RFR: 8280089: compiler/c2/irTests/TestIRAbs.java fails on some arches [v2]
Aleksey Shipilev
shade at openjdk.java.net
Tue Jan 18 08:11:03 UTC 2022
> This new test is apparently fails in GHA.
>
> The trouble is that AbsL parts of the test are arch-specific, as only {x86_64,aarch64,ppc64,s390}.ad have the match rules for AbsL. x86_32.ad does not have it, so C2 never actually emits AbsL in x86_32 case. AbsI, AbsF, AbsD are also not supported by arm.ad.
>
> I wonder if IR Test Framework might be able to disable subtests based on arch... It is probably all around easier to make this test x86_64 specific. It tests shared C2 code, so testing one architecture is probably enough.
>
> Open for other suggestions, though.
>
> Additional testing:
> - [x] Linux x86_64 fastdebug, test still passes
> - [x] Linux x86_32 fastdebug, test used to fail, now is skipped
Aleksey Shipilev has updated the pull request incrementally with one additional commit since the last revision:
Add AArch64 too
-------------
Changes:
- all: https://git.openjdk.java.net/jdk/pull/7116/files
- new: https://git.openjdk.java.net/jdk/pull/7116/files/3abe12c5..9286b510
Webrevs:
- full: https://webrevs.openjdk.java.net/?repo=jdk&pr=7116&range=01
- incr: https://webrevs.openjdk.java.net/?repo=jdk&pr=7116&range=00-01
Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod
Patch: https://git.openjdk.java.net/jdk/pull/7116.diff
Fetch: git fetch https://git.openjdk.java.net/jdk pull/7116/head:pull/7116
PR: https://git.openjdk.java.net/jdk/pull/7116
More information about the hotspot-compiler-dev
mailing list