On Sat, 31 Jan 2026 01:08:04 GMT, Yasumasa Suenaga <[email protected]> wrote:
>> test/hotspot/jtreg/serviceability/sa/TestJhsdbJstackMixedCore.java line 71:
>>
>>> 69: // Check whether the symbol of signal trampoline is available.
>>> 70: var libc = SATestUtils.getLibCPath();
>>> 71: if (!SATestUtils.isSymbolAvailable(libc, "__restore_rt")) {
>>
>> Do we have any reason to support the lookup of symbols not in libc? I'm just
>> wondering if `SATestUtils.isSymbolAvailable()` should just take the symbol
>> argument not the libc argument.
>
> I want to use `isSymbolAvailable()` in JDK-8376269 (PR #29400) to find the
> symbol relates to `gettimeofday()` in vDSO.
> (so I defined `getDebugInfo(()` as public to check vDSO debuginfo
> availability)
ok
>> test/lib/jdk/test/lib/SA/SATestUtils.java line 345:
>>
>>> 343: }
>>> 344:
>>> 345: private static boolean isSymbolAvailableInternal(String lib,
>>> String symbol) throws IOException {
>>
>> Should `lib` be named something more like `debugInfoPath`?
>
> This function is used for finding symbol from lib (not debuginfo) in L361.
Ok. I see now it is called in two places, once with a lib path and once with a
debuginfo path. And nm refers to the argument as `objfile`, but makes no
mention of what format `objfile` can be (presumably .o, .so, .debuginfo, or
executable).
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/29480#discussion_r2755598410
PR Review Comment: https://git.openjdk.org/jdk/pull/29480#discussion_r2755618737
