CVE-2018-18854 Use TreeMap instead of HashMap for JsObject key/value pairs, fixes #277 #280
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jrudolph
commented
Oct 30, 2018
| val regularTime = nanoBench { JsonParser(regularJson) } | ||
| val collidingTime = nanoBench { JsonParser(collidingJson) } | ||
|
|
||
| collidingTime / regularTime must be < 2L // speed must be in same order of magnitude |
There was a problem hiding this comment.
If you undo the fix, the ratio is ~ 100x slowdown for this test on my machine.
jrudolph
force-pushed
the
use-TreeMap-fixes-277
branch
from
c0b06dd
to
1a05a62
Compare
jrudolph
commented
Oct 30, 2018
| | "yes": 1, | ||
| | "no": 0 | ||
| | }, ["a", "b", null], false] | ||
| | "no": 0, |
There was a problem hiding this comment.
New rendering output is now ordered by the key Strings. However, that's not guaranteed anywhere so I fixed the test not to assume anything.
johanandren
approved these changes
Nov 7, 2018
|
LGTM |
…pairs, fixes spray#277 The problem is that with String's hashCode implementation it is too simple to create synthetic collisions. This allows an attacker to create an object with keys that all collide which leads to a performance drop for the HashMap just for creating the map in the first place. See scala/bug#11203 for more information about the underlying HashMap issue. For the time being, it seems safer to use a TreeMap which uses String ordering. Benchmarks suggest that using a TreeMap is only ~6% slower for reasonably sized JSON objects up to 100 keys. Benchmark for non-colliding keys: Benchmark (_size) (parser) Mode Cnt Score Error Units ExtractFieldsBenchmark.readSpray 1 HashMap thrpt 5 1195832.262 ± 64366.605 ops/s ExtractFieldsBenchmark.readSpray 1 TreeMap thrpt 5 1342009.641 ± 17307.555 ops/s ExtractFieldsBenchmark.readSpray 10 HashMap thrpt 5 237173.327 ± 70341.742 ops/s ExtractFieldsBenchmark.readSpray 10 TreeMap thrpt 5 233510.618 ± 69638.750 ops/s ExtractFieldsBenchmark.readSpray 100 HashMap thrpt 5 23202.016 ± 1514.763 ops/s ExtractFieldsBenchmark.readSpray 100 TreeMap thrpt 5 21899.072 ± 823.225 ops/s ExtractFieldsBenchmark.readSpray 1000 HashMap thrpt 5 2073.754 ± 66.093 ops/s ExtractFieldsBenchmark.readSpray 1000 TreeMap thrpt 5 1793.329 ± 43.603 ops/s ExtractFieldsBenchmark.readSpray 10000 HashMap thrpt 5 208.160 ± 7.466 ops/s ExtractFieldsBenchmark.readSpray 10000 TreeMap thrpt 5 160.349 ± 5.809 ops/s
jrudolph
force-pushed
the
use-TreeMap-fixes-277
branch
from
1a05a62
to
855b35e
Compare
jrudolph
changed the title
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The problem is that with String's hashCode implementation it is too simple to
create synthetic collisions. This allows an attacker to create an object with
keys that all collide which leads to a performance drop for the HashMap
just for creating the map in the first place. See scala/bug#11203
for more information about the underlying HashMap issue.
For the time being, it seems safer to use a TreeMap which uses String ordering.
Benchmarks suggest that using a TreeMap is only ~6% slower for reasonably sized JSON objects
up to 100 keys.
Benchmark for non-colliding keys:
/cc @plokhotnyuk