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
assert_lessthan 1 [count_message_lines $stdout "integrity check failed"]
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
kill_server $srv ;# let valgrind look for issues
}
Sanitize dump payload: handle remaining empty key when RDB loading and restore command (#9349) This commit mainly fixes empty keys due to RDB loading and restore command, which was omitted in #9297. 1) When loading quicklsit, if all the ziplists in the quicklist are empty, NULL will be returned. If only some of the ziplists are empty, then we will skip the empty ziplists silently. 2) When loading hash zipmap, if zipmap is empty, sanitization check will fail. 3) When loading hash ziplist, if ziplist is empty, NULL will be returned. 4) Add RDB loading test with sanitize.
2021-08-09 10:13:46 -04:00
foreach sanitize_dump {no yes} {
test {corrupt payload: load corrupted rdb with empty keys} {
set server_path [tmpdir "server.rdb-corruption-empty-keys-test"]
exec cp tests/assets/corrupt_empty_keys.rdb $server_path
start_server [list overrides [list "dir" $server_path "dbfilename" "corrupt_empty_keys.rdb" "sanitize-dump-payload" $sanitize_dump]] {
r select 0
assert_equal [r dbsize] 0
Sanitize dump payload: fix empty keys when RDB loading and restore command (#9297) When we load rdb or restore command, if we encounter a length of 0, it will result in the creation of an empty key. This could either be a corrupt payload, or a result of a bug (see #8453 ) This PR mainly fixes the following: 1) When restore command will return `Bad data format` error. 2) When loading RDB, we will silently discard the key. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-05 15:42:20 -04:00
Sanitize dump payload: handle remaining empty key when RDB loading and restore command (#9349) This commit mainly fixes empty keys due to RDB loading and restore command, which was omitted in #9297. 1) When loading quicklsit, if all the ziplists in the quicklist are empty, NULL will be returned. If only some of the ziplists are empty, then we will skip the empty ziplists silently. 2) When loading hash zipmap, if zipmap is empty, sanitization check will fail. 3) When loading hash ziplist, if ziplist is empty, NULL will be returned. 4) Add RDB loading test with sanitize.
2021-08-09 10:13:46 -04:00
verify_log_message 0 "*skipping empty key: set*" 0
verify_log_message 0 "*skipping empty key: list_quicklist*" 0
verify_log_message 0 "*skipping empty key: list_quicklist_empty_ziplist*" 0
verify_log_message 0 "*skipping empty key: list_ziplist*" 0
verify_log_message 0 "*skipping empty key: hash*" 0
verify_log_message 0 "*skipping empty key: hash_ziplist*" 0
verify_log_message 0 "*skipping empty key: zset*" 0
verify_log_message 0 "*skipping empty key: zset_ziplist*" 0
Replace all usage of ziplist with listpack for t_zset (#9366) Part two of implementing #8702 (zset), after #8887. ## Description of the feature Replaced all uses of ziplist with listpack in t_zset, and optimized some of the code to optimize performance. ## Rdb format changes New `RDB_TYPE_ZSET_LISTPACK` rdb type. ## Rdb loading improvements: 1) Pre-expansion of dict for validation of duplicate data for listpack and ziplist. 2) Simplifying the release of empty key objects when RDB loading. 3) Unify ziplist and listpack data verify methods for zset and hash, and move code to rdb.c. ## Interface changes 1) New `zset-max-listpack-entries` config is an alias for `zset-max-ziplist-entries` (same with `zset-max-listpack-value`). 2) OBJECT ENCODING will return listpack instead of ziplist. ## Listpack improvements: 1) Add `lpDeleteRange` and `lpDeleteRangeWithEntry` functions to delete a range of entries from listpack. 2) Improve the performance of `lpCompare`, converting from string to integer is faster than converting from integer to string. 3) Replace `snprintf` with `ll2string` to improve performance in converting numbers to strings in `lpGet()`. ## Zset improvements: 1) Improve the performance of `zzlFind` method, use `lpFind` instead of `lpCompare` in a loop. 2) Use `lpDeleteRangeWithEntry` instead of `lpDelete` twice to delete a element of zset. ## Tests 1) Add some unittests for `lpDeleteRange` and `lpDeleteRangeWithEntry` function. 2) Add zset RDB loading test. 3) Add benchmark test for `lpCompare` and `ziplsitCompare`. 4) Add empty listpack zset corrupt dump test.
2021-09-09 11:18:53 -04:00
verify_log_message 0 "*skipping empty key: zset_listpack*" 0
verify_log_message 0 "*empty keys skipped: 9*" 0
Sanitize dump payload: handle remaining empty key when RDB loading and restore command (#9349) This commit mainly fixes empty keys due to RDB loading and restore command, which was omitted in #9297. 1) When loading quicklsit, if all the ziplists in the quicklist are empty, NULL will be returned. If only some of the ziplists are empty, then we will skip the empty ziplists silently. 2) When loading hash zipmap, if zipmap is empty, sanitization check will fail. 3) When loading hash ziplist, if ziplist is empty, NULL will be returned. 4) Add RDB loading test with sanitize.
2021-08-09 10:13:46 -04:00
}
Sanitize dump payload: fix empty keys when RDB loading and restore command (#9297) When we load rdb or restore command, if we encounter a length of 0, it will result in the creation of an empty key. This could either be a corrupt payload, or a result of a bug (see #8453 ) This PR mainly fixes the following: 1) When restore command will return `Bad data format` error. 2) When loading RDB, we will silently discard the key. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-05 15:42:20 -04:00
}
}
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
test {corrupt payload: listpack invalid size header} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
r config set sanitize-dump-payload no
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
catch {
r restore key 0 "\x0F\x01\x10\x00\x00\x00\x00\x00\x00\x00\x02\x00\x00\x00\x00\x00\x00\x00\x02\x40\x55\x5F\x00\x00\x00\x0F\x00\x01\x01\x00\x01\x02\x01\x88\x31\x00\x00\x00\x00\x00\x00\x00\x09\x88\x32\x00\x00\x00\x00\x00\x00\x00\x09\x00\x01\x00\x01\x00\x01\x00\x01\x02\x02\x88\x31\x00\x00\x00\x00\x00\x00\x00\x09\x88\x61\x00\x00\x00\x00\x00\x00\x00\x09\x88\x32\x00\x00\x00\x00\x00\x00\x00\x09\x88\x62\x00\x00\x00\x00\x00\x00\x00\x09\x08\x01\xFF\x0A\x01\x00\x00\x09\x00\x45\x91\x0A\x87\x2F\xA5\xF9\x2E"
} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*Stream listpack integrity check failed*" 0
}
}
test {corrupt payload: listpack too long entry len} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
r config set sanitize-dump-payload no
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
catch {
Add stream consumer group lag tracking and reporting (#9127) Adds the ability to track the lag of a consumer group (CG), that is, the number of entries yet-to-be-delivered from the stream. The proposed constant-time solution is in the spirit of "best-effort." Partially addresses #8737. ## Description of approach We add a new "entries_added" property to the stream. This starts at 0 for a new stream and is incremented by 1 with every `XADD`. It is essentially an all-time counter of the entries added to the stream. Given the stream's length and this counter value, we can trivially find the logical "entries_added" counter of the first ID if and only if the stream is contiguous. A fragmented stream contains one or more tombstones generated by `XDEL`s. The new "xdel_max_id" stream property tracks the latest tombstone. The CG also tracks its last delivered ID's as an "entries_read" counter and increments it independently when delivering new messages, unless the this read counter is invalid (-1 means invalid offset). When the CG's counter is available, the reported lag is the difference between added and read counters. Lastly, this also adds a "first_id" field to the stream structure in order to make looking it up cheaper in most cases. ## Limitations There are two cases in which the mechanism isn't able to track the lag. In these cases, `XINFO` replies with `null` in the "lag" field. The first case is when a CG is created with an arbitrary last delivered ID, that isn't "0-0", nor the first or the last entries of the stream. In this case, it is impossible to obtain a valid read counter (short of an O(N) operation). The second case is when there are one or more tombstones fragmenting the stream's entries range. In both cases, given enough time and assuming that the consumers are active (reading and lacking) and advancing, the CG should be able to catch up with the tip of the stream and report zero lag. Once that's achieved, lag tracking would resume as normal (until the next tombstone is set). ## API changes * `XGROUP CREATE` added with the optional named argument `[ENTRIESREAD entries-read]` for explicitly specifying the new CG's counter. * `XGROUP SETID` added with an optional positional argument `[ENTRIESREAD entries-read]` for specifying the CG's counter. * `XINFO` reports the maximal tombstone ID, the recorded first entry ID, and total number of entries added to the stream. * `XINFO` reports the current lag and logical read counter of CGs. * `XSETID` is an internal command that's used in replication/aof. It has been added with the optional positional arguments `[ENTRIESADDED entries-added] [MAXDELETEDID max-deleted-entry-id]` for propagating the CG's offset and maximal tombstone ID of the stream. ## The generic unsolved problem The current stream implementation doesn't provide an efficient way to obtain the approximate/exact size of a range of entries. While it could've been nice to have that ability (#5813) in general, let alone specifically in the context of CGs, the risk and complexities involved in such implementation are in all likelihood prohibitive. ## A refactoring note The `streamGetEdgeID` has been refactored to accommodate both the existing seek of any entry as well as seeking non-deleted entries (the addition of the `skip_tombstones` argument). Furthermore, this refactoring also migrated the seek logic to use the `streamIterator` (rather than `raxIterator`) that was, in turn, extended with the `skip_tombstones` Boolean struct field to control the emission of these. Co-authored-by: Guy Benoish <guy.benoish@redislabs.com> Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-23 15:34:58 -05:00
r restore key 0 "\x0F\x01\x10\x00\x00\x00\x00\x00\x00\x00\x02\x00\x00\x00\x00\x00\x00\x00\x02\x40\x55\x55\x00\x00\x00\x0F\x00\x01\x01\x00\x01\x02\x01\x88\x31\x00\x00\x00\x00\x00\x00\x00\x09\x88\x32\x00\x00\x00\x00\x00\x00\x00\x09\x00\x01\x00\x01\x00\x01\x00\x01\x02\x02\x89\x31\x00\x00\x00\x00\x00\x00\x00\x09\x88\x61\x00\x00\x00\x00\x00\x00\x00\x09\x88\x32\x00\x00\x00\x00\x00\x00\x00\x09\x88\x62\x00\x00\x00\x00\x00\x00\x00\x09\x08\x01\xFF\x0A\x01\x00\x00\x09\x00\x40\x63\xC9\x37\x03\xA2\xE5\x68"
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
} err
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
assert_equal [count_log_message 0 "crashed by signal"] 0
assert_equal [count_log_message 0 "ASSERTION FAILED"] 1
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
}
}
test {corrupt payload: listpack very long entry len} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
r config set sanitize-dump-payload no
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
catch {
Add stream consumer group lag tracking and reporting (#9127) Adds the ability to track the lag of a consumer group (CG), that is, the number of entries yet-to-be-delivered from the stream. The proposed constant-time solution is in the spirit of "best-effort." Partially addresses #8737. ## Description of approach We add a new "entries_added" property to the stream. This starts at 0 for a new stream and is incremented by 1 with every `XADD`. It is essentially an all-time counter of the entries added to the stream. Given the stream's length and this counter value, we can trivially find the logical "entries_added" counter of the first ID if and only if the stream is contiguous. A fragmented stream contains one or more tombstones generated by `XDEL`s. The new "xdel_max_id" stream property tracks the latest tombstone. The CG also tracks its last delivered ID's as an "entries_read" counter and increments it independently when delivering new messages, unless the this read counter is invalid (-1 means invalid offset). When the CG's counter is available, the reported lag is the difference between added and read counters. Lastly, this also adds a "first_id" field to the stream structure in order to make looking it up cheaper in most cases. ## Limitations There are two cases in which the mechanism isn't able to track the lag. In these cases, `XINFO` replies with `null` in the "lag" field. The first case is when a CG is created with an arbitrary last delivered ID, that isn't "0-0", nor the first or the last entries of the stream. In this case, it is impossible to obtain a valid read counter (short of an O(N) operation). The second case is when there are one or more tombstones fragmenting the stream's entries range. In both cases, given enough time and assuming that the consumers are active (reading and lacking) and advancing, the CG should be able to catch up with the tip of the stream and report zero lag. Once that's achieved, lag tracking would resume as normal (until the next tombstone is set). ## API changes * `XGROUP CREATE` added with the optional named argument `[ENTRIESREAD entries-read]` for explicitly specifying the new CG's counter. * `XGROUP SETID` added with an optional positional argument `[ENTRIESREAD entries-read]` for specifying the CG's counter. * `XINFO` reports the maximal tombstone ID, the recorded first entry ID, and total number of entries added to the stream. * `XINFO` reports the current lag and logical read counter of CGs. * `XSETID` is an internal command that's used in replication/aof. It has been added with the optional positional arguments `[ENTRIESADDED entries-added] [MAXDELETEDID max-deleted-entry-id]` for propagating the CG's offset and maximal tombstone ID of the stream. ## The generic unsolved problem The current stream implementation doesn't provide an efficient way to obtain the approximate/exact size of a range of entries. While it could've been nice to have that ability (#5813) in general, let alone specifically in the context of CGs, the risk and complexities involved in such implementation are in all likelihood prohibitive. ## A refactoring note The `streamGetEdgeID` has been refactored to accommodate both the existing seek of any entry as well as seeking non-deleted entries (the addition of the `skip_tombstones` argument). Furthermore, this refactoring also migrated the seek logic to use the `streamIterator` (rather than `raxIterator`) that was, in turn, extended with the `skip_tombstones` Boolean struct field to control the emission of these. Co-authored-by: Guy Benoish <guy.benoish@redislabs.com> Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-23 15:34:58 -05:00
# This will catch migrated payloads from v6.2.x
r restore key 0 "\x0F\x01\x10\x00\x00\x00\x00\x00\x00\x00\x02\x00\x00\x00\x00\x00\x00\x00\x02\x40\x55\x55\x00\x00\x00\x0F\x00\x01\x01\x00\x01\x02\x01\x88\x31\x00\x00\x00\x00\x00\x00\x00\x09\x88\x32\x00\x00\x00\x00\x00\x00\x00\x09\x00\x01\x00\x01\x00\x01\x00\x01\x02\x02\x88\x31\x00\x00\x00\x00\x00\x00\x00\x09\x88\x61\x00\x00\x00\x00\x00\x00\x00\x09\x88\x32\x00\x00\x00\x00\x00\x00\x00\x09\x9C\x62\x00\x00\x00\x00\x00\x00\x00\x09\x08\x01\xFF\x0A\x01\x00\x00\x09\x00\x63\x6F\x42\x8E\x7C\xB5\xA2\x9D"
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
} err
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
assert_equal [count_log_message 0 "crashed by signal"] 0
assert_equal [count_log_message 0 "ASSERTION FAILED"] 1
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
}
}
test {corrupt payload: listpack too long entry prev len} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
catch {
r restore key 0 "\x0F\x01\x10\x00\x00\x00\x00\x00\x00\x00\x02\x00\x00\x00\x00\x00\x00\x00\x02\x40\x55\x55\x00\x00\x00\x0F\x00\x01\x01\x00\x15\x02\x01\x88\x31\x00\x00\x00\x00\x00\x00\x00\x09\x88\x32\x00\x00\x00\x00\x00\x00\x00\x09\x00\x01\x00\x01\x00\x01\x00\x01\x02\x02\x88\x31\x00\x00\x00\x00\x00\x00\x00\x09\x88\x61\x00\x00\x00\x00\x00\x00\x00\x09\x88\x32\x00\x00\x00\x00\x00\x00\x00\x09\x88\x62\x00\x00\x00\x00\x00\x00\x00\x09\x08\x01\xFF\x0A\x01\x00\x00\x09\x00\x06\xFB\x44\x24\x0A\x8E\x75\xEA"
} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*Stream listpack integrity check failed*" 0
}
}
Santize dump payload: fix crash when stream with duplicate consumes (#9918) When rdb creates a consumer without determining whether it exists in advance, it may return NULL and crash if it encounters corrupt data with duplicate consumers.
2021-12-08 11:11:57 -05:00
test {corrupt payload: stream with duplicate consumers} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
catch {
r restore key 0 "\x0F\x00\x00\x00\x00\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x00\x00\x00\x02\x04\x6E\x61\x6D\x65\x2A\x4C\xAA\x9A\x7D\x01\x00\x00\x00\x04\x6E\x61\x6D\x65\x2B\x4C\xAA\x9A\x7D\x01\x00\x00\x00\x0A\x00\xCC\xED\x8C\xA7\x62\xEE\xC7\xC8"
} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*Duplicate stream consumer detected*" 0
r ping
}
}
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
test {corrupt payload: hash ziplist with duplicate records} {
# when we do perform full sanitization, we expect duplicate records to fail the restore
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch { r RESTORE _hash 0 "\x0D\x3D\x3D\x00\x00\x00\x3A\x00\x00\x00\x14\x13\x00\xF5\x02\xF5\x02\xF2\x02\x53\x5F\x31\x04\xF3\x02\xF3\x02\xF7\x02\xF7\x02\xF8\x02\x02\x5F\x37\x04\xF1\x02\xF1\x02\xF6\x02\x02\x5F\x35\x04\xF4\x02\x02\x5F\x33\x04\xFA\x02\x02\x5F\x39\x04\xF9\x02\xF9\xFF\x09\x00\xB5\x48\xDE\x62\x31\xD0\xE5\x63" } err
assert_match "*Bad data format*" $err
}
}
Replace all usage of ziplist with listpack for t_hash (#8887) Part one of implementing #8702 (taking hashes first before other types) ## Description of the feature 1. Change ziplist encoded hash objects to listpack encoding. 2. Convert existing ziplists on RDB loading time. an O(n) operation. ## Rdb format changes 1. Add RDB_TYPE_HASH_LISTPACK rdb type. 2. Bump RDB_VERSION to 10 ## Interface changes 1. New `hash-max-listpack-entries` config is an alias for `hash-max-ziplist-entries` (same with `hash-max-listpack-value`) 2. OBJECT ENCODING will return `listpack` instead of `ziplist` ## Listpack improvements: 1. Support direct insert, replace integer element (rather than convert back and forth from string) 3. Add more listpack capabilities to match the ziplist ones (like `lpFind`, `lpRandomPairs` and such) 4. Optimize element length fetching, avoid multiple calculations 5. Use inline to avoid function call overhead. ## Tests 1. Add a new test to the RDB load time conversion 2. Adding the listpack unit tests. (based on the one in ziplist.c) 3. Add a few "corrupt payload: fuzzer findings" tests, and slightly modify existing ones. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-10 02:18:49 -04:00
test {corrupt payload: hash listpack with duplicate records} {
# when we do perform full sanitization, we expect duplicate records to fail the restore
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
fix string escaping in corrupt-dump test to support TCL8.5 (#9824) TCL8.5 can't handle cases where part of the string is escaped and part of it isn't, if there's a single char that needs escaping, we need to escape the whole string.
2021-11-22 05:30:06 -05:00
catch { r RESTORE _hash 0 "\x10\x17\x17\x00\x00\x00\x04\x00\x82\x61\x00\x03\x82\x62\x00\x03\x82\x61\x00\x03\x82\x64\x00\x03\xff\x0a\x00\xc0\xcf\xa6\x87\xe5\xa7\xc5\xbe" } err
Replace all usage of ziplist with listpack for t_hash (#8887) Part one of implementing #8702 (taking hashes first before other types) ## Description of the feature 1. Change ziplist encoded hash objects to listpack encoding. 2. Convert existing ziplists on RDB loading time. an O(n) operation. ## Rdb format changes 1. Add RDB_TYPE_HASH_LISTPACK rdb type. 2. Bump RDB_VERSION to 10 ## Interface changes 1. New `hash-max-listpack-entries` config is an alias for `hash-max-ziplist-entries` (same with `hash-max-listpack-value`) 2. OBJECT ENCODING will return `listpack` instead of `ziplist` ## Listpack improvements: 1. Support direct insert, replace integer element (rather than convert back and forth from string) 3. Add more listpack capabilities to match the ziplist ones (like `lpFind`, `lpRandomPairs` and such) 4. Optimize element length fetching, avoid multiple calculations 5. Use inline to avoid function call overhead. ## Tests 1. Add a new test to the RDB load time conversion 2. Adding the listpack unit tests. (based on the one in ziplist.c) 3. Add a few "corrupt payload: fuzzer findings" tests, and slightly modify existing ones. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-10 02:18:49 -04:00
assert_match "*Bad data format*" $err
}
}
Fix false positive leak reported by GCC ASAN (#9816) Leak found by the corrupt-dump-fuzzer when using GCC ASAN, which seems to falsely report leaks on pointers kept only on the stack when calling exit. Instead we now use _exit on panic / assert to skip these leak checks. Additionally, check for sanitizer warnings in the corrupt-dump-fuzzer between iterations, so that when something is found we know which test to relate it too (and it prints reproduction command list)
2021-11-21 11:47:10 -05:00
test {corrupt payload: hash listpack with duplicate records - convert} {
# when we do NOT perform full sanitization, but we convert to hash, we expect duplicate records panic
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r config set hash-max-listpack-entries 1
r debug set-skip-checksum-validation 1
fix string escaping in corrupt-dump test to support TCL8.5 (#9824) TCL8.5 can't handle cases where part of the string is escaped and part of it isn't, if there's a single char that needs escaping, we need to escape the whole string.
2021-11-22 05:30:06 -05:00
catch { r RESTORE _hash 0 "\x10\x17\x17\x00\x00\x00\x04\x00\x82\x61\x00\x03\x82\x62\x00\x03\x82\x61\x00\x03\x82\x64\x00\x03\xff\x0a\x00\xc0\xcf\xa6\x87\xe5\xa7\xc5\xbe" } err
Fix false positive leak reported by GCC ASAN (#9816) Leak found by the corrupt-dump-fuzzer when using GCC ASAN, which seems to falsely report leaks on pointers kept only on the stack when calling exit. Instead we now use _exit on panic / assert to skip these leak checks. Additionally, check for sanitizer warnings in the corrupt-dump-fuzzer between iterations, so that when something is found we know which test to relate it too (and it prints reproduction command list)
2021-11-21 11:47:10 -05:00
assert_equal [count_log_message 0 "crashed by signal"] 0
assert_equal [count_log_message 0 "listpack with dup elements"] 1
}
}
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
test {corrupt payload: hash ziplist uneven record count} {
Fix false positive leak reported by GCC ASAN (#9816) Leak found by the corrupt-dump-fuzzer when using GCC ASAN, which seems to falsely report leaks on pointers kept only on the stack when calling exit. Instead we now use _exit on panic / assert to skip these leak checks. Additionally, check for sanitizer warnings in the corrupt-dump-fuzzer between iterations, so that when something is found we know which test to relate it too (and it prints reproduction command list)
2021-11-21 11:47:10 -05:00
# when we do NOT perform full sanitization, but shallow sanitization can detect uneven count
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
Fix false positive leak reported by GCC ASAN (#9816) Leak found by the corrupt-dump-fuzzer when using GCC ASAN, which seems to falsely report leaks on pointers kept only on the stack when calling exit. Instead we now use _exit on panic / assert to skip these leak checks. Additionally, check for sanitizer warnings in the corrupt-dump-fuzzer between iterations, so that when something is found we know which test to relate it too (and it prints reproduction command list)
2021-11-21 11:47:10 -05:00
r config set sanitize-dump-payload no
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
r debug set-skip-checksum-validation 1
catch { r RESTORE _hash 0 "\r\x1b\x1b\x00\x00\x00\x16\x00\x00\x00\x04\x00\x00\x02a\x00\x04\x02b\x00\x04\x02a\x00\x04\x02d\x00\xff\t\x00\xa1\x98\x36x\xcc\x8e\x93\x2e" } err
assert_match "*Bad data format*" $err
}
}
Fixed some typos, add a spell check ci and others minor fix (#8890) This PR adds a spell checker CI action that will fail future PRs if they introduce typos and spelling mistakes. This spell checker is based on blacklist of common spelling mistakes, so it will not catch everything, but at least it is also unlikely to cause false positives. Besides that, the PR also fixes many spelling mistakes and types, not all are a result of the spell checker we use. Here's a summary of other changes: 1. Scanned the entire source code and fixes all sorts of typos and spelling mistakes (including missing or extra spaces). 2. Outdated function / variable / argument names in comments 3. Fix outdated keyspace masks error log when we check `config.notify-keyspace-events` in loadServerConfigFromString. 4. Trim the white space at the end of line in `module.c`. Check: https://github.com/redis/redis/pull/7751 5. Some outdated https link URLs. 6. Fix some outdated comment. Such as: - In README: about the rdb, we used to said create a `thread`, change to `process` - dbRandomKey function coment (about the dictGetRandomKey, change to dictGetFairRandomKey) - notifyKeyspaceEvent fucntion comment (add type arg) - Some others minor fix in comment (Most of them are incorrectly quoted by variable names) 7. Modified the error log so that users can easily distinguish between TCP and TLS in `changeBindAddr`
2021-06-10 08:39:33 -04:00
test {corrupt payload: hash duplicate records} {
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
# when we do perform full sanitization, we expect duplicate records to fail the restore
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch { r RESTORE _hash 0 "\x04\x02\x01a\x01b\x01a\x01d\t\x00\xc6\x9c\xab\xbc\bk\x0c\x06" } err
assert_match "*Bad data format*" $err
}
}
Sanitize dump payload: handle remaining empty key when RDB loading and restore command (#9349) This commit mainly fixes empty keys due to RDB loading and restore command, which was omitted in #9297. 1) When loading quicklsit, if all the ziplists in the quicklist are empty, NULL will be returned. If only some of the ziplists are empty, then we will skip the empty ziplists silently. 2) When loading hash zipmap, if zipmap is empty, sanitization check will fail. 3) When loading hash ziplist, if ziplist is empty, NULL will be returned. 4) Add RDB loading test with sanitize.
2021-08-09 10:13:46 -04:00
test {corrupt payload: hash empty zipmap} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch { r RESTORE _hash 0 "\x09\x02\x00\xFF\x09\x00\xC0\xF1\xB8\x67\x4C\x16\xAC\xE3" } err
assert_match "*Bad data format*" $err
verify_log_message 0 "*Zipmap integrity check failed*" 0
}
}
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
test {corrupt payload: fuzzer findings - NPD in streamIteratorGetID} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch {
r RESTORE key 0 "\x0F\x01\x10\x00\x00\x01\x73\xBD\x68\x48\x71\x00\x00\x00\x00\x00\x00\x00\x00\x40\x42\x42\x00\x00\x00\x18\x00\x03\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x00\x01\x02\x01\x00\x01\x00\x01\x01\x01\x00\x01\x05\x01\x02\x01\x00\x01\x01\x01\x01\x01\x82\x5F\x31\x03\x05\x01\x02\x01\x00\x01\x02\x01\x01\x01\x02\x01\x48\x01\xFF\x03\x81\x00\x00\x01\x73\xBD\x68\x48\x71\x02\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x81\x00\x00\x01\x73\xBD\x68\x48\x71\x00\x01\x00\x00\x01\x73\xBD\x68\x48\x71\x00\x00\x00\x00\x00\x00\x00\x00\x72\x48\x68\xBD\x73\x01\x00\x00\x01\x01\x05\x41\x6C\x69\x63\x65\x72\x48\x68\xBD\x73\x01\x00\x00\x01\x00\x00\x01\x73\xBD\x68\x48\x71\x00\x00\x00\x00\x00\x00\x00\x00\x09\x00\x80\xCD\xB0\xD5\x1A\xCE\xFF\x10"
r XREVRANGE key 725 233
}
assert_equal [count_log_message 0 "crashed by signal"] 0
assert_equal [count_log_message 0 "ASSERTION FAILED"] 1
}
}
test {corrupt payload: fuzzer findings - listpack NPD on invalid stream} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch {
r RESTORE _stream 0 "\x0F\x01\x10\x00\x00\x01\x73\xDC\xB6\x6B\xF1\x00\x00\x00\x00\x00\x00\x00\x00\x40\x42\x42\x00\x00\x00\x18\x00\x03\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x00\x01\x02\x01\x00\x01\x00\x01\x01\x01\x00\x01\x05\x01\x02\x01\x1F\x01\x00\x01\x01\x01\x6D\x5F\x31\x03\x05\x01\x02\x01\x29\x01\x00\x01\x01\x01\x02\x01\x05\x01\xFF\x03\x81\x00\x00\x01\x73\xDC\xB6\x6C\x1A\x00\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x81\x00\x00\x01\x73\xDC\xB6\x6B\xF1\x00\x01\x00\x00\x01\x73\xDC\xB6\x6B\xF1\x00\x00\x00\x00\x00\x00\x00\x00\x4B\x6C\xB6\xDC\x73\x01\x00\x00\x01\x01\x05\x41\x6C\x69\x63\x65\x3D\x6C\xB6\xDC\x73\x01\x00\x00\x01\x00\x00\x01\x73\xDC\xB6\x6B\xF1\x00\x00\x00\x00\x00\x00\x00\x00\x09\x00\xC7\x7D\x1C\xD7\x04\xFF\xE6\x9D"
r XREAD STREAMS _stream 519389898758
}
assert_equal [count_log_message 0 "crashed by signal"] 0
assert_equal [count_log_message 0 "ASSERTION FAILED"] 1
}
}
test {corrupt payload: fuzzer findings - NPD in quicklistIndex} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch {
r RESTORE key 0 "\x0E\x01\x13\x13\x00\x00\x00\x10\x00\x00\x00\x03\x12\x00\xF3\x02\x02\x5F\x31\x04\xF1\xFF\x09\x00\xC9\x4B\x31\xFE\x61\xC0\x96\xFE"
Replace ziplist with listpack in quicklist (#9740) Part three of implementing #8702, following #8887 and #9366 . ## Description of the feature 1. Replace the ziplist container of quicklist with listpack. 2. Convert existing quicklist ziplists on RDB loading time. an O(n) operation. ## Interface changes 1. New `list-max-listpack-size` config is an alias for `list-max-ziplist-size`. 2. Replace `debug ziplist` command with `debug listpack`. ## Internal changes 1. Add `lpMerge` to merge two listpacks . (same as `ziplistMerge`) 2. Add `lpRepr` to print info of listpack which is used in debugCommand and `quicklistRepr`. (same as `ziplistRepr`) 3. Replace `QUICKLIST_NODE_CONTAINER_ZIPLIST` with `QUICKLIST_NODE_CONTAINER_PACKED`(following #9357 ). It represent that a quicklistNode is a packed node, as opposed to a plain node. 4. Remove `createZiplistObject` method, which is never used. 5. Calculate listpack entry size using overhead overestimation in `quicklistAllowInsert`. We prefer an overestimation, which would at worse lead to a few bytes below the lowest limit of 4k. ## Improvements 1. Calling `lpShrinkToFit` after converting Ziplist to listpack, which was missed at #9366. 2. Optimize `quicklistAppendPlainNode` to avoid memcpy data. ## Bugfix 1. Fix crash in `quicklistRepr` when ziplist is compressed, introduced from #9366. ## Test 1. Add unittest for `lpMerge`. 2. Modify the old quicklist ziplist corrupt dump test. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-11-24 06:34:13 -05:00
} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
}
}
Replace all usage of ziplist with listpack for t_hash (#8887) Part one of implementing #8702 (taking hashes first before other types) ## Description of the feature 1. Change ziplist encoded hash objects to listpack encoding. 2. Convert existing ziplists on RDB loading time. an O(n) operation. ## Rdb format changes 1. Add RDB_TYPE_HASH_LISTPACK rdb type. 2. Bump RDB_VERSION to 10 ## Interface changes 1. New `hash-max-listpack-entries` config is an alias for `hash-max-ziplist-entries` (same with `hash-max-listpack-value`) 2. OBJECT ENCODING will return `listpack` instead of `ziplist` ## Listpack improvements: 1. Support direct insert, replace integer element (rather than convert back and forth from string) 3. Add more listpack capabilities to match the ziplist ones (like `lpFind`, `lpRandomPairs` and such) 4. Optimize element length fetching, avoid multiple calculations 5. Use inline to avoid function call overhead. ## Tests 1. Add a new test to the RDB load time conversion 2. Adding the listpack unit tests. (based on the one in ziplist.c) 3. Add a few "corrupt payload: fuzzer findings" tests, and slightly modify existing ones. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-10 02:18:49 -04:00
test {corrupt payload: fuzzer findings - encoded entry header reach outside the allocation} {
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r debug set-skip-checksum-validation 1
catch {
r RESTORE key 0 "\x0D\x19\x19\x00\x00\x00\x16\x00\x00\x00\x06\x00\x00\xF1\x02\xF1\x02\xF2\x02\x02\x5F\x31\x04\x99\x02\xF3\xFF\x09\x00\xC5\xB8\x10\xC0\x8A\xF9\x16\xDF"
Replace all usage of ziplist with listpack for t_hash (#8887) Part one of implementing #8702 (taking hashes first before other types) ## Description of the feature 1. Change ziplist encoded hash objects to listpack encoding. 2. Convert existing ziplists on RDB loading time. an O(n) operation. ## Rdb format changes 1. Add RDB_TYPE_HASH_LISTPACK rdb type. 2. Bump RDB_VERSION to 10 ## Interface changes 1. New `hash-max-listpack-entries` config is an alias for `hash-max-ziplist-entries` (same with `hash-max-listpack-value`) 2. OBJECT ENCODING will return `listpack` instead of `ziplist` ## Listpack improvements: 1. Support direct insert, replace integer element (rather than convert back and forth from string) 3. Add more listpack capabilities to match the ziplist ones (like `lpFind`, `lpRandomPairs` and such) 4. Optimize element length fetching, avoid multiple calculations 5. Use inline to avoid function call overhead. ## Tests 1. Add a new test to the RDB load time conversion 2. Adding the listpack unit tests. (based on the one in ziplist.c) 3. Add a few "corrupt payload: fuzzer findings" tests, and slightly modify existing ones. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-10 02:18:49 -04:00
} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
}
}
test {corrupt payload: fuzzer findings - invalid ziplist encoding} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {
r RESTORE _listbig 0 "\x0E\x02\x1B\x1B\x00\x00\x00\x16\x00\x00\x00\x05\x00\x00\x02\x5F\x39\x04\xF9\x02\x86\x5F\x37\x04\xF7\x02\x02\x5F\x35\xFF\x19\x19\x00\x00\x00\x16\x00\x00\x00\x05\x00\x00\xF5\x02\x02\x5F\x33\x04\xF3\x02\x02\x5F\x31\x04\xF1\xFF\x09\x00\x0C\xFC\x99\x2C\x23\x45\x15\x60"
} err
assert_match "*Bad data format*" $err
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
}
}
test {corrupt payload: fuzzer findings - hash crash} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
r RESTORE _hash 0 "\x0D\x19\x19\x00\x00\x00\x16\x00\x00\x00\x06\x00\x00\xF1\x02\xF1\x02\xF2\x02\x02\x5F\x31\x04\xF3\x02\xF3\xFF\x09\x00\x38\xB8\x10\xC0\x8A\xF9\x16\xDF"
r HSET _hash 394891450 1635910264
r HMGET _hash 887312884855
}
}
test {corrupt payload: fuzzer findings - uneven entry count in hash} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r debug set-skip-checksum-validation 1
Replace all usage of ziplist with listpack for t_hash (#8887) Part one of implementing #8702 (taking hashes first before other types) ## Description of the feature 1. Change ziplist encoded hash objects to listpack encoding. 2. Convert existing ziplists on RDB loading time. an O(n) operation. ## Rdb format changes 1. Add RDB_TYPE_HASH_LISTPACK rdb type. 2. Bump RDB_VERSION to 10 ## Interface changes 1. New `hash-max-listpack-entries` config is an alias for `hash-max-ziplist-entries` (same with `hash-max-listpack-value`) 2. OBJECT ENCODING will return `listpack` instead of `ziplist` ## Listpack improvements: 1. Support direct insert, replace integer element (rather than convert back and forth from string) 3. Add more listpack capabilities to match the ziplist ones (like `lpFind`, `lpRandomPairs` and such) 4. Optimize element length fetching, avoid multiple calculations 5. Use inline to avoid function call overhead. ## Tests 1. Add a new test to the RDB load time conversion 2. Adding the listpack unit tests. (based on the one in ziplist.c) 3. Add a few "corrupt payload: fuzzer findings" tests, and slightly modify existing ones. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-10 02:18:49 -04:00
catch {
r RESTORE _hashbig 0 "\x0D\x3D\x3D\x00\x00\x00\x38\x00\x00\x00\x14\x00\x00\xF2\x02\x02\x5F\x31\x04\x1C\x02\xF7\x02\xF1\x02\xF1\x02\xF5\x02\xF5\x02\xF4\x02\x02\x5F\x33\x04\xF6\x02\x02\x5F\x35\x04\xF8\x02\x02\x5F\x37\x04\xF9\x02\xF9\x02\xF3\x02\xF3\x02\xFA\x02\x02\x5F\x39\xFF\x09\x00\x73\xB7\x68\xC8\x97\x24\x8E\x88"
} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
}
}
test {corrupt payload: fuzzer findings - invalid read in lzf_decompress} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch { r RESTORE _setbig 0 "\x02\x03\x02\x5F\x31\xC0\x02\xC3\x00\x09\x00\xE6\xDC\x76\x44\xFF\xEB\x3D\xFE" } err
assert_match "*Bad data format*" $err
}
}
test {corrupt payload: fuzzer findings - leak in rdbloading due to dup entry in set} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch { r RESTORE _setbig 0 "\x02\x0A\x02\x5F\x39\xC0\x06\x02\x5F\x31\xC0\x00\xC0\x04\x02\x5F\x35\xC0\x02\xC0\x08\x02\x5F\x31\x02\x5F\x33\x09\x00\x7A\x5A\xFB\x90\x3A\xE9\x3C\xBE" } err
assert_match "*Bad data format*" $err
}
}
Sanitize dump payload: fix empty keys when RDB loading and restore command (#9297) When we load rdb or restore command, if we encounter a length of 0, it will result in the creation of an empty key. This could either be a corrupt payload, or a result of a bug (see #8453 ) This PR mainly fixes the following: 1) When restore command will return `Bad data format` error. 2) When loading RDB, we will silently discard the key. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-05 15:42:20 -04:00
test {corrupt payload: fuzzer findings - empty intset} {
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
Sanitize dump payload: fix empty keys when RDB loading and restore command (#9297) When we load rdb or restore command, if we encounter a length of 0, it will result in the creation of an empty key. This could either be a corrupt payload, or a result of a bug (see #8453 ) This PR mainly fixes the following: 1) When restore command will return `Bad data format` error. 2) When loading RDB, we will silently discard the key. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-05 15:42:20 -04:00
catch {r RESTORE _setbig 0 "\x02\xC0\xC0\x06\x02\x5F\x39\xC0\x02\x02\x5F\x33\xC0\x00\x02\x5F\x31\xC0\x04\xC0\x08\x02\x5F\x37\x02\x5F\x35\x09\x00\xC5\xD4\x6D\xBA\xAD\x14\xB7\xE7"} err
assert_match "*Bad data format*" $err
r ping
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
}
}
Replace all usage of ziplist with listpack for t_zset (#9366) Part two of implementing #8702 (zset), after #8887. ## Description of the feature Replaced all uses of ziplist with listpack in t_zset, and optimized some of the code to optimize performance. ## Rdb format changes New `RDB_TYPE_ZSET_LISTPACK` rdb type. ## Rdb loading improvements: 1) Pre-expansion of dict for validation of duplicate data for listpack and ziplist. 2) Simplifying the release of empty key objects when RDB loading. 3) Unify ziplist and listpack data verify methods for zset and hash, and move code to rdb.c. ## Interface changes 1) New `zset-max-listpack-entries` config is an alias for `zset-max-ziplist-entries` (same with `zset-max-listpack-value`). 2) OBJECT ENCODING will return listpack instead of ziplist. ## Listpack improvements: 1) Add `lpDeleteRange` and `lpDeleteRangeWithEntry` functions to delete a range of entries from listpack. 2) Improve the performance of `lpCompare`, converting from string to integer is faster than converting from integer to string. 3) Replace `snprintf` with `ll2string` to improve performance in converting numbers to strings in `lpGet()`. ## Zset improvements: 1) Improve the performance of `zzlFind` method, use `lpFind` instead of `lpCompare` in a loop. 2) Use `lpDeleteRangeWithEntry` instead of `lpDelete` twice to delete a element of zset. ## Tests 1) Add some unittests for `lpDeleteRange` and `lpDeleteRangeWithEntry` function. 2) Add zset RDB loading test. 3) Add benchmark test for `lpCompare` and `ziplsitCompare`. 4) Add empty listpack zset corrupt dump test.
2021-09-09 11:18:53 -04:00
test {corrupt payload: fuzzer findings - zset ziplist entry lensize is 0} {
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
Replace all usage of ziplist with listpack for t_zset (#9366) Part two of implementing #8702 (zset), after #8887. ## Description of the feature Replaced all uses of ziplist with listpack in t_zset, and optimized some of the code to optimize performance. ## Rdb format changes New `RDB_TYPE_ZSET_LISTPACK` rdb type. ## Rdb loading improvements: 1) Pre-expansion of dict for validation of duplicate data for listpack and ziplist. 2) Simplifying the release of empty key objects when RDB loading. 3) Unify ziplist and listpack data verify methods for zset and hash, and move code to rdb.c. ## Interface changes 1) New `zset-max-listpack-entries` config is an alias for `zset-max-ziplist-entries` (same with `zset-max-listpack-value`). 2) OBJECT ENCODING will return listpack instead of ziplist. ## Listpack improvements: 1) Add `lpDeleteRange` and `lpDeleteRangeWithEntry` functions to delete a range of entries from listpack. 2) Improve the performance of `lpCompare`, converting from string to integer is faster than converting from integer to string. 3) Replace `snprintf` with `ll2string` to improve performance in converting numbers to strings in `lpGet()`. ## Zset improvements: 1) Improve the performance of `zzlFind` method, use `lpFind` instead of `lpCompare` in a loop. 2) Use `lpDeleteRangeWithEntry` instead of `lpDelete` twice to delete a element of zset. ## Tests 1) Add some unittests for `lpDeleteRange` and `lpDeleteRangeWithEntry` function. 2) Add zset RDB loading test. 3) Add benchmark test for `lpCompare` and `ziplsitCompare`. 4) Add empty listpack zset corrupt dump test.
2021-09-09 11:18:53 -04:00
catch {r RESTORE _zsetbig 0 "\x0C\x3D\x3D\x00\x00\x00\x3A\x00\x00\x00\x14\x00\x00\xF1\x02\xF1\x02\x02\x5F\x31\x04\xF2\x02\xF3\x02\xF3\x02\x02\x5F\x33\x04\xF4\x02\xEE\x02\xF5\x02\x02\x5F\x35\x04\xF6\x02\xF7\x02\xF7\x02\x02\x5F\x37\x04\xF8\x02\xF9\x02\xF9\x02\x02\x5F\x39\x04\xFA\xFF\x09\x00\xAE\xF9\x77\x2A\x47\x24\x33\xF6"} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*Zset ziplist integrity check failed*" 0
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
}
}
test {corrupt payload: fuzzer findings - valgrind ziplist prevlen reaches outside the ziplist} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
Replace ziplist with listpack in quicklist (#9740) Part three of implementing #8702, following #8887 and #9366 . ## Description of the feature 1. Replace the ziplist container of quicklist with listpack. 2. Convert existing quicklist ziplists on RDB loading time. an O(n) operation. ## Interface changes 1. New `list-max-listpack-size` config is an alias for `list-max-ziplist-size`. 2. Replace `debug ziplist` command with `debug listpack`. ## Internal changes 1. Add `lpMerge` to merge two listpacks . (same as `ziplistMerge`) 2. Add `lpRepr` to print info of listpack which is used in debugCommand and `quicklistRepr`. (same as `ziplistRepr`) 3. Replace `QUICKLIST_NODE_CONTAINER_ZIPLIST` with `QUICKLIST_NODE_CONTAINER_PACKED`(following #9357 ). It represent that a quicklistNode is a packed node, as opposed to a plain node. 4. Remove `createZiplistObject` method, which is never used. 5. Calculate listpack entry size using overhead overestimation in `quicklistAllowInsert`. We prefer an overestimation, which would at worse lead to a few bytes below the lowest limit of 4k. ## Improvements 1. Calling `lpShrinkToFit` after converting Ziplist to listpack, which was missed at #9366. 2. Optimize `quicklistAppendPlainNode` to avoid memcpy data. ## Bugfix 1. Fix crash in `quicklistRepr` when ziplist is compressed, introduced from #9366. ## Test 1. Add unittest for `lpMerge`. 2. Modify the old quicklist ziplist corrupt dump test. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-11-24 06:34:13 -05:00
catch {r RESTORE _listbig 0 "\x0E\x02\x1B\x1B\x00\x00\x00\x16\x00\x00\x00\x05\x00\x00\x02\x5F\x39\x04\xF9\x02\x02\x5F\x37\x04\xF7\x02\x02\x5F\x35\xFF\x19\x19\x00\x00\x00\x16\x00\x00\x00\x05\x00\x00\xF5\x02\x02\x5F\x33\x04\xF3\x95\x02\x5F\x31\x04\xF1\xFF\x09\x00\x0C\xFC\x99\x2C\x23\x45\x15\x60"} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
}
}
test {corrupt payload: fuzzer findings - valgrind - bad rdbLoadDoubleValue} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch { r RESTORE _list 0 "\x03\x01\x11\x11\x00\x00\x00\x0A\x00\x00\x00\x01\x00\x00\xD0\x07\x1A\xE9\x02\xFF\x09\x00\x1A\x06\x07\x32\x41\x28\x3A\x46" } err
assert_match "*Bad data format*" $err
}
}
test {corrupt payload: fuzzer findings - valgrind ziplist prev too big} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
Replace ziplist with listpack in quicklist (#9740) Part three of implementing #8702, following #8887 and #9366 . ## Description of the feature 1. Replace the ziplist container of quicklist with listpack. 2. Convert existing quicklist ziplists on RDB loading time. an O(n) operation. ## Interface changes 1. New `list-max-listpack-size` config is an alias for `list-max-ziplist-size`. 2. Replace `debug ziplist` command with `debug listpack`. ## Internal changes 1. Add `lpMerge` to merge two listpacks . (same as `ziplistMerge`) 2. Add `lpRepr` to print info of listpack which is used in debugCommand and `quicklistRepr`. (same as `ziplistRepr`) 3. Replace `QUICKLIST_NODE_CONTAINER_ZIPLIST` with `QUICKLIST_NODE_CONTAINER_PACKED`(following #9357 ). It represent that a quicklistNode is a packed node, as opposed to a plain node. 4. Remove `createZiplistObject` method, which is never used. 5. Calculate listpack entry size using overhead overestimation in `quicklistAllowInsert`. We prefer an overestimation, which would at worse lead to a few bytes below the lowest limit of 4k. ## Improvements 1. Calling `lpShrinkToFit` after converting Ziplist to listpack, which was missed at #9366. 2. Optimize `quicklistAppendPlainNode` to avoid memcpy data. ## Bugfix 1. Fix crash in `quicklistRepr` when ziplist is compressed, introduced from #9366. ## Test 1. Add unittest for `lpMerge`. 2. Modify the old quicklist ziplist corrupt dump test. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-11-24 06:34:13 -05:00
catch {r RESTORE _list 0 "\x0E\x01\x13\x13\x00\x00\x00\x10\x00\x00\x00\x03\x00\x00\xF3\x02\x02\x5F\x31\xC1\xF1\xFF\x09\x00\xC9\x4B\x31\xFE\x61\xC0\x96\xFE"} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
}
}
test {corrupt payload: fuzzer findings - lzf decompression fails, avoid valgrind invalid read} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch {r RESTORE _stream 0 "\x0F\x02\x10\x00\x00\x01\x73\xDD\xAA\x2A\xB9\x00\x00\x00\x00\x00\x00\x00\x00\xC3\x40\x4B\x40\x5C\x18\x5C\x00\x00\x00\x24\x00\x05\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x00\x20\x01\x00\x01\x20\x03\x00\x05\x20\x1C\x40\x07\x05\x01\x01\x82\x5F\x31\x03\x80\x0D\x40\x00\x00\x02\x60\x19\x40\x27\x40\x19\x00\x33\x60\x19\x40\x29\x02\x01\x01\x04\x20\x19\x00\xFF\x10\x00\x00\x01\x73\xDD\xAA\x2A\xBC\x00\x00\x00\x00\x00\x00\x00\x00\xC3\x40\x4D\x40\x5E\x18\x5E\x00\x00\x00\x24\x00\x05\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x00\x20\x01\x06\x01\x01\x82\x5F\x35\x03\x05\x20\x1E\x17\x0B\x03\x01\x01\x06\x01\x40\x0B\x00\x01\x60\x0D\x02\x82\x5F\x37\x60\x19\x80\x00\x00\x08\x60\x19\x80\x27\x02\x82\x5F\x39\x20\x19\x00\xFF\x0A\x81\x00\x00\x01\x73\xDD\xAA\x2A\xBE\x00\x00\x09\x00\x21\x85\x77\x43\x71\x7B\x17\x88"} err
assert_match "*Bad data format*" $err
}
}
test {corrupt payload: fuzzer findings - stream bad lp_count} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch { r RESTORE _stream 0 "\x0F\x01\x10\x00\x00\x01\x73\xDE\xDF\x7D\x9B\x00\x00\x00\x00\x00\x00\x00\x00\x40\x42\x42\x00\x00\x00\x18\x00\x03\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x00\x01\x02\x01\x00\x01\x00\x01\x01\x01\x00\x01\x56\x01\x02\x01\x22\x01\x00\x01\x01\x01\x82\x5F\x31\x03\x05\x01\x02\x01\x2C\x01\x00\x01\x01\x01\x02\x01\x05\x01\xFF\x03\x81\x00\x00\x01\x73\xDE\xDF\x7D\xC7\x00\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x81\x00\x00\x01\x73\xDE\xDF\x7D\x9B\x00\x01\x00\x00\x01\x73\xDE\xDF\x7D\x9B\x00\x00\x00\x00\x00\x00\x00\x00\xF9\x7D\xDF\xDE\x73\x01\x00\x00\x01\x01\x05\x41\x6C\x69\x63\x65\xEB\x7D\xDF\xDE\x73\x01\x00\x00\x01\x00\x00\x01\x73\xDE\xDF\x7D\x9B\x00\x00\x00\x00\x00\x00\x00\x00\x09\x00\xB2\xA8\xA7\x5F\x1B\x61\x72\xD5"} err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: fuzzer findings - stream bad lp_count - unsanitized} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
r RESTORE _stream 0 "\x0F\x01\x10\x00\x00\x01\x73\xDE\xDF\x7D\x9B\x00\x00\x00\x00\x00\x00\x00\x00\x40\x42\x42\x00\x00\x00\x18\x00\x03\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x00\x01\x02\x01\x00\x01\x00\x01\x01\x01\x00\x01\x56\x01\x02\x01\x22\x01\x00\x01\x01\x01\x82\x5F\x31\x03\x05\x01\x02\x01\x2C\x01\x00\x01\x01\x01\x02\x01\x05\x01\xFF\x03\x81\x00\x00\x01\x73\xDE\xDF\x7D\xC7\x00\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x81\x00\x00\x01\x73\xDE\xDF\x7D\x9B\x00\x01\x00\x00\x01\x73\xDE\xDF\x7D\x9B\x00\x00\x00\x00\x00\x00\x00\x00\xF9\x7D\xDF\xDE\x73\x01\x00\x00\x01\x01\x05\x41\x6C\x69\x63\x65\xEB\x7D\xDF\xDE\x73\x01\x00\x00\x01\x00\x00\x01\x73\xDE\xDF\x7D\x9B\x00\x00\x00\x00\x00\x00\x00\x00\x09\x00\xB2\xA8\xA7\x5F\x1B\x61\x72\xD5"
catch { r XREVRANGE _stream 638932639 738}
assert_equal [count_log_message 0 "crashed by signal"] 0
assert_equal [count_log_message 0 "ASSERTION FAILED"] 1
}
}
test {corrupt payload: fuzzer findings - stream integrity check issue} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch { r RESTORE _stream 0 "\x0F\x02\x10\x00\x00\x01\x75\x2D\xA2\x90\x67\x00\x00\x00\x00\x00\x00\x00\x00\xC3\x40\x4F\x40\x5C\x18\x5C\x00\x00\x00\x24\x00\x05\x01\x00\x01\x4A\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x00\x20\x01\x00\x01\x20\x03\x00\x05\x20\x1C\x40\x09\x05\x01\x01\x82\x5F\x31\x03\x80\x0D\x00\x02\x20\x0D\x00\x02\xA0\x19\x00\x03\x20\x0B\x02\x82\x5F\x33\xA0\x19\x00\x04\x20\x0D\x00\x04\x20\x19\x00\xFF\x10\x00\x00\x01\x75\x2D\xA2\x90\x67\x00\x00\x00\x00\x00\x00\x00\x05\xC3\x40\x56\x40\x60\x18\x60\x00\x00\x00\x24\x00\x05\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x00\x20\x01\x06\x01\x01\x82\x5F\x35\x03\x05\x20\x1E\x40\x0B\x03\x01\x01\x06\x01\x80\x0B\x00\x02\x20\x0B\x02\x82\x5F\x37\x60\x19\x03\x01\x01\xDF\xFB\x20\x05\x00\x08\x60\x1A\x20\x0C\x00\xFC\x20\x05\x02\x82\x5F\x39\x20\x1B\x00\xFF\x0A\x81\x00\x00\x01\x75\x2D\xA2\x90\x68\x01\x00\x09\x00\x1D\x6F\xC0\x69\x8A\xDE\xF7\x92" } err
assert_match "*Bad data format*" $err
}
}
test {corrupt payload: fuzzer findings - infinite loop} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
r RESTORE _stream 0 "\x0F\x01\x10\x00\x00\x01\x75\x3A\xA6\xD0\x93\x00\x00\x00\x00\x00\x00\x00\x00\x40\x42\x42\x00\x00\x00\x18\x00\x03\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x00\x01\x02\x01\x00\x01\x00\x01\x01\x01\x00\x01\x05\x01\x02\x01\x00\x01\x01\x01\x01\x01\x82\x5F\x31\x03\xFD\x01\x02\x01\x00\x01\x02\x01\x01\x01\x02\x01\x05\x01\xFF\x03\x81\x00\x00\x01\x75\x3A\xA6\xD0\x93\x02\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x81\x00\x00\x01\x75\x3A\xA6\xD0\x93\x00\x01\x00\x00\x01\x75\x3A\xA6\xD0\x93\x00\x00\x00\x00\x00\x00\x00\x00\x94\xD0\xA6\x3A\x75\x01\x00\x00\x01\x01\x05\x41\x6C\x69\x63\x65\x94\xD0\xA6\x3A\x75\x01\x00\x00\x01\x00\x00\x01\x75\x3A\xA6\xD0\x93\x00\x00\x00\x00\x00\x00\x00\x00\x09\x00\xC4\x09\xAD\x69\x7E\xEE\xA6\x2F"
catch { r XREVRANGE _stream 288270516 971031845 }
assert_equal [count_log_message 0 "crashed by signal"] 0
assert_equal [count_log_message 0 "ASSERTION FAILED"] 1
}
}
Replace all usage of ziplist with listpack for t_hash (#8887) Part one of implementing #8702 (taking hashes first before other types) ## Description of the feature 1. Change ziplist encoded hash objects to listpack encoding. 2. Convert existing ziplists on RDB loading time. an O(n) operation. ## Rdb format changes 1. Add RDB_TYPE_HASH_LISTPACK rdb type. 2. Bump RDB_VERSION to 10 ## Interface changes 1. New `hash-max-listpack-entries` config is an alias for `hash-max-ziplist-entries` (same with `hash-max-listpack-value`) 2. OBJECT ENCODING will return `listpack` instead of `ziplist` ## Listpack improvements: 1. Support direct insert, replace integer element (rather than convert back and forth from string) 3. Add more listpack capabilities to match the ziplist ones (like `lpFind`, `lpRandomPairs` and such) 4. Optimize element length fetching, avoid multiple calculations 5. Use inline to avoid function call overhead. ## Tests 1. Add a new test to the RDB load time conversion 2. Adding the listpack unit tests. (based on the one in ziplist.c) 3. Add a few "corrupt payload: fuzzer findings" tests, and slightly modify existing ones. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-10 02:18:49 -04:00
test {corrupt payload: fuzzer findings - hash ziplist too long entry len} {
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r debug set-skip-checksum-validation 1
Replace all usage of ziplist with listpack for t_hash (#8887) Part one of implementing #8702 (taking hashes first before other types) ## Description of the feature 1. Change ziplist encoded hash objects to listpack encoding. 2. Convert existing ziplists on RDB loading time. an O(n) operation. ## Rdb format changes 1. Add RDB_TYPE_HASH_LISTPACK rdb type. 2. Bump RDB_VERSION to 10 ## Interface changes 1. New `hash-max-listpack-entries` config is an alias for `hash-max-ziplist-entries` (same with `hash-max-listpack-value`) 2. OBJECT ENCODING will return `listpack` instead of `ziplist` ## Listpack improvements: 1. Support direct insert, replace integer element (rather than convert back and forth from string) 3. Add more listpack capabilities to match the ziplist ones (like `lpFind`, `lpRandomPairs` and such) 4. Optimize element length fetching, avoid multiple calculations 5. Use inline to avoid function call overhead. ## Tests 1. Add a new test to the RDB load time conversion 2. Adding the listpack unit tests. (based on the one in ziplist.c) 3. Add a few "corrupt payload: fuzzer findings" tests, and slightly modify existing ones. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-10 02:18:49 -04:00
catch {
r RESTORE _hash 0 "\x0D\x3D\x3D\x00\x00\x00\x3A\x00\x00\x00\x14\x13\x00\xF5\x02\xF5\x02\xF2\x02\x53\x5F\x31\x04\xF3\x02\xF3\x02\xF7\x02\xF7\x02\xF8\x02\x02\x5F\x37\x04\xF1\x02\xF1\x02\xF6\x02\x02\x5F\x35\x04\xF4\x02\x02\x5F\x33\x04\xFA\x02\x02\x5F\x39\x04\xF9\x02\xF9\xFF\x09\x00\xB5\x48\xDE\x62\x31\xD0\xE5\x63"
} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*integrity check failed*" 0
Sanitize dump payload: validate no duplicate records in hash/zset/intset If RESTORE passes successfully with full sanitization, we can't affort to crash later on assertion due to duplicate records in a hash when converting it form ziplist to dict. This means that when doing full sanitization, we must make sure there are no duplicate records in any of the collections.
2020-11-02 02:35:37 -05:00
}
}
Skip OOM-related tests on incompatible platforms. (#9386) We only run OOM related tests on x86_64 and aarch64, as jemalloc on other platforms (notably s390x) may actually succeed very large allocations. As a result the test may hang for a very long time at the cleanup phase, iterating as many as 2^61 hash table slots.
2021-08-18 09:00:22 -04:00
if {$run_oom_tests} {
Sanitize dump payload: fail RESTORE if memory allocation fails When RDB input attempts to make a huge memory allocation that fails, RESTORE should fail gracefully rather than die with panic
2020-11-22 14:22:49 -05:00
test {corrupt payload: OOM in rdbGenericLoadStringObject} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
catch { r RESTORE x 0 "\x0A\x81\x7F\xFF\xFF\xFF\xFF\xFF\xFF\xFF\x13\x00\x00\x00\x0E\x00\x00\x00\x02\x00\x00\x02\x61\x00\x04\x02\x62\x00\xFF\x09\x00\x57\x04\xE5\xCD\xD4\x37\x6C\x57" } err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: fuzzer findings - OOM in dictExpand} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch { r RESTORE x 0 "\x02\x81\x02\x5F\x31\xC0\x00\xC0\x02\x09\x00\xCD\x84\x2C\xB7\xE8\xA4\x49\x57" } err
assert_match "*Bad data format*" $err
r ping
}
}
Skip OOM-related tests on incompatible platforms. (#9386) We only run OOM related tests on x86_64 and aarch64, as jemalloc on other platforms (notably s390x) may actually succeed very large allocations. As a result the test may hang for a very long time at the cleanup phase, iterating as many as 2^61 hash table slots.
2021-08-18 09:00:22 -04:00
}
Replace all usage of ziplist with listpack for t_zset (#9366) Part two of implementing #8702 (zset), after #8887. ## Description of the feature Replaced all uses of ziplist with listpack in t_zset, and optimized some of the code to optimize performance. ## Rdb format changes New `RDB_TYPE_ZSET_LISTPACK` rdb type. ## Rdb loading improvements: 1) Pre-expansion of dict for validation of duplicate data for listpack and ziplist. 2) Simplifying the release of empty key objects when RDB loading. 3) Unify ziplist and listpack data verify methods for zset and hash, and move code to rdb.c. ## Interface changes 1) New `zset-max-listpack-entries` config is an alias for `zset-max-ziplist-entries` (same with `zset-max-listpack-value`). 2) OBJECT ENCODING will return listpack instead of ziplist. ## Listpack improvements: 1) Add `lpDeleteRange` and `lpDeleteRangeWithEntry` functions to delete a range of entries from listpack. 2) Improve the performance of `lpCompare`, converting from string to integer is faster than converting from integer to string. 3) Replace `snprintf` with `ll2string` to improve performance in converting numbers to strings in `lpGet()`. ## Zset improvements: 1) Improve the performance of `zzlFind` method, use `lpFind` instead of `lpCompare` in a loop. 2) Use `lpDeleteRangeWithEntry` instead of `lpDelete` twice to delete a element of zset. ## Tests 1) Add some unittests for `lpDeleteRange` and `lpDeleteRangeWithEntry` function. 2) Add zset RDB loading test. 3) Add benchmark test for `lpCompare` and `ziplsitCompare`. 4) Add empty listpack zset corrupt dump test.
2021-09-09 11:18:53 -04:00
test {corrupt payload: fuzzer findings - zset ziplist invalid tail offset} {
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
Replace all usage of ziplist with listpack for t_zset (#9366) Part two of implementing #8702 (zset), after #8887. ## Description of the feature Replaced all uses of ziplist with listpack in t_zset, and optimized some of the code to optimize performance. ## Rdb format changes New `RDB_TYPE_ZSET_LISTPACK` rdb type. ## Rdb loading improvements: 1) Pre-expansion of dict for validation of duplicate data for listpack and ziplist. 2) Simplifying the release of empty key objects when RDB loading. 3) Unify ziplist and listpack data verify methods for zset and hash, and move code to rdb.c. ## Interface changes 1) New `zset-max-listpack-entries` config is an alias for `zset-max-ziplist-entries` (same with `zset-max-listpack-value`). 2) OBJECT ENCODING will return listpack instead of ziplist. ## Listpack improvements: 1) Add `lpDeleteRange` and `lpDeleteRangeWithEntry` functions to delete a range of entries from listpack. 2) Improve the performance of `lpCompare`, converting from string to integer is faster than converting from integer to string. 3) Replace `snprintf` with `ll2string` to improve performance in converting numbers to strings in `lpGet()`. ## Zset improvements: 1) Improve the performance of `zzlFind` method, use `lpFind` instead of `lpCompare` in a loop. 2) Use `lpDeleteRangeWithEntry` instead of `lpDelete` twice to delete a element of zset. ## Tests 1) Add some unittests for `lpDeleteRange` and `lpDeleteRangeWithEntry` function. 2) Add zset RDB loading test. 3) Add benchmark test for `lpCompare` and `ziplsitCompare`. 4) Add empty listpack zset corrupt dump test.
2021-09-09 11:18:53 -04:00
catch {r RESTORE _zset 0 "\x0C\x19\x19\x00\x00\x00\x02\x00\x00\x00\x06\x00\x00\xF1\x02\xF1\x02\x02\x5F\x31\x04\xF2\x02\xF3\x02\xF3\xFF\x09\x00\x4D\x72\x7B\x97\xCD\x9A\x70\xC1"} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*Zset ziplist integrity check failed*" 0
Sanitize dump payload: fuzz tester and fixes for segfaults and leaks it exposed The test creates keys with various encodings, DUMP them, corrupt the payload and RESTORES it. It utilizes the recently added use-exit-on-panic config to distinguish between asserts and segfaults. If the restore succeeds, it runs random commands on the key to attempt to trigger a crash. It runs in two modes, one with deep sanitation enabled and one without. In the first one we don't expect any assertions or segfaults, in the second one we expect assertions, but no segfaults. We also check for leaks and invalid reads using valgrind, and if we find them we print the commands that lead to that issue. Changes in the code (other than the test): - Replace a few NPD (null pointer deference) flows and division by zero with an assertion, so that it doesn't fail the test. (since we set the server to use `exit` rather than `abort` on assertion). - Fix quite a lot of flows in rdb.c that could have lead to memory leaks in RESTORE command (since it now responds with an error rather than panic) - Add a DEBUG flag for SET-SKIP-CHECKSUM-VALIDATION so that the test don't need to bother with faking a valid checksum - Remove a pile of code in serverLogObjectDebugInfo which is actually unsafe to run in the crash report (see comments in the code) - fix a missing boundary check in lzf_decompress test suite infra improvements: - be able to run valgrind checks before the process terminates - rotate log files when restarting servers
2020-08-14 09:05:34 -04:00
}
}
test {corrupt payload: fuzzer findings - negative reply length} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
r RESTORE _stream 0 "\x0F\x01\x10\x00\x00\x01\x75\xCF\xA1\x16\xA7\x00\x00\x00\x00\x00\x00\x00\x00\x40\x42\x42\x00\x00\x00\x18\x00\x03\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x00\x01\x02\x01\x00\x01\x00\x01\x01\x01\x00\x01\x05\x01\x02\x01\x00\x01\x01\x01\x01\x01\x14\x5F\x31\x03\x05\x01\x02\x01\x00\x01\x02\x01\x01\x01\x02\x01\x05\x01\xFF\x03\x81\x00\x00\x01\x75\xCF\xA1\x16\xA7\x02\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x81\x00\x00\x01\x75\xCF\xA1\x16\xA7\x01\x01\x00\x00\x01\x75\xCF\xA1\x16\xA7\x00\x00\x00\x00\x00\x00\x00\x01\xA7\x16\xA1\xCF\x75\x01\x00\x00\x01\x01\x05\x41\x6C\x69\x63\x65\xA7\x16\xA1\xCF\x75\x01\x00\x00\x01\x00\x00\x01\x75\xCF\xA1\x16\xA7\x00\x00\x00\x00\x00\x00\x00\x01\x09\x00\x1B\x42\x52\xB8\xDD\x5C\xE5\x4E"
catch {r XADD _stream * -956 -2601503852}
catch {r XINFO STREAM _stream FULL}
assert_equal [count_log_message 0 "crashed by signal"] 0
assert_equal [count_log_message 0 "ASSERTION FAILED"] 1
}
}
test {corrupt payload: fuzzer findings - valgrind negative malloc} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {r RESTORE _key 0 "\x0E\x01\x81\xD6\xD6\x00\x00\x00\x0A\x00\x00\x00\x01\x00\x00\x40\xC8\x6F\x2F\x36\xE2\xDF\xE3\x2E\x26\x64\x8B\x87\xD1\x7A\xBD\xFF\xEF\xEF\x63\x65\xF6\xF8\x8C\x4E\xEC\x96\x89\x56\x88\xF8\x3D\x96\x5A\x32\xBD\xD1\x36\xD8\x02\xE6\x66\x37\xCB\x34\x34\xC4\x52\xA7\x2A\xD5\x6F\x2F\x7E\xEE\xA2\x94\xD9\xEB\xA9\x09\x38\x3B\xE1\xA9\x60\xB6\x4E\x09\x44\x1F\x70\x24\xAA\x47\xA8\x6E\x30\xE1\x13\x49\x4E\xA1\x92\xC4\x6C\xF0\x35\x83\xD9\x4F\xD9\x9C\x0A\x0D\x7A\xE7\xB1\x61\xF5\xC1\x2D\xDC\xC3\x0E\x87\xA6\x80\x15\x18\xBA\x7F\x72\xDD\x14\x75\x46\x44\x0B\xCA\x9C\x8F\x1C\x3C\xD7\xDA\x06\x62\x18\x7E\x15\x17\x24\xAB\x45\x21\x27\xC2\xBC\xBB\x86\x6E\xD8\xBD\x8E\x50\xE0\xE0\x88\xA4\x9B\x9D\x15\x2A\x98\xFF\x5E\x78\x6C\x81\xFC\xA8\xC9\xC8\xE6\x61\xC8\xD1\x4A\x7F\x81\xD6\xA6\x1A\xAD\x4C\xC1\xA2\x1C\x90\x68\x15\x2A\x8A\x36\xC0\x58\xC3\xCC\xA6\x54\x19\x12\x0F\xEB\x46\xFF\x6E\xE3\xA7\x92\xF8\xFF\x09\x00\xD0\x71\xF7\x9F\xF7\x6A\xD6\x2E"} err
assert_match "*Bad data format*" $err
r ping
}
}
Sanitize dump payload: excessive free on dup zset fields (#8189)
2020-12-14 10:10:31 -05:00
test {corrupt payload: fuzzer findings - valgrind invalid read} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {r RESTORE _key 0 "\x05\x0A\x02\x5F\x39\x00\x00\x00\x00\x00\x00\x22\x40\xC0\x08\x00\x00\x00\x00\x00\x00\x20\x40\x02\x5F\x37\x00\x00\x00\x00\x00\x00\x1C\x40\xC0\x06\x00\x00\x00\x00\x00\x00\x18\x40\x02\x5F\x33\x00\x00\x00\x00\x00\x00\x14\x40\xC0\x04\x00\x00\x00\x00\x00\x00\x10\x40\x02\x5F\x33\x00\x00\x00\x00\x00\x00\x08\x40\xC0\x02\x00\x00\x00\x00\x00\x00\x00\x40\x02\x5F\x31\x00\x00\x00\x00\x00\x00\xF0\x3F\xC0\x00\x00\x00\x00\x00\x00\x00\x00\x00\x09\x00\x3C\x66\xD7\x14\xA9\xDA\x3C\x69"} err
assert_match "*Bad data format*" $err
r ping
}
}
Sanitize dump payload: fix empty keys when RDB loading and restore command (#9297) When we load rdb or restore command, if we encounter a length of 0, it will result in the creation of an empty key. This could either be a corrupt payload, or a result of a bug (see #8453 ) This PR mainly fixes the following: 1) When restore command will return `Bad data format` error. 2) When loading RDB, we will silently discard the key. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-05 15:42:20 -04:00
test {corrupt payload: fuzzer findings - empty hash ziplist} {
Fix test issues from introduction of HRANDFIELD (#8424) * The corrupt dump fuzzer found a division by zero. * in some cases the random fields from the HRANDFIELD tests produced fields with newlines and other special chars (due to \ char), this caused the TCL tests to see a bulk response that has a newline in it and add {} around it, later it can think this is a nested list. in fact the `alpha` random string generator isn't using spaces and newlines, so it should not use `\` either.
2021-01-31 05:13:45 -05:00
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
Sanitize dump payload: fix empty keys when RDB loading and restore command (#9297) When we load rdb or restore command, if we encounter a length of 0, it will result in the creation of an empty key. This could either be a corrupt payload, or a result of a bug (see #8453 ) This PR mainly fixes the following: 1) When restore command will return `Bad data format` error. 2) When loading RDB, we will silently discard the key. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-05 15:42:20 -04:00
catch {r RESTORE _int 0 "\x04\xC0\x01\x09\x00\xF6\x8A\xB6\x7A\x85\x87\x72\x4D"} err
assert_match "*Bad data format*" $err
r ping
Fix test issues from introduction of HRANDFIELD (#8424) * The corrupt dump fuzzer found a division by zero. * in some cases the random fields from the HRANDFIELD tests produced fields with newlines and other special chars (due to \ char), this caused the TCL tests to see a bulk response that has a newline in it and add {} around it, later it can think this is a nested list. in fact the `alpha` random string generator isn't using spaces and newlines, so it should not use `\` either.
2021-01-31 05:13:45 -05:00
}
}
Corrupt stream key access to uninitialized memory (#8681) the corrupt-dump-fuzzer test found a case where an access to a corrupt stream would have caused accessing to uninitialized memory. now it'll panic instead. The issue was that there was a stream that says it has more than 0 records, but looking for the max ID came back empty handed. p.s. when sanitize-dump-payload is used, this corruption is detected, and the RESTORE command is gracefully rejected.
2021-03-24 05:33:49 -04:00
test {corrupt payload: fuzzer findings - stream with no records} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
r restore _stream 0 "\x0F\x01\x10\x00\x00\x01\x78\x4D\x55\x68\x09\x00\x00\x00\x00\x00\x00\x00\x00\x40\x42\x42\x00\x00\x00\x18\x00\x02\x01\x01\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x00\x01\x02\x01\x00\x01\x00\x01\x01\x01\x00\x01\x05\x01\x03\x01\x3E\x01\x00\x01\x01\x01\x82\x5F\x31\x03\x05\x01\x02\x01\x50\x01\x00\x01\x01\x01\x02\x01\x05\x23\xFF\x02\x81\x00\x00\x01\x78\x4D\x55\x68\x59\x00\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x81\x00\x00\x01\x78\x4D\x55\x68\x47\x00\x01\x00\x00\x01\x78\x4D\x55\x68\x47\x00\x00\x00\x00\x00\x00\x00\x00\x9F\x68\x55\x4D\x78\x01\x00\x00\x01\x01\x05\x41\x6C\x69\x63\x65\x85\x68\x55\x4D\x78\x01\x00\x00\x01\x00\x00\x01\x78\x4D\x55\x68\x47\x00\x00\x00\x00\x00\x00\x00\x00\x09\x00\xF1\xC0\x72\x70\x39\x40\x1E\xA9" replace
catch {r XREAD STREAMS _stream $}
assert_equal [count_log_message 0 "crashed by signal"] 0
assert_equal [count_log_message 0 "Guru Meditation"] 1
}
}
Improvements to corrupt payload sanitization (#9321) Recently we found two issues in the fuzzer tester: #9302 #9285 After fixing them, more problems surfaced and this PR (as well as #9297) aims to fix them. Here's a list of the fixes - Prevent an overflow when allocating a dict hashtable - Prevent OOM when attempting to allocate a huge string - Prevent a few invalid accesses in listpack - Improve sanitization of listpack first entry - Validate integrity of stream consumer groups PEL - Validate integrity of stream listpack entry IDs - Validate ziplist tail followed by extra data which start with 0xff Co-authored-by: sundb <sundbcn@gmail.com>
2021-08-05 15:56:14 -04:00
test {corrupt payload: fuzzer findings - quicklist ziplist tail followed by extra data which start with 0xff} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {
r restore key 0 "\x0E\x01\x11\x11\x00\x00\x00\x0A\x00\x00\x00\x01\x00\x00\xF6\xFF\xB0\x6C\x9C\xFF\x09\x00\x9C\x37\x47\x49\x4D\xDE\x94\xF5" replace
} err
assert_match "*Bad data format*" $err
verify_log_message 0 "*integrity check failed*" 0
}
}
test {corrupt payload: fuzzer findings - dict init to huge size} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch {r restore key 0 "\x02\x81\xC0\x00\x02\x5F\x31\xC0\x02\x09\x00\xB2\x1B\xE5\x17\x2E\x15\xF4\x6C" replace} err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: fuzzer findings - huge string} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {r restore key 0 "\x00\x81\x01\x09\x00\xF6\x2B\xB6\x7A\x85\x87\x72\x4D"} err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: fuzzer findings - stream PEL without consumer} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {r restore _stream 0 "\x0F\x01\x10\x00\x00\x01\x7B\x08\xF0\xB2\x34\x00\x00\x00\x00\x00\x00\x00\x00\xC3\x3B\x40\x42\x19\x42\x00\x00\x00\x18\x00\x02\x01\x01\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x00\x20\x10\x00\x00\x20\x01\x00\x01\x20\x03\x02\x05\x01\x03\x20\x05\x40\x00\x04\x82\x5F\x31\x03\x05\x60\x19\x80\x32\x02\x05\x01\xFF\x02\x81\x00\x00\x01\x7B\x08\xF0\xB2\x34\x02\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x81\x00\x00\x01\x7B\x08\xF0\xB2\x34\x01\x01\x00\x00\x01\x7B\x08\xF0\xB2\x34\x00\x00\x00\x00\x00\x00\x00\x01\x35\xB2\xF0\x08\x7B\x01\x00\x00\x01\x01\x13\x41\x6C\x69\x63\x65\x35\xB2\xF0\x08\x7B\x01\x00\x00\x01\x00\x00\x01\x7B\x08\xF0\xB2\x34\x00\x00\x00\x00\x00\x00\x00\x01\x09\x00\x28\x2F\xE0\xC5\x04\xBB\xA7\x31"} err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: fuzzer findings - stream listpack valgrind issue} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
r restore _stream 0 "\x0F\x01\x10\x00\x00\x01\x7B\x09\x5E\x94\xFF\x00\x00\x00\x00\x00\x00\x00\x00\x40\x42\x42\x00\x00\x00\x18\x00\x02\x01\x01\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x00\x01\x02\x01\x00\x01\x00\x01\x01\x01\x00\x01\x05\x01\x03\x01\x25\x01\x00\x01\x01\x01\x82\x5F\x31\x03\x05\x01\x02\x01\x32\x01\x00\x01\x01\x01\x02\x01\xF0\x01\xFF\x02\x81\x00\x00\x01\x7B\x09\x5E\x95\x31\x00\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x81\x00\x00\x01\x7B\x09\x5E\x95\x24\x00\x01\x00\x00\x01\x7B\x09\x5E\x95\x24\x00\x00\x00\x00\x00\x00\x00\x00\x5C\x95\x5E\x09\x7B\x01\x00\x00\x01\x01\x05\x41\x6C\x69\x63\x65\x4B\x95\x5E\x09\x7B\x01\x00\x00\x01\x00\x00\x01\x7B\x09\x5E\x95\x24\x00\x00\x00\x00\x00\x00\x00\x00\x09\x00\x19\x29\x94\xDF\x76\xF8\x1A\xC6"
catch {r XINFO STREAM _stream FULL }
assert_equal [count_log_message 0 "crashed by signal"] 0
assert_equal [count_log_message 0 "ASSERTION FAILED"] 1
}
}
test {corrupt payload: fuzzer findings - stream with bad lpFirst} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {r restore _stream 0 "\x0F\x01\x10\x00\x00\x01\x7B\x0E\x52\xD2\xEC\x00\x00\x00\x00\x00\x00\x00\x00\x40\x42\x42\x00\x00\x00\x18\x00\x02\xF7\x01\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x00\x01\x02\x01\x00\x01\x00\x01\x01\x01\x00\x01\x05\x01\x03\x01\x01\x01\x00\x01\x01\x01\x82\x5F\x31\x03\x05\x01\x02\x01\x01\x01\x01\x01\x01\x01\x02\x01\x05\x01\xFF\x02\x81\x00\x00\x01\x7B\x0E\x52\xD2\xED\x01\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x81\x00\x00\x01\x7B\x0E\x52\xD2\xED\x00\x01\x00\x00\x01\x7B\x0E\x52\xD2\xED\x00\x00\x00\x00\x00\x00\x00\x00\xED\xD2\x52\x0E\x7B\x01\x00\x00\x01\x01\x05\x41\x6C\x69\x63\x65\xED\xD2\x52\x0E\x7B\x01\x00\x00\x01\x00\x00\x01\x7B\x0E\x52\xD2\xED\x00\x00\x00\x00\x00\x00\x00\x00\x09\x00\xAC\x05\xC9\x97\x5D\x45\x80\xB3"} err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: fuzzer findings - stream listpack lpPrev valgrind issue} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
r restore _stream 0 "\x0F\x01\x10\x00\x00\x01\x7B\x0E\xAE\x66\x36\x00\x00\x00\x00\x00\x00\x00\x00\x40\x42\x42\x00\x00\x00\x18\x00\x02\x01\x01\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x00\x01\x02\x01\x00\x01\x00\x01\x01\x01\x00\x01\x1D\x01\x03\x01\x24\x01\x00\x01\x01\x69\x82\x5F\x31\x03\x05\x01\x02\x01\x33\x01\x00\x01\x01\x01\x02\x01\x05\x01\xFF\x02\x81\x00\x00\x01\x7B\x0E\xAE\x66\x69\x00\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x81\x00\x00\x01\x7B\x0E\xAE\x66\x5A\x00\x01\x00\x00\x01\x7B\x0E\xAE\x66\x5A\x00\x00\x00\x00\x00\x00\x00\x00\x94\x66\xAE\x0E\x7B\x01\x00\x00\x01\x01\x05\x41\x6C\x69\x63\x65\x83\x66\xAE\x0E\x7B\x01\x00\x00\x01\x00\x00\x01\x7B\x0E\xAE\x66\x5A\x00\x00\x00\x00\x00\x00\x00\x00\x09\x00\xD5\xD7\xA5\x5C\x63\x1C\x09\x40"
catch {r XREVRANGE _stream 1618622681 606195012389}
assert_equal [count_log_message 0 "crashed by signal"] 0
assert_equal [count_log_message 0 "ASSERTION FAILED"] 1
}
}
test {corrupt payload: fuzzer findings - stream with non-integer entry id} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {r restore _streambig 0 "\x0F\x03\x10\x00\x00\x01\x7B\x13\x34\xC3\xB2\x00\x00\x00\x00\x00\x00\x00\x00\xC3\x40\x4F\x40\x5C\x18\x5C\x00\x00\x00\x24\x00\x05\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x80\x20\x01\x00\x01\x20\x03\x00\x05\x20\x1C\x40\x09\x05\x01\x01\x82\x5F\x31\x03\x80\x0D\x00\x02\x20\x0D\x00\x02\xA0\x19\x00\x03\x20\x0B\x02\x82\x5F\x33\xA0\x19\x00\x04\x20\x0D\x00\x04\x20\x19\x00\xFF\x10\x00\x00\x01\x7B\x13\x34\xC3\xB2\x00\x00\x00\x00\x00\x00\x00\x05\xC3\x40\x56\x40\x61\x18\x61\x00\x00\x00\x24\x00\x05\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x00\x20\x01\x06\x01\x01\x82\x5F\x35\x03\x05\x20\x1E\x40\x0B\x03\x01\x01\x06\x01\x40\x0B\x03\x01\x01\xDF\xFB\x20\x05\x02\x82\x5F\x37\x60\x1A\x20\x0E\x00\xFC\x20\x05\x00\x08\xC0\x1B\x00\xFD\x20\x0C\x02\x82\x5F\x39\x20\x1B\x00\xFF\x10\x00\x00\x01\x7B\x13\x34\xC3\xB3\x00\x00\x00\x00\x00\x00\x00\x03\xC3\x3D\x40\x4A\x18\x4A\x00\x00\x00\x15\x00\x02\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x00\x20\x01\x40\x00\x00\x05\x60\x07\x02\xDF\xFD\x02\xC0\x23\x09\x01\x01\x86\x75\x6E\x69\x71\x75\x65\x07\xA0\x2D\x02\x08\x01\xFF\x0C\x81\x00\x00\x01\x7B\x13\x34\xC3\xB4\x00\x00\x09\x00\x9D\xBD\xD5\xB9\x33\xC4\xC5\xFF"} err
assert_match "*Bad data format*" $err
r ping
}
}
Sanitize dump payload: fix empty keys when RDB loading and restore command (#9297) When we load rdb or restore command, if we encounter a length of 0, it will result in the creation of an empty key. This could either be a corrupt payload, or a result of a bug (see #8453 ) This PR mainly fixes the following: 1) When restore command will return `Bad data format` error. 2) When loading RDB, we will silently discard the key. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-05 15:42:20 -04:00
test {corrupt payload: fuzzer findings - empty quicklist} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {
r restore key 0 "\x0E\xC0\x2B\x15\x00\x00\x00\x0A\x00\x00\x00\x01\x00\x00\xE0\x62\x58\xEA\xDF\x22\x00\x00\x00\xFF\x09\x00\xDF\x35\xD2\x67\xDC\x0E\x89\xAB" replace
} err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: fuzzer findings - empty zset} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {r restore key 0 "\x05\xC0\x01\x09\x00\xF6\x8A\xB6\x7A\x85\x87\x72\x4D"} err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: fuzzer findings - hash with len of 0} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {r restore key 0 "\x04\xC0\x21\x09\x00\xF6\x8A\xB6\x7A\x85\x87\x72\x4D"} err
assert_match "*Bad data format*" $err
r ping
}
}
Replace all usage of ziplist with listpack for t_hash (#8887) Part one of implementing #8702 (taking hashes first before other types) ## Description of the feature 1. Change ziplist encoded hash objects to listpack encoding. 2. Convert existing ziplists on RDB loading time. an O(n) operation. ## Rdb format changes 1. Add RDB_TYPE_HASH_LISTPACK rdb type. 2. Bump RDB_VERSION to 10 ## Interface changes 1. New `hash-max-listpack-entries` config is an alias for `hash-max-ziplist-entries` (same with `hash-max-listpack-value`) 2. OBJECT ENCODING will return `listpack` instead of `ziplist` ## Listpack improvements: 1. Support direct insert, replace integer element (rather than convert back and forth from string) 3. Add more listpack capabilities to match the ziplist ones (like `lpFind`, `lpRandomPairs` and such) 4. Optimize element length fetching, avoid multiple calculations 5. Use inline to avoid function call overhead. ## Tests 1. Add a new test to the RDB load time conversion 2. Adding the listpack unit tests. (based on the one in ziplist.c) 3. Add a few "corrupt payload: fuzzer findings" tests, and slightly modify existing ones. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-08-10 02:18:49 -04:00
test {corrupt payload: fuzzer findings - hash listpack first element too long entry len} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r debug set-skip-checksum-validation 1
r config set sanitize-dump-payload yes
catch { r restore _hash 0 "\x10\x15\x15\x00\x00\x00\x06\x00\xF0\x01\x00\x01\x01\x01\x82\x5F\x31\x03\x02\x01\x02\x01\xFF\x0A\x00\x94\x21\x0A\xFA\x06\x52\x9F\x44" replace } err
assert_match "*Bad data format*" $err
verify_log_message 0 "*integrity check failed*" 0
}
}
Sanitize dump payload: fix double free after insert dup nodekey to stream rax and returns 0 (#9399)
2021-08-20 03:37:45 -04:00
test {corrupt payload: fuzzer findings - stream double free listpack when insert dup node to rax returns 0} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r debug set-skip-checksum-validation 1
r config set sanitize-dump-payload yes
catch { r restore _stream 0 "\x0F\x03\x10\x00\x00\x01\x7B\x60\x5A\x23\x79\x00\x00\x00\x00\x00\x00\x00\x00\xC3\x40\x4F\x40\x5C\x18\x5C\x00\x00\x00\x24\x00\x05\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x00\x20\x01\x00\x01\x20\x03\x00\x05\x20\x1C\x40\x09\x05\x01\x01\x82\x5F\x31\x03\x80\x0D\x00\x02\x20\x0D\x00\x02\xA0\x19\x00\x03\x20\x0B\x02\x82\x5F\x33\xA0\x19\x00\x04\x20\x0D\x00\x04\x20\x19\x00\xFF\x10\x00\x00\x01\x7B\x60\x5A\x23\x79\x00\x00\x00\x00\x00\x00\x00\x05\xC3\x40\x51\x40\x5E\x18\x5E\x00\x00\x00\x24\x00\x05\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x00\x20\x01\x06\x01\x01\x82\x5F\x35\x03\x05\x20\x1E\x40\x0B\x03\x01\x01\x06\x01\x80\x0B\x00\x02\x20\x0B\x02\x82\x5F\x37\xA0\x19\x00\x03\x20\x0D\x00\x08\xA0\x19\x00\x04\x20\x0B\x02\x82\x5F\x39\x20\x19\x00\xFF\x10\x00\x00\x01\x7B\x60\x5A\x23\x79\x00\x00\x00\x00\x00\x00\x00\x00\xC3\x3B\x40\x49\x18\x49\x00\x00\x00\x15\x00\x02\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x00\x20\x01\x40\x00\x00\x05\x20\x07\x40\x09\xC0\x22\x09\x01\x01\x86\x75\x6E\x69\x71\x75\x65\x07\xA0\x2C\x02\x08\x01\xFF\x0C\x81\x00\x00\x01\x7B\x60\x5A\x23\x7A\x01\x00\x0A\x00\x9C\x8F\x1E\xBF\x2E\x05\x59\x09" replace } err
assert_match "*Bad data format*" $err
r ping
}
}
Prevent LCS from allocating temp memory over proto-max-bulk-len (#9817) LCS can allocate immense amount of memory (sizes of two inputs multiplied by each other). In the past this caused some possible security issues due to overflows, which we solved and also added use of `trymalloc` to return "Insufficient memory" instead of OOM panic zmalloc. But in case overcommit is enabled, it could be that we won't get the OOM panic, and zmalloc will succeed, and then we can get OOM killed by the kernel. The solution here is to prevent LCS from allocating transient memory that's bigger than `proto-max-bulk-len` config. This config is not directly related to transient memory, but using a hard coded value ad well as introducing a specific config seems wrong. This comes to solve an error in the corrupt-dump-fuzzer test that started in the daily CI see #9799
2021-11-21 07:30:20 -05:00
test {corrupt payload: fuzzer findings - LCS OOM} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r SETRANGE _int 423324 1450173551
catch {r LCS _int _int} err
assert_match "*Insufficient memory*" $err
r ping
}
}
Fix false positive leak reported by GCC ASAN (#9816) Leak found by the corrupt-dump-fuzzer when using GCC ASAN, which seems to falsely report leaks on pointers kept only on the stack when calling exit. Instead we now use _exit on panic / assert to skip these leak checks. Additionally, check for sanitizer warnings in the corrupt-dump-fuzzer between iterations, so that when something is found we know which test to relate it too (and it prints reproduction command list)
2021-11-21 11:47:10 -05:00
test {corrupt payload: fuzzer findings - gcc asan reports false leak on assert} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r debug set-skip-checksum-validation 1
r config set sanitize-dump-payload no
catch { r restore _list 0 "\x12\x01\x02\x13\x13\x00\x00\x00\x10\x00\x00\x00\x03\x00\x00\xF3\xFE\x02\x5F\x31\x04\xF1\xFF\x0A\x00\x19\x8D\x3D\x74\x85\x94\x29\xBD" }
catch { r LPOP _list } err
assert_equal [count_log_message 0 "crashed by signal"] 0
assert_equal [count_log_message 0 "ASSERTION FAILED"] 1
}
}
santize dump payload: fix carsh when zset with NAN score (#10002) `zslInsert` with a NAN score will crash the server. This one found by the `corrupt-dump-fuzzer`.
2021-12-26 04:40:11 -05:00
test {corrupt payload: fuzzer findings - lpFind invalid access} {
Fix invalid access in lpFind on corrupted listpack (#9819) Issue found by corrupt-dump-fuzzer test with ASAN. The problem was that lpSkip and lpGetWithSize could read the next listpack entry without validating that it's in range. Similarly even the memcmp in lpFind could do that and possibly crash on segfault and now they'll crash on assert first. The naive fix of using lpAssertValidEntry every time, resulted in 30% degradation in the lpFind benchmark of the unit test. The final fix with the condition at the bottom has no performance implications.
2021-11-22 08:30:00 -05:00
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r debug set-skip-checksum-validation 1
r config set sanitize-dump-payload no
r restore _hashbig 0 "\x10\x39\x39\x00\x00\x00\x14\x00\x06\x01\x06\x01\x03\x01\x82\x5F\x33\x03\x07\x01\x82\x5F\x37\x03\x00\x01\x00\x01\x04\x01\x04\x01\x09\x01\x82\x5F\x39\x03\x05\x01\x82\x5F\x35\x03\x08\x01\x08\x01\x01\x01\x82\x5F\x31\x03\x02\x01\xF0\x01\xFF\x0A\x00\x29\xD7\xE4\x52\x79\x7A\x95\x82"
catch { r HLEN _hashbig }
catch { r HSETNX _hashbig 513072881620 "\x9A\x4B\x1F\xF2\x99\x74\x6E\x96\x84\x7F\xB9\x85\xBE\xD6\x1A\x93\x0A\xED\xAE\x19\xA0\x5A\x67\xD6\x89\xA8\xF9\xF2\xB8\xBD\x3E\x5A\xCF\xD2\x5B\x17\xA4\xBB\xB2\xA9\x56\x67\x6E\x0B\xED\xCD\x36\x49\xC6\x84\xFF\xC2\x76\x9B\xF3\x49\x88\x97\x92\xD2\x54\xE9\x08\x19\x86\x40\x96\x24\x68\x25\x9D\xF7\x0E\xB7\x36\x85\x68\x6B\x2A\x97\x64\x30\xE6\xFF\x9A\x2A\x42\x2B\x31\x01\x32\xB3\xEE\x78\x1A\x26\x94\xE2\x07\x34\x50\x8A\xFF\xF9\xAE\xEA\xEC\x59\x42\xF5\x39\x40\x65\xDE\x55\xCC\x77\x1B\x32\x02\x19\xEE\x3C\xD4\x79\x48\x01\x4F\x51\xFE\x22\xE0\x0C\xF4\x07\x06\xCD\x55\x30\xC0\x24\x32\xD4\xCC\xAF\x82\x05\x48\x14\x10\x55\xA1\x3D\xF6\x81\x45\x54\xEA\x71\x24\x27\x06\xDC\xFA\xE4\xE4\x87\xCC\x81\xA0\x47\xA5\xAF\xD1\x89\xE7\x42\xC3\x24\xD0\x32\x7A\xDE\x44\x47\x6E\x1F\xCB\xEE\xA6\x46\xDE\x0D\xE6\xD5\x16\x03\x2A\xD6\x9E\xFD\x94\x02\x2C\xDB\x1F\xD0\xBE\x98\x10\xE3\xEB\xEA\xBE\xE5\xD1" }
}
}
fix invalid read on corrupt ziplist (#9831) If the last bytes in ziplist are corrupt and we decode from tail to head, we may reach slightly outside the ziplist.
2021-11-23 07:56:52 -05:00
test {corrupt payload: fuzzer findings - invalid access in ziplist tail prevlen decoding} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r debug set-skip-checksum-validation 1
r config set sanitize-dump-payload no
Replace ziplist with listpack in quicklist (#9740) Part three of implementing #8702, following #8887 and #9366 . ## Description of the feature 1. Replace the ziplist container of quicklist with listpack. 2. Convert existing quicklist ziplists on RDB loading time. an O(n) operation. ## Interface changes 1. New `list-max-listpack-size` config is an alias for `list-max-ziplist-size`. 2. Replace `debug ziplist` command with `debug listpack`. ## Internal changes 1. Add `lpMerge` to merge two listpacks . (same as `ziplistMerge`) 2. Add `lpRepr` to print info of listpack which is used in debugCommand and `quicklistRepr`. (same as `ziplistRepr`) 3. Replace `QUICKLIST_NODE_CONTAINER_ZIPLIST` with `QUICKLIST_NODE_CONTAINER_PACKED`(following #9357 ). It represent that a quicklistNode is a packed node, as opposed to a plain node. 4. Remove `createZiplistObject` method, which is never used. 5. Calculate listpack entry size using overhead overestimation in `quicklistAllowInsert`. We prefer an overestimation, which would at worse lead to a few bytes below the lowest limit of 4k. ## Improvements 1. Calling `lpShrinkToFit` after converting Ziplist to listpack, which was missed at #9366. 2. Optimize `quicklistAppendPlainNode` to avoid memcpy data. ## Bugfix 1. Fix crash in `quicklistRepr` when ziplist is compressed, introduced from #9366. ## Test 1. Add unittest for `lpMerge`. 2. Modify the old quicklist ziplist corrupt dump test. Co-authored-by: Oran Agra <oran@redislabs.com>
2021-11-24 06:34:13 -05:00
catch {r restore _listbig 0 "\x0e\x02\x1B\x1B\x00\x00\x00\x16\x00\x00\x00\x05\x00\x00\x02\x5F\x39\x04\xF9\x02\x02\x5F\x37\x04\xF7\x02\x02\x5F\x35\xFF\x19\x19\x00\x00\x00\x16\x00\x00\x00\x05\x00\x00\xF5\x02\x02\x5F\x33\x04\xF3\x02\x02\x5F\x31\xFE\xF1\xFF\x0A\x00\x6B\x43\x32\x2F\xBB\x29\x0a\xBE"} err
assert_match "*Bad data format*" $err
r ping
fix invalid read on corrupt ziplist (#9831) If the last bytes in ziplist are corrupt and we decode from tail to head, we may reach slightly outside the ziplist.
2021-11-23 07:56:52 -05:00
}
}
santize dump payload: fix carsh when zset with NAN score (#10002) `zslInsert` with a NAN score will crash the server. This one found by the `corrupt-dump-fuzzer`.
2021-12-26 04:40:11 -05:00
test {corrupt payload: fuzzer findings - zset zslInsert with a NAN score} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch {r restore _nan_zset 0 "\x05\x0A\x02\x5F\x39\x00\x00\x00\x00\x00\x00\x22\x40\xC0\x08\x00\x00\x00\x00\x00\x00\x20\x40\x02\x5F\x37\x00\x00\x00\x00\x00\x00\x1C\x40\xC0\x06\x00\x00\x00\x00\x00\x00\x18\x40\x02\x5F\x35\x00\x00\x00\x00\x00\x00\x14\x40\xC0\x04\x00\x00\x00\x00\x00\x00\x10\x40\x02\x5F\x33\x00\x00\x00\x00\x00\x00\x08\x40\xC0\x02\x00\x00\x00\x00\x00\x00\x00\x40\x02\x5F\x31\x00\x00\x00\x00\x00\x55\xF0\x7F\xC0\x00\x00\x00\x00\x00\x00\x00\x00\x00\x0A\x00\xEC\x94\x86\xD8\xFD\x5C\x5F\xD8"} err
assert_match "*Bad data format*" $err
r ping
}
}
solve corrupt dump fuzzer crash in streams (#10579) we had a panic in streamLastValidID when the stream metadata said it's not empty, but the rax is empty.
2022-04-14 01:29:35 -04:00
test {corrupt payload: fuzzer findings - streamLastValidID panic} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {r restore _streambig 0 "\x13\xC0\x10\x00\x00\x01\x80\x20\x48\xA0\x33\x00\x00\x00\x00\x00\x00\x00\x00\xC3\x40\x4F\x40\x5C\x18\x5C\x00\x00\x00\x24\x00\x05\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x00\x20\x01\x00\x01\x20\x03\x00\x05\x20\x1C\x40\x09\x05\x01\x01\x82\x5F\x31\x03\x80\x0D\x00\x02\x20\x0D\x00\x02\xA0\x19\x00\x03\x20\x0B\x02\x82\x5F\x33\x60\x19\x40\x2F\x02\x01\x01\x04\x20\x19\x00\xFF\x10\x00\x00\x01\x80\x20\x48\xA0\x34\x00\x00\x00\x00\x00\x00\x00\x01\xC3\x40\x51\x40\x5E\x18\x5E\x00\x00\x00\x24\x00\x05\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x00\x20\x01\x06\x01\x01\x82\x5F\x35\x03\x05\x20\x1E\x40\x0B\x03\x01\x01\x06\x01\x80\x0B\x00\x02\x20\x0B\x02\x82\x5F\x37\xA0\x19\x00\x03\x20\x0D\x00\x08\xA0\x19\x00\x04\x20\x0B\x02\x82\x5F\x39\x20\x19\x00\xFF\x10\x00\x00\x01\x80\x20\x48\xA0\x34\x00\x00\x00\x00\x00\x00\x00\x06\xC3\x3D\x40\x4A\x18\x4A\x00\x00\x00\x15\x00\x02\x01\x00\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x40\x10\x00\x00\x20\x01\x40\x00\x00\x05\x60\x07\x02\xDF\xFA\x02\xC0\x23\x09\x01\x01\x86\x75\x6E\x69\x71\x75\x65\x07\xA0\x2D\x02\x08\x01\xFF\x0C\x81\x00\x00\x01\x80\x20\x48\xA0\x35\x00\x81\x00\x00\x01\x80\x20\x48\xA0\x33\x00\x00\x00\x0C\x00\x0A\x00\x34\x8B\x0E\x5B\x42\xCD\xD6\x08"} err
assert_match "*Bad data format*" $err
r ping
}
}
Avoid valgrind fishy value warning on corrupt restore payloads (#10937) The corrupt dump fuzzer uncovered a valgrind warning saying: ``` ==76370== Argument 'size' of function malloc has a fishy (possibly negative) value: -3744781444216323815 ``` This allocation would have failed (returning NULL) and being handled properly by redis (even before this change), but we also want to silence the valgrind warnings (which are checking that casting to ssize_t produces a non-negative value). The solution i opted for is to explicitly fail these allocations (returning NULL), before even reaching `malloc` (which would have failed and return NULL too). The implication is that we will not be able to support a single allocation of more than 2GB on a 32bit system (which i don't think is a realistic scenario). i.e. i do think we could be facing cases were redis consumes more than 2gb on a 32bit system, but not in a single allocation. The byproduct of this, is that i dropped the overflow assertions, since these will now lead to the same OOM panic we have for failed allocations.
2022-07-13 02:14:38 -04:00
test {corrupt payload: fuzzer findings - valgrind fishy value warning} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
r debug set-skip-checksum-validation 1
catch {r restore _key 0 "\x13\x01\x10\x00\x00\x01\x81\xCC\x07\xDC\xF2\x00\x00\x00\x00\x00\x00\x00\x00\x40\x42\x42\x00\x00\x00\x18\x00\x02\x01\x01\x01\x02\x01\x84\x69\x74\x65\x6D\x05\x85\x76\x61\x6C\x75\x65\x06\x00\x01\x02\x01\x00\x01\x00\x01\x01\x01\x00\x01\x05\x01\x03\x01\x2C\x01\x00\x01\x01\x01\x82\x5F\x31\x03\x05\x01\x02\x01\x3C\x01\x00\x01\x01\x01\x02\x01\x05\x01\xFF\x02\xD0\x00\x00\x01\x81\xCC\x07\xDD\x2E\x00\x81\x00\x00\x01\x81\xCC\x07\xDC\xF2\x00\x81\x00\x00\x01\x81\xCC\x07\xDD\x1E\x00\x03\x01\x07\x6D\x79\x67\x72\x6F\x75\x70\x81\x00\x00\x01\x81\xCC\x07\xDD\x1E\x00\x02\x01\x00\x00\x01\x81\xCC\x07\xDD\x1E\x00\x00\x00\x00\x00\x00\x00\x00\x71\xDD\x07\xCC\x81\x01\x00\x00\x01\x01\x05\x41\x6C\x69\x63\x65\x58\xDD\x07\xCC\x81\x01\x00\x00\x01\x00\x00\x01\x81\xCC\x07\xDD\x1E\x00\x00\x00\x00\x00\x00\x00\x00\x0A\x00\x2F\xB0\xD1\x15\x0A\x97\x87\x6B"} err
assert_match "*Bad data format*" $err
r ping
}
}
sanitize dump payload: fix crash with empty set with listpack encoding (#11519) The following example will create an empty set (listpack encoding): ``` > RESTORE key 0 "\x14\x25\x25\x00\x00\x00\x00\x00\x02\x01\x82\x5F\x37\x03\x06\x01\x82\x5F\x35\x03\x82\x5F\x33\x03\x00\x01\x82\x5F\x31\x03\x82\x5F\x39\x03\x04\xA9\x08\x01\xFF\x0B\x00\xA3\x26\x49\xB4\x86\xB0\x0F\x41" OK > SCARD key (integer) 0 > SRANDMEMBER key Error: Server closed the connection ``` In the spirit of #9297, skip empty set when loading RDB_TYPE_SET_LISTPACK. Introduced in #11290
2022-11-20 05:12:15 -05:00
test {corrupt payload: fuzzer findings - empty set listpack} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
r debug set-skip-checksum-validation 1
catch {r restore _key 0 "\x14\x25\x25\x00\x00\x00\x00\x00\x02\x01\x82\x5F\x37\x03\x06\x01\x82\x5F\x35\x03\x82\x5F\x33\x03\x00\x01\x82\x5F\x31\x03\x82\x5F\x39\x03\x04\xA9\x08\x01\xFF\x0B\x00\xA3\x26\x49\xB4\x86\xB0\x0F\x41"} err
assert_match "*Bad data format*" $err
r ping
}
}
Avoid valgrind fishy value warning on corrupt restore payloads (#10937) The corrupt dump fuzzer uncovered a valgrind warning saying: ``` ==76370== Argument 'size' of function malloc has a fishy (possibly negative) value: -3744781444216323815 ``` This allocation would have failed (returning NULL) and being handled properly by redis (even before this change), but we also want to silence the valgrind warnings (which are checking that casting to ssize_t produces a non-negative value). The solution i opted for is to explicitly fail these allocations (returning NULL), before even reaching `malloc` (which would have failed and return NULL too). The implication is that we will not be able to support a single allocation of more than 2GB on a 32bit system (which i don't think is a realistic scenario). i.e. i do think we could be facing cases were redis consumes more than 2gb on a 32bit system, but not in a single allocation. The byproduct of this, is that i dropped the overflow assertions, since these will now lead to the same OOM panic we have for failed allocations.
2022-07-13 02:14:38 -04:00
Sanitize dump payload: ziplist, listpack, zipmap, intset, stream When loading an encoded payload we will at least do a shallow validation to check that the size that's encoded in the payload matches the size of the allocation. This let's us later use this encoded size to make sure the various offsets inside encoded payload don't reach outside the allocation, if they do, we'll assert/panic, but at least we won't segfault or smear memory. We can also do 'deep' validation which runs on all the records of the encoded payload and validates that they don't contain invalid offsets. This lets us detect corruptions early and reject a RESTORE command rather than accepting it and asserting (crashing) later when accessing that payload via some command. configuration: - adding ACL flag skip-sanitize-payload - adding config sanitize-dump-payload [yes/no/clients] For now, we don't have a good way to ensure MIGRATE in cluster resharding isn't being slowed down by these sanitation, so i'm setting the default value to `no`, but later on it should be set to `clients` by default. changes: - changing rdbReportError not to `exit` in RESTORE command - adding a new stat to be able to later check if cluster MIGRATE isn't being slowed down by sanitation.
2020-08-13 09:41:05 -04:00
} ;# tags
Reference in New Issue Copy Permalink