Add assertion in internalFindClassString #20501
Conversation
|
Related to #20444 |
runtime/vm/classsupport.c
Outdated
| U_8 localBuf[J9VM_PACKAGE_NAME_BUFFER_LENGTH]; | ||
| J9UTF8 *romClassName = J9ROMCLASS_CLASSNAME(result->romClass); | ||
|
|
||
| utf8Name = (U_8*)copyStringToUTF8WithMemAlloc(currentThread, className, J9_STR_NULL_TERMINATE_RESULT, "", 0, (char *)localBuf, J9VM_PACKAGE_NAME_BUFFER_LENGTH, &utf8Length); |
There was a problem hiding this comment.
You don't need to create utf8Name if the class is newly loaded by line 379:
result = internalFindClassInModule(currentThread, j9module, utf8Name, utf8Length, classLoader, options);
You can add new code right after line 379 to check the class name.
If the class is found by hashClassTableAtString(), you can add an else block at line 384, and use compareStringToUTF8() to do the class name comparison.
runtime/vm/classsupport.c
Outdated
| utf8Name[i] = '/'; | ||
| } | ||
| } | ||
| Assert_VM_true(J9UTF8_DATA_EQUALS( |
There was a problem hiding this comment.
Better to compare the class name length first before doing full class name comparison. . Also the class names should be captures in the failure case, something like;getStringUTF8Length() could help you get the length of a string
/* do class name comparison */
if (ClassNamesDiffers) {
Trc_VM_Class_Found_Has_Diff_Name(currentThread, className1, len1, className2, len2, whetherFromClassTable);
Trc_VM_Assert_ShouldNeverHappen();
}
There was a problem hiding this comment.
On my second thought, there is no need to do class name length comparison first. I've edited the previous comment.
Signed-off-by: Gengchen Tuo <[email protected]>
thallium
force-pushed
the
find-class-assert
branch
from
8d4a78d to
b59d906
Compare
| @@ -378,10 +378,42 @@ internalFindClassString(J9VMThread* currentThread, j9object_t moduleName, j9obje | |||
|
|
|||
| result = internalFindClassInModule(currentThread, j9module, utf8Name, utf8Length, classLoader, options); | |||
| } | |||
|
|
|||
| for (int i = 0; i < utf8Length; i++) { | |||
There was a problem hiding this comment.
There is no need to go through the for loop if the result is null or it is an array. The for loop can be moved after the null check and array check.
There was a problem hiding this comment.
You don't need this for loop if J9_STR_XLAT is already set in stringFlags.
This PR is not meant to be merged.