JIT and bpf2c use different verifier options #4324
Conversation
Alan-Jowett
requested review from
dthaler,
dv-msft,
LakshK98,
matthewige,
mikeagun,
mtfriesen,
poornagmsft,
saxena-anurag,
shankarseal and
vpidatala94
as code owners
Alan-Jowett
force-pushed
the
normalize_verifier_options
branch
from
60185ab to
0563be2
Compare
Signed-off-by: Alan Jowett <alanjo@microsoft.com>
Signed-off-by: Alan Jowett <alanjo@microsoft.com>
Alan-Jowett
force-pushed
the
normalize_verifier_options
branch
from
0563be2 to
7a61e04
Compare
|
Closing this PR as the only difference between the two paths are what the verifier reports. For the JIT case the verifier runs a second time with different options to print more information. |
Why close this? Shouldn't bpf2c also print more information? I'd think we'd want native mode to be as diagnosable as JIT mode. |
Signed-off-by: Alan Jowett <alanjo@microsoft.com>
|
Originally closed the PR (and was going to create a more constrainted one) as it had unrelated changes that was causing test failures. Rather than create a new PR, I will just remove the unrelated changes to see if I can avoid the test failures and scope this down to just changing the default verifier options. |
Signed-off-by: Alan Jowett <alanjo@microsoft.com>
Alan-Jowett
force-pushed
the
normalize_verifier_options
branch
from
9205f3b to
62cf60e
Compare
Signed-off-by: Alan Jowett <alanjo@microsoft.com>
Resolves: #4323
Description
JIT/Interpreter and the bpf2c path end up passing slightly different verifier options resulting in inconsistent behavior from the verifier. This change refactors the common
Testing
CI/CD
Documentation
No.
Installation
No.