-
Notifications
You must be signed in to change notification settings - Fork 13k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
configure: Disable rpass-valgrind when there's no valgrind #24859
Conversation
(rust_highfive has picked a reviewer for you, use r? to override) |
Hm I'm a little hesitant to do this as it will silently stop running all valgrind tests on the bots if we reimage and forget to install valgrind. |
I have another patch that makes configure error if valgrind isn't available: #24477. Personally, I do not like requiring valgrind to configure and build and might be in favor of this. |
I'd be in favour of tweaking this patch to make |
It looks like #24477 will fail by default if valgrind isn't present, which I don't believe to be the behaviour we want to expose to end users. I'll probably update this patch tonight but I don't have a particular horse in this race, eg I'd be perfectly content with an updated version of @brson's that won't break by default without valgrind. |
I updated the branch. It'll now fail if you do something like:
and don't have a valgrind, otherwise it will just disable the valgrind rpass tests |
r? @brson |
@alexcrichton Is the idea that we first land this, and then after it is landed, we update the bots ASAP to pass Or do we already pass |
Looks like we're currently passing |
ping @brson |
OK, so as I understand it this patch turns on valgrind-rpass by default if valgrind exists, but otherwise silently disables it. If --enable-valgrind is passed explicitly and valgrind does not exist then this emits a new error. @richo I'm sorry for the delay and hate to ask you to do more, but can you add a warning message to mk/main.mk saying that valgrind-rpass was disabled, similar to the existing case where it prints a status message when it is enabled. Since this case causes the build to miss some test coverage it's important to point that out I think. |
r=me after that nit is addressed |
Sorry for delay, added the log message. r? @brson |
@bors: r+ |
📌 Commit 01fc026 has been approved by |
@alexcrichton @pnkfelix What do you anticipate changing on the bots when this patch lands? I don't see it (maybe I don't actually understand this patch...). I think all that will happen is this will run on machines w/o valgrind and disable the valgrind-rpass tests. On closer inspection I don't understand how the valgrind-rpass code works. |
Err, wait yeah I think I might have goofed at some point in a rebase. Sorry this pr is weeks old and I don't really remember how it works either. Want to r- and I'll have a closer peek today? Apologies! |
EDIT: No, it does the correct thing (CFG_VALGRIND is populated by probing for a valgrind binary on PATH) So the only change to the bots should be to pass |
@brson I think we'd want to pass |
This stung me more than once in dev. Bonus DRY'ing up of configure that I did on my way past.
This stung me more than once in dev.
Bonus DRY'ing up of configure that I did on my way past.