On Wed, 9 Sep 2026 06:58:16 GMT, Yasumasa Suenaga <[email protected]> wrote:
> [JDK-8388696](https://bugs.openjdk.org/browse/JDK-8388696) (PR #31999) has > changed SA code to show error message if DWARF parser encounts unsupported > DWARF instructions. This caused a new error in > TestJhsdbJstackMixedWithXComp.java#xcomp. See JBS for details. > > This PR updates to return the result from > `DwarfParser::parse_dwarf_instructions` - `false` if DWARF parser failed such > as unknown instruction. `DW_CFA_def_cfa_expression` is one of cases of > failure because SA does not support it. > However `DW_CFA_def_cfa_expression` might be set for signal trampoline such > as `__restore_rt` on AMD64. Thus this PR ignores the failure in > `DwarfParser::parse_dwarf_instructions` if PC points signal trampoline. > > This PR passed serviceability/sa tests on both AMD64 and AArch64 Linux. > > --------- > - [x] I confirm that I make this contribution in accordance with the [OpenJDK > Interim AI Policy](https://openjdk.org/legal/ai). I think this is good. I had agreed with printing the error on unknown instruction, and not just in debug mode, as it seems very useful -- but see that tests using stderrShouldBeEmptyIgnoreVMWarnings() can fail. This looks like a reasonable way to handle it. There might be a few ways process_cie etc can now fail, which is new, but should be handled already. ------------- Marked as reviewed by kevinw (Reviewer). PR Review: https://git.openjdk.org/jdk/pull/32777#pullrequestreview-5155889458
