| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Mon, 15 Jun 2009 17:56:51 -0700
From: Trond Myklebust <trond.myklebust@....uio.no>
To: Steve Dickson <SteveD@...hat.com>
Cc: Linux NFS Mailing list <linux-nfs@...r.kernel.org>,
Linux Kernel Mailing list <linux-kernel@...r.kernel.org>,
Ingo Molnar <mingo@...hat.com>, Jason Baron <jbaron@...hat.com>
Subject: Re: [PATCH] sunrpc: Adding trace points to status routines (take 2)
On Tue, 2009-06-09 at 14:18 -0400, Steve Dickson wrote:
> Hey Trond,
>
> Here is an updated version of the patch I sent out a while
> back... all that has changed was the addtion of 'rpc_' to
> the name of the trace points which coincides with the
> other trace points naming convention.
>
> Here is a pointer to the previous email that has the
> fairly long winded explanation of how all this works:
>
> http://marc.info/?l=linux-kernel&m=124058933520183&w=2
>
> And below is the updated patch which is based on Ingo's
> linux-2.6-tip.git tree.
>
> Comments?
>
> steved.
>
> Author: Steve Dickson <steved@...hat.com>
> Date: Tue Jun 9 14:12:07 EDT 2009
>
> Adds three trace points to the status routines in the
> sunrpc state machine. The trace points will only execute
> in error conditions.
>
> The goal of these trace points it to make it easier to
> debug process that spin or get hung up in the NFS and
> related subsystems.
Would this replace the existing dprint_status()? How does it differ from
the existing DYNAMIC_DEBUG functionality that was merged in 2.6.30?
> Signed-off-by: Steve Dickson <steved@...hat.com>
>
> diff --git a/include/trace/events/sunrpc.h b/include/trace/events/sunrpc.h
> new file mode 100644
> index 0000000..f502eb4
> --- /dev/null
> +++ b/include/trace/events/sunrpc.h
> @@ -0,0 +1,86 @@
> +#if !defined(_TRACE_SUNRPC_H) || defined(TRACE_HEADER_MULTI_READ)
> +#define _TRACE_SUNRPC_H
> +
> +#include <linux/tracepoint.h>
> +
> +#undef TRACE_SYSTEM
> +#define TRACE_SYSTEM sunrpc
> +
> +TRACE_EVENT(rpc_call_status,
> +
> + TP_PROTO(struct rpc_task *task),
> +
> + TP_ARGS(task),
> +
> + TP_STRUCT__entry(
> + __field(u32, cl_prog)
> + __field(u32, cl_vers)
> + __field(u32, p_proc)
> + __field(int, tk_status)
> + ),
> +
> + TP_fast_assign(
> + __entry->cl_prog = task->tk_client->cl_prog;
> + __entry->cl_vers = task->tk_client->cl_vers;
> + __entry->p_proc = task->tk_client->cl_procinfo->p_proc;
> + __entry->cl_prog = task->tk_client->cl_prog;
You should only need to set entry->cl_prog once, right?
> + __entry->tk_status = task->tk_status;
> + ),
> +
> + TP_printk("%d[%d]:%d status=%d", __entry->cl_prog,
> + __entry->cl_vers,__entry->p_proc, __entry->tk_status)
> +);
> +
> +TRACE_EVENT(rpc_bind_status,
> +
> + TP_PROTO(struct rpc_task *task),
> +
> + TP_ARGS(task),
> +
> + TP_STRUCT__entry(
> + __field(u32, cl_prog)
> + __field(u32, cl_vers)
> + __field(u32, p_proc)
> + __field(int, tk_status)
> + ),
> +
> + TP_fast_assign(
> + __entry->cl_prog = task->tk_client->cl_prog;
> + __entry->cl_vers = task->tk_client->cl_vers;
> + __entry->p_proc = task->tk_client->cl_procinfo->p_proc;
> + __entry->cl_prog = task->tk_client->cl_prog;
Ditto
> + __entry->tk_status = task->tk_status;
> + ),
> +
> + TP_printk("%d[%d]:%d status=%d", __entry->cl_prog,
> + __entry->cl_vers,__entry->p_proc, __entry->tk_status)
> +);
> +TRACE_EVENT(rpc_connect_status,
> +
> + TP_PROTO(struct rpc_task *task, int status),
> +
> + TP_ARGS(task, status),
> +
> + TP_STRUCT__entry(
> + __field(u32, cl_prog)
> + __field(u32, cl_vers)
> + __field(u32, p_proc)
> + __field(int, tk_status)
> + ),
> +
> + TP_fast_assign(
> + __entry->cl_prog = task->tk_client->cl_prog;
> + __entry->cl_vers = task->tk_client->cl_vers;
> + __entry->p_proc = task->tk_client->cl_procinfo->p_proc;
> + __entry->cl_prog = task->tk_client->cl_prog;
Ditto
> + __entry->tk_status = status;
> + ),
> +
> + TP_printk("%d[%d]:%d status=%d", __entry->cl_prog,
> + __entry->cl_vers,__entry->p_proc, __entry->tk_status)
> +
> +);
> +#endif /* __TRACE_SUNRPC_H */
> +
> +/* This part must be outside protection */
> +#include <trace/define_trace.h>
> diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
> index 5abab09..b86ee2a 100644
> --- a/net/sunrpc/clnt.c
> +++ b/net/sunrpc/clnt.c
> @@ -37,6 +37,9 @@
> #include <linux/sunrpc/rpc_pipe_fs.h>
> #include <linux/sunrpc/metrics.h>
>
> +#include <linux/ftrace.h>
> +#define CREATE_TRACE_POINTS
> +#include <trace/events/sunrpc.h>
>
> #ifdef RPC_DEBUG
> # define RPCDBG_FACILITY RPCDBG_CALL
> @@ -957,6 +960,7 @@ call_bind_status(struct rpc_task *task)
> return;
> }
>
> + trace_rpc_bind_status(task);
> switch (task->tk_status) {
> case -ENOMEM:
> dprintk("RPC: %5u rpcbind out of memory\n", task->tk_pid);
> @@ -1038,6 +1042,7 @@ call_connect_status(struct rpc_task *task)
> return;
> }
>
> + trace_rpc_connect_status(task, status);
> switch (status) {
> /* if soft mounted, test if we've timed out */
> case -ETIMEDOUT:
> @@ -1141,6 +1146,7 @@ call_status(struct rpc_task *task)
> return;
> }
>
> + trace_rpc_call_status(task);
> task->tk_status = 0;
> switch(status) {
> case -EHOSTDOWN:
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists