-
Notifications
You must be signed in to change notification settings - Fork 30
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
Build failure on Debian unstable #19
Comments
Hm. I've built it with exactly that clang version (6.0.1-9) several times (on debian unstable) and I tried it just now again just to be sure. I just run:
And it works fine for me. I use separate checkouts/builds for all components for the fuzzer. |
On a different machine, also latest Debian unstable and also a dev machine, I tried with a fresh clone of curl-fuzzer in /tmp.
|
Just a guess... my 'make' is an alias to 'make -j$(nproc)'. Is there a problem with parallel builds ? Will check later, have to finish some work now :-| |
I don't believe there's a problem with parallel builds but obviously there could be something subtle going on. Make's usually pretty good at resolving dependencies in order. libstandaloneengine.a should be built for you if required; it's defined in the Automake at https://github.com/curl/curl-fuzzer/blob/master/Makefile.am#L61 and this generally works on the travis runners and such. I'd retry with a simple |
In curl-fuzzer/ a |
For me it looks like a missing dependency. I can see |
This fixes it for me - thanks for for your time. |
BTW, I now see memory leaks in curl - any interest in reporting those ? Or it that expected behavior ? |
@rockdaboot: I'm surprised that that patch works, because Could you make a similar |
NP |
So, from the broken log:
From the working log:
To my eyes those are identical, so I'm at a loss as to why one of them is working and one of them is not... did you do a clean build for the second one? |
Yes I did. I can tell you that the broken build didn't generate libstandaloneengine.a at all. You can see that in I now revert to the previous commit (hard resetting my changes) and try again with Though this looks good:
|
If I replace in Makefile.am
then Reverting the above brings back the issue. Looks like LIB_FUZZING_ENGINE is set outside the of 'make' - but I checked my environment, no such variable. |
As a PoC, I changed |
So the idea of Possibly there's some easy way around this. Perhaps a configuration option could be used instead to set this up. |
If not setting But the way LIB_FUZZING_ENGINE ?= is used in Makefile.am looks absolutely fine. Also, renaming that to show there is no environment issue and then using just = with success tells me that there must be a make issue (make 4.2.1-1.2). I did those changes directly in the generated Makefile to eliminate possible automake issues. BTW, |
|
Not sure what is going on, clang is 6.0.1-9 and used often for fuzzing other projects (also for OSS-Fuzz).
I naively followed the instructions (../curl was fetched from upstream a minute ago)
Possibly something is missing, but not to easily detectable from the above messages.
I appreciate any help / ideas :-)
The text was updated successfully, but these errors were encountered: