ICU-23251 Fix static analyzer bugs #3758

poliudian-iv · 2025-10-31T13:28:19Z

Bugs identified by the static analyzers when analyzing the chromium source code (part 3)

icu4c/source/i18n/rbt_pars.cpp - "Out of bound access to memory preceding the field 'preContext'"
icu4c/source/common/cstring.cpp - "Out of bound access to memory" if n == INT_MAX
icu4c/source/i18n/rematch.cpp - "Out of bound access to memory preceding the field 'd'" at Regex8BitSet::contains() when c == -1;
icu4c/source/i18n/number_longnames.cpp - "Out of bound access to memory" at getMixedUnitModifier
icu4c/source/common/messagepattern.cpp - "Out of bound access to memory preceding the field 'postContext'" when index < msg.length()
icu4c/source/common/ubidiln.cpp - "Out of bound access to memory" when limit==0
icu4c/source/i18n/simpletz.cpp - "The left operand of '<' is a garbage value"

TODO: Fill out the checklist below.

Checklist

Required: Issue filed: ICU-23251
Required: The PR title must be prefixed with a JIRA Issue number. Example: "ICU-23251 Fix xyz"
Required: Each commit message must be prefixed with a JIRA Issue number. Example: "ICU-23251 Fix xyz"
Issue accepted (done by Technical Committee after discussion)
Tests included, if applicable
API docs and/or User Guide docs changed or added, if applicable
Approver: Feel free to merge on my behalf

jira-pull-request-webhook · 2025-11-01T07:30:29Z

Hooray! The files in the branch are the same across the force-push. 😃

~ Your Friendly Jira-GitHub PR Checker Bot

roubert · 2025-11-08T02:46:47Z

icu4c/source/common/messagepattern.cpp


    // Set postContext to some of msg starting at index.
-    length=msg.length()-index;
+    length=msg.length()>index ? msg.length()-index : 0;


Is this something that could reasonably happen? If not, I suggest instead adding the length condition to the first if-statement in this function and just do nothing at all if index is out of range. You might even add an U_ASSERT() statement to assert that this function never gets called with an invalid index value.

roubert · 2025-11-08T03:03:14Z

icu4c/source/i18n/rbt_pars.cpp

                                          int32_t pos,
                                          UErrorCode& status)
 {
+    if (pos < 0) {


Same here, this is not something that should ever happen, assert (to get a nice human readable error message in debug builds if some bug causes it to happen even if it shouldn't) and then return immediately without doing anything.

roubert · 2025-11-08T03:26:29Z

icu4c/source/i18n/rematch.cpp

-                    const UnicodeSet &s = RegexStaticSets::gStaticSets->fPropSets[opValue];
-                    if (s.contains(c)) {
-                        success = !success;
+                if (c >= 0) {


This doesn't look right, UChar32 should be an unsigned data type, it shouldn't be possible for this value to be negative.

UChar32 is int32_t
And U16_NEXT can return U_SENTINEL == -1

Oh, I see, I mixed it up with UChar, sorry about that. But then there's a clearly defined sentinel value that you can test for here, U_SENTINEL, which will tell any future reader that this is a value that this variable really can be assigned. (And in case you have reason to fear that any other negative value could ever be assigned here, cover that case with a U_ASSERT.)

Testing for negative / non-negative is much spiffier, and used throughout the code base.
U_SENTINEL is negative for a reason -- for easy testing via if (c < 0).

roubert · 2025-11-11T01:35:49Z

icu4c/source/i18n/rematch.cpp

-                    const UnicodeSet &s = RegexStaticSets::gStaticSets->fPropSets[opValue];
-                    if (s.contains(c)) {
-                        success = !success;
+                if (c >= 0) {


Oh, I see, I mixed it up with UChar, sorry about that. But then there's a clearly defined sentinel value that you can test for here, U_SENTINEL, which will tell any future reader that this is a value that this variable really can be assigned. (And in case you have reason to fear that any other negative value could ever be assigned here, cover that case with a U_ASSERT.)

jira-pull-request-webhook · 2025-11-11T07:39:00Z

Notice: the branch changed across the force-push!

icu4c/source/common/messagepattern.cpp is different
icu4c/source/i18n/number_longnames.cpp is different
icu4c/source/i18n/rbt_pars.cpp is different
icu4c/source/i18n/rematch.cpp is different

View Diff Across Force-Push

~ Your Friendly Jira-GitHub PR Checker Bot

markusicu · 2025-11-11T16:50:08Z

icu4c/source/common/ubidiln.cpp

            /* search for the run limits and initialize visualLimit values with the run lengths */
            i=0;
-            do {
+            while(i<limit) {


line 578 says that limit>0. let's assert that if necessary, rather than change the logic.
after 26 years with this code as is, i doubt there is a real problem to be fixed.

Ok, if you wanna, i can add checking limit to start of function for checking argument pBiDi->trailingWSStart

markusicu · 2025-11-11T16:51:59Z

icu4c/source/i18n/rematch.cpp

-                    const UnicodeSet &s = RegexStaticSets::gStaticSets->fPropSets[opValue];
-                    if (s.contains(c)) {
-                        success = !success;
+                if (c != U_SENTINEL) {


Suggested change

if (c != U_SENTINEL) {

if (c >= 0) {

etc.

Actually, I am surprised that @aheninger used the validating macro but didn't check for well-formed input.

Also, is it ok to do nothing when the input is ill-formed? We shouldn't change this code without understanding what it does.

FYI: If it was ok to use a simple fallback, then we could use U16_NEXT_OR_FFFD().

jira-pull-request-webhook · 2025-11-13T09:26:21Z

Notice: the branch changed across the force-push!

icu4c/source/common/ubidiln.cpp is different
icu4c/source/i18n/rematch.cpp is different

View Diff Across Force-Push

~ Your Friendly Jira-GitHub PR Checker Bot

poliudian-iv changed the title ~~Fix static analyzer bugs~~ ICU-23251 Fix static analyzer bugs Nov 1, 2025

poliudian-iv force-pushed the analyzer_bugs_2 branch from 883a651 to 7955d20 Compare November 1, 2025 07:30

markusicu assigned roubert Nov 6, 2025

markusicu requested review from eggrobin, markusicu and richgillam November 6, 2025 17:38

roubert requested changes Nov 8, 2025

View reviewed changes

roubert requested changes Nov 11, 2025

View reviewed changes

poliudian-iv force-pushed the analyzer_bugs_2 branch from 7955d20 to e1a8cc6 Compare November 11, 2025 07:38

poliudian-iv requested a review from roubert November 11, 2025 09:15

markusicu requested changes Nov 11, 2025

View reviewed changes

ICU-23251 Fix static analyzer bugs

2d6871d

poliudian-iv force-pushed the analyzer_bugs_2 branch from e1a8cc6 to 2d6871d Compare November 13, 2025 09:26

Uh oh!

ICU-23251 Fix static analyzer bugs #3758

Are you sure you want to change the base?

ICU-23251 Fix static analyzer bugs #3758

Conversation

poliudian-iv commented Oct 31, 2025 • edited by markusicu Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Checklist

Uh oh!

jira-pull-request-webhook bot commented Nov 1, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jira-pull-request-webhook bot commented Nov 11, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jira-pull-request-webhook bot commented Nov 13, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

poliudian-iv commented Oct 31, 2025 •

edited by markusicu

Loading