Hi Ilya, On Thu, May 16, 2019 at 02:04:06PM +0500, ???? ??????? wrote: > Hello, > > I'm going to enable address sanitizer in travis-ci. > > x86_64: (known leak) > https://travis-ci.org/chipitsine/haproxy-1/jobs/533196875 > > ppc64le: > https://travis-ci.org/chipitsine/haproxy-1/jobs/533196874 > > can someone have a look at ppc64le ?
I had a quick look at the raw logs and am not seeing any exploitable information there. I'm not against using such mechanisms in the future but only once we manage to make sure they can reliably be used, which is still far from being the case. Analysing false positive reports in general is taking a bit too much of my time, I'd rather start to have a look once we are certain these are real bugs, and then we can safely enable these features to detect new regressions. For now we're testing gcc and clang in fact, which is nice, but I already can't find enough time to work on haproxy. By the way I'd also like to fix the 1 or 2 regtests which randomly fail on the MacOS Travis tests, because I already got used to consider that red was the normal color there :-( Apparently the MacOS VMs are very slow and regularly show huge latency spikes causing some timeouts to trigger in random order. Probably we need to increase these timeouts again and move them to the slow group :-/ Thanks! Willy