reassign 997051 src:jemalloc 5.2.1-3 thanks On Mon, Oct 25, 2021 at 09:17:45AM +0200, Johannes Schauer Marin Rodrigues wrote: > > > I guess one fix or workaround would be to add support for the SYS_openat > > > syscall as a fallback. Luckily this has already been done upstream: > > > > > > https://github.com/jemalloc/jemalloc/commit/6924f83cb21f75e1c892d8f469500e12f1a3f5a7#diff-e2003bd99a76acf15d071c2fd49cfaeefae69debe6fc304a86f104b662da986c > > > > I have rebuilt jemalloc with this patch on riscv64. I confirm that it fixes > > the testsuite of fakechroot. > > Aurelien, I have no idea how you found this bug but thanks a ton for analyzing > it and finding the right fix!! :D
Indeed, thank you so much Aurelien! > Faidon, maybe jemalloc could add a test that checks for this so that this > problem does not show up in the tests of an unrelated package? Testing for this would be hard, as it's a bit of a special case - not sure how the test would look exactly. Also, I wouldn't say that this is an "unrelated" packaged entirely. While it's true that the two packages have no interdependencies, fundamentally what happens here with fakechroot and jemalloc in that test is that we're preloading two different libraries that both override glibc functions. These kind of problems are not entirely unexpected. As long as fakechroot tests remain functional/enabled, we should be able to catch those and address them (and through autopkgtest, perhaps even catch them on subsequent jemalloc uploads?). I'll upload a 5.2.1-4 with the backported upstream fix that Aurelien identified momentarily. Thanks again! Best, Faidon
