03-16-2019, 07:41 PM
Hello Folks. I'm having problems since the ending of last year I think. RPCS3 simply will not start. I thought I ignore it, but since you guys made significant progress I kinda wanted to figure it out, so decided to invest some time to debug the situation. To put things into perspective, I use linux, gentoo, all things up to date. Kernel is 5.0.2, nvidia drivers 418.43, vulkan layers 1.1.101... and I think that's about it. The rest is included in your project. So got yesterday a new set of sources. Configured with: cmake ../rpcs3 -G "Unix Makefiles" -DUSE_ALSA=ON -DUSE_PULSE=ON -DUSE_LIBEVDEV=ON -DUSE_VULKAN=ON -DUSE_SYSTEM_LIBPNG=ON -DUSE_SYSTEM_FFMPEG=ON -D CMAKE_C_COMPILER="/usr/lib/ccache/bin/distcc" -D CMAKE_CXX_COMPILER="/usr/lib/ccache/bin/distcc" -D CMAKE_C_COMPILER_ARG1="gcc-8.3.0" -D CMAKE_CXX_COMPILER_ARG1="g++-8.3.0" -D CMAKE_C_FLAGS="-mtune=nehalem -march=nehalem -O2 -pipe" -D CMAKE_CXX_FLAGS="-mtune=nehalem -march=nehalem -O2 -pipe"
Everything compiles ok, but when I try to start it, it just does nothing. Spikes the cpu for a fraction of a second, then does nothing. I did a preliminary strace on the binary, and it seems it goes in a nanosleep loop wait for something.
recvmsg(6, {msg_name=NULL, msg_namelen=0, msg_iov=[{iov_base="U\2\230\1<\317\t\17\3\24\4\0\20\0\0\0\0\0\0\24\24\24\24\24\0\0\3\37i\2\0\0", iov_len=4096}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 32
This check is not run by the parent pid, so I guess it's somewhere in 3rdparty thingies. I did a grep for nanosleep in the sources. There's only a few instances, some in ffmpeg (which is my main suspect atm) and some in llvm. I thought I should ask first before messing around in the sources. Any ideas are appreciated.
Funny enough, the windows version works fine with wine on the same machine, but it seems silly to run it like this
The appImage provided by you guys has the same issue (nanosleep loop).
I guess I should also mention I'm running gnome-3.28 on top of X-1.20.4 and mesa 19. I don't know how much this matters, but I thought I should specify.
Everything compiles ok, but when I try to start it, it just does nothing. Spikes the cpu for a fraction of a second, then does nothing. I did a preliminary strace on the binary, and it seems it goes in a nanosleep loop wait for something.
recvmsg(6, {msg_name=NULL, msg_namelen=0, msg_iov=[{iov_base="U\2\230\1<\317\t\17\3\24\4\0\20\0\0\0\0\0\0\24\24\24\24\24\0\0\3\37i\2\0\0", iov_len=4096}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 32
This check is not run by the parent pid, so I guess it's somewhere in 3rdparty thingies. I did a grep for nanosleep in the sources. There's only a few instances, some in ffmpeg (which is my main suspect atm) and some in llvm. I thought I should ask first before messing around in the sources. Any ideas are appreciated.
Funny enough, the windows version works fine with wine on the same machine, but it seems silly to run it like this
The appImage provided by you guys has the same issue (nanosleep loop).
I guess I should also mention I'm running gnome-3.28 on top of X-1.20.4 and mesa 19. I don't know how much this matters, but I thought I should specify.