On Mon, 4 Feb 2019 08:58:37 +0200 Slavomir Kaslev <kaslevs@xxxxxxxxxx> wrote: > This patchset adds support for tracing guest kernels to trace-cmd. IIRC, you mentioned I should ignore these because you had another series coming? Let me know if that's not what you said. Also, I looked at the splice code wrt. the buffer percentage. The splice code will return if either the file descriptor has O_NONBLOCK or the splice command has SPLICE_F_NONBLOCK set. Or do you think that we need to do something in case the flag changes? To make everything non-block, we need to send a single to the code, which should cause the code to start again. Once the trace_fd is set to O_NONBLOCK the splice fd should return immediately when there's nothing it can read left. -- Steve > > Changes in v5: > - add FIFOs transport for tracing data > - fixed a bug in tracecmd_msg_wait_close >
|