From: David Wilder <dwilder@us.ibm.com>
To: prasad@linux.vnet.ibm.com
Cc: linux-kernel@vger.kernel.org, mathieu.desnoyers@polymtl.ca,
hunt@redhat.com, michaele@au1.ibm.com, dave@linux.vnet.ibm.com
Subject: Re: [RFC Patch 1/1] debugfs_printk and debugfs_dump interface
Date: Thu, 24 Apr 2008 14:23:16 -0700 [thread overview]
Message-ID: <4810FA44.9060402@us.ibm.com> (raw)
In-Reply-To: <20080421124329.GA14118@in.ibm.com>
debugfs_print makes a great improvement to the trace interface. I
missing a couple of bugs the first time I reviewed it. Please see my
comments in-line.
Dave..
> +static inline struct trace_info *init_trace_interface(struct
> + debugfs_printk_data *dpk)
> +{
> + struct trace_info *ti;
> +
> + dpk->exists = trace_exists(dpk->parent_dir, dpk->dir, &ti);
> +
> + switch(dpk->exists) {
> +
> + case TRACE_PARENT_DIR_EXISTS:
> + case TRACE_PARENT_DIR_ABSENT:
> + if(!dpk->buf_size)
> + dpk->buf_size = DEFAULT_TRACE_BUF_SIZE;
> + if(!dpk->sub_buf_size)
> + dpk->sub_buf_size = DEFAULT_TRACE_SUB_BUF_NR;
> + if(!dpk->flags)
> + dpk->flags = TRACE_FLIGHT_CHANNEL;
> + ti = trace_setup(dpk->parent_dir, dpk->dir,
> + dpk->buf_size, dpk->sub_buf_size, dpk->flags);
> + printk(KERN_INFO "Trace interface %s setup through "
> + "debugfs_printk\n",
> + ti->dir->d_iname);
> + if (IS_ERR(ti)) {
> + printk(KERN_ERR "Trace interface could not be "
> + "initialised\n");
> + return PTR_ERR(ti);
> + }
> + /* Fall through */
> + case TRACE_DIR_EXISTS:
> + if (ti->state != TRACE_RUNNING) {
You should start the trace only if the current state is TRACE_SETUP (I
think). See my comment in debugfs_print()
if (ti->state == TRACE_SETUP)
> + trace_start(ti);
> + }
> + }
> + return ti;
> +}
> +
> +/*
> + * debugfs_printk - A function to write into the debugfs mount 'directly'
> + * using 'trace' infrastructure
> + * @dpk - Structure containing info such as parent_dir and directory
> + * @format - String containing format string specifiers
> + * @ap - List of arguments
> + */
> +int debugfs_printk(struct debugfs_printk_data *dpk, char *format, ...)
> +{
> + int ret;
> + struct trace_info *ti;
> + va_list(ap);
> + unsigned long flags;
> +
> + va_start(ap, format);
> +
> + ti = init_trace_interface(dpk);
init_trace_interface() alway sets trace->state to TRACE_RUNNING . The
results is that the user is prevented from stopping the trace. You can
see this in fork_new_trace.
$ cat debug/trace_new_example/do_fork/state
running
$ echo stop > debug/trace_new_example/do_fork/state
$ cat debug/trace_new_example/do_fork/state
running
> +
> + /* Now do the actual printing */
> + /* Take an RCU Lock over the trace_info state */
> + rcu_read_lock();
> + /* Take a spinlock for the global buffer used by relay */
> + if (dpk->flags & TRACE_GLOBAL_CHANNEL)
> + spin_lock_irqsave(&trace_lock, flags);
> + ret = trace_printf(ti, format, ap);
> + if (dpk->flags & TRACE_GLOBAL_CHANNEL)
> + spin_unlock_irqrestore(&trace_lock, flags);
> + rcu_read_unlock();
> +
> + va_end(ap);
> + return ret;
> +}
> +EXPORT_SYMBOL(debugfs_printk);
> +
> +/*
> + * debugfs_printk - A function to write into the debugfs mount 'directly'
> + * using 'trace' infrastructure
> + * @dpk - Structure containing info such as parent_dir and directory
> + * @output - Data that needs to be output
> + * @output_len - Length of the output data
> + */
> +int debugfs_dump(struct debugfs_printk_data *dpk, const void *output,
> + const int output_len)
> +{
> + struct trace_info *ti;
> + char *record;
> +
> + ti = init_trace_interface(dpk);
Same issue as debugfs_printf,
You should also check for trace_running as you did for trace_printf.
> +
> + /* Now do the actual printing */
> + rcu_read_lock();
> + record = relay_reserve(ti->rchan, output_len);
> + if (record)
> + memcpy(record, output, output_len);
> + else
> + return -ENOMEM;
> + rcu_read_unlock();
> +
> + return 0;
> +}
> +EXPORT_SYMBOL(debugfs_dump);
next prev parent reply other threads:[~2008-04-24 21:23 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-04-15 11:14 [RFC Patch 0/1] Enhancements to 'trace' infrastructure K. Prasad
2008-04-15 11:26 ` [RFC Patch 1/1] debugfs_printk and debugfs_dump interface K. Prasad
2008-04-16 21:25 ` David Wilder
2008-04-21 12:43 ` K. Prasad
2008-04-21 17:14 ` Randy Dunlap
2008-04-22 5:22 ` K. Prasad
2008-04-22 14:53 ` Randy Dunlap
2008-04-23 7:59 ` K. Prasad
2008-04-24 21:23 ` David Wilder [this message]
2008-04-25 7:42 ` K. Prasad
2008-04-25 8:19 ` K. Prasad
2008-04-28 0:50 ` Michael Ellerman
2008-04-28 4:48 ` K. Prasad
2008-04-28 4:54 ` Michael Ellerman
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4810FA44.9060402@us.ibm.com \
--to=dwilder@us.ibm.com \
--cc=dave@linux.vnet.ibm.com \
--cc=hunt@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mathieu.desnoyers@polymtl.ca \
--cc=michaele@au1.ibm.com \
--cc=prasad@linux.vnet.ibm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox