Re: [tip:tracing/urgent] tracing: Fix trace_marker output

From: Carsten Emde
Date: Sun Dec 06 2009 - 08:10:40 EST


Olof,

> On Sun, Nov 22, 2009 at 10:24:48AM +0000, tip-bot for Carsten Emde wrote:
>> Commit-ID: c13d2f7c3231e873f30db92b96c8caa48f100f33
>> Gitweb: http://git.kernel.org/tip/c13d2f7c3231e873f30db92b96c8caa48f100f33
>> Author: Carsten Emde <Carsten.Emde@xxxxxxxxx>
>> AuthorDate: Mon, 16 Nov 2009 20:56:13 +0100
>> Committer: Steven Rostedt <rostedt@xxxxxxxxxxx>
>> CommitDate: Tue, 17 Nov 2009 09:19:06 -0500
>>
>> tracing: Fix trace_marker output
>>
>> When a string was written to <debugfs>/tracing/trace_marker, some
>> strange characters appeared in the trace output instead of the
>> string, since a vprint function erroneously called a vararg print
>> function with a va_list argument. This patch fixes the problem and
>> simplifies the related code.
>
> [...]
>
>> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
>> index 03c7fd5..12b49ca 100644
>> --- a/kernel/trace/trace.c
>> +++ b/kernel/trace/trace.c
>> @@ -1361,10 +1361,11 @@ int trace_array_vprintk(struct trace_array *tr,
>> pause_graph_tracing();
>> raw_local_irq_save(irq_flags);
>> __raw_spin_lock(&trace_buf_lock);
>> - len = vsnprintf(trace_buf, TRACE_BUF_SIZE, fmt, args);
>> -
>> - len = min(len, TRACE_BUF_SIZE-1);
>> - trace_buf[len] = 0;
>> + if (args == NULL) {
>> + strncpy(trace_buf, fmt, TRACE_BUF_SIZE);
>> + len = strlen(trace_buf);
>> + } else
>> + len = vsnprintf(trace_buf, TRACE_BUF_SIZE, fmt, args);
>
>
> Comparing a va_list with NULL is bogus. It's supposed to be treated like
> an opaque type and only be manipulated with va_* accessors.
>
> I wouldn't really care much, but it broke builds on some ARM platforms:
>
> kernel/trace/trace.c: In function 'trace_array_vprintk':
> kernel/trace/trace.c:1364: error: invalid operands to binary == (have 'va_list' and 'void *')
> kernel/trace/trace.c: In function 'tracing_mark_write':
> kernel/trace/trace.c:3349: error: incompatible type for argument 3 of 'trace_vprintk'
Having looked at other possible solutions, I find that the best thing to do
is to revert this part of the patch. Sorry that I didn't check it on ARM
platforms.

Carsten.

-=-

This patch partly reverts c13d2f7c3231e873f30db92b96c8caa48f100f33 and
re-installs Steven's original mark_printk() mechanism.

Signed-off-by: Carsten Emde <C.Emde@xxxxxxxxx>

Index: linux-2.6-tip/kernel/trace/trace.c
===================================================================
--- linux-2.6-tip.orig/kernel/trace/trace.c
+++ linux-2.6-tip/kernel/trace/trace.c
@@ -1361,11 +1361,7 @@ int trace_array_vprintk(struct trace_arr
pause_graph_tracing();
raw_local_irq_save(irq_flags);
__raw_spin_lock(&trace_buf_lock);
- if (args == NULL) {
- strncpy(trace_buf, fmt, TRACE_BUF_SIZE);
- len = strlen(trace_buf);
- } else
- len = vsnprintf(trace_buf, TRACE_BUF_SIZE, fmt, args);
+ len = vsnprintf(trace_buf, TRACE_BUF_SIZE, fmt, args);

size = sizeof(*entry) + len + 1;
buffer = tr->buffer;
@@ -3320,6 +3316,16 @@ tracing_entries_write(struct file *filp,
return cnt;
}

+static int mark_printk(const char *fmt, ...)
+{
+ int ret;
+ va_list args;
+ va_start(args, fmt);
+ ret = trace_vprintk(0, fmt, args);
+ va_end(args);
+ return ret;
+}
+
static ssize_t
tracing_mark_write(struct file *filp, const char __user *ubuf,
size_t cnt, loff_t *fpos)
@@ -3346,7 +3352,7 @@ tracing_mark_write(struct file *filp, co
} else
buf[cnt] = '\0';

- cnt = trace_vprintk(0, buf, NULL);
+ cnt = mark_printk("%s", buf);
kfree(buf);
*fpos += cnt;

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/