Re: RFR 8067105: Socket returned by ServerSocket.accept() is inherited by child process on Windows
On 28/01/2015 21:40, Chris Hegarty wrote: I’ll make the change, add a test, and then update the webrev. On second thoughts, it shouldn't need to be changed because the child SOCKET is created before AcceptEx is used and so should already have the inherit flag set correctly. Has there been any consideration given to writing a test for this? I thought we had one or two tests for this already. -Alan
Re: RFR 8067105: Socket returned by ServerSocket.accept() is inherited by child process on Windows
On 29 Jan 2015, at 09:21, Alan Bateman wrote: > On 28/01/2015 21:40, Chris Hegarty wrote: >> I’ll make the change, add a test, and then update the webrev. >> > On second thoughts, it shouldn't need to be changed because the child SOCKET > is created before AcceptEx is used and so should already have the inherit > flag set correctly. Thanks for checking this. > Has there been any consideration given to writing a test for this? I thought > we had one or two tests for this already. There are existing tests for client/connecting sockets, but none for accepted sockets. The test, and check, for accepted sockets is quite different from that of the ones already checked in for client/connecting sockets, so I added a new test for this. Updated webrev: http://cr.openjdk.java.net/~chegar/8067105/webrev.01/webrev/ -Chris. > -Alan
Re: RFR 8067105: Socket returned by ServerSocket.accept() is inherited by child process on Windows
On 29/01/2015 15:39, Chris Hegarty wrote: : Has there been any consideration given to writing a test for this? I thought we had one or two tests for this already. There are existing tests for client/connecting sockets, but none for accepted sockets. The test, and check, for accepted sockets is quite different from that of the ones already checked in for client/connecting sockets, so I added a new test for this. Updated webrev: http://cr.openjdk.java.net/~chegar/8067105/webrev.01/webrev/ Thanks, the additional test coverage is good (although I'm somewhat surprised we didn't have several tests the accepted socket case too). -Alan
Re: Backporting the TCP loopback fast path (Windows) improvement to OpenJDK v7
On 26/01/2015 13:48, Seán Coffey wrote: Valeriy, those changes look fine to me and I've verified that they run through testing fine also. webrev here for reference : http://cr.openjdk.java.net/~coffeys/webrev.8060170.7u.v3/webrev/ Can I have a jdk7u reviewer for this change ? There is a race in socket0 with the OS version check. Would it be simpler to drop that just have WASOPNOTSUPP and WSAEINVAL cause the option to be disabled? -Alan
