Add no bytecode flush flags to call-with-arraylike-or-spread tests

These tests depend on predictable optimize/deoptimize movements.
--no-stress-flush-bytecode and --no-flush-bytecode ensure the feedback
vector is not flushed, causing unexpected deoptimizations.

Bug: v8:12328, v8:12311
Change-Id: I2eb7575e8c943a92c907831d4d66d2b0c9716869
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3232825
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Commit-Queue: Shu-yu Guo <syg@chromium.org>
Cr-Commit-Position: refs/heads/main@{#77478}
This commit is contained in:
Shu-yu Guo 2021-10-19 16:37:26 -07:00 committed by V8 LUCI CQ
parent cac5e4bfb3
commit 4fb132b855
7 changed files with 28 additions and 0 deletions

View File

@ -9,6 +9,10 @@
// and assertions would fail. We prevent re-runs.
// Flags: --nostress-opt --no-always-opt
// These tests do not work well if we flush the feedback vector, which causes
// deoptimization.
// Flags: --no-stress-flush-bytecode --no-flush-bytecode
// Tests for optimization of CallWithSpread and CallWithArrayLike.
// This test is in a separate file because it invalidates protectors.

View File

@ -9,6 +9,10 @@
// and assertions would fail. We prevent re-runs.
// Flags: --nostress-opt --no-always-opt
// These tests do not work well if we flush the feedback vector, which causes
// deoptimization.
// Flags: --no-stress-flush-bytecode --no-flush-bytecode
// The test relies on optimizing/deoptimizing at predictable moments, so
// it's not suitable for deoptimization fuzzing.
// Flags: --deopt-every-n-times=0

View File

@ -9,6 +9,10 @@
// and assertions would fail. We prevent re-runs.
// Flags: --nostress-opt --no-always-opt
// These tests do not work well if we flush the feedback vector, which causes
// deoptimization.
// Flags: --no-stress-flush-bytecode --no-flush-bytecode
// The test relies on optimizing/deoptimizing at predictable moments, so
// this is not suitable for deoptimization fuzzing.
// Flags: --deopt-every-n-times=0 --interrupt-budget=1024

View File

@ -9,6 +9,10 @@
// and assertions would fail. We prevent re-runs.
// Flags: --nostress-opt --no-always-opt
// These tests do not work well if we flush the feedback vector, which causes
// deoptimization.
// Flags: --no-stress-flush-bytecode --no-flush-bytecode
// Some of the tests rely on optimizing/deoptimizing at predictable moments, so
// this is not suitable for deoptimization fuzzing.
// Flags: --deopt-every-n-times=0

View File

@ -9,6 +9,10 @@
// and assertions would fail. We prevent re-runs.
// Flags: --nostress-opt --no-always-opt
// These tests do not work well if we flush the feedback vector, which causes
// deoptimization.
// Flags: --no-stress-flush-bytecode --no-flush-bytecode
// Some of the tests rely on optimizing/deoptimizing at predictable moments, so
// this is not suitable for deoptimization fuzzing.
// Flags: --deopt-every-n-times=0

View File

@ -9,6 +9,10 @@
// and assertions would fail. We prevent re-runs.
// Flags: --nostress-opt --no-always-opt
// These tests do not work well if we flush the feedback vector, which causes
// deoptimization.
// Flags: --no-stress-flush-bytecode --no-flush-bytecode
// Some of the tests rely on optimizing/deoptimizing at predictable moments, so
// this is not suitable for deoptimization fuzzing.
// Flags: --deopt-every-n-times=0

View File

@ -9,6 +9,10 @@
// and assertions would fail. We prevent re-runs.
// Flags: --nostress-opt --no-always-opt
// These tests do not work well if we flush the feedback vector, which causes
// deoptimization.
// Flags: --no-stress-flush-bytecode --no-flush-bytecode
// Some of the tests rely on optimizing/deoptimizing at predictable moments, so
// this is not suitable for deoptimization fuzzing.
// Flags: --deopt-every-n-times=0