Hide Forgot
When running remote-viewer through valgrind, I got this trace: ==13623== 48 bytes in 2 blocks are definitely lost in loss record 7,424 of 13,299 ==13623== at 0x4A06409: malloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) ==13623== by 0x6B52971: handle_events (io.c:1940) ==13623== by 0x6B536F2: libusb_handle_events_timeout_completed (io.c:2126) ==13623== by 0x6B537DE: libusb_handle_events (io.c:2202) ==13623== by 0x5C18A69: spice_usb_device_manager_usb_ev_thread (usb-device-manager.c:1149) ==13623== by 0x3D17C6EA04: g_thread_proxy (gthread.c:798) ==13623== by 0x3134A07F32: start_thread (pthread_create.c:309) ==13623== by 0x31342F4EAC: clone (clone.S:111) This is with libusbx-1.0.17-1.fc20.x86_64 (happened with 1.0.16 as well). I couldn't spot an obvious leak looking at the code. I can reliably reproduce it, so let me know if more tests are needed. I'm logging this here to not forget, let me know if this should go elsewhere.
Hi, I believe this may be fixed by the 3 commits I've just pushed to libusbx git: https://github.com/libusbx/libusbx/commits/master Can you build libusbx from master and give that a try? Thanks & Regards, Hans
Hmm just tried to reproduce, but I did not manage to reproduce, even with f20 libusbx, and I did not leave enough details about the exact setup I was using :( Let's just say it's fixed, and I'll reopen if I see that again.