New issue
Advanced search Search tips

Issue 867791 link

Starred by 1 user

Issue metadata

Status: Assigned
Owner:
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Linux
Pri: 3
Type: Bug



Sign in to add a comment

Float-cast-overflow in clampTo

Project Member Reported by ClusterFuzz, Jul 26

Issue description

Detailed report: https://clusterfuzz.com/testcase?key=6240122966900736

Fuzzer: inferno_twister
Job Type: linux_ubsan_chrome
Platform Id: linux

Crash Type: Float-cast-overflow
Crash Address: 
Crash State:
  clampTo
  clampTo
  clampTo<int,
  
Sanitizer: undefined (UBSAN)

Regressed: https://clusterfuzz.com/revisions?job=linux_ubsan_chrome&range=551565:563900

Reproducer Testcase: https://clusterfuzz.com/download?testcase_id=6240122966900736

Issue filed automatically.

See https://github.com/google/clusterfuzz-tools for more information.
 
Project Member

Comment 1 by ClusterFuzz, Jul 26

Components: Blink>Internals>WTF
Labels: Test-Predator-Auto-Components
Automatically applying components based on crash stacktrace and information from OWNERS files.

If this is incorrect, please apply the Test-Predator-Wrong-Components label.
Components: -Blink>Internals>WTF Blink>Compositing Blink>Paint
Labels: -Pri-2 Test-Predator-Wrong-Components Pri-3
Cc: fmalita@chromium.org f...@opera.com
Status: Available (was: Untriaged)
So maybe we have to check for nan in the clamp code?
If we invert one of the min/max conditionals it would catch/clamp NaN also.

But it's unclear to me how we could end up with NaN in this case, so maybe there's an issue higher up the stack.

I guess we should either decide NaNs are valid input for clamping, and handle them, or assert the inputs are not NaN.
I agree we should fix this up the stack. Leaving it open for someone to look into that.
Owner: f...@opera.com
Status: Assigned (was: Available)

Sign in to add a comment