Summary: | x11/alacritty crashes on exit with wayland on CURRENT | ||
---|---|---|---|
Product: | Ports & Packages | Reporter: | shamaz.mazum |
Component: | Individual Port(s) | Assignee: | Mikael Urankar <mikael> |
Status: | Closed Overcome By Events | ||
Severity: | Affects Only Me | CC: | 2khramtsov, jbeich, mikael |
Priority: | --- | Flags: | linimon:
maintainer-feedback?
(mikael) |
Version: | Latest | ||
Hardware: | Any | ||
OS: | Any |
Description
shamaz.mazum
2021-01-14 11:29:26 UTC
Can you reproduce with MALLOC_CONF=junk:false per malloc(3) or WITH_MALLOC_PRODUCTION=yes per src.conf(5)? 0x5a is a value assigned to deallocated memory likely to catch use-after-free. Bisecting led to alacritty v0.5.0 (since ports r543906) via https://github.com/alacritty/alacritty/commit/04f0bcaf54ed. Yes, with MALLOC_CONF=junk:false all is OK. Should I report it to alacritty developers then? (In reply to shamaz.mazum from comment #2) > Should I report it to alacritty developers then? Yep as upstream may provide additional ideas. However, if it's not reproducible on Linux you'd need to debug further. There're many differences to account: - FreeBSD -CURRENT enables debugging (like junk:true) by default - jemalloc maybe more unforgiving to use-after-free than glibc malloc - epoll-shim (as used by libwayland) may behave slightly different than native epoll - Tests from wayland-rs (wayland-client crate) fail to build on FreeBSD - Other wayland-rs consumers like ion-shell or veloren don't crash Can you test the patch at https://github.com/Smithay/smithay-clipboard/issues/32 ? (taken from https://github.com/alacritty/alacritty/issues/4702) comment #4, no this does not help. Must be unrelated problem. Can you reproduce after 0.7.2 aka ports r566130? Nevermind, "env MALLOC_CONF=junk:true alacritty" still crashes here. > ports r566130 Typo: should be ports r566128 (In reply to Jan Beich from comment #8) I can't reproduce this, but I have a different environment (poudriere log with my make.conf can be seen in 254009). |