From: Dave Airlie Subject: Re: Re: Issue with RPC buffer allocation.. Date: Wed, 29 Jan 2003 06:34:40 +0000 (GMT) Sender: nfs-admin@lists.sourceforge.net Message-ID: References: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Return-path: To: NFS@lists.sourceforge.net In-Reply-To: Errors-To: nfs-admin@lists.sourceforge.net List-Help: List-Post: List-Subscribe: , List-Id: Discussion of NFS under Linux development, interoperability, and testing. List-Unsubscribe: , List-Archive: Okay I switched on the memory debugger stuff and it isn't a kernel issue.. I've messed up a free somewhere else... but it takes ages for the kernel to find out without the debugging on :-) sorry for wasting time ... Dave. On Wed, 29 Jan 2003, Dave Airlie wrote: > > of course I included the wrong piece of log .. read 677 and 678 .. > > RPC: 677 new task procpid 21 > RPC: 677 rpc_execute flgs 1 > RPC: 677 reserved req c7e51618 xid 7cc932a4 > RPC: 677 xprt_reserve returns 0 > RPC: allocated buffer c11f6000 498 > RPC: 677 xprt_transmit c11f6000(7cc932a4) > RPC: 677 xprt_cwnd_limited cong = 512 cwnd = 320 > RPC: 677 TCP write queue full > RPC: 677 sleep_on(queue "xprt_sending" time 4808) > RPC: 677 added to queue c7e51044 "xprt_sending" > RPC: 678 new task procpid 21 > RPC: 678 rpc_execute flgs 1 > RPC: 678 reserved req c7e514f8 xid 7cc932a5 > RPC: 678 xprt_reserve returns 0 > RPC: allocated buffer c11f6000 498 > RPC: 678 xprt_transmit c11f6000(7cc932a5) > RPC: 678 xprt_cwnd_limited cong = 512 cwnd = 320 > RPC: 678 TCP write queue full > RPC: 678 sleep_on(queue "xprt_sending" time 4815) > RPC: 678 added to queue c7e51044 "xprt_sending" > > Dave. > > On Wed, 29 Jan 2003, Dave Airlie wrote: > > > > > Take a look at the log below, taken from one of my systems, running 2.4.20 > > + some patches by me - is the behaviour below possible? my patches do two > > things, lock in all memory pages using VM_LOCKED for user space apps, and > > CRC the pages of every image loaded in .. and I've disabled any threads my > > code runs, and the other code seems trivial... > > > > task 676 gets allocated a buffer at c11f6000, and hits congestion and gets > > put on a queue > > task 677 get allocated the same buffer? even though 676 should have it .. > > > > Dave. > > > > > > RPC: 675 new task procpid 21 > > RPC: 675 rpc_execute flgs 1 > > RPC: 675 reserved req c7e51588 xid 7cc932a2 > > RPC: 675 xprt_reserve returns 0 > > RPC: allocated buffer c7e09000 498 > > RPC: 675 xprt_transmit c7e09000(7cc932a2) > > RPC: 675 xprt_cwnd_limited cong = 0 cwnd = 320 > > RPC: xprt_sendmsg(120) = 120 > > RPC: 675 xmit complete > > RPC: 675 sleep_on(queue "xprt_pending" time 4787) > > RPC: 675 added to queue c7e5105c "xprt_pending" > > RPC: 675 setting alarm for 270 ms > > RPC: wake_up_next(c7e51050 "xprt_resend") > > RPC: wake_up_next(c7e51044 "xprt_sending") > > RPC: 676 new task procpid 21 > > RPC: 676 rpc_execute flgs 1 > > RPC: 676 reserved req c7e51078 xid 7cc932a3 > > RPC: 676 xprt_reserve returns 0 > > RPC: allocated buffer c11f6000 498 > > RPC: 676 xprt_transmit c11f6000(7cc932a3) > > RPC: 676 xprt_cwnd_limited cong = 256 cwnd = 320 > > RPC: xprt_sendmsg(120) = 120 > > RPC: udp_data_ready... > > RPC: udp_data_ready client c7e51000 > > RPC: 676 received reply > > RPC: 676 xmit complete > > RPC: 676 sleep_on(queue "xprt_pending" time 4799) > > RPC: 676 added to queue c7e5105c "xprt_pending" > > RPC: 676 setting alarm for 270 ms > > RPC: wake_up_next(c7e51050 "xprt_resend") > > RPC: 677 new task procpid 21 > > RPC: 677 rpc_execute flgs 1 > > RPC: 677 reserved req c7e51618 xid 7cc932a4 > > RPC: 677 xprt_reserve returns 0 > > RPC: allocated buffer c11f6000 498 > > RPC: 677 xprt_transmit c11f6000(7cc932a4) > > RPC: 677 xprt_cwnd_limited cong = 512 cwnd = 320 > > RPC: 677 TCP write queue full > > RPC: 677 sleep_on(queue "xprt_sending" time 4808) > > RPC: 677 added to queue c7e51044 "xprt_sending" > > > > > > -- David Airlie, Software Engineer http://www.skynet.ie/~airlied / airlied@skynet.ie pam_smb / Linux DecStation / Linux VAX / ILUG person ------------------------------------------------------- This SF.NET email is sponsored by: SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See! http://www.vasoftware.com _______________________________________________ NFS maillist - NFS@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/nfs