platform/posix: Unbreak fuzzer support #9107
Conversation
andyross
requested review from
dbaluta,
kv2019i,
lbetlej,
lgirdwood,
marc-hb,
mmaka1 and
plbossart
as code owners
|
Put this together in a rush, please review carefully. Not complicated, just code motion. |
Yes, @andyross please try |
|
Heh, indeed. The fuzz.sh delta was sitting un-git-added in my tree still. Oops. |
|
(Oh, and the checkpatch whine isn't really fixable: the symbols in question don't appear in Zephyr headers, they're still whitebox entry points in the native_posix arch. In the medium term we'll need to move to "native_sim", which is evolved architecture, and it has a better specified API.) |
kv2019i
left a comment
kv2019i
left a comment
There was a problem hiding this comment.
Thanks, looks good now! The previously passing CI checks now pass, so expectation is the remaining checks will pass as well (as the fuzz changes won't affect the on-device SOF tests).
|
@andyross Can you rebase, there's a small conflict with in CMakeLists.txt as one of the posix platform files was removed in another PR today. We need to get clean run of the CI to merge, but I checked the faiiling case and it's a fail in "16_00_TestSdwTransferLoopback" test case, so cannot possibly be related to your PR. If the case fails after rebase, I'll ping our CI folks to check. |
|
@lgirdwood wrote:
Yes, that's how this was discovered -> https://github.com/thesofproject/sof/actions/runs/8972185189/job/24639608648?pr=9098 when doing a Zephyr update |
Upstream Zephyr moved the LLVM fuzzer entry point out of the arch layer and made it an app responsibility, so we broke. Add back the support here that got removed. Fixes thesofproject#9101 Signed-off-by: Andy Ross <andyross@google.com>
|
Rebased. |
|
And the tests now pass as expected, merging! |
4 participants
Upstream Zephyr moved the LLVM fuzzer entry point out of the arch layer and made it an app responsibility, so we broke. Add back the support here that got removed.
Fixes #9101